Magmatism and the causes of continental break-up

Magmatism and the Causes of Continental Break-up

Geological Society Special Publications Series Editor J. BROOKS

GEOLOGICAL SOCIETY SPECIAL PUBLICATION NO 68

Magmatism and the Causes of Continental Break-up

E D I T E D BY

B. C. S T O R E Y British Antarctic Survey, Cambridge, UK

T. A L A B A S T E R School of the Environment, University of Sunderland, UK

R. J. P A N K H U R S T British Antarctic Survey, Nottingham, UK

1992 Published by The Geological Society London

THE GEOLOGICAL SOCIETY The Society was founded in 1807 as the Geological Society of London and is the oldest geological society in the world. It received its Royal Charter in 1825 for the purpose of 'investigating the mineral structure of the Earth'. The Society is Britain's national learned society for geology with a Fellowship of 6965 (1991). It has countrywide coverage and approximately 1000 members reside overseas. The Society is responsible for all aspects of the geological sciences including professional matters. The Society has its own publishing house which produces the Society's international journals, books and maps, and which acts as the European distributor for publications of the American Association of Petroleum Geologists. Fellowship is open to those holding a recognized honours degree in geology or cognate subject and who have at least two years relevant postgraduate experience, or who have not less than six years relevant experience in geology or a cognate subject. A Fellow who has not less than five years relevant postgraduate experience in the practice of geology may apply for validation and, subject to approval, may be able to use the designatory letters C. Geol (Chartered Geologist). Further information about the Society is available from the Membership Manager, The Geological Society, Burlington House, Piccadilly, London W1V 0JU, UK. Published by The Geological Societyfrom: The Geological Society Publishing House Unit 7 Brassmill Enterprise Centre Brassmill Lane Bath BA1 3JN UK

(Orders: Tel. 0225 445046

Distributors USA AAPG Bookstore PO Box 979 Tulsa Oklahoma 74101-0979 USA (Orders: Tel. (918) 584-2555 Fax (918) 584-0469)

Fax 0225 442836) First published 1992 (~The Geological Society 1992. All rights reserved. No reproduction, copy or transmission of this publication may be made without written permission. No paragraph of this publication may be reproduced, copied or transmitted save with the provisions of the Copyright Licensing Agency, 90 Tottenham Court Road, London WlP 9HE, UK. Users registered with Copyright Clearance Center, 27 Congress Street, Salem, MA 01970, USA: the item-free code for this publication is 0305-8719/92 $03.50. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library ISBN 0-903317-83-4

Printed and bound by The Cromwell Press, Broughton Gifford, Melksham

Australia Australian Mineral Foundation 63 Conyngham Street Glenside South Australia 5065 Australia (Orders: Tel. (08) 379-0444 Fax (08) 379-4634) India Affiliated East-West Press PVT Ltd G-l/16 Ansari Road New Delhi 110 002 India (Orders: Tel. (11) 327-9113 Fax (11) 331-2830) Japan Kanda Book Trading Co. Tanikawa Building 3-2 Kand Surugadai Chiyoda-Ku Tokyo 101 Japan (Orders: Tel. (03) 3255-3497 Fax (03) 3255-3495)

Contents Preface

vii

Magma generation and break-up processes WHrm, R. S. Magmatism during and after continental break-up

1

ComN, M. F. & ELDHOLM,O. Volcanism and continental break-up: a global compilation of large igneous provinces

17

MENZmS,M. A. The lower lithosphere as a major source for continental flood basalts: a re-appraisal

31

SAUNDERSA. D., STOREY,M., KEwr, R. W. & NORRY,M. 3. Consequences of plume-lithosphere interactions

41

GIBSON,S. A., THOMPSON,R. N., LEAT,P. T., DICKIN,A. P., MORRISON,M. A., HENDRY, G.L. & MITCHELL,J. G. Asthenosphere-derived magmatism in the Rio Grande rift, western USA: implications for continental break-up

61

BAILEY,D. K. Episodic alkaline igneous activity across Africa: implications for the causes o f continental break-up

91

ANDERSON,D. L., Yo-SHENZHANG& TANIMOTO,T. Plume heads, continental lithosphere, flood basalts and tomography

99

Bcrrr, M. H. P. The stress regime associated with continental break-up

125

Early stages of Gondwana break-up Cox, K. G. Karoo igneous activity, and the early stages of the break-up of Gondwanaland

137

STOREY,B. C., ALABASTER,T., HOLE,M. J., PANKHURST,R. J. & WEVER,H. E. Role of subduction-plate boundary forces during the initial stages of Gondwana break-up: evidence from the proto-Pacific margin of Antarctica

149

ELUOT, D. H. Jurassic magmatism and tectonism associated with Gondwanaland break-up: an Antarctic perspective

165

BREWER,T. S., HERGT,J. M., HAWKESWORTH,C. J., REX, D. & STOREY,B. C. Coats Land dolerites and the generation of Antarctic continental flood basalts

185

RAPELA,C. W. & PANKHURST,R. J. The granites of northern Patagonia and the Gastre Fault System in relation to the break-up of Gondwana

209

South Atlantic opening

HAWKESWORTH,C. J., GALLAGHER,K., KELLEY,S., MANTOVANI,M., PEATE,D. W., REGELOUS,M. & ROGERS,N. W. Paran~i magmatism and the opening of the South Atlantic

221

WILSON,M. Magmatism and continental rifting during the opening of the South Atlantic Ocean: a consequence of Lower Cretaceous super-plume activity?

241

LIGHT,M. P. R., MASLANYJ,M. P. & BANKS,N. L. New geophysical evidence for extensional tectonics on the divergent margin offshore Namibia

257 ~

vi

com~ms

Northwest Indian Ocean and Gulf of Aden

DEVEY,C. W. & STEPHENS,W. E. Deccan-related magmatism west of the SeychellesIndia rift

271

MENZIES, M. A., BAKER,J., BOSENCE,D., DART,C., DAVISON,I., HURFORD,A., AL' KADASI,M. MCCLAY,K., NICHOLS,G., AL'SUBBARY,A. ~ YELLAND,A. The timing of magmatism, uplift and crustal extension: preliminary observations from Yemen

293

North Atlantic opening SKOGSEID,J., PEDERSEN,T., ELDHOLM,O. • LARSEN,B. T. Tectonism and magmatism during NE Atlantic continental break-up: the Vering Margin

305

LARSEN,L. M., PEDERSEN,A. K., PEDERSEN,G. K. & PIASECKI,S. Timing and duration of Early Tertiary volcanism in the North Atlantic: new evidence from West Greenland

321

GILL,R. C. O., PEDERSEN,A. K. & LARSEN,J. G. Tertiary picrites in West Greenland: melting at the periphery of a plume?

335

HOLM,P. M., HALD,N. & NIELSEN,T. F. D. Contrasts in composition and evolution of Tertiary CFBs between West and East Greenland and their relations to the establishment of the Icelandic mantle plume

349

LARSEN,H. C. & MARCUSSEN,C. Sill-intrusion, flood basalt emplacement and deep crustal structure of the Scoresby Sund region, East Greenland

365

Joy, A. M. Right place, wrong time: anomalous post-fiR subsidence in sedimentary basins around the North Atlantic Ocean

387

Preface The association between fithospheric extension, continental break-up, mantle plumes and massive bursts of igneous activity is well recognized, but their causal relationship remains controversial. According to active mantle hypotheses, rifting is initiated by doming above a mantle plume. Alternative hypotheses consider magmatism as a passive response to lithospheric stretching and rifting with the chance unroofing of a plume only enhancing lithospheric failure and producing abnormally large volumes of basaltic magmatism. Some models combine aspects of both active and passive hypotheses and it is the arrival of a new plume beneath lithosphere already under tension that causes it to split and form a new ocean. The active and passive hypotheses highlight important differences in the relative timing of rifting, magmatism and uplift. Consequently, this debate should be resolved and the main aim of this volume is to integrate relevant tectonic, geochemical and geophysical data which will lead to a better understanding of the causal relationships between magmatism and continental break-up. The first section of the volume is concerned mainly with models of magma generation and breakup processes. Critical to the debate is the origin of the large continental flood basalt provinces and the difficult task of interpreting geochemical signatures. The remaining sections present examples from the geological record. They provide essential feedback to the models and it is clear that some may need to be modified. The debate is, however, by no means over and many of the problems discussed in this volume will be the focus of continuing research for some time. In conclusion the many people who contributed to the production of this volume, including the staff of the Geological Society Publishing House are gratefully acknowledged. We are very grateful to the referees for their careful reviews and for responding quickly to our requirements, and to contributors for making an effort to meet our deadlines. Staff of the British Antarctic Survey, in particular Gill McDonnell, gave much help and time to make the conference on which this volume is based a success and they are warmly thanked. Financial support received from The Royal Society, Shell UK Exploration and Production, Amerada Hess Ltd, Esso Exploration and Production UK Ltd, Intera ECL Petroleum Technologies and ARK Geophysics Ltd underpinned the success of the meeting by enabling keynote speakers to be invited. Bryan C. Storey Tony Alabaster Robert J. Pankhurst

Magmatism during and after continental break-up R. S. W H I T E

Bullard Laboratories, Madingley Road, Cambridge CB3 0EZ, UK Abstract: Magmatism accompanying continental break-up is caused primarily by decompression melting of the underlying mantle as it wells up beneath the rift. The amount of melt produced depends mainly on the temperature of the asthenospheric mantle and on the rate of rifting. Break-up above normal mantle generates only small amounts of melt, reaching a maximum of 7 km in fully oceanic crust. If extension lasts for 10 Ma or more, as on many such 'non-volcanic' continental margins, then still less melt is generated because the slowly upwelling mantle cools by conductive heat loss. Break-up above abnormally hot mantle surrounding mantle plumes generates much larger melt volumes and 'volcanic' continental margins. The largest melt volumes occur when rifting is above a newly initiated mantle plume, with its transient high excess temperatures and flow rates. This creates flood basalt provinces. The resultant basalt may flow distances of over 1000 km, both as surface flows and as dykes and sills intruded laterally in the crust. Only about one-quarter of the melt reaches the surface, the remainder underplating or intruding the lower crust. The excess gravitational potential provided by mantle plumes assists rifling, though plumes do not always cause continental break-up.

Magmatic activity, to a greater or lesser degree, invariably accompanies the formation of new ocean basins by continental break-up. This is a consequence of the decompression and partial melting of the mantle as it wells up beneath the stretched and thinned continental lithosphere. However, the amount of melt that is emplaced on the continental margin varies greatly in different areas. On 'volcanic' continental margins huge melt volumes may be emplaced during continental break-up, with a total of typically 10 million km 3 of new igneous rock being produced in as little as one or two million years or even less. At the other end of the spectrum are so-called 'non-volcanic' margins where only very small amounts of melting are produced as continental break-up proceeds. In this paper I consider mainly the magmatism which occurs on volcanic continental margins due to the presence of thermal anomalies in the mantle, although for completeness I summarize results from 'non-volcanic' margins in the following section. Previous papers discuss the mechanism whereby mantle decompression beneath rifts can produce large quantities of melt very rapidly (e.g. White & McKenzie 1989), and those arguments will not be repeated in detail here. Instead I concentrate on some of the issues related to this hypothesis. In particular I discuss three main areas which affect our interpretation of melt generation during and after continental break-up. They are: the temporal relationship between rifting and magmatism (can we usefully distinguish cause and effect?); the transient conditions that accompany

the initiation of new mantle plumes; and the processes of melt intrusion in the crust and melt flow across the surface of the earth that allow melt to be redistributed over huge distances from the region beneath the rift where it was generated. These questions are all pertinent to the debate over how magmatism and continental break-up are connected, and to the question of how we may relate the spatial and temporal patterns of igneous rocks found in and adjacent to continental riffs to the processes in the mantle that caused partial melting in the first place.

Non-volcanic continental margins The volumes and distribution of melt generated on the so-called 'non-volcanic' continental margins, that formed well away from any thermal anomalies in the mantle, provide a reference against which to compare the voluminous magmatism on the volcanically active margins. The difference between volcanic and non-volcanic margins is illustrated by the crustal cross-sections shown in Fig. 1, one taken from the volcanic Hatton Bank margin in the northern North .Atlantic (after Fowler et al. 1989 and Morgan et al. 1989), and the other from the non-volcanic Goban Spur margin in the eastern North Atlantic (from Horsefield et al. in press). The continental crust thins and breaks to oceanic crust over a similar distance of 50-80 km in each case: the main difference between the two types of margin is that the volcanic Hatton margin exhibits a layer of extruded lava flows up to 4 km thick and

From STOREY,B. C., ALABASTER,T. & PANKHURST,R. J. (eds), 1992, Magmatism and the Causes of ContinentalBreak-up, Geological Society Special Publication No. 68, pp. 1-16.

2

R.S. WHITE

(a) HATTON BANK VOLCANIC MARGIN WEST 0 i

EAST ..... D i p p i n g

10-

OCEANIC

...

. .

refle'ctors

Sedime~..:.~;..__

s.o

~,..~~

/ _./

..

"

\

J"

Heavily

tO" Z/IlllIIlIII/,//IIIIIiI~

\

Intruded

CONTINENTAL

\

or underplated

\a

~ ~ . o

.....

MANTLE

. ' i~

~'s"---------- ~

/

~

'

,.o

30-

35-

'

'

'

'

I

0

'"" '

'

'

50

'

'

km

I

l

,

,

'

,

100

150

(b) GOBAN SPUR NON-VOLCANIC MARGIN ~EST

EAST ,,,,, ,,

Tilted Sediments

10-

. - ~ ......................... ~:,. . . . . .

~ .

.

.

.

......" ...... . . - " ~ " .

.

.

•.'.. •

6.8

"~,::::: . . . . . . . .

~..;

~

~

:

'

....................

~"~ ...................

.

.

.................................. ,..,~

....

,,,,//,/~.~].

20"

"

"/

.......... s. o ....

.. . . . . . . . . . . . . . . . . . . .

"........................ "'"',.,.,,.,

..........

,

...... ~'",,

" ...... "'""

.......... ..........".......................

/

/

,...,

.... ,.,~..........~--:-_':~--~.. ............. ~ . . . ~ . . . . . . . . . . . . . . ....~ . . . . . . . . . . . . . . . . . . . . . 56 . . . . . ?;. ......

"..'.'Z':;:';."..............

",.'"'"'''""

..., 1 5 "

l-

.

,,,

"'~....".............................. .......

OCEANIC ............

block

Basalt

\

v

fault

7

%

"

........

6,41

...................s.s

~

& m

...............

CONTINENTAL

...............

....... 6 e ,

a 25.

MANtLe M O H o ......._._

3o. 35

"

,

0

. . . . v'

~

',

t

50

'

'"

'

' km

'

I

100

• '

' ......

'

'

150

Fig. 1. Cross-sections at the same scales with velocity contours from wide-angle seismic profiles across (a) typical volcanic continental margin west of Hatton Bank (from White etal. 1987; Morgan etal. 1989), and (b) a typical non-volcanic margin on the Goban Spur (from Horsefield et al. in press). Vertical exaggeration is 2.3:1.

MAGMATISM DURING AND AFTER BREAK-UP a prism of lower crustal intrusions, or underplating up to 10 km thick. There is some evidence of volcanism on the 'non-volcanic' Goban Spur margin, but it is restricted to a small pond of extruded basalts only 300 m thick and some lower crustal layering which may represent melt intruded at the time of continental break-up (White 1990; Horsefield et al. in press), or which may be inherited from the pre-break-up history (Peddy et al. 1989). Igneous volumes aside, the other main difference between volcanic and non-volcanic margins is that tilted fault blocks are invariably present in the upper crust of non-volcanic rifted margins, but are not generally seen on their volcanic counterparts (e.g. Fig. 1). The reason may be in part because the extensive volcanism on

~~

~ -20

3O

j~ -18

26,

p-e p-4

20

MagnmU 18.

10

p-a

6

O__ 1200

13,00

1400

1800

PotenUe! Temperature °(3

Fig. 2. Thickness of melt generated by instantaneous adiabatic decompression of asthenospheric mantle over a range of potential temperatures using the method of McKenzie & Bickle (1988), but with an entropy change on melting of 400 J (kg K) -~ rather than the value of 250 J (kg K) -1 that they used, and allowing for compaction of the residual mantle as melt is removed. The schematic temperature history of the Iceland mantle plume is shown based on the variation in thickness of melt generated during initial rifting and subsequent seafloor spreading, assuming pure shear rifting (from White 1992a). High mantle temperatures at the time of continental break-up were associated with the initiation of the Iceland plume, with a subsequent decrease in mantle temperature as the plume matured. Present day temperatures in the area surrounding the mantle plume remain higher than the normal mantle temperature elsewhere which generates 7 km of oceanic crust. Curves of melt generation for thinning by different beta factors are shown for an initial lithospheric thickness appropriate to the thermal plate model of approximately 129 km.

3

the volcanic margins obscures and overprints any tilted fault blocks that may start to form; it may also be that the intrusion of large volumes of melt causes the crust on volcanic margins to behave more ductilely as it becomes heated and weakened by the melt. In general, the transition from unstretched continental crust to fully oceanic crust occurs over a broader region on non-volcanic margins than on their volcanic counterparts, although in some cases such as the Goban Spur illustrated in Fig. 1, the transition on non-volcanic margins may be just as narrow. Both pure shear and simple shear models have been proposed to explain the structure of non-volcanic rifted margins. Recent evidence from seismic reflection (Keen et al. 1989; Sibuet et al. 1990; Sibuet 1992), seismic refraction (Horsefield et al. in press), heat flow and subsidence measurements (Louden et al. 1991) are consistent with pure shear models, and in this paper I assume that continental tiffing is approximated by bulk pure shear. Clearly in the brittle upper layers some deformation is locally by simple shear, but this will not greatly influence our discussion of melt generation processes, because the basic lithospheric mechanism is pure shear and the melt is generated at depths beneath and near the base of the lithosphere. As the lithosphere is stretched and thinned, decompression of the underlying asthenospheric mantle allows some partial melting to occur. When there is extreme stretching, as there is at oceanic spreading centres, the underlying mantle can well up to the base of the crust. At oceanic spreading centres this generates a very consistent thickness of melt, which bleeds upwards from the mantle and solidifies to form the oceanic crust with a mean thickness of 7.1+0.8 km (White et al. in press). This provides a calibration for the normal potential temperature of the asthenospheric mantle (Fig. 2): it is close to 1300°C. On continental margins above normal temperature mantle, if the stretching is sufficiently rapid that the underlying mantle does not cool down significantly by conductive heat loss as it wells up, some melt will be produced as the mantle decompresses once the lithosphere has been thinned by a factor of about three or more (Fig. 2). However, if the stretching and thinning occurs over a time interval of around 10 Ma or more, the amount of melt generated is reduced considerably compared to the instantaneous stretching case illustrated in Fig. 2, because the rising mantle can lose heat by conduction (Bown, pers. comm. 1992). It is not uncommon for the stretching phase to last 10 Ma or more before final continental break-up, so this may

4

R.S. WHITE

(a) explain why only very small volumes of igneous Normnl Atlantic rock are found on 'non-volcanic' continental Oceanic Crust -40% margins, despite considerable thinning by stretching which in some places reduces the continental crust to a thickness of less than 4 kin, L -30% only about half the thickness of normal oceanic Oceanic Crust near crust (Whitmarsh et al. 1986, 1990; Horsefield k:elend Plume -20% 1992; White 1992a). On dried continental margins it is always difficult to be certain, using geophysical methods, as to how much melt has been added to • the crust, and how much of the measured crustal 10 16 o thickness represents residual continental crust, albeit possibly heavily intruded. This problem (b) -40% does not arise with igneous crust generated at an oceanic spreading centre, because the entire crustal section from the top of the basement to the Moho must be new igneous material generated by partial melting of the underlying mantle. -20% Oceanic Crust near ~!:!:i:!:i:i! Because the Moho is a readily identifiable seis- .on-Vole,hie u , r e . |~::~!~i~::~::l mic boundary, the oceanic crustal thickness can be measured easily using wide-angle seismic ~::j d'::']:.~ techniques. Results of such seismic measurements along the non-volcanic continental maro o ;o ,; Igneous Crustal Thickness, km gins of the North Atlantic show that the oceanic crust is abnormally thin immediately adjacent to Fig. 3. Histograms showing thickness of oceanic crust the continent-ocean transition (Fig. 3b): instead of having a normal thickness of about 7 km, it is formed adjacent to (a) volcanic continental margins and (b) non-volcanic continental margins in the North typically only 5-6 km thick, in places reducing to as little as 2 km thick (Ginzburg et al. 1985; Atlantic (from White 1992a). Histogram of normal oceanic crustal thicknesses, for comparison, is from Sibuet et al. 1990; Whitmarsh et al. 1990; Horsemature crust in the North Atlantic away from the field 1992; White 1992a; Pinheiro et al. 1992). influence of hot-spots, fracture zones and active The simplest explanation for the consistently spreading centres. thin oceanic crust adjacent to the North Atlantic non-volcanic margins is that it is caused by very slow spreading. Plate reconstructions and seafloor spreading magnetic anomalies suggest that the continent, which allowed the mantle welling full spreading rates were only 15-22 mm per up under the rift to cool down by conduction. year immediately after opening of the North AtWhen continental break-up finally occurred, it lantic. Very slow seafloor spreading, particularly was this unusually cool mantle which decompressed beneath the initial oceanic rift and thus at the lower end of this range, can reduce markgenerated less melt than the subsequent mantle edly the volume of melt generated and thus produce abnormally thin oceanic crust (Reid & welling up beneath the fully developed oceanic rift. Jackson 1991; White et al. in press). So the unIn summary, continental break-up above usually thin crust adjacent to the North Atlantic margins may be primarily a result of very slow normal temperature mantle is invariably acseafloor spreading rather than a universal fea- companied by some igneous activity, which may manifest itself by surface lava flows or by ture of non-volcanic margins. However, off the sill intrusion. However, if the duration of preGalicia Bank continental margin where the seafloor spreading rate of 22 mm per year was break-up rifting extends over a period of 10 Ma sufficiently high to generate normal thickness or more, conductive heat loss by the asthenospheric mantle welling up under the thinning oceanic crust, Horsefield (1992) has shown that continental lithosphere is likely to be sufficient the oceanic crustal thickness is reduced by 1-3 to decrease significantly the amount of melt genkm over a 15 km wide region adjacent to the continental-ocean transition, beyond which it erated, compared to that resulting from inincreases to normal thickness. This may be stantaneous stretching. This may in some circumstances allow the continental crust to be explained by the relatively long duration of prebreak-up lithospheric stretching and thinning of thinned to less than the normal oceanic crustal !

):::::::::~,oJ~zm

i

MAGMATISM DURING AND AFTER BREAK-UP thickness without extensive accompanying volcanism, and may cause the oceanic crust immediately adjacent to the continent-ocean transition to be thinner than normal.

Volcanic continental margins The main features of volcanic continental margins are their relative narrowness and the immense volumes of igneous rock produced during continental break-up. Much of the melt, perhaps reaching 70% of the total produced, is underplated beneath or intruded into the lower continental crust (Fig. la). Since it is beyond the reach of the drill, the only ways the underplated material can be identified are through wide angle seismic measurements, through the gravity anomaly these dense igneous rocks create, or, indirectly, through the effects on the observed subsidence history of adding large volumes of material to the crust (White & McKenzie 1989). The portion that is erupted often creates characteristic seaward dipping reflector packages on the margin which can be seen on seismic reflection profiles. In some instances (e.g. the Paran~i, the Deccan and the Karoo), the extruded rocks flow over the adjacent continental hinterland to form extensive flood basalt provinces covering more than one million square kilometres. In other cases (e.g. North Atlantic Tertiary Igneous Province), there is much less surface flow and the majority of the extrusive basalts remained on the continental margins and are now underwater. The reason for these differences is probably related to the topography at the time of continental break-up and eruption. Where the adjacent land was relatively flat, as in the case of the Paran~i and the Deccan flood basalt provinces, the melt could flow long distances land-ward on the surface. Where there were many deep basins as was the case with the Rockall Basin, the HattonRockall Basin, the Faeroes Basins, the MOre Basin and the V0ring Basin on the eastern side of the North Atlantic break-up, the melt could not flow across them to reach the northwest European mainland. The huge volumes of melt generated on volcanic continental margins can be attributed most readily to enhanced mantle temperatures in the underlying asthenospheric mantle at the time of break-up, caused by the presence of mantle plumes. As Fig. 2 shows, the amount of melt generated is very sensitive to the mantle temperature. An increase in the mantle temperature of 200°C above normal quadruples the amount of melt created by decompression.

5

Plume shapes Mantle plumes are not steady-state features. From time to time new plumes initiate and old plumes die. The Iceland plume is a good example of a new plume which initiated around 62 Ma, whilst the Bermuda plume is an example of a mantle plume which appears to have died. Nor do mantle plumes exhibit steady-state flow during their lifetime. They initiate as boundary-layer instabilities, though it is still debated as to whether the origin of the instabilities is at the core-mantle boundary, or at the upper-lower mantle boundary. The initial instability exhibits considerably increased mass and heat transport than does the subsequent flow. Many, though by no means all, volcanic continental rifts are associated with the initiation of new mantle plumes and the attendant transient abnormally high mantle temperatures. The volcanic products of plumes vary on timescales ranging from a few days, reflecting flow conditions in the vents; through a few years to thousands of years, caused by magma chamber processes in the crust and possibly by flow conditions of porosity waves in the mantle; to millions and tens of millions of years, reflecting varying flow in the mantle. Plumes exhibit a range of strengths: some are cooler than others, some have lower mass fluxes than others. Lastly, mantle plumes are not always axisymmetric in plan view, even near the surface. It is often convenient to model them as such, and indeed many hot spots such as Hawaii can be explained well by axisymmetric mantle flow. But high Rayleigh number convection, particularly if it has some internal heating, commonly exhibits spoke-like patterns (Parsons & Richter 1981; Houseman 1990). These rising sheets tend to coalesce to form restricted regions of vertical flow which are approximated near the surface by cylindrical upwelling, but nevertheless rising sheets of hot material sometimes reach the surface. If this behaviour occurs in the earth it would sometimes generate linear regions of volcanism which would exhibit essentially the same age of volcanism along their lengths. The 1600 km long Cameroon line, which crosses from the African continent to the Equatorial Atlantic Ocean and which has been shown to arise from asthenospheric mantle sources (Fitton 1987), is a possible example. The mainly Cenozoic alkaline volcanism of the Cameroon Line may be caused by a presently active mantle sheet, but has also been interpreted as due to reactivation of a layer of mantle at the base of the lithosphere modified during the Late Jurassic-Early Cretaceous by the St Helena plume (Wilson & Guiraud 1992).

6

R.S. WHITE

However, even if the Cenozoic volcanism represents reactivation of a metasomatized layer, it would still require a linear thermal anomaly to trigger it. Other possible examples of the effects of mantle sheets are the Line Islands (Schlanger et al. 1984) in the Pacific and Rodrigues Ridge in the Indian Ocean. In the North Atlantic Tertiary Igneous Province (Fig. 4) there is a northwest-southeast trend of the earliest volcanism associated with the arrival of the Iceland plume. It is marked at the southeastern end by dykes which cut across northern Britain into the North Sea, and at the northwestern end by picritic lavas in west Greenland and Baffin Island, in the vicinity of the Davis Strait (Clarke 1970; Clarke & Upton 1971). It is unknown whether the magmatism connects across central Greenland, because the cover of ice obscures the geology. The ages of igneous activity on this northwest--southeast axis are the oldest in the Province, extending back to 62 Ma and pre-dating the continental break-up by several million years (Mussett 1988). Many of the lavas, particularly in west Greenland, are primitive picritic basalts. Subsequent rifting occurred in a direction orthogonal to this trend. It may well be that the initial northwest--southeast trend of the volcanic centres was caused in the early history of the Iceland hot spot by a rising mantle sheet with this orientation, but that the mantle flow subsequently focused to the nar-

¢¢I 11.1 I%.

0

rower plume that presently exists beneath Iceland. The final direction of the North Atlantic rift may reflect another, orthogonal rising sheet of mantle or it may be controlled primarily by regional stress fields. The total volume of melt produced along the initial northwest-southeast trend is only small, and is consistent with the melting that would occur in a rising sheet without the extra decompression that would produce more massive melting under rifted and thinned lithosphere. The largest volumes were generated in the vicinity of the Davis Strait, where the pre-existing stretching responsible for the Labrador Sea had already produced thinner lithosphere.

Melt distribution One of the main observational constraints on magmatism associated with continental breakup is simple mapping of where the melt is now found, and the distribution of different rock types. The occurrence of thick underplated igneous sequences on rifted continental margins is indicative of the presence of anomalously hot mantle beneath the lithosphere at the time of break-up. However, it is dangerous to equate the outcrop area of basalts and dykes with the extent of the underlying thermal anomaly in the mantle, because basalt may flow huge distances

Fig. 4. Reconstruction of the northern North Atlantic at 55 Ma, shortly after the onset of seafloor spreading. The shaded area shows the extent of the lava flows and sills of the Tertiary Igneous Province emplaced during continental break-up (from Smythe 1983; Larsen 1984; Roberts etal. 1984; Uruski & Parson 1985; Skogseid & Eldholm 1987; Larsen & Jakobsd6ttir 1988; Mutter et al. 1988; Spence et al. 1989; White & McKenzie •1989): most lie underwater at present and have been identified from seismic reflection and refraction profiles, with some drill control. The known extent of dykes of the same age in and around Britain and east Greenland is also shown (from Fahrig 1987; Dewey & Windley 1988; Upton 1988): the extent of dykes offshore and beneath the ice of mainland Greenland is almost completely unknown and it is likely that dykes are widespread throughout these areas. The reconstruction uses an equal area Lambert stereographic projection and encompasses an area with a diameter of 3000 km.

MAGMATISM DURING AND AFI"ER BREAK-UP from its sources, both on the surface as flows and within the crust as dykes.

Surface flows

Surface basalt flows may extend many hundreds of kilometres from the source, provided the supply of magma is sufficiently large and there are no major topographic barriers to block the flow. Some of the best estimates for the extent and volume of individual eruptive flows come from the Columbia River Basalt Group, largely because it is still well exposed because of its relatively young age of 17-15 Ma. It has been known since the 1970s that individual flows of up to 700 km 3 exist in the Roza Member (e.g. Swanson et al. 1975, 1979; Hooper 1988), and that they flow many hundreds of kilometres. Recent work suggests that the largest flows in the Grande Ronde Basalts exceed 2000 km 3 and probably approach 3000 km 3in volume (Tolan et al. 1989). Extrusion was from linear vent systems 100-200 km long, and individual flows must have been extruded in periods of between a few days and a week (Shaw & Swanson 1970; Martin 1989). Individual flows have been mapped over distances of more than 750 km from the vents (Tolan et al. 1989). In comparison with other flood basalt provinces such as the Deccan, and the Paramt (Figs 5 & 6), the Columbia River Basalt is an order of

~,

==-~°" 107 km 3) LIPs are oceanic plateaus (Coffin & Eldholm 1991): Ontong Java and Kerguelen. Aside from

Aden Traps Alpha Ridge Austral Seamounts Bermuda Rise Broken Ridge Cape Verde Rise Caribbean Hood Basalts Caroline Seamounts Ceara Rise Chagos-Laccadive Ridge Columbia River Basalt Conrad Rise Crozet Plateau Cuvier Plateau Deccan Traps Del Carlo Rise Eauripik Rise East Mariana Basin Etendeka Ethiopian Flood Basalts Galapagos Hawaiian-Emperor Seamounts Hess Rise Iceland/Faeroe--GreenlandRidge Karoo Kerguelen Plateau Line Islands Lord Howe Rise Seamounts Louisville Ridge Madagascar Ridge Madeira Rise Magellan Rise Magellan Seamounts Manihiki Plateau Marcus Wake Seamounts Marquesas Islands Mashall Gilbert Seamounts Mascarene Plateau

LIP ADEN ALPH AUST BERM BROK CAPE CARI CARO CEAR CHAG COLR CONR CROZ CUVI DECC DELC EAUR EMAR ETEN ETHI GALA HAWA HESS ICEL KARO KERG LINE LORD LOUI MADA MADE MAGR MAGS MANI MARC MARQ MARS MASC

Abbreviation (Fig. 1) CFB SR/OP* SMT OP SR OP CFB/OBFB SMT OP SRt CFB OP OP OP CFB OP OP OBFB CFB CFB SMT SMT OP OP/SRt CFB OP SMT SMT SMT SRt OP OP SMT OP SMT SMT SMT OPt

Type Yes ? No No No No ? No No No No No No Yes Yes ? No No Yes Yes No No No No Yes No No No No ? No No No No No No No Yes

Spatial or temporal association with continental break-up?

Table 1. Large igneous provinces emplaced over the past 250 Ma (Volcanic passive margins in Table 2)

Mohr & Zanettin 1988 Asudeh etal. 1988 Crough 1978 Detrick et al. 1986 MacKenzie 1984 Courtney & White 1986 Bowland & Rosencrantz 1988 Mattey 1982 Supko & Perch-Nielsen 1977 Duncan 1990 Reidel & Hooper 1989 Diament & Goslin 1986 Goslin & Diament 1987 Larsen etal. 1979 Mahoney 1988 Goslin & Diament 1987 Den etal. 1971 Abrams et al. in press Cox 1988 Mohr & Zanettin 1988 Christie et al. 1992 Detrick & Crough 1978 Vallier etal. 1983 Vogt 1974 Cox 1988 Houtz etaL 1977 Sandweli & Renkin 1988 Wellman & McDougaU 1974 Lonsdale 1988 Sinhaetal. 1981 Peirce & Barton 1991 Winterer et o2. 1973 Iwabuchi 1984 Winterer eta/. 1974 Heezen etal. 1973 Fischer etal. 1987 Schlanger et al. 1981 Duncan 1990

Reference

t-

0

0

t'3

/:

MAUD MIDP NATU NAUR NEWE NINE NAVP ONTO OSBO PARA PIGA RAJM RIOG ROOR SHAT SIBE SIER TAHI TASM TAUM WALL WALV

OP SMT OP OBFB SMT SR CFB OP OP CFB OBFB CFB OPt OP OP CFB OP SMT SMT SMT OP SRt

No No Yes No No No Yes No No Yes No ? No No No No No No No No No No

Barker etal. 1988 Winterer & Metzler 1984 Coleman etal. 1982 Shipley et al. in press Duncan 1984 Peirce etal. 1989 Upton 1988; Dickin 1988 Hussong et al. 1979 Sclater & Fisher 1974 Picciriilo etaL 1988 Abrams etal. in press Mahoney etal. 1983 Gamboa & Rabinowitz 1984 Monahan etal. 1984 Den etal. 1969 Zolotukhin & Al'mukhamedov 1988 Kumar 1979 Duncan & McDougal11976 McDougall & Duncan 1988 Duncan & Clague 1985 Symonds & Cameron 1977 Rabinowitz & LaBrecque 1979

CFB, continefital flood basalt; OBFB, ocean basin flood basalt; OP, oceanic plateau; SMT, seamount; SR, submarine ridge. *Referred to in the literature as both a submarine ridge and an oceanic plateau. tOceanic plateaus or submarine ridges which can be tied to LIPs originating during break-up, but for which volcanism post-dates break-up.

Maud Rise Mid-Pacific Mountains Naturaliste Plateau Nauru Basin New England Seamounts Ninetyeast Ridge North Atlantic Volcanic Province Ontong Java Plateau Osborn Knoll Parand Flood Volcanism Pigafetta Basin Raj mahal Traps Rio Grande Rise Roo Rise Shatsky Rise Siberian Traps Sierra Leone Rise Tahiti Tasmantid Seamounts Tuamotu Archipelago Wallaby Plateau Walvis Ridge

22

M.F. COFFIN & O. ELDHOLM

Table 2. Volcanic passive margins (including marginal plateaus) Continent/Subcontinent/ Microcontinent Africa

Antarctica

Australia Greenland

India North America

Northwest Europe

Seychelles South America

Location Abutment Plateau Angola Plain Cape Basin Gulf of Guinea Mozambique Basin Astrid Ridge Explora Wedge GunnerusRidge Weddell Sea Wilkes Land Cuvier Plateau Scott Plateau Morris Jesup Riset Northeast Greenland Southeast Greenland Southwest Greenland Kerala Basin Kutch Basin Baltimore Canyon Trough Carolina Trough Newfoundland Ridge Sohm Abyssal Plain* Bear Island Margin Hatton Bank Jan Mayen Ridge Lofoten Margin MOre Margin VOting Margin Yermak Plateaut Seychelles Bank Argentine Margin Brazilian Margin Falkland Plateau

Reference Hinz 1981 Hinz & Block 1991 Hinz 1981 Rosendahl etal. 1991 DeBuyl & Flores 1986 RoeseretaL 1990 Hinz & Krause 1982 Roeser et aL 1990 Hinz & Krause 1982 Eittrem etal. 1985 Mutter etal. 1988 Hinz 1981 Feden etal. 1979 Hinzetal. 1987 Larsen & Jakobsdottir 1988 Chalmers 1991 Hinz 1981 Hinz 1981 Talwani etal. 1991 Austin etal. 1990 Grant 1977 Hinz & Popovici 1989 Faleide etal. 1988 RobertsetaL 1984 Skogseid & Eldholm 1987 Eldholm etaL 1979 Smythe 1983 Hinz & Weber 1976 Feden et al. 1979 E. Belle pers. comm. 1990 Hinz 1990 Hinz etal. 1988 D. Roberts pets. comm. 1990

*Appears to lie entirely within oceanic crust. tVolcanism continued following break-up.

isotopic evidence on the southernmost Kerguelen Plateau raising the possibility of subcrustal, continental lithospheric contamination (Alibert 1991), both LIPs appear to have been constructed in isolation from continental crust (Cande et al. 1989), and thus probably neither was closely involved in continental break-up. S u b m a r i n e ridges

Submarine ridges are elongated, steep-sided elevations of the seafloor which may be of continental or oceanic origin; in this compilation we only consider the latter. They are commonly characterized by varying topography, and those of oceanic origin may be created either on or off the axes of spreading centers.

Many oceanic submarine ridges follow emplacement of continental flood basalt provinces or oceanic plateaus in. time. A prime example is Faeroe-Iceland-Greenland Ridge (Talwani & Eldholm 1977; Bott 1985) of the North Atlantic Volcanic Province (Morton & Parson 1988). Transient volcanism associated with the breakup of NW Europe and Greenland abated over a broad area shortly after seafloor spreading commenced, but in the Iceland region magmatic activity continued vigorously and produced this submarine ridge. O t h e r examples include the Chagos-Laccadive Ridge Which shares the source, now beneath R6union, which created the Deccan Traps (e.g. Duncan & Richards 1991); Broken Ridge and Ninetyeast Ridge, which together with Kerguelen Plateau had a

LARGE IGNEOUS PROVINCES common source (e.g. Mahoney et al. 1983); and Walvis Ridge, which originated from the same source as the Etendeka province (e.g. O'Connor & Duncan 1990). An oceanic submarine ridge which appears to be linked both spatially and temporally with continental breakup, that of Madagascar and Antartica, is Madagascar Ridge (Sinha et aL 1981), as may its conjugate, the Del Carlo Rise (Goslin & Diament 1987). S e a m o u n t groups

Seamounts are local elevations of the seafloor which are either flat-topped (guyot) or peaked. These marie volcanoes may be discrete, arranged in a linear or random grouping; or connected at their bases and aligned along the ridge or rise. No major seamount groups are spatially or temporally related to break-up of any continents, although such associations would be expected by analogy with submarine ridges. Ocean basin f l o o d basalts

Ocean basin flood basalts are extensive igneous provinces in deep ocean basins which appear to distinctly post-date creation of underlying oceanic igneous basement. This class of oceanic LIP is the least studied of all: of four known examples worldwide, only one, the onshore portion of the Caribbean flood basalt province, appears to be related to continental break-up (e.g. Bowland & Rosencrantz 1988). Volcanism coveting c. 106 km2in the Nauru Basin (Shipley etal. in press) correlates temporally with the emplacement of the Ontong Java Plateau (Tarduno et aL 1991); this may also hold true for the Pigafetta and East Mariana basin flood basalts (Abrams et al. in press). LIPs and oceanic break-up Just as temporal and spatial relationships can be documented for some continental flood basalts and the break-up of continents, break-up of oceanic lithosphere (e.g. Mammerickx & Sandwell 1986) can be related to some oceanic LIPs. The best documented case is break-up between Kerguelen Plateau and Broken Ridge (Mutter & Cande 1983), which occurred when the Kerguelen hotspot was situated at the intersection of the southern terminus of Ninetyeast Ridge, the western end of Broken Ridge, and the northwestern edge of the Kerguelen Plateau. Mammerickx & Sandweil (1986) also proposed that the Line Islands were related to a jump in

23

the spreading center between oceanic lithosphere of Pacific and Phoenix plates. LIP groupings Several hotspot sources can be tied through Early Cretaceous and younger plate reconstructions to many LIPs (e.g. Morgan 1981; Duncan & Richards 1991), including: Iceland: North Atlantic Volcanic Province, Greenland and NW Europe volcanic margins, Faeroe-Iceland-Greenland Ridge. Kerguelen: Bunbury Basalt (Australia), Naturaliste Plateau, Rajmahal Traps, Kerguelen Plateau/Broken Ridge, Ninetyeast Ridge, northernmost Kerguelen Plateau. R6union: Deccan Traps, western Indian volcanic margins, Chagos-Laccadive Ridge, Mascarene Plateau, Mauritius. Tristan da Cunha: Paran,~i and Etendeka basalts, SW African (Abutment Plateau, Angola Plain, Cape Basin, Gulf of Guinea) volcanic margins, Rio Grande Rise, Waivis Ridge. The Iceland source has been active for at least 60 Ma (e.g. Upton 1988), the Kerguelen source for 135 Ma (e.g. Davies et al. 1989), the Rrunion source for 65 Ma (e.g. Courtillot etaL 1988), and the Tristan de Cunha source for 120 Ma (e.g. Piccirillo et al. 1988). Each of these hotspots was temporally and spatially associated with continental break-ups, but their longevity is difficult to reconcile with existing models of LIP origin. Persistent versus transient volcanism

The LIP groupings above, which include the best-studied volcanic passive margins, exemplify long-lived magmatic sources in the mantle, sources which initially and over relatively short (c. 1 Ma) intervals transfer huge volumes of marie rock into the crust, but which later transfer material at a far lesser rate, albeit for relatively long (c. 100 Ma) intervals. In contrast, transient magmatic activity in the mantle is exemplified by many other volcanic passive margins (Fig. 1), along which great volumes of igneous rock are emplaced during and immediately subsequent to continental break-up, but for which no volcanism, aside from that associated with seafloor spreading, persists thereafter. These observations of transient and persistent magma sources suggest that more than one model may be required to explain LIPs in general, and volcanic margins in particular.

24

M.F. COFFIN & O. ELDHOLM

Models for LIP origin and emplacement The original hotspot/mantle plume concept (Wilson 1963; Morgan 1971, 1981) has been developed further recently (Courtney & White 1986; Mahoney 1987; Richards et al. 1989; White & McKenzie 1989; Campbell & Griffiths 1990; Griffiths & Campbell 1990, 1991), generating intense interest in LIPs. Although models for structure and temporal evolution of mantle plumes vary considerably, a commonly observed feature is the capability of a plume to generate large quantities of melt by decompression of upwelling, thermally anomalous mantle. Where the thermal anomaly is associated with continental break-up, the cause-and-effect of which is hotly debated (e.g. Duncan & Richards 1991; Hill 1991; Hill et al. 1992), it may induce the formation of a volcanic margin distinguished by transient magmatism over a wide region. Moreover, the hot, narrow focus of the upwelling mantle may also persist to create an oceanic plateau, a submarine ridge, or a seamount chain on oceanic lithosphere. If the plume initially surfaces through oceanic lithosphere, an oceanic plateau may form, and as the plate migrates over the focus of upwelling a submarine ridge and/or seamounts may be constructed. The mantle plume model, in various forms, is widely supported because it represents the most plausible mechanism for explaining the large amounts of thermal energy required by massive melting anomalies. Neither much of the midplate igneous activity nor any of the topographic swells in the western North Atlantic and eastern North American region, however, are easily reconciled with a simple hotspot or plume (Vogt 1991). One possibility is that these features are caused by shallow mantle convection controlled by vertical thermal boundaries possibly related to episodic midplate stress intensification. Another idea (Mutter et al. 1988) is that volcanic passive margins originate or are augmented by small-scale convective circulation within a narrow conduit of hot, upwelling asthenosphere bounded by cold, old lithosphere. A further feature of recent mantle, plume models, as opposed to Morgan's (1981) ideas, is that plumes and plate kinematics are unrelated phenomena. Some examples, e.g. HawaiianEmperor seamount chain, strongly support this view, but some LIPs, e.g. the Ontong Java Plateau specifically and the Early Cretaceous Pacific volcanic events, in general, are so large that they probably reflect first-order modifications of mantle dynamics. Emplacement of the latter LIPs may be connected to changes in spreading rates in the Early Cretaceous Pacific

(W. I. Morgan pers. comm. 1990), and Early Cretaceous mantle plume activity may be genetically related to large-scale variations in magnetic reversal frequency (Vogt 1972; Larson 1991a, b; Larson & Olson 1991). At least four different models have been proposed for asthenospheric behavior in the formation of volcanic passive margins: (i) a broad plume head impinging on the base of the lithosphere (e.g. Richards et al. 1989; Griffiths & Campbell 1990,1991; Duncan & Richards 1991); (ii) a steady-state, large mantle plume head lying beneath extending lithosphere (e.g. White & McKenzie 1989); (iii) hot zones of upper mantle exploiting either initially weak lithosphere or lithosphere weakened by plate reorganizations (e.g. Anderson et al. this volume). (iv) mantle convectively overturning ('secondary convection') close to the conjugate trailing edges of cold, thick lithosphere (e.g. Mutter et al. 1988). The first three models may also be used to explain, in varying degrees, the origin of classes of LIPs other than volcanic passive margins. The first model is based primarily on the results of laboratory experiments, and describes a convective instability which develops at the coremantle boundary (D" layer), eventually detaching and rising towards the surface. This plume has a large, hot head over a narrow stem. When the plume head impinges on the lithosphere, the continent ruptures and excess volcanism results. This model suggests 'active' rifting, i.e., strain is transferred from the plume to lithospheric plates. The second model, based on observations of crustal structures along volcanic margins and on petrologic modeling, proposes a plumederived thermal anomaly beneath unrifted lithosphere. Heating of the overlying lithosphere could weaken it, but only when independently-driven extension occurs would the anomalously hot asthenosphere adiabatically upwell to produce excess volcanism. This model has both 'active' and 'passive' elements, and has been used to explain continental flood basalts and volcanic passive margins. The third model, rooted in seismic tomography, relies on a thermally, chemically, and isotopically heterogeneous asthenosphere. Plate reorganizations and cratonic zones of weakness allow hot regions of the upper mantle to surface and produce LIPs. The fourth model, based on seismic observations from North Atlantic volcanic margins and fluid dynamic modeling, suggests that the asthenosphere convects on the scale of lithospheric

LARGE IGNEOUS PROVINCES

25

thickness only where lithosphere is rifting, allow- This may suggest that plumes originate in the ing large volumes of mantle to rise and melt via lower mantle. Furthermore, longevity of hotspot adiabatic decompression. This model is entirely sources is hard to reconcile with an exclusive 'passive', with plate separation driving second- plume model in a convecting mantle. Why, for ary, small-scale convection. It requires little example, should the thin plume stem be mainlithospheric extension prior to a sudden break- tained over 100 Ma or more? Also problematic in up, whereas new data suggest that lithospheric a convecting upper mantle is the apparent fixity of extension affects a wide area around the incipient hotspots within two distinct groups, an Atlantic/ Indian and a Pacific, over the past 90 Ma (Miiller plate boundary over 15-20 Ma prior to break-up (Skogseid et al. this volume). et al. 1991). One might envision a convecting mantle ocThe four models are not mutually exclusive; casionally being penetrated by a plume originatcomponents of them may contribute to emplaceing deep in the mantle. This would lead to a ment of volcanic passive margins on regional heterogeneous upper asthenosphere in which scales and to emplacement of a single volcanic margin segment. This is exemplified in the North t e m p e r a t u r e , composition, and fluid content vary regionally. These factors, together with the Atlantic by Skogseid et al. (this volume), who history and rate of Iithospheric extension prior interpret Late Cretaceous-Palaeocene rifting to continental separation, would then determine and rnagmatism to be compatible with the arrival the amount, timing, and position of igneous of the Iceland plume in the upper mantle, and rocks emplaced during break-up. A heterogenwith its subsequent impingement on lithosphere eous asthenosphere would produce the observed under pre-existing regional stress. excess magma at, and in the vicinity of, the crustal focus of deep mantle plumes. On the other Concluding discussion hand, it would also allow formation of volcanic passive margins by increased melt production Study of LIPs is at a nascent stage. The majority away from a plume if rifting occurs within a reof continental flood basalt provinces known can gion of 'abnormal' asthenosphere. Nevertheless, be correlated temporally and spatially with conadditional observational data and modeling are tinental break-up; however the vast Siberian and clearly required to adequately address the origin Columbia River provinces cannot be linked with and evolution of LIPs and their relationship to continental lithospheric separation (Table 1). continental and oceanic break-up. Volcanic passive margins may prove to be the most common type df LIP. As offshore data improve, more and more margins appear to be volWe thank the US Science Advisory Committee and canic (Fig. 1; Table 2), whereas most hotspots on the Joint Oceanographic Institutions for supporting the Earth's surface have probably been identhe Large Igneous Provinces Workshop in Woods tified (Morgan 1981). Thus the proportion of Hole, Massachusetts, 4-6 November 1990. We thank volcanic margins which can be tied to continall participants of and contributors to that workshop ental flood basalt provinces, and to hotspots/ for fomenting much of the work presented herein. We are grateful to H. C. Larsen and P. A. Floyd for mantle plumes, becomes smaller. Some oceanic conscientious and constructive reviews, and to B. C. plateaus and submarine ridges can be linked to Storey for editorial comments. The senior author was continental break-up, but the majority cannot supported by a research scholarship from the Nor(Table 1). No seamount groups can be corwegian Research Council for Science and the related to continental break-up. Of four ocean Humanities, JOI award 16-90 and the sponsors of the basin flood basalt provinces, only the Caribbean PLATES project. The University of Texas at Austin may have a relationship to continental break-up. Institute of Geophysics contribution No. 895. Some LIP emplacements have been related to oceanic break-up. Thus correlations exist between various LIPs References and continental and oceanic break-up, but ABRAMS,L. J., ]_.ARSON,R. L., SttlPLE¥, T. H. & LANcausal mechanisms have yet to be well docuCELOT, Y. In press. Cretaceous volcanic semented. The active plume head, steady-state quences and Jurassic oceanic crust in the East plume, or hot mantle models could account for all Mariana and Pigafetta basins of the Western LIPs; the secondary convection model could apPacific. In: PRINGLE, M. S. & SAGER, W. W. ply exclusively to some volcanic passive margins. (eds). The Mesozoic Pacific. American GeoMantle tomography suggests, however, that physical Union Monograph Series. subduction, not plume upwelling, is the domin- ALIaERT, C. 1991. Mineralogy and geochemistry of a ant upper mantle process (e.g. Dziewonski & basalt from Site 738: implications for the tectonic history of the southernmost part of the Kerguelen Woodhouse 1987; Anderson et al. this volume).

26

M.F. COFFIN & O. ELDHOLM Plateau. In: BARRON,J., LARSEN, B. et al. Pro-

ceedings of the Ocean Drilling Program, Scienafic Results, 119. Ocean Drilling Program, College Station, TX, 293-298. ANDERSON, D. L., ZHANG, Y. S. & TANIMOTO, T. 1992. Plume heads, continental lithosphere, flood basalts and tomography. This volume. ASUDEH, I., GREEN, A. G. & FORSYTH, D. A. 1988. Cgnadian expedition to study the Alpha Ridge complex: results of the seismic refraction survey. Geophysical Journal, 92, 283-301. AUSTIN, J. A., Jr., STOFFA, P. L., PHILLIPS,J. D., OH, J., SAWYER,D. S., PURDY,G. M., REITER, E. & MAKRIS, J. 1990. Crustal structure of the Southeast Georgia embayment-Carolina trough: preliminary results of a composite seismic image of a continental suture (?) and a volcanic passive margin. Geology, 18, 1023-1027. BARKER,P. F., KENNETT,J. P. et al. 1988. Proceedings

of the Ocean Drilling Program, Initial Reports, 113. Ocean Drilling Program, College Station, TX. Boa-r, M. P. 1985. Plate tectonic evolution of the Icelandic transverse ridge and adjacent regions. Journal of Geophysical Research, 90, 9953-9960. BOWLAND, C. L. & ROSENCRANTZ, E. 1988. Upper crustal structure of the western Colombian Basin, Caribbean Sea. Geological Society of America Bulletin, 100, 534-546. CAMPBELL, [. H. & GRIFFrHS, R. W. 1990. Implications of mantle plume structure for the evolution of flood basalts. Earth and Planetary Science Letters, 99, 79-93. CANDE, S. C., LABRECQUE,J. L., LARSON,R. L., PITMANIII, W. C., GOLOVCHENKO,X. & HAXBY,W. F. 1989. Magnetic lineations of the World's ocean basins (chart). American Association of Petroleum Geologists, Tulsa, OK. CHALMERS,J. A. 1991. New evidence on the structure of the Labrador Sea/Greenland continental mar-

~n. Journal of the Geological Society, London, 148, 899-908. C-'SRISTm, D. M.,DUNCAN, R. A., MCBIRNEY,A. R., RICHARDS, M. A., WHITE, W. M., HARPP, K. S. & Fox, C. G. 1992. Drowned islands downstream from the Galapagos hotspot imply extended speciation times. Nature, 355, 246-248. COFFIN, M. F. & ELDHOLM,O. (eds) 1991. Large Igne-

ous Provinces: JOIIUSSAC Workshop Report. The University of Texas at Austin Institute for Geophysics Technical Report, 114. COLEMAN, P., MICHAEL, P. & MUTTER, J. 1982. The origin of the Naturaliste Plateau, SE Indian Ocean: implications from dredged basalts.

Journal of the Geological Society of Australia, 29, 457-468. COURTILLOTr, V., FI~RAUD, G., MALUSKI, H., VANDAMME, D., MOREAU, M. G. & BESSE, J. 1988. Deccan flood basalts and the Cretaceous/Tertiary boundary. Nature, 333, 843-846. COURTNEY, R. C. & WHITE, R. S. 1986. Anomalous heat flow and geoid across the Cape Verde Rise: evidence for dynamic support from a thermal plume in the mantle. Geophysical Journal of the Royal Astronomical Soceity, 87, 815-867.

Cox, K. G. 1988. The Karoo Province. In: MAc. DOUGALL,J. D. (ed.) Continental Flood Basalts. Kluwer, Dordrecht, The Netherlands, 239-271. CROUGH, S. T. 1978. Thermal origin of mid-plate hotspot swells. Geophysical Journal of the Royal Astronomical Society, 55, 4451-4469. DAVIES, H. L., SUN, S.-S., FREY, F. A., GAUTIER,I., MCCULLOCH, M. T., PRICE, R. C., BASSIAS,Y., KLOOTWlIK, C. T. & LECLAIRE, L. 1989. Basalt basement from the Kerguelen Plateau and the trail of a DUPAL plume. Contributions to Mineralogy and Petrology, 103, 457-469. DEBUYL, M. & FLORES, G. 1986. The southern Mozambique Basin: the most promising hydrocarbon province offshore East Africa. In: HALBOUTY,M. T. (ed.). Future Petroleum Provinces of the World. American Association of Petroleum Geologists Memoir, 40, 399-425. DEN, N., LUDWIG, W. J., MURUACHI,S., EWlNG, J., Ho'rrA, H., EDGAR, N., YOSHII, T., ASANUMA, T., HAGIWARA,K., SATO, T. & ANDO, S. 1969. Seismic refraction measurements in the northwest Pacific Basin. Journal of Geophysical Research, 74, 1421-1434. - - , ASANUMA,T., YOSHII, T., KUBOTERA, ,~. & HAGIWARA, K. 1971. Sediments and structure of the Eauripik-New Guinea Rise. Journal of Geophysical Research, 76, 4711-4723. D~'mcg, R. S. & CROUGH, S. T. 1978. Island subsidence, hot spots, and lithospheric thinning. Journal of Geophysical Research, 83,1236-1244. , VON HERZEN, R. P., PARSONS,B., SANDWELL,D. & DOOGHERTY,M. 1986. Heat flow observations on the Bermuda Rise and thermal models of midplate swells. Journal of Geophysical Research, 91, 3701-3723. DIAMENT, M. & GOSLIN,J. 1986. Emplacement of the Marion Dufresne, Lena, and Ob seamounts (south Indian Ocean) from a study of isostasy. Tectonophysics, 121,253-262. DICKIN, A. P. 1988. The North Atlantic Tertiary Province. In: MACDOUGALL,J. D. (ed.). Continental Flood Basalts. Kluwer, Dordrecht, The Netherlands, 111-149. DUNCAN, R. A. 1984. Age progressive volcanism in the New England Seamounts and the opening of the central Atlantic Ocean. Journal of Geophysical Research, 89, 9980-9990. 1990. The volcanic record of the Rrunion hotspot. In: DUNCAN, R. A., BACKMAN, J., PETERSON, L. C. et al. Proceedings of the Ocean Drilling Program, Scientific Results, 115. Ocean Drilling Program, College Station, TX, 3-10. 8£ CLAGUE, D. 1985. Pacific plate motions recorded by linear volcanic chains. In: NAIRN, A., STEHLI, F. & UYEDA, S. (eds) The Ocean Basins

and Margins, Vol. 7A, The Pacific Ocean. Plenum, New York, 89-121. & McDOUGALL, I. 1976. Linear volcanism in French Polynesia. Journal of Volcanology and Geothermal Research, l, 197-227. & RICHARDS, M. A. 1991. Hotspots, mantle plumes, flood basalts, and true polar wander. Reviews of Geophysics, 29( 1), 31-50.

LARGE IGNEOUS PROVINCES DZIEWONSKI, A. M. & WOODHOUSE, J. H. 1987. Global images of the Earth's interior. Science, 236, 37-48. ErlTREM, S. L., HAMPTON, M. A. & CHILDS, J. R. 1985. Seismic-reflection signature of Cretaceous continental breakup on Wilkes Land margin, Antarctica. Science, 229, 1082-1084. ELDHOLM, O. 1991. Magmatic-tectonic evolution of a volcanic rifted margin. In: MEYER, A. W., DAVIES,T. A. & WISE, S. W. (eds). Evolution of Mesozoic and Cenozoic Continental Margins. Marine Geology, 102, 43-61. ~, SUNDVOR,E. & MYHRE, A. 1979. Continental margin off Lofoten-Vester/llen, northern Norway. Marine Geophysical Researches, 4, 3-35. ~, THIEDE,J. & TAYLOR,E. 1989. Evolution of the Vering volcanic margin. In: ELDHOLM, O., THIEDE, J., TAYLOR, E. et al. Proceedings of Ocean Drilling Program, Scientific Results, 104. Ocean Drilling Program, College Station, TX, 1033-1065. FALEIDE, J. I., MYHRE, A. i . & ELDHOLM,O. 1988. Early Tertiary volcanism at the western Barents Sea margin. In: MORTON,A. D. & PARSON,L. i . (eds). Early Tertiary Volcanism and the Opening of the NE Atlantic. Geological Society, London, Special Publication, 39, 135-146. FEDEN, R. H., VOGT, P. R. & FLEMING, H. S. 1979. Magnetic and bathymetric evidence for the 'Yermak hot spot' northwest of Svalbard in the Arctic Basin. Earth and Planetary Science Letters, 44, 18-38. FISCHER, K., McNuTr, M. & SHURE,L. 1987. Thermal and mechanical constraints on the lithosphere beneath the Marquesas swell. Nature, 322, 733-736. GAMBOA,L. & RABINOWITZ,P. 1984. The evolution of the Rio Grande Rise in the southwest Atlantic Ocean. Marine Geology, 58, 35-58. GASS, I. G., CHAPMAN, D. S., POLLACK, H. N. & THORPE, R. S. 1978. Geological and geophysical parameters of mid-plate volcanism. Philosophical Transactions of the Royal Society of London, A288, 581-597. GOSLIN, J. & DIAMENT,M. 1987. Mechanical and thermal isostatic response of the Del Carlo Rise and Crozet Bank (southern Indian Ocean) from altimetry data. Earth and Planetary Science Letters, 84, 285-294. GRANT, A. C. 1977. Multichannel seismic reflection profiles of the continental crust beneath the Newfoundland Ridge. Nature, 270, 22-25. GRIFF1THS, R. W. & CAMPBELL,I. H. 1990. Stirring and structure in mantle starting plumes. Earth and Planetary Science Letters, 99, 66-78. & 1991. Interaction of mantle plume heads with the Earth's surface and onset of small-scale convection. Journal of Geophysical Research, 96, 18295-18310. HEEZEN, B. C., MA'I'I'HEWS, J. L., CATALANO, R., NATLAND, J., COOGAN, A., THARP, M. & RAWSON, M. 1973. Western Pacific guyots. In: HEEZEN, B. C., MACGREGOR, I. D. et al. Initial Reports of the Deep Sea Drilling Project, 20. US

27

Government Printing Office, Washington, DC, 653-724. HILL, R. I. 1991. Starting plumes and continental break-up. Earth and Planetary Science Letters, 104, 398-416. , CAMPBELL,I. H., DAVIES,G. F. & GRIFFITHS,R. W. 1992. Mantle plumes and continental tectonics. Science, 256, 186-193. HINz, K. 1981. A hypothesis on terrestrial catastrophes: wedges of very thick oceanward dipping layers beneath passive margins--their origin and paleoenvironment significance. Geologisches Jahrbuch, E22, 345-363. 1990. The Argentine eastern continental margin: structure and geological development. American Association of Petroleum Geologists Bulletin, 74, 675-676. & BLOCK,M. 1991. Deep seismic reflection studies in the Gambia Plain, in the eastern South Atlantic, and in the Angola Plain with M. V. Bin HaM511. Bundesanstalt fiir Geowissenschaften und Rohstoffe Report, 109.045. & KRAUSE, W. 1982. The continental margin of Queen Maud Land, Antarctica: Seismic sequences, structural elements, and geological development. Geologisches Jahrbuch, E23, 17-41. • & PoPovIcl, A. 1989. Report on a multichannel seismic reconnaissance survey in the Newfoundland Basin and in the Sohm abyssal plain with S.V. Prospekta. Bundesanstalt fiir Geowissenschaften und Rohstoffe Report 105.848. & WEBER,J. 1976. Zum geologischenAufbau des Norwegischen Kontinentalrandes und der Barents-See nach reflexionsseismischen Messungen. Erd61 und Kohle, Erdgas, Petrochemie, Compendium, 75/76, 3-29. & MUTIER, J. C., ZEHNDER,C. M. & NGT STUDY GROUP 1987. Symmetric conjugation of continent-ocean boundary structures along the Norwegian and East Greenland Margins. Marine and Petroleum Geology, 3, 166-187. . . et al. 1988. Report on a multichannel seismic reconnaissance survey of the Argentine eastern continental margin by S.V. Explora. Bundesanstalt fiir Geowissenschaften und Rohstoffe Report, 102.371. HOUTZ, R. E., HAVES,D. E. & MARKL, R. G. 1977. Kerguelen Plateau bathymetry, sediment distribution, and crustal structure. Marine Geology, 25, 95-130. HUSSONG, D. M., WIPPERMAN,L. K. & KROENKE,L. W. 1979. The crustal structure of the Ontong Java and Manihiki plateaus. Journal of Geophysical Research, 84, 6003-6010. IWABUCm,Y. 1984. General Bathymetric Chart of the Oceans (GEBCO). Canadian Hydrographic Service, Ottawa, 5.06. KUMAR, N. 1979. Origin of 'paired' aseismic ridges: Ceara and Sierra Leone rises in the equatorial and the Rio Grande Rise and Walvis Ridge in the South Atlantic. Marine Geology, 30, 175-191. LARSEN, H. C. & MARCUSSEN,C. 1992. Sill-intrusion, flood basalt emplacement and deep crustal structure of the Scoresby Sund region, East Greenland. This volume.

28

M.F. COFFIN & O. ELDHOLM

d~ JAKOBSDOTFIR, S. 1988. Distribution, crustal properties, and significance of seaward-dipping sub-basement reflectors off E. Greenland. In: MORTON, A. D. & PARSON, L. M. (eds). Early Tertiary Volcanism and the Opening of the NE Atlantic. Geological Society, London, Special Publication, 39, 95-114. LARSON, R. L. 1991a. Geological consequences of superplumes. Geology, 19, 963-966. 1991b. Latest pulse of Earth: evidence for a midCretaceous superplume. Geology, 19, 547-550. ~'& OLSON, P. 1991. Mantle plumes control magnetic reversal frequency. Earth and Planetary Science Letters, 107, 437-447. & MLrrTER, J. C., DIEBOLD,J. B., CARPENTER,G. B. & SYMONDS,P. 1979. Cuvier Basin: a product of ocean crust formation by Early Cretaceous rifting off Western Australia. Earth and Planetary Science Letters, 45, 105-114. LONSDALE, P. 1988. Geography and history of the Louisville hotspot chain in the Southwest Pacific. Journal o[ Geophysical Research, 93, 3078-3104. MACOOUGALL, J. D. (ed.) 1988. Continental Flood Basalts. Kluwer, Dordrecht, The Netherlands. MACKENZIE, K. 1984. Crustal structure and realistic seismic data. PhD Thesis, University of California, San Diego. MAHONEY, J. J. 1987. An isotopic survey of Pacific oceanic plateaus: implications for their nature and origin. In: KEAT[NG,B. (ed.). Seamounts, Islands, and Atolls. American Geophysical Union Monograph, 43, 207-220.. 1988. Deccan Traps. In: MACDOUGALL,J. D. (ed.). Continental Flood Basalts. Kluwer, Dordrecht, The Netherlands, 151-194. ~, MACDOUGALL, J. D., LUGMAIR, G. W. & GOPALAN,K. 1983. Kerguelen hot spot source for Rajmahal Traps and Ninetyeast Ridge? Nature, 303, 385-389. MAMMERICKX,J. & SANDWELL,D. 1986. Rifting of old oceanic lithosphere. Journal of Geophysical Research, 91, 1975-1988. MA'rrEY, D. 1982. Minor and trace element geochemistry of volcanic rocks from Truk, Ponape, and Kusaie, eastern Caroline Islands: evolution of a young hot spot trace across old oceanic crust. Contributions to Mineralogy and Petrology, 80, 1-13. McDOUGALL, I. & DUNCAN, R. A. 1988. Age progressive volcanism in the Tasmantid seamounts. Earth and Planetary Science Letters, 89, 207-220. MCKENZm, D. P. & SCLATER,J. G. 1971. The evolution of the Indian Ocean since the late Cretaceous. Geophysical Journal of the Royal Astronomical Society, 25, 437-528. McNuTr, M. 1990. Deep causes of h0tspots. Nature, 346, 701-702. ~, FISCHER, K., KRUSE, S. & NATLAND,J. 1989. The origin of the Marquesas fracture zone ridge and its implications for the nature of hot spots. Earth and Planetary Science Letters, 9 I, 3S 1-393. MOHR, P. & ZANNETrtN,B. 1988. The Ethiopian flood basalt province. In: MACDOUGALL,J. D. (ed.). Continental Flood Basalts. Kluwer, Dordrecht, The Netherlands,63-110.

MONAHAN, D., FALCONER, R. K. H. & THARP, M. 1984. General Bathymetric Chart of the Oceans (GEBCO). Canadian Hydrographic Service, Ottawa, Canada, 5.10. MORGAN,W. J. 1971. Convection plumes in the lower mantle. Nature, 230, 42-43. 1981. Hotspot tracks and the opening of the Atlantic and Indian Oceans. In: EMILIANI,C. (ed.). The Sea, Volume 7, The Oceanic Lithosphere. Wiley-Interscience, New York, 443-487. MORTON, A. C. & PARSON, L. M. (eds) 1988. Early Tertiary Volcanism and the Opening of the NE Atlantic. Geological Society, London, Special Publication, 39. MOLLER, R. D., ROYER, J.-Y. & LAWVER,L. A. 1991. Evidence for hotspot group motion in the Late Cretaceous/Early Tertiary. Geological Society of America, Abstracts with Program, 1991 Annual Meeting, 23/5, A318. MtrrlER, J. C. & CANDE,S. C. 1983. The early opening between Broken Ridge and Kerguelen Plateau. Earth and Planetary Science Letters, 65,369-376. , BUCK,W. R. • ZEHNDER,C. M. 1988. Convective partial melting. 1. A model for the formation of thick basaltic sequences during the initiation of spreading. Journal of Geophysical Research, 93, 1031-1048. O'CONNOR, J. M. & DUNCAN,R. A. 1990. Evolution of the Walvis Ridge-Rio Grande Rise hot spot system: implications for African and South American plate motions over plumes. Journal of Geophysical Research, 95, 17,475-17,502. OKAL, E. A. & BATIZA,R. 1987. Hotspots: the first 25 years. In: KEATING,B., FRYER, P., BATIZA,R. & BOEHLERT, G. W. (eds) Seamounts, Islands, and Atolls. American Geophysical Union Monograph, 43, 1-11. PEIRCE, C. & BARTON,P. J. 1991. Crustal structure of the Madeira-Tore Rise, easternNorth Atlantic results of a DUBS wide-angle and normal incidence seismic experiment in the Josephine Seamount region. Geophyiscal Journal International, 106, 357-378. PEIRCE, J., WEtSSEL,J. et al. 1989. Proceedings of the Ocean Drilling Program, Initial Reports, 121. Ocean Drilling Program, College Station, TX. PICCIRILLO, E. M . , MELFI, A. - J-., (~OMINCHIARAMONTI, P., BELLIENI, G., ERNESTO, i . , MARGUES,L. S., NARDY,A. J. R., PACCA,I. G., ROtSENBERG,A. & STOLFA,D. 1988. Continental flood volcanism from the Paranfi Basin (Brazil). In: MACDOUGALL,J. D. (ed.). Continental Flood Basalts. Kluwer, Dordrecht, The Netherlands, 195-238. POLLACK,H. N., GASS, I. G., THORPE, R. S. & CHAPMAN, D. S. 1981. On the vulnerability of lithospheric plates to mid-plate volcanism: reply to comments by P. R. Vogt. Journal of Geophysical Research, 86, 961-966. RABINOWITZ, P. D. & LABRECQUE, J. 1979. The Mesozoic South Atlantic Ocean and evolution of its continental margins. Journal of Geophysical Research, 84, 5,973-6,001. REIDEL, S. P. & HOOPER, P. R. 1989. Volcanism and

LARGE IGNEOUS PROVINCES tectonism in the Columbia River flood-basalt province. Geological Society of America Special Paper, 239, 386 pp. I~NNE, P. R. & BASU, A. R. 1991. Rapid eruption of the Siberian Traps flood basalts at the PermoTriassic boundary. Science, 253, 176-179. RICHARDS,M. A., DUNCAN,R. A. & COURTILLOT,V. E. 1989. Flood basalts and hot spot tracks: plume heads and tails. Science, 246, 103-107. ROBERTS, D. G., BACKMAN, J., MORTON, A. C., MURRAY, J. W. & KEENE, J. B. 1984. Evolution of volcanic rifted margins: synthesis of Leg 81 results on the west margin of the Rockall Plateau. In: ROBERTS,D. G., SCHNITKER,D. et al. Initial Reports of Deep Sea Drilling Project, 81. US Govt. Printing Office, Washington, 883-911. ROESER, H. A., BARGELOT,H.-O., FRITSCH,J., HINZ, K., KEWITSCH, P., MEYER, H. & SCHRECKENBERGER, B. 1990. Geophysical investigations on the crustal structure off East Antarctica between longitude O* and 400E: final report of the PFS Polarstern cruise A N T VIII/6. Bundesanstalt ffir Geowissensehaften und Rohstoffe Report, 106.883. ROSENDAHL, B. R., GROSCHEL-BECKER,H., MEYERS, J. & KACZMARICK,K. 1991. Deep seismic reflection study of a passive margin, southeastern Gulf of Guinea. Geology, 19, 291-295. SANDWELL,D. T. & RENKIN, M. L. 1988. Compensation of swells and plateaus in the North Pacific: no direct evidence for mantle convection. Journal of Geophysical Research, 93, 2775-2783. SCHLANGER,S. O., CAMPBELL,J. F., HAGGERTY,J. A. & PREMOLI-SILVA,1. 1981. Cretaceous volcanism and Eocene failed atolls in the Radak Chain: implications for the geological history of the Marshall Islands. LOS Transactions of the American Geophysical Union, 62, 1075. SCLATER,J. G. tY¢.FISHER,R. L. 1974. Evolution of the east central Indian Ocean with emphasis on the tectonic setting of the Ninetyeast Ridge. Geological Society of America Bulletin, 85,683-702. SHIPLEY,T. H., ABRAMS,L. J., LANCELOT,Y. & LARSON, R. L. (In press). Late Jurassic-Early Cretaceous oceanic crust, mid-Cretaceous volcanic sequences and seismic stratigraphy of the Nauru Basin, Western Pacific. In: PRINGLE, M. S. & SAGER, W. W. (eds). The Mesozoic Pacific. American Geophysical Union Monograph Series. SINHA, M., LOUDEN, K. & PARSONS, B. 1981. The crustal structure of the Madagascar Ridge. Geophysical Journal of the Royal Astronomical Society, 66, 351-377. SKOGSEID, J. & ELDHOLM, O. 1987. Early Cenozoic crust at the Norwegian continental margin and the conjugate Jan Mayen Ridge. Journal of Geophysical Research, 92, 11471-11491. ~, PEDERSEN,T. & ELDHOLM,O. 1992. Tectonism and magmatism during NE Atlantic continental break-up: the Voring Margin. This volume. SLEEP, N. H. 1990. Hotspots and mantle plumes: some phenomenology. Journal of Geophysical Research, 95, 6715-6736.

29

SMYTHE, D. K. 1983. Faeroe-Shetland Escarpment and continental margin north of the Faeroes. In: Bo'rr, M. H. P., SAXOV, S., TALWANI, M. & THIEDE, J. (eds). Structure and Development of the Greenland-Scotland Ridge-New Methods and Concepts. Plenum, New York, 109-119. SUPKO, P. R. & PERCH-NIELSEN, K. 1977. General synthesis of Central and South Atlantic drilling results. Leg 39, Deep Sea Drilling Project. In: SuPKO, P. R., PERCH-NIELSEN,K. et al. Initial Reports of the Deep Sea Drilling Project, 39. US Government Printing Office, Washington, DC, 1099-1132. SYKES, L. R. 1978. Intraplate seismicity, reactivation of preexisting zones of weakness, alkaline magmatism, and other tectonism postdating continental fragmentation. Reviews of Geophysics and Space Physics, 16, 621-688. SYMONDS,P. A. & CAMERON,P. J. 1977. The structure and stratigraphy of the Canarvon Terrace and Wallaby Plateau. Australian Petroleum Exploration Association Journal, 17, 30-4I. TALWANI, M. & ELDHOLM,O. 1977. Evolution of the Norwegian-Greenland Sea. Geological Society of America Bulletin, 88, 969-999. ~, EWING, J., SHERIDAN, R. E., MUSSER, D. L., GLOVER, L., III, HOLBROOK, S. & PURDY, M. 1991. EDGE lines off the U.S. Mid-Atlantic margin and the East Coast Magnetic Anomaly. L O S Transactions American Geophysical Union, 72, 274. TARDUNO, J. A., SLITER, W. V., KROENKE, L., LEG KIE, M., MAYER, H., MAHONEY, J. J., MUSGRAVE,R., STOREY,M. & WINTERER,E. L. 1991. Rapid formation of the Ontong Java Plateau by Aptian mantle plume volcanism. Science, 254, 399-403. UPTON, B. G. J. 1988. History of Tertiary igneous activity in the N Atlantic borderlands. In: MORTON, A. D. & PARSON,L. M. (eds). Early Tertiary Volcanism and the Opening of the NE Atlantic. Geological Society, London, Special Publication, 39, 429-453. VALLIER, T. L., DEAN, W. E., REA, D. K. & THIEDE, J. 1983. Geologic evolution of Hess Rise, central North Pacific Ocean. Geological Society of America Bulletin, 94, 1289-1307. VOGT, P. R. 1972. Evidence for global synchronism in mantle plume convection and possible significance for geology. Nature, 240, 338-342. -. 1974. The Iceland phenomenon: imprints of a hot spot on the ocean crust, and implications for flow below the plates. In: KRISTJANSSON,L. (ed.). Geodynamics of Iceland and the North Atlantic Area. Reidel, Hingham, MA, 105-126. 1981. On the applicability of thermal conduction models to mid-plate volcanism: comments on a paper by Gass et al. Journal of Geophysical Research, 86, 950-960. 1991. Bermuda and Appalachian rises: common non-hotspot processes. Geology, 19, 41-44. WELLMAN,P. & McDOUGALL,I. 1974. Cainozoic igneous activity in eastern Australia. Tectonophysics, 23, 49-65.

30

M. F. COFFIN & O. E L D H O L M

WHITE, R. S. & MCKENZIE,D. P. 1989. Magmatism at rift zones: The generation of volcanic continental margins and flood hasalts. Journal of Geophysical Research, 94, 7685-7729. WILSON, J. T. 1963. A possible origin of the Hawaiian Islands. Canadian Journal of Physics, 41, 863-870. WINTERER, E. L. & MELTZER, C. -V. 1984. Origin and subsidence of guyots in Mid-Pacific Mountains. Journalof Geophysical Research, 89, 9969-9979.

, EYING, J. I. et al. 1973. Initial Reports of the Deep Sea Drilling Project, 17. US Government Printing Office, Washington, DC. , LONSDALE,P., MATHtEWS, J. & ROSENDAHL,B. 1974. Structure and acoustic stratigraphy of Manihiki Plateau. Deep Sea Research, 21,793-814. ZOLOTUKHIN, V. V. ,g" AL'MUKHAMEDOV,A. I. 1988. Traps of the Siberian Platform. In: MACDOUGALL,J. D. (ed.). Continental Flood Basalts. Kluwer, Dordrecht, The Netherlands, 273-310.

The lower lithosphere as a major source for continental flood basalts: a re-appraisal MARTIN A. MENZIES

Department of Geology, Royal Holloway and Bedford New College, University o f London, Egham, Surrey TW20 OEX, UK Abstract: Continental flood basalt provinces (cfb) have isotopic ratios different from midocean ridge basaits, characteristics that may be due to (a) interaction of asthenospheric melts with old crust, (b) melting of enriched continental lower lithosphere, (c) upwelling of deep mantle plumes containing recycled components, (d) mixing of enriched and depleted mantle sources, or (e) combinations of these processes. It is apparent that several aspects of the chemistry of cfb are subcrustal in origin and therefore crustal contamination cannot be invoked as a generally applicable hypothesis. In addition wholesale extraction of cfb from the lower lithosphere is unlikely because the lithospheric mantle is believed to be a rather thin (< 150 km) reservoir of cold, anhydrous, granular peridotite. Dry melting of such peridotites would not produce a normal 'basalt' because they have already experienced a melting event. Metasomatism or enrichment processes can enhance the chemical budget of the lower lithosphere thus providing an adequate reservoir for small volume alkaline and potassic melts. However no conclusive evidence exists that such modification of the lower lithosphere is widespread enough to generate a laterally continuous, wet, enriched reservoir. It would appear that cfb require a dominant sub-lithospheric component (plume) to account for the large volumes of magma, in some cases produced over a short time period (e.g. Deccan). If so, the heterogeneity observed in mantle-derived magmas within flood basalt provinces reflects inhomogeneities within plumes rather than relatively shallow inhomogeneities in the lithosphere. The apparent concentration of depleted (STSr/Srsr < 0.7045) cfb in the northern hemisphere and enriched cfb (STSr/arSr ~' 0.7045) in the southern hemisphere may be linked to the preponderance of enriched plumes in the Southern Oceans.

Since the early work of Geikie (1987) and Washington (1922), plateau or continental flood basalts (cfb) have been thought of as sub-crustal in origin because of their size and apparent relationship to dike injection within the crust. The first attempts to define the origin of cfb were by Thompson et al. (1972) and McDougall (1976) who believed that cfb originated in the mantle. Thompson (1977) deduced that certain cfb were extruded at rates equivalent to mid-ocean ridge basalts. To account for these large volumes, Thompson (1977) and Swanson et al. (1975) proposed that cfb had an origin similar to oceanic basalts and that any modification of their trace element or isotopic ratio was due to digestion of variable quantities of crust during upwelling or storage of the magmas in the crust. Whilst this is undoubtedly the case for several cfb, including the British Tertiary, many others retain evidence of mantle heterogeneity. A classic example is the Snake River Plain, Idaho, where Leeman (1975) reported enriched tholeiities with high 87Sr/86Sr ratios that lacked unequivocal evidence for wholesale crustal contamination. Consequently, Leeman (1975) proposed that

these enriched tholeiitic melts were derived from the continental keel or sub-continental mantle, a proposition supported by Brooks & Hart (1978) who argued that flood basalts inherited age information from their lithospheric mantle source region. This suggestion of a shallow source for cfb contrasted with that proposed by De Paolo & Wasserburg (1979) and Wasserburg & DePaolo (1979). According to their neodymium isotopic data cfb originated in an undifferentiated (chondritic) source similar to the source of ocean island basalts (OIB) in the lower mantle. Despite these sub-lithospheric models for the genesis of cfb, evidence from mantle xenoliths helped galvenize models involving enriched continental lithosphere (Hawkesworth et al. 1983; Menzies et al. 1983; Carlson 1984; McDougall 1988; Ellam et al. 1991). Isotopic and elemental data on basaltborne and kimberlite-borne xenoliths led to a widespread acceptance that much of the lower lithosphere was chemically enriched and as such a suitable reservoir for cfb (Basu & Tatsumoto 1980; Menzies & Murthy 1980; Menzies & Hawkesworth 1987; Nixon 1987). Pioneering work

From STOREY,B. C., ALABASTER,T. & PANKHURST,R. J. (eds), 1992, Magmatismand the Causes of Continental Break-up, Geological Society Special Publication No. 68, pp. 31-39.

31

32

M.A. MENZIES

by Hawkesworth e t a l . (1983) helped to integrate xenolith and volcanic rock data and to invoke the participation of lithospheric mantle in the production of the Karoo flood basalt province of South Africa. They convincingly demonstrated that much of the elemental and isotopic variability in the Karoo was sub-crustal in origin, a significant conclusion since much of the argument throughout the 1960s and 1970s centred on whether or not cfb retained any evidence of processes other than high level crustal contamination. While it is now generally accepted that crustal contamination is an important process in cfb genesis (e.g. British Tertiary), equally important is the proposition that all cfb provinces do not originate in MORB mantle and that many cfb provinces (e.g. Deccan, Ferrar, Paran~i, Karoo; Fig. 1) require a source similar to, or more enriched than, OIB. Such enriched

sources are sub-crustal and the debate is now centred on whether or not they are located within the continental lithosphere or sub-lithospheric sources (e.g. McKenzie & O'Nions 1983; Sun & McDonough 1989; White & McKenzie 1989). Before embarking on an assessment of the nature of the continental lower lithosphere, it is worth summarizing some of the physical and chemical characteristics of cfb provinces. Individual flood basalt eruptions are very large in volume with variable but high rates of extrusion. While it took approximately 1 Ma to produce the Deccan cfb province, which covers half a million square kilometres (Mahoney 1988), cfb in Yemen-Ethiopia, in close proximity to the Afar plume, were erupted over 10-15 Ma (Chiesa et al. 1989; Fig. 1). Cfb are too magnesium poor and or silica saturated to be primary magmas

(2O Ma)

RIVER (, 17 Me)

~1~)

KEY: ETENOEKA Enriched d b

°°°°°°

o

~EUImON

•

Depleted plume

0

Enriched plume

•

Depleted xenoliths

[]

Enriched xenoliths Cratons

PARANA (leo Me)

(181-1711Me)

TRISTAN

o

IK)UVET

:

d

KERGRUELEN

r'l FERRAR GROUP (I?S Me)

/"io

Fig. 1. Distribution ofcfb provinces in relation to mantle plumes and cratons. Enriched cfb (STSr/S6Sr~ 0.7045) a r e (1) older (45-193 Ma); (2) mainly confined to the southern hemisphere (i.e. Karoo, Etendeka, Parawi) or originated in the southern hemisphere (e.g. Deccan), and (3) related to enriched plumes (e.g. Tristan, Rdunion, Bouvet). Depleted cfb (87Sr/86Sr < 0.7045) are (1) younger (0-60 Ma) with the exception of the Siberian cfb (230 Ma); (2) mainly confined to the northern hemisphere, and (3) related to depleted plumes (Iceland, Columbia River, Afar, Siberia). The distribution of time-integrated enriched (STSr/86Sr~, 0.7045) and depleted (STSr/ 8t'Sr < 0.7045) xenoliths does not appear to relate to the existence of depleted or enriched cfb. Aspects of the elemental and isotopic geochemistry of cfb better reflects the global chemical variability of deep mantle plumes. Data sources: Cfb/cratons--MacDougall (1988 and references therein), Wilson (1989), Hergt et al. (1991), Sharma et al. (1991); plumes--Anderson (1989); xenoliths--Irving & Carlson (1991), Nixon (1987 and references therein), Menzies & Hawkesworth (1987 and references therein), Menzies (1990) and Menzies et al. (1992).

CFB SOURCE generated from lherzolitic mantle. Fractionation must have occurred during the evolution of cfb (Cox 1980) to account for the range in rock types from picrites, alkaline and tholeiitic basalts to rhyolites and ignimbrites. Overall phenocrysts in the basaltic rocks of cfb tend to be anhydrous (plagioclase, clinopyroxene and olivine) and CaO-MgO correlations reveal the importance of clinopyroxene in the genesis of efb either during partial melting or fractional crystallization processes. In addition; a comparison of phenocryst assemblages in MORB and cfb indicates a predominance of plagioclase phenocrysts in cfb (Wilson 1989). Cfb are heterogeneous for Sr, Nd and Pb isotopes and the range in these isotopes is beyond the range for most OIB. While cfb related to the break-up of Gondwanaland are enriched (i.e. S7Sr/86Sr ~> 0.7045), cfb in the northern hemisphere are depleted (i.e. STSr/a6Sr < 0.7045, Fig. 1). In this paper, data pertinent to the nature of the sub-continental lithospheric mantle will be briefly reviewed and it will become apparent that this reservoir cannot act as a widespread homogeneous source for cfb magmatism because it is either (a) too dry and refactory on a large scale, or (b) too wet and inhomogeneous on a local scale.

Lithosphere thickness The thickness of the lithosphere is important in any assessment of the viability of the lithospheric mantle as a major mantle reservoir that can supply large quantities of magma to the Earth's surface. Seismic tomography indicates that some of the thickest lithosphere on Earth exists under the Archaean (> 2500 Ma) cratons (Fig. 1), reaching 200 km beneath the Kaapvaal craton of South Africa (Anderson 1989). Studies of micro-inclusions in diamonds substantiate this assertion, and, more importantly, indicate that the lithosphere had reached a thickness of 150200 km in the first billion years or so of Earth's history (Richardson e t al. 1984; Boyd & Gurney 1986). However, it must be remembered that most continental flood basalts are erupted through tectonically active regions (Fig. 1) located within post-Archaean lithosphere, which is thinner (< 150 kin) than Archaean lithosphere. Therefore, the relevance of the physical (and chemical) properties of the Archaean lithosphere to cfb is perhaps rather limited. Even in the case of the Karoo volcanic rocks in South Africa (Fig. 1), it has been shown that their genesis may have involved transfer of magmas through post-Archaean circum-cratonic (and not Archaean) lithosphere (Hill 1991). Es-

33

sentially the lithosphere is so thin in regions of active tectonism and cfb magmatism that it is an unlikely source for such voluminous tholeiitic magmas (Anderson 1990). On the eastern margins of the southern Red Sea the lithosphere is almost 90 km thick and < 50 km of that is lithospheric mantle (Gettings et al. 1986). While erosion of the lithosphere may have removed some 30-100 km in the last 30 Ma, the youngest volcanic rocks (< 10 Ma) in Yemen and Ethiopia were produced at a time when the lithosphere was so thin as to constitute a minor reservoir. Dry lithosphere Detailed studies of kimberlite-borne and basaltborne xenoliths indicate that much of the lower lithosphere consists of anhydrous peridotites dominated by olivine, orthopyroxene, clinopyroxene and garnet. The Archaean lithosphere comprises lherzolite containing small, isolated pockets of low calcium harzburgite (Schulze 1991) and overall Archaean peridotites are more magnesian than their post-Archaean equivalents (Boyd 1989). In contrast postArchaean lithosphere comprises less magnesian peridotites dominated by lherzolites and harzburgites (Boyd 1989). For example, in eastern Australia (Fig. 1) Cr-diopside spinel lherzolites represent 80-90% of all basalt-borne xenoliths (O'ReiUy & Griffin 1987) and spinel lherzolites comprise > 95% of the xenoliths from the Massif Central and Languedoc regions of France (Downes 1987; Fig. 1). Peridotites that underplate the crust are traditionally thought of as melt residues. For example, Archaean peridotites from the lower lithosphere are believed to represent high temperature residues formed by extraction of komatiitic melts. Equilibrium experiments (Cani11991) support the contention that granular garnet peridotites from beneath the Kaapvaal craton, South Africa (Fig. 1) are residua after komatiite extraction. Lherzolites beneath post-Archaean terrains are interpreted as melt residues produced by extraction of basaltic melts in contrast to the komatiitic melts extracted from Archaean lithospheric peridotites (Boyd 1989). However recent studies indicate that the refractory nature of peridotites can also be explained by melt-rock interaction (Kelemen 1990) particularly in the case of orthopyroxene-rich peridotites with elevated LREE contents. If we accept that the lower lithosphere consists of a mixture of melt residues with low contents of Rb and the light REE and reacted peridotites with variable Rb and light REE contents, then it

34

M.A. MENZIES

is important to note that such material would have highly variable a7Sr/S6Sr and 143Nd/144Nd isotopic ratios on a relatively small scale. Consequently, any cfb extracted from anhydrous lower lithosphere, in the last 200 Ma, would tend to have an isotopic composition similar to that of the lithosphere. If the lower lithosphere is mainly composed of residual peridotites then the situation is worse in that the residue and recently extracted melts would have low S7Sr/a6Sr and high 143Nd/144Nd ratios. However in the case of most cfb the composition of the cfb does not match that of the lithosphere as exemplified by basalt- and kimberlite-borne xenoliths (Fig. 1). Some 60 Ma ago the depleted (87Sr/a6Sr < 0.7045) cfb of the North Atlantic were erupted through lithosphere similar to that found beneath northwest Scotland and Greenland (Scott-Smith 1987; Menzies & Halliday 1988). While the Hebridean-Greenland craton appears to be underlain by enriched spinel-bearing mantle the lower lithosphere beneath the surrounding mobile belts is heterogeneous for Sr isotopes. In fact the cfb of the North Atlantic are much more depleted than the majority of the recovered xenoliths. In contrast the enriched (87Sr/a6Sr > 0.7045) cfb of the Ferrar are associated with heterogeneous depleted and locally enriched lithosphere if the basalt-borne xenoliths of the Ross Sea area can be taken as representative. However one cannot assume that production of large amounts of tholeiitic magma is an intrinsic property of the protolith that constitutes much of the lower lithosphere. Kushiro (1973) carried out dry melting experiments on a granular garnet lherzolite (cold equilibrated lower lithosphere) from Bultfontein, South Africa, and produced liquids which had low Fe/Mg, low alkalies and high Cr contents relative to normal basaltic composition. Several other attempts to generate basaltic melts from granular garnet and spinel lherzolites (Yoder 1976) have failed, mainly because the peridotites in question had already been depleted of their basaltic component. Consequently, the anhydrous protolith that underplates the crust is an Unlikely source of basaltic magmas in any great abundance. The residue has to be chemically modified before it can generate melts of broadly basaltic composition.

Wet lithosphere A significant amount of research over the last ten to fifteen years has concentrated on xenoliths that contain hydrous phases. To help put their relative abundance into perspective it is worth noting that of the 2000 basalt-borne xenoliths studied at one of the classic amphibole peridotite

localities on Nunivak island, Alaska, '~ 1% contained unequivocal amphibole (Francis 1976). Moreover within the Massif Central and Languedoc regions of France amphibole- and mica-bearing xenoliths are rare and restricted to a few localities (Downes 1987). Finally, of the thousands of kimberlite-borne xenoliths studied from the many kimberlite pipes of South Africa, only one pipe contains significant amounts of amphibole-beating peridotites and within that pipe 10% of the xenoliths contain richterite (Erlank et al. 1987). Although some 90% of the xenoliths were hydrated, it is not known to what extent the hydration (growth of mica) reflects primary (pre-entrainment) or secondary (entrainment) processes. While the distribution of amphibole within orogenic massifs confirms the observation from xenoliths that amphibole is rare, it also gives us some insight into the possible mechanisms that may be responsible for the growth of amphibole. Pyroxenite and hornblendite veins (15 cm) in orogenic massifs are occasionally surrounded by reaction aureoles that tend to have hydrated the anhydrous wall rock close to the vein (< 10 cm). In such metasomatic fronts less than 1% amphibole has crystallized within minor fractures or as a result of short-range porous flow (Bodinier et al. 1990). Therefore, amphibole is rather restricted in amount and distribution. If we take the presence of Fe-Ti pyroxenite veins as some indication of the potential for growth of small amounts of amphibole in the lithospheric wall rock, it is of interest to note that of the 2213 basalt-borne xenoliths studied in the southwestern USA only 18% contain Fe-Ti augite (igneous) pyroxenite veins (Wilshire et al. 1988). If one accepts xenoliths as representative of the lower lithosphere, one must conclude that the growth of amphibole in the lithosphere is rather limited. In several instances amphibole has grown as a result of the upwelling of mantlederived melts, a process that has led to transformation of the dry reduced protolith. Melts tend to be transferred by crack propagation through the lithosphere and amphibole, mica, apatite and titanates grow within these fractures/conduits or immediately adjacent to them due to short range melt percolation. The transformed mantle is referred to as 'enriched mantle' due to the introduction of incompatible elements. With time the low Sm/Nd (titanates, amphibole, mica, clinopyroxene) and high Rb/Sr (mica) ratios within the new minerals 'age' to produce high S7Sr/~Sr and low 143Nd/144Nd ratios, very different from the initial isotopic composition of a residual protolith (i.e. low STSr/S6Srand high 143Nd/ 144Nd). It is this aspect of the continental lower lithosphere that is most appealing when search-

CFB SOURCE ing for a source region for isotopicaUy enriched cfb magmas. As a result of metasomatism and enrichment processes, the 'aged' isotopes required for the production of cfb are resident within the mechanical boundary layer of the lithosphere. However, if one uses the 143Nd/ 144Nd ratio as a measure of amount and possible age of enrichment, xenolith data indicate that kimberlite-borne xenoliths and diamond inclusions from beneath the Archaean cratons are the most enriched xenoliths (e.g. Greenland; Scotland; Wyoming; Kaapvaal) (Fig. 1). In contrast, xenoliths from beneath circum-cratonic mobile belts are less enriched and have the isotopic characteristics of 'oceanic' mantle (e.g. western USA; western Europe; eastern China; eastern Australia) (Fig. 1). However most cfb provinces are erupted through post-Archaean 'oceanic' lithosphere (Fig. 1) so it is rather unlikely that they would encounter an isotopically enriched reservoir, of sufficient size, en route to the surface. It is apparent from the study of alkali basalts and potassic volcanics that enriched regions within the lithospheric mantle beneath circum-cratonic regions are probably small in size. Wet melting of the lithosphere containing < 1% amphibole will produce melts very different in composition from those generated by dry melting. Early work by Green (1973) showed that pargasite was a major subsolidus phase to nearly 30 kbar, after which its breakdown sharply depressed the solidus (pyrolite +0.2% water). This work also demonstrated that wet melting of undepleted compositions at 20 kbar produced olivine tholeiites and quartz normative basaltic andesites/quartz tholeiites at 10 kbar. More recently Foley (1991) studied the stability of fluor- and hydroxy-endmembers of pargasite and K-richterite and extended the depth range for K-richterite bearing assemblages to 350 km and fluor-pargasite to 1300°C and 35 kbar. Foley (1991) discussed the melting of amphibole in the subcontinental lithosphere and stressed that these experimental results are of greatest significance to basalt genesis in regions where amphibole is concentrated in veins and not in the regions where amphibole is a minor phase. Partial melting of such wet veined mantle would, according to Foley (1991), consume most of the vein material first, but due to the different stability of F- and OH-endmembers, amphibole or mica would still be stable when most of the vein had melted. The presence of stable hydrous phases in the vein and wall rock would facilitate melting of the adjacent wall rock. Some anhydrous peridotites have been chemically modified such that they are enriched in magmatophile elements and can produce basaltic melts.

35

Sheared garnet lherzolite from Thaba Putsoa (Kushiro 1973) produced basaltic liquids of quartz tholeiitic composition at 10-15 kbar and olivine tholeiite at 20 kbar at temperatures of 20-50 ° above the solidus. This occurred because the sheared peridotite was less depleted than the dry, granular peridotites. One can conclude from this section that parts of the sub-continental mantle are hydrous and susceptible to wet melting. However, the sparseness of amphibole must be borne in mind when wet melting models are applied to the lower lithosphere.

Enriched lithosphere When regions of the lithosphere that underlie cfb provinces like the Ferrar, Deccan and Karoo are sampled by alkaline basalts, the entrained xenoliths do not reveal the presence of laterally continuous enriched reservoirs, but rather a localized distribution of enriched and depleted xenoliths that may even be restricted to one diatreme (Fig. 1). Within the pockets of enriched mantle wet melting would be restricted to the vein conduits and the immediately adjacent wall rock, a factor that would tend to produce alkaline melts with an isotopic chemistry similar to the vein minerals. Genesis of cfb must require more widespread melting of the lithospheric mantle and to what extent this is viable is unknown given the apparently restricted distribution of hydrous phases. Moreover the composition of the melts produced by wet melting of the lithospheric mantle tends to reflect the nature of the veins, which are themselves the crystallization products of the high pressure equivalents of alkaline basalts, kimberlites and lamproites. Consequently, the wet melts tend to be alkaline or potassic in character since they are produced by melting of amphibole-mica-apatite-bearing assemblages, a feature that contrasts with the dominantly tholeiitic or andesitic nature of efb magmas. Recently the extreme isotopic heterogeneity in mantle xenoliths was used to argue that enriched continental lower lithosphere was a source for small volume alkaline and potassic melts (Smith 1983). Moreover, theoretical considerations indicated the melting of metasomes adequately explained that origin of potassic and alkaline magmas within continental regions (McKenzie 1989). However, the genesis of such small volume melts within the continental lithosphere by wet melting must be reconciled with the need to produce large amounts of tholeiitic melt (i.e. cfb) from within the same reservoir by dry melting. If one accepts that cfb are capable of inheriting their isotopic signature by interac-

36

M.A. MENZIES

tion with enriched lithosphere or by melting of the lithosphere, one would expect to find evidence for contamination of melts with enriched mantle wall rock in suites of mantle xenoliths or in orogenic massifs. This does not appear to be the case, as re~,ealed by detailed studies of meltwall rock reaction zones (Bodinier et al. 1990). Detailed studies of elemental and isotopic effects adjacent to veins reveals extreme fractionation of elements and isotopic heterogeneity (Downes et al. 1991; Macpherson et al. unpublished data) on a scale similar to that observed in basalt-borne and kimberlite-borne xenoliths. Clearly magma transfer by crack propagation through mantle peridotite has produced reaction zones within the surrounding wall rock but little or no evidence for changes in magma chemistry due to 'stoping' of mantle rocks. Since melt conduits are armoured against contamination from the surrounding lithospheric mantle by reaction zones (metasomatic and enrichment fronts), such chemical buffers are complex hybrids that essentially protect the melt from inheriting the isotopic characteristics of the surrounding lithospheric mantle. It is, therefore, difficult to envisage a situation where large amounts of melt could be transferred rapidly through the lithosphere by crack propagation while simultaneously acquiring the chemical characteristics of the lithosphere.

Sub-lithospheric sources, plumes and cfb Swanson et al. (1975) and Thompson (1977) believed that cfb originated in a mantle source similar to that of oceanic basalts with the added complication of high level assimilation of crustal rocks. Several isotopic studies substantiated this assertion and proposed that sublithospheric mantle sources were primarily responsible for the production of cfb (DePaolo & Wasserburg 1979; Wasserburg & DePaolo 1979; DePaolo 1988; Sharma et al. 1991; Lassiter & DePaolo 1991). DePaolo (1988) reviewed the available isotopic data on cfb and noted that in the case of the British Tertiary crustal contamination had modified the composition of a mantle-derived magma. He proposed that the uncontaminated magma was derived from a source similar to that of the East Greenland-Baffin Island magmas (i.e. Iceland plume). In the case of the Columbia River and the Karoo magmas, DePaolo (1988) invoked the participation of a primitive mantle source (lower mantle plume?). Geological and geochemical data on worldwide cfb (McDougaU 1988; McKenzie & O'Nions 1991) support a sublithospheric origin for cfb either in the asthenosphere or lower mantle. More specifically in the case of the British Tertiary (North Atlantic), the

Deccan, the Paran~i and the Karoo (Fig. 1), McDougall (1988) believed that the tholeiitic magmas were 'probably produced primarily by' melting of upwelling asthenosphere'. Significant contributions to cfb from deeper, possibly lower, mantle sources are indicated by the similarity in chemistry between cfbs and ocean island basalts (Thompson et al. 1983; McDougall 1988; Anderson 1989) and the close spatial relationship between cfb and plumes (White & McKenzie 1989). In Fig. 1 the distribution of cfb and plumes is shown in relation to craton boundaries and the possible composition of the lower lithosphere as indicated by xenoliths. It should be noted that some of the oldest and most controversial Phanerozoic cfb (i.e. Paran~i, Etendeka, Karoo and Ferrar) are located in the southern hemisphere, a part of the world where oceanic volcanic rocks are known to be isotopically enriched due to the presence of enriched plumes. In contrast some of the youngest cfb (i.e. Snake River Plain, Afar, Iceland) are located in the northern hemisphere close to depleted plumes. If the isotopic geochemistry of cfb was more related to the composition of the lower lithosphere one might expect a closer spatial relationship to the isotopic composition of xenoliths (Fig. 1). In general this is not the case. While time-integrated enrichments are observed in the subcratonic lithosphere of the western USA, Arabia, South Africa and Greenland, only in South Africa (Karoo cfb) and Idaho (Snake River Plain cfb) are enriched sources proposed for cfb. It has already been stressed that many cfb tend to be erupted through post-Archaean lithosphere, thus avoiding any time-integrated enriched reservoirs. Finally, it seems paradoxical that when seeking a large uniform reservoir for cfb the thin lithosphere is chosen as a suitable reservoir but deep mantle plumes are preferred as a source for OIB which are, in general, volumetrically smaller than cfb. While it is likely that within ocean islands there may be contributions from the lithosphere in the form of small volume alkaline melts (e.g. post-erosional series on Hawaii) the bulk of cfb must originate from sublithospheric sources.

Conclusion Any hypothesis that argues for a significant contribution from the lower lithosphere in the genesis of cfb must address the following issues. (a) In regions where cfb are erupted the lithosphere may be too thin to act as a major reservoir. Some cfb provinces are associated with lithosphere thicknesses equivalent to that beneath

CFB SOURCE ocean islands like Hawaii. In the case of OIB the bulk of the volcanic rocks are believed to be sublithospheric in origin. (b) The thickest and most chemically enriched reservoirs exist in the lower lithosphere beneath Archaean crust. However most cfb are erupted through post-Archaean crust, or on the edge of cratons and as such would avoid these enriched reservoirs. (c) Dry melting of a melt residue similar to that which constitutes part of the lower lithosphere does not produce 'basaltic' melts. Melt residua have a 'depleted' isotopic composition (87Sr/86Sr -----N-MORB; 14"3Nd/144Nd> NM O R B ) very different from that required to produce either 'enriched' (S7Sr/S6Sr > 0.7045) or 'depleted' cfb (grSr/86Sr = OIB). (d) Metasomatism or ingress of small volume melts can provide the vital minerals to facilitate wet melting of the lower lithosphere but xenolith data indicate that the modal amount of amphibole is a per cent or so. Wet melting experiments are of the greatest significance to basalt genesis in regions where amphibole is concentrated in veins. Such veined mantle is believed to be the source of small volume alkaline and potassic magmas, a fact that would have to be reconciled with extraction of voluminous cfb from the same source. (e) On the basis of xenoliths, flood basalt provinces appear to be underlain by highly variable lithospheric mantle that is extremely heterogeneous on a small scale. Moreover adjacent to most cfb provinces the lithospheric mantle is depleted. (f) Vein-wall rock studies indicate that melts transferred by crack propagation through the lower lithosphere tend to modify the adjacent peridotite and not vice-versa. Thus it is rather unlikely that silicate melts (alkaline to tholeiitic composition) would inherit the chemical identity of the lithosphere during transEort to the surface. (g) While enriched (~lSr/a6Sr -> 0.7045) plumes appear to be spatially associated with enriched cfb and depleted (87Sr/86Sr < 0.7045) plumes with depleted cfb (Fig. 1), the distribution of enriched and depleted xenoliths seem to be more constrained by the presence of cratons and mobile belts respectively.

The development of ideas relating to the genesis of cfb provinces have benefited greatly from discussions with D. Anderson, K. G. Cox, C. J. Hawkesworth, D. McKenzie and M. F. ThirlwaU. M. Norry is thanked for constructive comments on an earlier version of this paper.

37

References ANDERSON,D. L. 1989. Theory of the Earth. Blackwell Scientific Publishers, Boston. 1990. Prospecting for the reservoirs Basalt, Heat and Isotope. Magmatism and Continental Break-up. Geological Society Programme with Abstracts. BASU, A. R. & TATSUMOTO, M. 1980. Nd isotopes in selected mantle-derived rocks and minerals and their implications for mantle evolution. Contributions to Mineralogy and Petrology, 75, 43-54. BODINIER,J. L., VASSEUR,G., VERNIERES,J., DUPUY, C. & FABRIES, J. 1990. Mechanisms of mantle metasomatism: geochemical evidence from the Lherz orogenic peridotite. Journal of Petrology, 31,597-628. BoYI), F. R. 1989. Compositional distinction between oceanic and cratonic lithosphere. Earth and Planetary Science Letters, 96, 15-26. & GURNEV,J. J. 1986. Diamonds and the African lithosphere. Science, 239, 472-477. BROOKS, C. & HART, S. R. 1978. Rb-Sr mantle isochrons and variations in the chemistry of Gondwanaland's lithosphere. Nature, 271, 220-223. CANIL,D. 1991. Experimental evidence for the exsolution of cratonic peridotite from high temperature harzburgite. Earth and Planetary Science Letters, 106, 64-72. CARLSON,R. W. 1984. Isotopic constraints on Columbia River flood basalt genesis and the nature of the subcontinental mantle. Geochimica et Cosmochimica Acta, 48, 2357-2372. CHIESA,S., CIVETI'A,L. DE Fir~o, M., LA VOLPE,L. & ORSl, G. 1989. The Yemen Trap Series: Genesis and evolution of a continental flood basalt province. Journal of Volcanology and Geothermal Research, 4, 307-314. Cox, K. G. 1980. A model for flood basalt volcanism. Journal of Petrology, 21,629-650. DEPAOLO, D. J. 1988. Neodymium isotope geochemistry. Springer-Verlag, Berlin. & WASSERaUR6, G. J., 1979. Neodymium isotopes in flood basalts from the Siberian Platform and inferences about their mantle sources.

Proceedings of the National Academy of Science USA, 76, 3056-3060. DOWNES, H. 1987. Relationship between geochemistry and texture type in spinel lherzolites, Massif central and Languedoc, France. In: NIXON,P. N. (ed.). Mantle Xenoliths. John Wiley and Sons, England, 125-134. , BODINIER, J.-L., THIRLWALL,M. F., LORAND, J.-P., & FABRmS,J. 1991. REE and Sr-Nd geochemistry of eastern Pyrenean peridotite massifs: sub-continental lithospheric mantle modified by continental magmatism. Journal of Petrology, Lherzolites Special Volume, 97-116. ELLAM, R. M., CARLSON,R. W. & SHmEY,S. B. 1991. Re-Os isotope constraints on flood basalt genesis. Transactions of the American Geophysical Union, 72, 579.

38

M.A. MENZIES

ERLANK,A. J., WATERS,F. G., HAWKESWORTH,C. J.. HAGGERTY, S. E., ALLSOPP,H. L., RICKARD,R. S. & MENZlES, M. A. 1987. Evidence for mantle metasomatism from the Kimberley Pipes, South Africa. In: MENZIES,M. A. & HAWKESWORTH,C. J. (eds). Mantle Metasomatism. Academic Press, London, 221-311. FOLEY, S. 1991. High pressure stability of the fluorand hydroxy- endmembers of pargasite and Krichterite. Geochimica et Cosmochimica Acta, 55, 2689-2694. Frtmqos, D. M. 1976. Amphibole pyroxenite xenoliths: cumulate or replacement phenomena from the upper mantle, Nunivak Island, Alaska. Contributions to Mineralogy ad Petrology, 58, 51-61. GEIKIE, A. 1897. The Ancient Volcanoes of Great Britain, 2. London. GETHNGS, M. E., BLANKJr., H. R., MOONEY,W. D. & HEALY,J. H. 1986. Crustal structure of southwestern Saudi Arabia. Journal of Geophysical Research, 91, 6491-6512. GREEN, D. H. 1973. Experimental melting studies on a model upper mantle composition at high pressure under water saturated and water undersaturated conditions. Earth and Planetary Science Letters, 19, 37-53. HILL, R. I. 1991. Starting plumes and continental break-up. Earth and Planetary Science Letters, 104, 398-416. HAWKESWORTH, C. J., ERLANK, A. J., MARSH, J. S., MENZIES, M. A. ,g' VAN CALSTEREN, P. 1983. Evolution of the continental lithosphere: evidence from volcanics and xenoliths in southern Africa. In: HAWKESWORTH,C. J. & NORRY, M. (eds). Continental Basalts and Mantle Xenoliths. Shiva, England, 111-138. HERGT, J., PEATE, D. W. & HAWKESWORTH,C. J. 1991. The petrogenesis of Mesozoic Gondwana low Ti flood basalts. Earth and Planetary Science Letters, 105, 127-133. IRWNG, A. J. & CARLSON, R. W. 1991. Mantle xenoliths in potassic magmas from Montana: Sr, Nd and Os isotopic constraints on the evolution of the Wyoming craton lithosphere. CPRM Special Publication, 2/91,183-185. KELEMEN,P. B. 1990. Reaction between ultramafic rock and fractionating basaltic magma I. Phase relations, the origin of calc-alkaline magma series, and the formation of discordant dunite. Journal of Petrology, 31, 51-98. KusHmo, I. 1973. Partial melting of garnet lherzoliths from kimberlite at high pressures. In: NIXON, P. H. (ed.). Lesotho Kimberlites. Lesotho National Development Corporation, Maseru, Lesotho, 294-299. LASSn'ER, J. C. & DEPAOLO, D. J. 1991. Isolation of mantle plume isotopic signatures by analysis of temporal variations in continental flood basalts. Transactions of the American Geophysical Union, 72, 579-580. LEEM~, W. P. 1975. Radiogenic tracers applied to basalt genesis in the Snake River Plain Yellowstone National Park region - evidence for a 2.7-

b.y.-old upper mantle keel (abs.). Geological Society of America Abstracts with Programs, 7, 1165. MAHONEY, J. J. 1988. Deccan Traps. In: MACDOUGALL,L D. (ed.). Continental Flood Basalts. Kluwer, Holland, 151-194. McDOUGALL, I. 1976. Geochemistry and origin of the Columbia River Group, Oregon and Washington. Geological Society of America Bulletin, 87, 777-792. , 1988. Continental Flood Basalts. Kluwer, Holland. MCKENZlE, D. 1989. Some remarks on the movement of small melt fractions in the mantle. Earth and Planetary Science Letters, 95, 53-72. & O'NIoNs, R. K. 1983. Mantle reservoirs and ocean island basalts. Nature, 301,229-231. & ,1991. Partial melt distributions from inversion of rare earth concentrations. Journal of Petrology, 32, 1021-1092. MENZmS, M. A. (ed.) 1990. Continental Mantle. Monograph in Geology and Geophysics number 16. Oxford University Press, Oxford, England. & HALLIDAY,A. J. 1988. Lithospheric mantle domains beneath the Archean and Proterozoic crust of Scotland. Journal of Petrology, Lithosphere Special Issue, 275-302. & HAWKESWORTH, C.. J. (eds) 1987. Mantle Metasomatism. Academic Press, London. , LEEMAN, W. P. & HAWKESWORTH,C. J. 1983. Isotope geochemistry of Cenozoic volcanic rocks reveals mantle heterogeneity below western U.S.A. Nature, 303, 205-209. & MURTHY,V. R. 1980. Enriched mantle: Nd and Sr isotopes in diopsides from kimberlite nodules. Nature, 282 634-636. , THIRLWALL,M. F., WEIMING,F. & ZHANG,M. 1992. Depleted and enriched lithosphere beneath eastern China - evidence from Quaternary alkaline volcanic rocks and their xenoliths. Transactions of the American Geophysical Union, 73, 324. NIXON, P. H. (ed.) 1987. Mantle Xenoliths. J. Wiley and Sons, England. O'REmLY, S. Y. & GRIFHN, W. L. 1987. Eastern Australia---4000 kilometres of mantle samples. In: NIXON, P. H. (ed.). Mantle xenoliths. John Wiley, England, 267-280. RICHARDSON,S. H., GURNEY,J. J., ERLANK,A. E. & HARRIS, J. W. 1984. Origin of diamonds in old enriched mantle. Nature, 310, 198-202. Scmn.z~, D. J. 1991. Low Ca garnet harzburgite xenoliths from southern Africa: abundance, composition and bearing on the structure and evolution of the suberatonic lithosphere. CPRM Special Publication 2/91,350-352. SCOTT-SMrrH, B. 1987. Greenland. In: NIXON, P. H. (ed.). Mantle Xenoliths. John Wiley and Sons, England, 23-40. SHARMA,M., BASU,A. R. & NESTERENKO,G. V. 1991. Nd-Sr isotopes, petrochemistry and origin of the Siberian flood basalt USSR. Geochemica et Cos= mochimica Acta, 55, 1183-1192. SMrrH, C. B. 1983. Pb, Sr and Nd isotopic evidence for

CFB SOURCE sources of African Cretaceous kimberlites. Na-

ture, 304, 51-54. SON, S-S & MCDONOUGH,W. F. 1989. Chemical and isotopic systematics of oceanic basalts: implications for mantle composition and processes. In: SAUNDERS, A. D. & NORRY, M. J. (eds). Magmatism in the Ocean Basins. Geological Society, London, Special Publication, 42, 313-345. SWANSON,D. A., WRiorrr, T. L. & HELZ, R. T. 1975. Linear vent systems and estimated rates of magma production and eruption for the Yakima Basalt on the Columbia Plateau. American Journal of Science, 275, 877-905. THOMPSON, R. N. 1977. Columbia/Snake River- Yellowstone magmatism in the context of western U.S.A. Cenozoic geodynamics. Tectonophysics, 39, 621-636. , Essos, J. & DUNHAM,A. C. 1972. Major element chemical variations in the Eocene lavas of the Isle of Skye, Scotland. Journal of Petrology, 13, 219-253. , MORRISON, M. A., DICKIN, A. P. & HENDRY, G. L. 1983. Continental flood basalts.., arachnids rule OK? In: HAWKESWORTH,C. J. & NORRY,

39

M. J. (eds). Continental basalts and mantle xenoliths. Nantwich, Shiva 158-185. W^SSE~URO, G. J. & DEPAOLO, D. J. 1979. Models of earth structure inferred from neodymium and strontium isotopic abundances. Proceedings of the National Academy of Sciences USA, 76, 3594-3598. WASHINCTON, H. S. 1922. Deccan traps and other plateau basalts. Bulletin of the Geological Society of America, 33, 765-804. WHITE, R. & MCKENZIE, D. 1989. Magmatism at rift zones: The generation of volcanic continental margins and flood basalts. Journal of Geophysical Research, 94, 7685-7729. WmSmRE, H. G., MEYER, C. E., NAKATA, J. K., CALl(, L. C., SHERVAIS,J. W., NIELSON,J. E. SCHWARZ~ANN,E. C. 1985. Mafic and ultramafic rocks of the western United States. United States Geological Survey, 85. WmsoN, M. 1989. Igneous Petrogenesis. Unwin Hyman, London. YODER, H. S. 1976. Generation of Basaltic Magmas. National Academy of Sciences, Washington DC.

Consequences of plume-lithosphere interactions A. D. SAUNDERS,

M. S T O R E Y , R. W. K E N T & M. J. N O R R Y

Department of Geology, University of Leicester, Leicester LE1 7RH Abstract: Splitting or thinning of lithosphere above a mantle plume can result in voluminous

melt generation, leading to the formation of large igneous provinces, or LIPs. Examples of LIPs include continental flood basalt provinces and oceanic plateaus. Basaltic samples from the Ontong Java Plateau, Nauru Basin and Manihiki Plateau, which are among the largest of the LIPs, have isotopic compositions within the range of ocean island basalts. The majority of continental basalts, however, record a trace element and isotopic contribution from the lithosphere through which they have erupted. We are thus unable to reconcile the available compositional data with models which derive the isotopic and large-ion lithophile element-enriched character of continental flood basalts solely from sub-lithospheric mantle plume sources. A combination of mantle sources is indicated, with the thermal energy being supplied by voluminous melts from a plume, and the lithospheric components in continental flood basalts being inherited by contamination of plume-derived melts by low melting point hydrous and carbonated fractions in the lithosphere. Successive injection of plume-derived melts serves to heat the lithosphere, reducing its viscosity and making it susceptible to rupture if allowed by regional plate forces. Furthermore, the lithosphere, including the mechanical boundary layer, may be thinned by thermal stripping from below, allowing the plume mantle to ascend and decompress further. Such a system has the potential for positive feedback leading to rapid melt generation. While we do not exclude recent models of LIP formation which require the sudden impact of a new mantle plume, we favour a model whereby the thermal anomaly builds gradually, incubating beneath a steady-state lithospheric cap.

Rupture of lithosphere enables the ascent, decompression and melting of asthenospheric mantle. The volume of melt produced will be a function of the temperature of the mantle and the degree of extension of the overlying lithosphere (or lithosphere thickness). Oceanic crust

(Hawaii?) 0

O/B & other intraplate volcanics

•

g ~

0

0

z

Ocean Plateaus & Continental Flood Basalts (Iceland)

~" (CapeVerde?)

+

(Bouvet)

~ No surface magmatism? >150 km

a:

0 km Lithosphere Thickness

Fig. 1. The relationship between lithosphere (mechanical boundary layer) thickness, mantle potential temperature and magma volumes as predicted by the models of McKenzie & Bickle (1988), Watson & McKenzie (1991). For simplicity, it is assumed that variations in the composition and volatile content of the mantle do not exert a large control on melting.

is the normal expression of this process, where extension factors are infinite, and the asthenosphere mantle temperatures are 'normal' (McKenzie & Bickle 1988; Fig. 1). If the mantle temperatures are abnormally high, for example above a plume, then the volume of melt produced at the extensional plate boundary may be much greater, resulting in the production of thickened oceanic crust (e.g. Iceland, Walvis Ridge, or seaward-dipping reflector sequences along plate margins). Large volumes of basalt may also be erupted on thinned lithosphere, continental or oceanic, forming continental flood basalts and oceanic plateaus. These voluminous provinces, which include flood basalts, oceanic plateaus, aseismic ridges and seawarddipping reflectors, constitute some of the most dramatic expressions of terrestrial magmatism. Many are erupted in a short period of time, thus testifying to very rapid eruption rates, and they provide strong evidence for episodicity of the thermal structure of the Earth's mantle. In this paper we shall use the term 'large igneous provinces', or LIPs (Coffin & Eldholm 1991), for these basaltic provinces (Fig. 2). Most if not all LIPs can be related to mantle plumes (e.g. Morgan 1971; White & McKenzie 1989). In many instances the mantle plume can be back-tracked to a hot-spot which is presently active, albeit on a much reduced scale compared

From STOREY,B. C., ALABASTER,T. & PANKHURST,R. J. (eds), 1992, Magmatismand the Causes of ContinentalBreak-up, Geological Society Special Publication No. 68, pp. 41-60.

41

A.D. SAUNDERS ETAL.

42

-

,..%,

°

.~

JJ~'" ~ • 30°N

'berian

I f

Columbia~111

- "~ I='-'~

River/,

1,~,

~ : . , ~r~.-Ii,= l . _

~ .

~

a~

° (--4 ~,~Z',~.~3adbbean

~ _ ~ Nauru i3asin 0° 'e/~.,m,,~..-~,i • ~ ~A ~ ' . . I q F ', I i Manihiki Plateau ~J ~. Ontong Java Plateau

I1' ~'~.~"

%

/" 4~" " t

~ ~"~i_-~ '~.

I~ •

._

~

. ~ "~i,, IF k "~

-

~ J

. Pa'aran @,J i ,--.. ~ ! f ~ ~(

"

~I ,-4,k~k / ' ~ ~

,¢.L~r',_~'~-J__ ~ ~--z

~" j

I

~ _~-~

-,.s---~

I,kr

180"

.. ,~,., P '

~-~1~

"~

Deccan . Raima~o~.~-' ~_.,Deccan

"~..~ Afar m'~ I \ / = I ) Karoo / ~ t ' ~

~dlJl~# • "',

o~~

\

.i

Rio Grande * "~ Rise . .~ i.I Crozet M Rise Wa!vls Plateau'4bu, ~ ,~ HKlge

o=01,, - i

,

L .arge Igneous Provinces

Traps

!,J

fi

o,

~

Gaussberg

,',t~ 90°W

00

90°E

Fig. 2. Present-day location of known large igneous provinces (LIPs), modified from Coffin & Eldholrn (1991). Seaward-dipping reflector sequences, which represent a substantial volcanic contribution to some continental margins, are not shown.

with the ancestral LIP. This may be due to the waning activity of the plume, or to the burial of the plume beneath increasingly thick lithosphere, thus reducing its capability for melt generation. This association between LIPs and plumes enables us to use the erupted basalts to test models of the interactions between the plume and the lithospheric cap. For example, what proportion of the melts preserved in the LIP originate in the lithosphere, plume-mantle, or adjacent asthenospheric-mantle? If we can recognize a lithospheric component, how is it incorporated: as small volume contaminants of plume-derived melts, or as wholesale melting of the lithosphere? If the latter, how is thermal energy transferred from the plume into the lithosphere? By using a combination of basalt chemistry, regional studies of uplift and extensional events, and accurate dating of magmatic events, we may be able to deduce the onset of plume-lithosphere interactions and determine whether plumes ascend rapidly and impact with the lithosphere (Richards et al. 1989; Griffiths & Campbell 1990) or accumulate more gradually and incubate beneath the lithosphere (Kent 1991). In this paper we develop a model whereby a vigorous plume incubates to produce a large

thermal anomaly beneath a lithospheric cap. He,it is transferred into the lithosphere, initially conductively but eventually, as the plume head begins to melt, convectively. The thermal boundary layer at the base of the lithosphere will be thinned, allowing ascent and decompression of the thermal anomaly. Any precursor low melting point fractions within the lithosphere may be mobilized, and could contaminate plumederived melts. Progressive heating and melt injection into the mechanical boundary layer will assist thermal and mechanical erosion from its basal layers. The system has the potential to develop positive feedback, with sudden and voluminous melt release when allowed by regional extension of the lithosphere. Plume-lithosphere

interactions

Mantle plumes are significant carriers of heat and material within the Earth's interior (Morgan 1971; Davies 1988; Sleep 1990). The efficacy of this convective system is limited, however, by the cool, rigid lithospheric boundary layer with which most plumes collide and interact. As a consequence, the heat transfer mechanism becomes much less efficient, changing from predominantly convective to mainly conductive,

PLUME-LITHOSPHERE INTERACTIONS

43

Domingof lithosphere Mobilisationof / low-temperature(hydrous Thermal& mechanical & carbonated?)melt -- erosionof MBL fractions~ / . . . . . -,-,,.~-,-, .,..,.,-,-. :,,:,:. . . . ,,,,,,,,,, • ,,,,,,,,,,,,,,,,,:~,,,, ,,,,,, ,,,,,,,,,,,>>>,,>,,,,,,,,,,,.

• A

~ J~J I A

Plumefromthe lowermantle

2000kin

Preexisting 'thinspot'

Contemporaneous Extension

Ii:i:i:i:!:i:!:i:i:I:::ii:i:!?)ili?i!ii!iiiiiiiii!::iiii!i:!i_ ::::::::::::::::::::::::::::::::::i:i, ,:,:,:,> ::,...,.,, /

Mal i;7 ....

B

MaC,n,

Mantle Plume

Basalts

3. Impacting plume head

2. Ascending ~ thermal entrains lower mantle

~

4. Decompression in plume head

I

5. Narrow 'followthrough' plume stem

~ 1. Instability in D" layer on core-mantle boundary

Fig. 3. Schematic diagrams of plume-lithosphere interactions. (a) Incubation of a plume head (potential temperature (Tp) of the plume core is shown to be 270°C greater than ambient mantle) beneath a thick mechanical boundary layer (> 100 km). Extensive melting will result only if the lithosphere is extended and thinned (McKenzie & Bickle 1988; White & McKenzie 1989); (b) As for (a) but with the thermal anomaly channelled by pre-existing topography at the base of the lithosphere, or by contemporaneous extension, resulting in extensive decompression melting (White & McKenzie 1989; Thompson & Gibson 1991); (e) Impacting plumehead 'start-up' model of Campbell & Griffiths (1990), Griffiths & Campbell (1990) and Richards et al. (1989). According to this model, if the plume temperature is sufficiently high (ATe > 300°C?) melting may precede lithosphere extension; picrites are erupted above the hottest part of the plume.

44

A.D. SAUNDERS E T A L .

unless a melt phase is present. The lithospheric cap may be viewed as controlling the evolution of the plume, both by trapping the plume head beneath the boundary layer and by restricting melting (Watson & McKenzie 1991) (Fig. 3a). This plume head may, according to some models, achieve a diameter of 1000 km (Courtney& White 1986) or even 2000 km (White & McKenzie 1989) beneath a stationary plate. Alternatively, the plume may be channelled towards zones of thin lithosphere (for example, an active spreading ridge, or an asthenospheric ridge beneath a 'thinspot') where melting can occur (Morgan 1978; Schilling 1991; Storey etal. 1988; Thompson & Gibson 1991); or the excess heat within the plume may be exploited by contemporaneous extension of the overlying lithosphere (Fig. 3b). In the model outlined above, the plume behaves passively, responding to structural changes in the lithosphere. Alternatively, the plume may play a much more active role. A large 'start-up' plume, originating as a convective instability near the core-mantle boundary, may intersect its melting curve at sufficient depth to enable it to undergo extensive melting prior to lithosphere extension (Richards et al. 1989; Griffiths & Campbell 1990) (Fig. 3c). In addition to producing large igneous provinces, such a vigorous event may even promote continental breakup and dispersion (Morgan 1971, 1972, 1981) (but see Hill 1991). The large volumes of LIPs, and the apparent short duration of their main emplacement, have been used as evidence in support of the plume impact model (e.g. Campbell & Griffiths 1990). As we argue below, however, correlation between the magmatic event and the timing of the impact of the ascending plume may not be valid in all provinces. Other predicted consequences of a plume impinging on the base of the lithosphere include: (i) doming, caused by the dynamic uplift above the plume (Courtney & White 1986), by thermal expansion of the lithosphere (Sleep 1990), and by magmatic underplating (Cox 1989); (ii) thermal and mechanical erosion of the base of the lithosphere, and the incorporation of this material in the asthenosphere and deeper mantle (Withjack 1979; Olson et al. 1988; Storey et al. 1989; Mahoney et al. 1989, 1991, in press b); (iii) injection of plume-generated melts into the lithosphere and consequential changes in both the composition (Hart et al. 1986; Menzies 1990) and the theology of the lithosphere (e.g. Withjack 1979; Spohn & Schubert 1982, 1983; Olson et al. 1988); (iv) storage of 'fossilized' plume mantle in the base of the lithosphere or upper asthenosphere and its tapping by subsequent

extension (Halliday et al. 1988); and (v) thermal mobilization of low-temperature melt fractions within the lithosphere (McKenzie 1989). Clearly, the subsequent thermal and compositional evolution of the lithosphere will vary strongly between a model which predicts longterm incubation of a plume head, and one which invokes a sudden impact event. All continent-based LIPs show compositional evidence for the involvement of continental lithosphere, crust or lithospheric mantle, in at least part of their eruptive sequences, and as such they are distinct from their oceanic counterparts. There is still considerable debate, however, about how the lithospheric component is incorporated. Does this component reflect substantial (c. 10%) melting of the lithospheric mantle (Hawksworth et al. 1988; this volume), or is it merely a contaminant of plume- or asthenosphere-derived melts? The requirement for a powerful thermal input from a plume or hot mantle source, but a compositional input from the lithosphere in production of continental flood basalts, may be resolved by transporting the thermal energy in voluminous, incompatible-element-dilute, melt fractions from the decompressing plume head, and hybridizing these liquids with small volume, incompatibleelement-concentrated, melt fractions from the lithosphere.

Evidence for plume involvement in LIP formation Three types of evidence may be used to argue for the involvement of mantle plumes in LIP formation: topographic, thermal and compositional. The causative link between continental flood basalts and present-day hotspots was noted by Burke & Dewey (1973) and Morgan (1971, 1981), and several recent studies have refined these early proposals (e.g. Campbell & Griffiths 1990; Griffiths & Campbell 1990; Richards et al. 1989; White & McKenzie 1989). Topographic

evidence

Topographically, several LIPs can be linked to extant mantle hotspots and, by implication, mantle plumes, often via aseismic ridges (Duncan 1978, 1981; McDougall & Duncan 1980; Morgan 1982; Richards et al. 1989; White & McKenzie 1989; Duncan & Richards 1991). Of the continental flood basalt provinces, the clearest examples of this are provided by the North Atlantic Tertiary Province-Greenland/ Faeroes Ridge-Iceland, the Deccan-Chagos/ Laccadive Ridge-La Rtunion Island, the

PLUME-LITHOSPHERE INTERACTIONS Etendeka/ParanA-Walvis Ridge/Rio Grande Rise-Tristan da Cuhna Island, the RajmahalNinetyeast Ridge-Kerguelen Islands, and the Madagascar-Madagascar Ridge-Marion Island systems. All of these examples were characterized by significant plate motion following the initial burst of magmatism, thus producing a distinctive plume trail. Other continental flood basalt provinces do not have such readily distinguished plume trails, for example the Columbia River Basalts (Yellowstone plume?) and the Karoo (the Crozet or Marion Island plume?). A few provinces have no obvious plume associated with them: the Siberian Traps and the Ferrar Province, for example. Of the Cretaceous oceanic plateaus, only the Kerguelen Plateau has been assigned a mantle plume (the Kerguelen plume) with any confidence, although the Ontong Java and Caribbean Plateaus may have formed above the Louisville and Galapagos plumes, respectively (Duncan & Hargraves 1984; Mahoney & Spencer 1991). The absence of a plume trail or obvious extant hotspot does not imply that a plume was not involved; it probably reflects the fact that, because of subsequent plate configurations, it has not been detected. In addition to the direct link between plumes and LIPs, there is the topographic evidence of doming in the region of some LIPs (Kent 1991; Kent et al. in press) which indirectly supports the notion of dynamic support and magmatic underplating above a thermal anomaly. Thermal evidence

The volume of LIPs, and the relatively short duration of their activity, necessitates a powerful thermal input. High mantle temperatures alone may not be sufficient, however, because if the lithosphere is too thick, insufficient melt will be produced unless the plume mantle is much hotter than that predicted by current models (e.g. McKenzie & Bickle 1988; Watson & McKenzie 1991). Ideally, a combination of heat and low pressure (= extension) is required (White & McKenzie 1989); this is supported by the geochemical data (see below, and Ellam 1992). That oceanic plateaus represent the melting products of large volumes of thermally enhanced mantle is also not disputed. An oceanic crustal (melt) thickness of 36 km (an estimated crustal thickness for the Ontong Java Plateau: Furomoto et al. 1970) represents a mantle potential temperature (Tp) in excess of 1500°C at a /3factor of infinity, according to the parameterization of McKenzie & Bickle (1988). The large volume of the magma generated in these provinces necessitates very large mantle volumes (for

45

example, 50x 106 km 3 of basaltic magma implies a source volume of about 500x 106 km 3, assuming 10% melting; the thermal volume is likely to be much larger). Even if plateau production is extended to several tens of millions of years, the high throughput of mantle material requires a deep-rooted (lower mantle) source to the thermal anomaly. o 0

soo .

.

,

.

i

~ooo .

.

.

.

I

.

15oo T "C .

.

.

I

~

|

i

~!~]:::: []ii]i]iii::!]iii[]ii[iii]]~]]i]i::i]]ii]i]i]::~ii][]]il]]i]]] []]]]]iii]]::ii]::]]]]::i]]]]::::li] :.:.:.:.:.:.:.:.::.:.::.:.:==================================================== ..: ======================!:=== e

~

::::::::::::::::::::::::::::

~

i::i~i~~~~ ~ ' ~ . :

::::::::::::::::::::::::::::::::::::::::::::::::::

:~~ii~~~~~!~;i~ii

======================== ::: : : . . - t ~

:!Boundary :!:i:!:i:i:i:i:ili:i:i!i:iiiiiiii~:i!i~iiiiiiii!!!iiii ii Layer ?i????!i]!i!i;!?i!]?iii!?!;ii!??ili!i~:~i!i!iiii]i?i!i!i;iii~!!iii!i?]!i!ii!il am :!~:?~:::~:: km ~

~

~

Fig. 4. Lithosphere terminology and steady-state thermal conditions used in this paper (McKenzie & Bickle 1988). Mantle solidus temperatures are from McKenzie & Bickle (1988) and Wyllie (1987). Area outlined by dashed line is expanded in Fig. 9; ATp = 300°Crepresents the possible increase in potential temperature in the adiabatic.mantle during the development of a mantle plume. The mechanism of formation of such large thermal anomalies is uncertain. As mentioned in the introduction, they may represent the impact of large thermals at the base of the lithosphere. Alternatively, the anomaly may incubate beneath the lithospheric cap. Such incubation can produce a mushroom-shaped thermal anomaly surprisingly rapidly. For example, with a steadystate plume flux rate of 12 km 3 a -1, which is thought to be the plume rate supplying the Hawaiian system (Schilling 1991), it will take approximately 20 Ma to develop a plume head with a thickness of 80 km, a diameter of 2000 km, and a volume of about 250x 106 km 3. This is not inconceivable in the situation of a slowly-moving plate. Extension of the lithosphere then causes decompression and melting. Several workers have noted the short duration of flood basalt activity in the Deccan Province (possibly less than 1 Ma: Courtillot et al. 1986), the Siberian Traps (< 3 Ma: Renne & Basu 1991) and the Columbia River Province (the main Grande Ronde formation was erupted in less than 2 Ma: Hooper 1988). Tarduno et al. (1991) have suggested that the bulk of the Ontong Java Plateau and adjacent Nauru Basin

46

A.D. SAUNDERS E T A L .

flood basalts were emplaced within a very short period of time (c. 3 Ma). It is not clear, however, if the pulse of major activity was of such a short duration in all LIPs, because reliable age constraints are not yet available (Kent et al. in press). The short duration of flood basalt activity implies high magma effusion rates, estimated to be 0.06 km 3 a -1 for the Grande Ronde basalts (Hooper 1988), and up to 1.5 km 3 a -1 for the Deccan Traps (Richards et al. 1989). Total magma supply rates of between 8 and 22 km 3 a -1 are proposed for the Ontong Java Plateau and Nauru Basin on the basis of a 3 Ma formation duration (Tarduno et al. 1991). For comparison, the total Hawaiian magma flux rate is estimated at 0.16 km 3 a -1 (Watson & McKenzie 1991). The arguments about high rates of magma production, and the short duration of magmatism in LIPS, are equally valid for the incubation model as they are for the impact model. The difference is that in the impact model, the arrival of the plume head is closely linked to the volcanism, whereas in the incubation model, the plume head may build slowly and only erupt when the lithosphere is sufficiently thin. It is difficult to see how, in the plume impact model, significant volumes of melt can be generated simultaneously with impact unless the lithosphere is already sufficiently thin (for example, if the plume impacts with young oceanic lithosphere or a spreading axis). The advantage of the incubation model is that it enables significant thermal transfer from the plume to the lithosphere, if the plate is moving sufficiently slowly. Further work on the relationships between doming, magmatism and extension, is urgently required to test these alternative models. Compositional evidence

The compositional evidence (trace element and isotopic) for plume involvement in LIP formation varies from province to province. Few continental provinces show unambiguous chemical evidence for incorporation of plume material in their basaltic magmas. Some, such as the Ferrar province, show none (e.g. Hergt etal. 1991). This may partly reflect a lack of suitable sampling, or the way in which plume-derived magmas interact with the lithosphere. It may also reflect difficulties in recognizing the plume component: there is a wide range of compositional types being sampled by present-day ocean islands (Figs 5-7). Plume or asthenospheric components characterized by low 87Sr/86Sr ratios, high eNd values and oceanic Pb isotopes have been reported in the Ambenali Formation of the Deccan (Light-

foot & Hawkesworth 1988; Mahoney etal. 1982; Mahoney 1988) (Figs 5 & 6), tholeiites from the East Coast of Madagascar (Mahoney et al. 1991; Storey etal. unpublished data), and the youngest parts of the Lower Lavas and the successions along the Blosseville Coast in eastern Greenland (Holm 1988; Hogg et al. 1989). An asthenospheric or plume-like component ('Cl' of Carlson & Hart 1988) is found in a few early basalts belonging to the Imnaha basalts of the Columbia River Basalt Group. In all of these provinces, where some basalt sequences have been related to a plume (viz. La R6union, Marion, and Iceland), the mantle plume has an isotopic composition which enables it to be distinguished from lithospheric components. In other provinces, the plume signature may be less easily distinguished from the putative lithospheric input. For example, basalts from the Tristan da Cunha plume have a 'Dupal' character (elevated 2°spb/2°4pb and 87Sr/a6Sr ratios), resembling basalts from the Paran~i province (Hawkesworth et al. 1986), but difficult to distinguish in any proposed mixing arrays. No unequivocal plume or asthenosphere mantle signature have been detected in the Karoo or Fertar tholeiites, although Ellam & Cox (1991) have proposed that depleted asthenospheric melts have mixed with lithosphere-derived lamproitic melts to explain the array of eNd values in the Karoo Nuanetsi potassic picrites. We have plotted data for basalts from a range of oceanic and continental large igneous provinces in Figs 5 to 8. It is interesting to note that in Pb and Nd isotope space, the Ontong Java, Nauru Basin and Manihiki Plateaus have similar compositions to basalts from the Ambenali formation of the Deccan. Oceanic plateau basalts (and the limited data for seaward-dipping reflectors) show a much more restricted range of isotopic and trace element compositions than their continental counterparts and, with the exception of basalts from Sites 738, 747 and 750 from the Kerguelen Plateau, they fall within the range of oceanic island basalts (Figs 5 & 6). This may be due to the limitations of the data base, but it also suggests that the plume and mantle melting processes are homogenizing otherwise diverse sources. It may also suggest that ocean plateau basalts are derived from plumes with deep-seated origins, for example the lower mantle, which appears to be a prerequisite from the volume evidence. In terms of trace element patterns (Fig. 7), the basalts from oceanic plateaus and some continental LIPs show lower La/Lu ratios than the intraplate ocean island basalts (e.g. La R6union, Tristan da Cunha or St Helena: Fig. 7). This

PLUME-LITHOSPHERE INTERACTIONS

47

/

.....- ~.~.;."

it8738,

15.7

~

~~'~'~

A

@/~eulae; Louisville~ ~ j Y / Ft6union / ~ i ~ i y

+

| '

15.6

~

15.5

-

I

:-'~-~.y

Oceanic Plateaus

",

I

NB:NauruBasin _

~

~ ~ ~ a v ~ t a ' ~

15.4 '

:

ManihikiPlateau

I~MoR~ +r~stan 5ii++++~, 18

I

I

19

I

15.8

I. . . . . .

~"

; ] Nuanetsi~~ 15.6

[

KP:Kemue,?P,ate?u !! I

20

=°epb/~Pb ~R/

16.0

I

21

-

B

........ /

arana S

'~N~',,~,~

22

BushePoladpur

I

i

.....

i

~ c e a n i c Paran~~ ~ - - ~ ~ Plateaus

j

22 15.2 16

17

18

19

20

21

~pb/2°4pb Fig. 5. Present-day 2°TPb/~Pb versus 2°6pb/2mpb for (a) oceanic plateau basalts, ocean island basalts and midocean ridge basaits (MORB); and (b) continental flood basalts and picrites. Note the much greater spread of data in the continental basalts, which has been interpreted as lithosphere input: either from the crust (e.g. in the Bushe-Poladpur basalts from the Deccan) or from subcrustal lithospheric mantle (e.g. Mahabaleshwar, Deccan, and the high-Ti Urubici basalts from Paran,'t). Abbreviations and data sources: KP: Kerguelen Plateau (Storey et al. 1992; Site 738 from Alibert 1992); OJ: Ontong Java Plateau; NB: Nauru Basin; MP: Manihiki Plateau (aU from Mahoney et aL in press a); fields for ocean island basalts and MORB from Mahoney & Spencer (1991); Deccan data (Bushe, Poladpur, Panhala, Ambenali and Mahabaleshwar) from Lightfoot & Hawkesworth (1988); Paranfi from Hawkesworth etal. (1986); Karoo (Nuanetsi Picrites) from Ellam & Cox (1989). The terms Urubici and Gramado are from Peate (1989) and correspond to high- and low-Ti basalt types, respectively.

difference is particularly pronounced in the comparison between ocean islands and oceanic plateaus, and it probably reflects the extent and pressure of melting during partial melting. The fiat to slightly light-rare-earth-enriched character of ocean plateau basalts is consistent with extensive melting in the spinel peridotite stability field (Mahoney et al. in press a; Saunders et al. in press), whereas intraplate ocean islands erupt basalts which were generated in the garnet stability field (EUam 1992).

Unaltered flood basalts with isotopic signatures indicating derivation from plume mantle, namely, the Ambenali Formation from Deccan, the Prince of Wales Mountains, Greenland, and the plume-related tholeiites from Eastern Madagascar, all have mantle-normalized Th/Nb, Rb/Nb and Ba/Nb ratios of less than one (Fig. 7), consistent with derivation from their associated plumes. Note, however, that many of these basalts have low to very low K/Nb ratios.While this is not a problem with the Greenland basalts

48

A.D. SAUNDERS E T A L . neath Marion Island do not have a potassium 'anomaly'), and in other samples from the Deccan (e.g. the Mahabaleshwar Formation). Note that 'normar type mid-ocean ridge basalts have K/Nb ratios of about 250 (Sun & McDonough 1989), and although some ocean islands (such as St Helena) have low K/Nb ratios (down to 140), to our knowledge no ocean island basalts have K/Nb ratios as low as 50. We cannot currently account for this anomaly; it implies either that potassium has been selectively removed during

because Icelandic basalts have a similar, low K/ Nb ratio it is difficult to account for the low K/Nb ratios in the Ambenali basalts (range 83 to 160, average 118: Lightfoot & Hawkesworth 1988) when La R6union Island basalts have K/Nb ranging from 148 to 333 (average 270) (Fisk et al. 1988). This discrepancy can be clearly seen by comparing the profiles plotted in Fig. 7; it also occurs in some basalts from eastern Madagascar, where K/Nb ratios as low as 50 are found (like La R6union, basalts from the putative plume be-

F

~;~i

~ ~ ~

LOuisville

! \ ~ i ~ ~) 0(._ ,,,~.~/.~,, J-- ~ ~ \ ~K P KerguelenIs t. 5 / , , ,

I

17

I

I

I

,

18

I

I

,

I

~ OceanPlateaus --I ~'B: NauruBasin / OJP:OntongJavaPlat.i MP:ManihikiPlateau | KP: KerguelenPlateau/

,,,

I

19

,

,

= ,.J

20

,

21

=oepb/204pb

22

10

Mahl A

I'"O Par= (Urul

-10

Nuanetsi Picrites "20

,

16

,

,

~

I

17

,

,~

,

,

Bushe I

18

,

i

=

J

I

19

. . . .

I

20

. . . .

21

pb/ Pb Fig. 6. ENd versus 2°epb/2°4pbfor (a) ocean plateaus and (b) continental flood basalts. Data sources as for Fig. 5. Note that oceanic plateaus, with the exception of some basalts from the Kerguelen Plateau, fall well within the range of ocean island basalts. Continental basalts, however, show much greater diversity, reflecting a contribution from the continental lithosphere.

PLUME-LITHOSPHERE INTERACTIONS

49

100-~

ts

¢~ 50 ~

~_~ 10-~ • • . o

E m o3

Tristan da Cunha St Helena Reunion Iceland

-No=.o

', i

i

i

i

i

~

i

i

i

i

v

i

i

i

i

!

v

f

RbBaThNbTa K LaCeSr Nd P Zr Hf Eu 13 Y YbLu 100":

Oceanic Plateau Basalts • • o .

c50o3

100-__ -

Kerguelen Plateau Ontong Java Plateau Manihiki Plateau Nauru Basin

Continental Flood Basalte (Plume Affinity?)

50--

0 E 10-

10-

o i

5-

m ,

Q.

o Madagascar: East Coast Ferrotholeiite • Greenland: Prince Wales Mountains o Deccan: Ambenali Basalts

E Or) 1

i

i

i

i

i

i

1

!

I

!

i

v

!

i

i

i

!

,

i

100~

i

i

i

I

i

,

Continental Flood Basalts 100-

,

i

,

,

,

i

~

i

,

,

Continental Flood Basalts

)

~-

,

RbBaThNbTa K LaCeSrNd P Zr Hf Eu 13 Y YbLu

RbBaThNbTa K LaCeSr Nd P Zr Hf Eu 13 Y '~Lu

)

50-

° ~

o

10-

Q-

5

E

105

o Parana Urubici Basalts o Madagascar: East Coast Tholeiite • Deccan: Mahabaleshwar Basalts

e~

E O3 1

1

i

v

!

i

]

i

i

|

~

!

i

i

!

i

~

i

• Parana: Gramado Basalts o Deccan: Bushe Basalts o Madagascar: East Coast Basalt v

RbBaThNbTa K LaCeSr Nd P ZrHf Eu13 Y YbLu

i

!

i

i

u

i

T

!

i

i

u

u

!

!

i

u

i

!

RbBaThNbTa K LaCeSr Nd P Zr Hf Eu13 Y YbLu

Fig. 7. Multi-element diagrams for basalts from ocean islands and various LIPs plotted on the basis of environment (oceanic versus continental), and the possible involvement of source materials (plume mantle, lithosphere mantle and continental crust). Note the variation of Th/Ta, Ba/Ta and K/Ta ratios (abundances of Ba and K in oceanic plateau basaits may be affected by secondary alteration). The position of K is indicated by the dashed line to highlight the low K abundances in some continental flood basalts. La R6union, Iceland and Tristan da Cunha are shown for comparison with associated flood basalt provinces (Deccan, Greenland and Parana, respectively). Datasources: Kerguelen Plateau: Storey et al. (1992); Ontong Java Plateau: Mahoney et al. (in press a); Nauru Basin: Saunders (1985); Manihiki Plateau: Saunders & Marriner (unpublished data). Deccan (averages values for Ambenali, Madabaleshwar, Bushe and Poladpur): Lightfoot & Hawkesworth (1988); Greenland (Prince of Wales Mountains): Hogg et al. (1989); Madagascar: East Coast samples MAN 90-43 (plume-related), MAN 90-8 (mantle lithosphere affinity) and MAN 90-35 (crustally contaminated) from Storey & Saunders (unpublished data); Paranfi Urubici ('high-Ti') and Gramado ('low-Ti') basalts are averages from Peate (1989); St Helena: Chaffey et al. (1989); La R6union: Fisk et aL (1988); Iceland: Wood (1978); Tristan da Cunha: Weaver et al. (1987). Normalizing values from Sun & McDonough (1989).

50

A.D. SAUNDERS E T A L . ~

CO ,.J

O-J P & N B

KP

Deccan (Ambena

Ranges of

La/Nb ratios in Ocean Plateau Basalts

Deccan (Mahabaleshwar \

Deccan (Bushe)

p

Primordial Mantle

0.1 Paran~ ( Parand (

Madagascar (East Coast)

Karoo Nuanetsi Picrites

0.1

.=

0.01

0.01

-

Basin & Range

. . . . . . L1 l

I

I I i

Ocean Island Basalts

I

I

I

1

La/Nb

I

Basalts

La/Nb

,,1,o

I' i'f'l'l''

10

Fig. 8. La/Ba versus La/Nb ratio in oceanic and continental basalts. Ocean island basalts (OIB) lie within a comparatively tight cluster, and the dispersion to higher La/Nb ratios may represent the effects of lithosphere contamination (as has been proposed for temporal variations in compositions of the Basin and Range basalts, inset, from Fitton et al. 1991). Data sources as for Fig. 7. La/Ba ratios for ocean plateau basalts are susceptible to secondary alteration, and are therefore not plotted, but data for Ontong Java and Nauru Basin indicate a ratio of between 0.2 and 0.4 (Saunders 1985; Mahoney et al. in press a). Field for ocean island basalts from Fitton et al. (199:). alteration (but inclusions in phenocrysts also have .low K contents); that the plume compositions have changed with time; that K is preferentially held back in the source during the formation of the flood basalts (unlikely); or that the ascending plume-derived melts have undergone exchange with K-depleted lithospheric rrmntie. These observations on oceanic versus continental LIPs allow us to draw an important conclusion. If oceanic plateaus are derived from plumes originating within the lower mantle, then the .much greater range of isotopic and trace element compositions seen in continental flood basalts shows that continental lithosphere must have played a role in the formation of the latter. We thus reject the observation made by Campbell & Griffiths (1990) that continental flood basalts have not undergone significant interaction with the lithosphere. Certainly, lower

mantle material appears to be required as the thermal source for LIPs, but the trace element and isotopic properties are inherited, in varying proportions, from both the plume and from the lithosphere. Why some continental flood basalt provinces should show evidence of plume involvement, while in others no plume signature can be detected by trace element or isotopic methods, is probably a function of the eruption location, the composition of the plume, and the composition of the lithosphere. The Ambenali basalts were erupted onto the edge of the Indian continent, thus erupting through strongly extended lithosphere; a similar argument could apply to the plume-like basalts from the East Coast of Madagascar. Magmatic armouring of feeder dykes may protect the ascending magmas from possible contaminants in the lithosphere (Mahoney 1988).

PLUME-LITHOSPHERE INTERACTIONS

Evidence for lithosphere involvement in the formation of LIPS On the one hand, the lithosphere acts as a thermal trap or blanket to ascending plume mantle, preventing it and hot asthenosphere from decompressing and melting. On the other hand, low melting point fractions within the lithosphere (especially, but not necessarily uniquely, the continental lithosphere) may contaminate plume-derived melts; indeed, it has been proposed that wholesale melting of the lithosphere may be responsible for flood basalt formation. It is this debate which is the topic of this and the next section. Structural and thermal aspects There is considerable debate as to what constitutes 'lithosphere', and of what the lithosphere is made. We broadly follow the nomenclature and model of McKenzie & Bickle (1988) for the definition, structure and thermal conditions of the steady-state lithosphere (Fig. 4). For the purposes of modelling (below) we assume an initial steady-state adiabatic gradient within the convecting asthenosphere, with a potential temperature (Tp) of 1280°C, and steady-state conductive geotherm within the lithosphere. These geotherms are horizontal averages; Klein & Langmuir (1987) have shown that the upper mantle beneath spreading centres is thermally heterogeneous, and seismic tomography has demonstrated the existence of broad thermal highs beneath much of the central Pacific region. Extrapolating these seismic data to actual temperature excesses is, however, difficult. For the purposes of this study, we have therefore retained the horizontally averaged value for the asthenosphere of McKenzie & Bickle (1988). The lithosphere comprises two main layers: the rigid, cool mechanical boundary layer, and the upper part of the transient thermal boundary layer. The thermal boundary layer represents a major change in mantle viscosity, flow velocity and thermal gradients. It also represents the zone where convection is replaced by mainly conductive heat transfer, and it is the portion of the lithosphere and upper asthenosphere most likely to be stripped or congealed as the temperature in the adiabatic mantle fluctuates. The mechanical boundary layer is not only insulated and thus protected from the conve/:ting upper mantle by a thermal boundary layer, but it may also be compositionally (mineralogically, chemically and isotopically) and structurally distinct from the underlying mantle, particularly beneath the continents (e.g. Jordan 1981).

51

This distinction reflects the complex history of the sub-continental lithosphere, possibly involving aggregation of residue following komatiite extraction during the Archaean and early Proterozoic (Boyd 1989), and the subsequent reenrichment of the lithosphere by ascending metasomatic melts (see below).

Compositional evidence for lithosphere involvement It is primarily the mechanical boundary layer, with its potentially long isolation time, that is invoked to account for the isotopic and chemical characteristics of some continental basalts. As with the definition of its structure, much debate also surrounds the composition of the continental mechanical (or compositional) boundary layer. Evidence comes from nodules in basalts and kimberlites, and the composition of the basalts themselves. Many continental flood basalts record significant lithospheric inputs from crustal and subcrustal sources. Thus, in addition to the plume component discussed above, two other important components can be identified. A crustal component shown by the high 87Sr/a6Sr, high 2°6pb/~°4pb and high La/Nb and Th/Nb ratios seen in basalts of the Bushe Formation from the Deccan, a group of low-Ti basalts from the East Coast of Madagascar, the Saddle Mountain basalts of the Columbia River Province, and the Gramado (Iow-Ti) basalts from the Paran,'t province (Figs 5-8) (Carlson & Hart 1988; Lightfoot & Hawkesworth 1988; Hawkesworth et al. 1988; Peate 1989; Storey et al. unpublished data). Note in particular the distinctive trace element patterns of these basalts (Fig. 7). The second component, a mantle lithosphere component, is seen in basalts of the Mahabaleshwar Formation of the Deccan, the Urubici (high titanium) basalts from the Paran~t Province, a suite of low 2°6pb/2°4pb-ratio basalts from Madagascar, and the Vandsfaldsdalen Formation of eastern Greenland (Holm 1988; Lightfoot & Hawkesworth 1988; Hawkesworth et al. 1988; Peate 1989; Storey et al. unpublished data). Isotopically, these basalts are highly variable, but are characterized by low eNd, moderate a7Sr/a6Sr, and very low to intermediate 2°6pb/ 2°4pb ratios. In terms of trace elements, they have mantle-normalized Th/Nb ratios slightly greater than 1, which distinguishes them from the plume-related group (Fig. 7), but which is much lower than the Th/Nb ratio in crustailycontaminated basalts. The involvement of a mantle lithosphere component is, perhaps, the most conjectural. Most flood basalts are tholeiites, and with elevated

52

A.D. SAUNDERS E T A L .

SiO2 contents, and it is often difficult to elim- despite the isotopic evidence suggesting that the Mahabaleshwar melts contain a substantial inate the effects of crustal contamination. For example, the combination of low 2°6pb/E°4pband lithospheric component (e.g. Fig. 6) (Lightfoot & Hawkesworth 1988). low eNd could, in theory, be inherited from anGiven that ancient, precursor melts percolatcient granulite crustal rocks with long term low U/Pb and Sm/Nd ratios. However, to signific- ing from the asthenosphere or a subduction zone antly alter the isotopic and trace element ratios are likely to be hydrous and carbonated, it is not in the parental basaltic liquid would require unreasonable to assume that small melt fracassimilation to a much greater extent than is tions, or their precipitates, reside in amphibole allowed by available models. It is much more and phlogopite (Hawkesworth et al. 1990). effective to control the ratios in the source, However, the volume of these hydrous phases in where the absolute incompatible trace element the lithospheric mantle is unknown. They may abundances are much lower, than in a basaltic be localized, either as frozen melts or as precipmelt (see, for example, Cox 1983, Hawkesworth itates and cumulates on the walls of narrow fissures, together with other exotica (apatite? et al. 1988; Gill et al. 1988; Holm 1988; Ellam & REE-, Ti- and Nb-bearing minerals such as the Cox 1989, 1991). Important evidence for the involvement of LIMA titanates?) (Menzies et al. 1987). Such mantle lithosphere comes from undersaturated material, if reactivated, could be a suitable parmelts, such as lamproites, which could not have ent for lamproitic magmatism, and may well act obtained their chemical character from assimila- as a contaminant of plume-derived magmas tion of crust (Fraser et al. 1985), and from mantle (Ellam & Cox 1991; Storey et al. 1989). Howxenoliths (e.g. Menzies & Murthy 1980). The ever, there is uncertainty as to whether these combination of low eNd, moderate 87Sr/S6Sr, hydrous minerals would lower the solidus tempmantle 3180 and low 2°6pb/2°4pb, high 2°7pb/2°6pb erature of enough of the lithosphere to form and 2°8pb/2°6pb ratios is strong evidence that the flood basalts, as suggested by Hawkesworth et melts are derived from a reservoir that has been al. (this volume). Can the lithosphere melt in isolated from the convecting mantle for a long sufficient quantity to supply the major oxides to period of time, and that this reservoir has under- form continental flood basalts? One main regone time-integrated increases in Rb/Sr, Nd/Sm, quirement needs to be satisfied; that sufficient Pb/Th and Th/U ratios and a decrease in U/Pb heat can be injected into the lithosphere, or that ratio (e.g. Hawkesworth et al. 1990). Other it can be sufficiently decompressed, to take it to characteristics of this source may include in- its solidus temperature, hydrous or anhydrous. creased La/Nb ratios, and a general increase in Conduction alone is unlikely to achieve this, incompatible element contents (e.g. Fitton et al. given the low thermal diffusivity of silicates, in a 1991; Thompson et al. 1989; Weaver & Tarney reasonable length of time. 1983) (Figs 7 & 8). These enrichments may originate as small- Sub-continental lithosphere in the oceanic scale melt fractions percolating either from the asthenosphere (Green 1971; McKenzie 1989; realm? Menzies et al. 1987), from dehydrating subduct- Removal of the mechanical boundary layer by ing slabs (Fitton etal. 1991; Saunders etal. 1991) •plumes, and its entrainment within the mantle is or from ancient plumes (Hart et al. 1986; Men- a possible consequence of plume-lithosphere zies 1990). In the case of subduction zones, we interactions, and it may be an important would expect that the fluids would have en- mechanism for returning evolved isotopic signahanced large-ion lithophile/high field strength tures to the circulating mantle system. Isotopicelement ratios (e.g. La/Nb or K/Ti), whereas in ally, Dupal-type basalts erupted on Tristan da the other two examples, the fluids would have Cunha resemble high-Ti basalts from the Paranfi originated from an OIB- or MORB-like source. of southern Brazil, but trace element data rule Thus, depending on its history and, in particular, out a direct match between the Tristan and the the nature of the last enrichment event, a given Urubici Paranfi basalts (Hawkesworth et al. volume of mechanical boundary layer is likely to 1986, this volume). Mahoney e t a l . (1989, 1991, be compositionally different from any other vol- in press b) have suggested that the anomalously ume and, with time, to develop distinct isotopic low 2°6pb/2°4pb ratios found in basalts from the characteristics. It may explain, for example, why Southwest Indian Ridge at 40°E have been inpronounced increases in La/Nb ratio are ob- herited from mobilized Madagascan lithosphere, served in Basin & Range basaIts (Fitton et al. and that a chemical 'spike' on the Central Indian 1991 ), but not in Deccan Mahabaleshwar basalts Ridge comes from the Indian lithosphere. Sim(Lightfoot & Hawkesworth 1988) (Figs 7 & 8), ilarly, Storey et al. (1992) suggest that the low

PLUME-LITHOSPHERE INTERACTIONS 2°6pb/2°4pb ratios found in basalts from Sites 738, 747 and 750 from the Kerguelen Plateau result from remobilization of the sub-Gondwana lithosphere by the Kerguelen plume before dispersal of Gondwana. In all of these examples, the lithosphere may have been thermally and tectonically stripped from the base of the respective continental fragments. Several authors have invoked recycling of sub-continental lithosphere to account for mantle components (HIMU, EM1 and EM2) found in some ocean island basalts (e.g. McKenzie & O'Nions 1983; Zindler & Hart 1986; Hart et al. 1986).

Plume--lithosphere interaction: thermal and compositional consequences of plume incubation From their study of the Hawaiian system, Watson & McKenzie (1991) have suggested that the bulk of the magma originates within the plume, rather than within the lithospheric mantle. A critical factor which emerges from their study is the thickness of the mechanical boundary layer; melting in the plume is effectively prevented if the mechanical boundary layer thickness exceeds about 100 km (Fig. 9). A further important result of their model is that the extent of melting within the mechanical boundary layer is limited to a few hundred metres, assuming thermal transfer via conduction. Sleep (1987), in his study of the Hawaiian system, also predicts that the degree of lithospheric heating and melting is limited; substantial transfer of heat into the base of the lithosphere would produce doming due to thermal expansion downstream of the actual hot mantle jet. The Hawaiian system, while providing information on the dynamic evolution of the lithosphere as it passes across a hot mantle jet, is probably atypical of plume-lithosphere interactions associated with flood basalts where, with the exception of the Deccan, the plate motions were small. Consequently, there is the potential for long-term build up of heat beneath the lithospheric cap. But can we constrain how long, and would heat transfer be sufficient to melt the mechanical boundary layer? Initially, we take a simple model of a plate moving at various rates across a stationary plume head, which is expanding from a diameter of zero to 2000 km within 20 Ma. This assumes a steady vertical plume flux rate of about 12 km 3 a -1, if the plume head is about 80 km thick. If the plate is moving sufficiently slowly, say about 1 cm a -1, then a point at the base of the lithosphere will be in contact with hot plume mantle for periods well in excess of 50 Ma. Indeed, with

53

Temperature (=C) 1ooo 1500 I i , I , , / , I , i , ::'" " M ~ ' h a ~ , ~ a i " ~:.?~,::::f:::::::::::::::::li:::::::::::::::l

0

:" Boundary Layer 3:,:,,~:,l,:,:,~,:,:,;,:,:-[,:,:,:,:,:-:-:,l

===============================================================

~i," E ~,,100 r~. a

180

%

?~ Ma,,,., .,.,., il,,,,,,O~.,.,, ,,1,,,,-, ,,-.-.,.I

--.

I!i

....

.....

....-.

i,:.:. I

t

i'll, i ,ii i iii!iiiliiii!!i'

1.2 1.0

.~ 0,8 0.6 ~, 0.4

~ 0.2 n" 0.0 o

3o

r~

~

t2o

Mechanical Boundary LayerThickness(kin)

Fig. 9. Temperature-depth diagram illustrating the changes in the lithosphere geotherm following the development of a plume with a potential temperature of 1580°C (i.e. ATp = 300°C). Initial parameters, including L(i) (= initial lithosphere thickness) from Fig, 4. In the absence of hydrous components, melting occurs at M. Times refer to period after heat transfer into the mechanical boundary layer begins; at 0 Ma it is assumed that the thermal boundary layer has been almost removed. Convective heat transfer rates to the plume-mechanical boundary layer boundary are assumed to be infinite relative to the conductive heat transfer in the lithosphere, and the thickness of the mechanical boundary layer is assumed to be infinite. Thermal diffusivity = 10-6 m2 s "t. Lower graph: melt production versus mechanical boundary layer thickness, from Watson & McKenzie (1991).

a plume radius expansion rate of 2 cm a -l the plate will be in contract with the plume mantle indefinitely, although clearly the thermal influence will decrease downstream as the heat is transferred into the lithospheric plate. This is distinct from the Hawaiian situation where the plate is moving across the plume at a rate of 8 cm a- 1 and rapidly moves downstream of the plume head.

54

A.D. SAUNDERS E T A L .

A full treatment of the heat transfer mechanism from the plume head into the lithosphere is beyond the scope of this paper but is presently being evaluated at Leicester utilizing finite-element modelling programs. The problem can be divided into several sections as follows.

Conduction

Although silicates have low thermal conductivities and diffusivities (Clark 1966), given the length of time that a slowly-moving plate may be in contact with a hot plume head, substantial heat transfer into the base of the mechanical boundary layer may occur (Fig. 9). Precise modelling of the thermal profile between the plume head and the lithosphere is very difficult, not least because of uncertainties in the viscosity and convective patterns of the uppermost plume mantle (compare, for example, Spohn & Schubert 1982; Yuen & Feitout 1985; Griffiths & Campbell 1991; Leitch et al. 1991; Watson & McKenzie 1991). However, we agree with McKenzie (1989) and Watson & McKenzie (1991) that this mechanism alone is unlikely to melt large volumes of lithospheric mantle, although it may mobilize pre-existing low-melting point hydrous and carbonated material within the lithosphere. Given suitable thermal and structural conditions these melts may reach the surface but initially, at least, they are likely to ascend and freeze as they lose heat to their surroundings. Conduction is likely to raise the isothermal surfaces in the lowest parts of the lithosphere only. Conduction-induced melting within the lithosphere serves to redistribute the heat upwards but will not introduce significant new heat. Furthermore, because the melt volumes are likely to be very small, and at a temperature only slightly above the hydrated and/or carbonated solidus, they will have small heatcarrying capacity, although they may serve to reduce the mantle solidus temperature (McKenzie 1985, 1989). Convection

Convection is the most effective, and dominant, mechanism of heat transfer within the asthenosphere. Convective processes are responsible for transferring heat to the thermal boundary layer, where conduction takes over. The efficiency of this convective transfer depends on the viscosity of the uppermost asthenosphere, which in turn depends on the temperature and the presence or absence of a melt fraction, as mentioned above. If the convection is sluggish, the rate of

heat supply to the base of the lithosphere will be low. Lowering the viscosity, either by increasing the temperature, or by having a melt fraction present (two-phase flow), will dramatically increase the convection rate. Indeed, two-phase flow within the mantle is an efficient heat transfer mechanism, initiating heat transfer to the surface via magmatism. In the high temperature conditions of a plume head the mantle will intersect its dry solidus temperature, when it has been allowed to decompress sufficiently. According to the calculations of Watson & McKenzie (1991), the mechanical boundary layer needs to be less than about 100 km thick before significant melting of a plume of Tp = 1525°C can occur. However, if the plume is hotter, or has a lower solidus temperature, this value is an underestimate. Indeed, the plumes associated with flood basalts may well be the most vigorous; many LIPs have erupted highMg lavas (Storey et al. 1991) and not all of these can be mantle remelts (Elliott et al. 1991). Interestingly, the observation that many picrites are erupted through continental lithosphere indicates that either there has been extensive melting within the plume head (Gill et al. 1988), despite a lithospheric cap or, if the picrites formed within the lithosphere, that substantial amounts of heat must have been transferred into the lithosphere. Melts generated within the plume head will separate from the matrix and migrate upwards. It is not clear what will happen as they attempt to traverse the lithosphere; it depends on their composition and volume. However, the following scenario may not be unrealistic. Initial small-degree melt fractions formed close to the hydrous solidus will freeze as they enter the cooler mechanical boundary layer. They will presumably mix with, and possibly mobilize, any ancient hydrous and carbonated fractions already present in the lithosphere, but vertical migration of this assemblage will be limited; at this stage, there is unlikely to be sufficient thermal energy to drive the liquids to the surface, unless the lithosphere fractures to allow rapid melt expulsion along deep-rooted fissures. Such melts may well be emplaced as highly enriched, hydrous and carbonated lamproites and Type II kimberlites. Successive melts expelled from the plume head will become more voluminous when the plume head moves upwards into the space vacated by the thermally-softened and thermally-eroded lithosphere. Once the anhydrous solidus has been intersected, these melts will not necessarily become hotter unless phase exhaustion (garnet or clinopyroxene) occurs. Em-

PLUME-LITHOSPHERE INTERACTIONS placement of these melts into the lithosphere will slowly raise the geotherm, and force carbonated and hydrous melt fractions and isothermal surfaces to move upwards. At this stage, melt production is still at relatively great depth, within the garnet stability field, and is likely to be alkaline and heavy-REE-depleted. At the same time, the overlying lithosphere could, if the plume is sufficiently large, be uplifting and doming and undergoing localized extension. This extension could, presumably, be assisted by the rheological changes occurring in the lithosphere, as the strength--depth profile is changed due to increasing temperature. Eventually, through the agencies of thermal erosion and lithosphere extension, the ascending mantle will finally reach the level (35-50 km depth) where melting is in the spinel-stability field, and the liquids generated will be tholeiitic in character. The source of these liquids will be within the hot plume head and in the decompressing and heated lithosphere, where melting may be aided by the presence of hydrous phases. These tholeiites will erupt to form a large igneous province; some may be trapped in large magma chambers on the Moho (Cox 1980). The evolution of the system depends on the activity of the plume, and the extensional stresses within the lithosphere. If the plume remains vigorously active, penetration of the lithosphere by successive melts and conductively- and convectively-transferred heat will cause further thermal erosion of the lithosphere (Withjack 1979); the system could then develop positive feedback. Furthermore, the stress profile and the viscous strength of the lithosphere will change as the thermal gradient is increased (e.g. Kusznir & Park 1987). In particular, the ability of the lower lithosphere, especially the olivinedominated mantle portion, to sustain stress will be substantially reduced. Combined with the formation of a dome above the dynamic plume column, and vertical and lateral thermal expansion of the heated lithosphere, this stress reduction could lead to gravitational collapse and thinning above the plume. Although this may not necessarily initiate or drive plate fragmentation (Hill 1991), such weakening of the lithosphere may focus or channel extensional stresses through a plate which is being extended by external forces. If the plate is unable to fragment, for example if it is surrounded by divergent or strongly compressional margins, internal plate deformation around the thermally-affected area may occur (for example, the TransbaikaI-Mongolia area which is a region of substantial uplift and neotectonics associated with MioccneRecent volcanism) (Khutorskoy & Yarmoluk 1989).

55

An extreme situation may be where a near-stationary supercontinent (Gondwana) is underpinned by several hot plumes. Thus, supercontinents may trap and enable coalescence of plumes that would otherwise be able to reach the surface. Extension of the lithosphere, allowing the hot plume mantle and hot lithosphere to decompress, is probably the important trigger in sudden melt generation, as proposed by White & McKenzie (1989). However, complete separation is not required, for example the Columbia River Basalts, Siberian Traps, and as evidenced by the occurrence of most continental flood basalts on the flanks of the oceans, rather than in the basins themselves. The important point made here is that melt emplacement into the mechanical boundary layer serves to raise the thermal energy of that system, more so than simple conduction, so that when flood basalts do form via decompressive melting, lithospheric material may be incorporated in the final melts. Furthermore, progressive heating of the lithosphere makes it more likely to fail above the hotspots when regional extension ultimately occurs, and it enables melts to traverse the lithosphere without freezing. Finally, thermal enhancement of the mechanical boundary layer will assist in its stripping and delamination, and eventual incorporation of this mantle in the asthenosphere or plume head. Some of this material may be incorporated in the flood basalts, and some may be entrained in the mantle and be sampled within the oceanic realm (Hawkesworth et al. 1986; Shirey et al. 1987; Mahoney et al. 1989, 1991, in press b; Storey et al. 1989, 1992). Concluding statements There is little doubt that plumes are responsible for supplying the thermal energy required for most LIP formation. In some cases it is not possible to trace the plume responsible for the magmatism, and in the case of the Ferrar Province it is difficult to envisage the form of the plume responsible for such a linear basaltic province. An alternative to the predominating involvement of a plume is to suggest that the solidus of the lithosphere is reduced by the breakdown of amphibole or phlogopite (Hawkesworth et al. this volume). The main problem here is that (a) melts from such mantle will by hydrous, and would be expected to crystallize hydrous phases; these are not, generally, seen in continental flood basaits or their hypabyssai equivalents; and (b) the decompression required would also lead to rapid decompression and melting of the underlying asthenosphere. Again, though, a thermal anomaly is required as a heat source; otherwise,

56

A.D. SAUNDERS ETAL.

we would expect to find LIPs developed along all continental margins. In this contribution we have attempted to demonstrate that the large isotopic and trace element differences between continental and ocean LIP basalts are due to incorporation of lithospheric material in the former, and that there is no need to invoke lower mantle to explain the compositions of continental flood basalts as suggested by Campbell & Griffiths (1990). We do not preclude involvement of lower mantle material in LIP formation, as it is required for the large material and thermal input, but its composition is more closely represented by some oceanic plateaus and aseismic ridges. Incorporation of lithospheric mantle in continental magmas is probably achieved by a combination of assimilation of low-temperature assemblages inherited from previous enrichment events and incorporation of mechanical boundary layer mantle in the decompressing mantle r6gime as extension occurs. The thermal input required to change the temperature and rheology of the mechanical boundary layer is unlikely to be achieved by conduction alone; convection involving injection of melts appears to be required. Significant modification of the thermal structure of the lithosphere is less likely with the plume impact model. The proposed short duration between impact and melt generation (< 10 Ma) rules out substantial heat transfer, at least via conduction. On the other hand, plume incubation, which is virtually a necessity in any Earth model which invokes plumes and thick lithosphere, can build the thermal structure required for flood basalt genesis. It accords better with the linkage between flood basalts and supercontinent breakup; and the progressive tapping of yo0nger flood basalt provinces as the Atlantic unzipped northwards (how did start-up plumes leaving the core-mantle boundary know that the Atlantic was opening northwards?). It less satisfactorily explains the rapid outburst of the Ontong Java Plateau, where a large thermal anomaly does appear to have been suddenly exploited. Clearly, deciding between the impact and incubation models is important for understanding the relationships between plumes and lithosphere, and not least for the forces acting on plates prior to, during, and after continent breakup. Long-term incubation beneath a nearstationary supercontinent (e.g. Kent 1991) could lead to the development of large, coalescing thermal anomalies which thermally erode the base of the lithosphere and affect its rheology by successive melt injection. In the extreme case

the lithosphere may be substantially thinned, as possibly happened to the Indian sub-continent, where the lithosphere may be only 55 km thick (Negy et al. 1986), owing to its successive exposure to the Kerguelen (c. 120 Ma), Marion (c. 90 Ma) and R6union (c. 65 Ma) plumes. Future studies could concentrate on determining age data for a wide range (geographically and temporally) of rock types from a variety of LIPs. This is vital if we are to determine the thermal history of a magmatic province; in particular, the occurrence of alkaline magmas before an episode of flood basalt activity may indicate precursor thermal effects of an incubating plume (Kent et al. in press). Extrapolating these data to the thermal history of the underlying plume requires juxtaposition of radiometric, structural and sedimentological data to monitor the contemporaneous evolution of the lithosphere. Geochemical data for erupted and intruded magmas still offer a powerful opportunity to monitor the relative contributions from the crust, lithospheric mantle and plume mantle. We thank N. Rogers (Open University) and D. McKenzie (Cambridge) for their helpful reviews, and J. Tarney for useful discussions. D. Peate (Durham) kindly allowed us to refer to data in his PhD thesis. M.S. and R.W.K. are supported by NERC (GR3F/484 and GT4/89/GS/055, respectively). References

ALmERT, C. 1992. Mineralogy and geochemistry of a basalt from Leg 119, Site 738: implications for the tectonic history of the southern part of the Kerguelen Plateau. In: BARRON,J. & LARSON,B: et al. (eds) Proceedings of the Ocean Drilling Program, Scientific Results 119. Ocean Drilling Program, College Station, Texas, 293-298. BOVD,F. R. 1989. Compositional distinction between oceanic and cratonic lithosphere. Earth and Planetary Science Letters, 96, 15-26. BURKE, K. C. & DEWEY,J. F. 1973. Plume generated triple junctions. Key indicators in applying plate tectonics to old rocks. Journal of Geology, 81, 403-433. CAMPBELL, I. n . & GRIFFITHS, R. W. 1990. Implications of mantle plume structure for the evolution of flood basalts. Earth and Planetary Science Letters, 99, 79-93. CARLSON,R. W. & HART, W. K. 1988. Flood basalt volcanism in the northwestern United States. In: McDOUGALL, J. D. (ed.) Continental Flood Basahs. Kluwer Academic Publishers, Dordrecht, Netherlands, 35-62. CHAFFEY, D. J., CLIFF,R. A. & WILSON,B. M. 1989. Characterization of the St Helena magma source. ln: SAUNDERS,A. O. & NORRY,M. J. (eds) Magmatism in the Ocean Basins. Geological Society, London, Special Publication, 42, 257-276.

PLUME-LITHOSPHERE INTERACTIONS CLARK, S. P. 1966. Thermal Conductivity. In: CLARKE, S. P. (ed.) Handbook of Physical Constants. Geological Society of America, Menlo Park, California, 459-482. COFFIN, M. F. & ELDHOLM,O. 1991. Large Igneous

Provinces: JOI/USSAC Workshop Report. The University of Texas at Austin Institute for Geophysics Technical Report No. 114. COURTILLOT, V., BESSE,J., VANDDAMME,D., MONTINGNY, R., JAEGGER, J.-J. 8/. CAPETTA, H. 1986. Deccan flood basalts at the Cretaceous/Tertiary boundary? Earth and Planetary Science Letters, 80, 361-374. COURTNEY, R. d~ WHITE, R. 1986. Anomalous heat flow and geoid across the Cape Verde Rise: evidence for dynamic support from a thermal plume in the mantle. Geophysical Journal Royal Astronomical Society, 87, 815-867. Cox, K. G. 1980. A model for flood basalt volcanism. Journal of Petrology, 21,629-650. 1983. The Karoo Province of Southern Africa: origin of trace element enrichment patterns. In: HAWKESWORTH, C. J. & NORRY, M. J. (eds) Continental Basalts and Mantle Xenoliths. Shiva, Nantwich, Cheshire, UK, 139-157. 1989. The role of mantle plumes in the development of continental drainage patterns. Nature, 342, 873-876. DAVIES, G. F. 1988. Ocean bathymetry and mantle convection 1. large-scale flow and hotspots. Journal of Geophysical Research, 93, 467-480. DUNCAN, R. A. 1978. Geochronology of basalts from the Ninetyeast Ridge and continental dispersion in the eastern Indian Ocean. Journal of Volcanology and Geothermal Research, 4, 283-305. 1981. Hotspots in the southern oceans--an absolute frame of reference for motion of the Gondwana continents. Tectonophysics, 74, 29-42. d~ HARGRAVES,R. B. 1984. Plate tectonic evolution of the Caribbean region in the mantle reference frame. In: BONINi, W. E., HARGRAVES, R. B. & SHAGAN,R. (eds) The Caribbean-South

American Plate Boundary and Regional Tectonics. Geological Society of America Memoir, 162, 81-93. & RICHARDS, M. A. 1991. Hotspots, mantle plumes, flood basaits and true polar wander. Review of Geophysics, 29, 31-50. ELLAM, R. M. 1992. Lithospheric thickness as a control on basalt geochemistry. Geology, 20, 153-156. -& Cox, R. G. 1989. A Proterozoic lithospheric source for Karoo magmatism: evidence from the Nuanetsi picrites. Earth and Planetary Science Letters, 92,207-218. & - 1991. An interpretation of Karoo picrite basalts in terms of interaction between asthospheric magmas and the mantle lithosphere. Earth and Planetary Science Letters, 105, 330-342. ELLIOTT, T. R., HAWKESWORTH,C. J. & GRONVOLD, K. 1991. Dynamic melting of the Iceland plume. Nature, 351,201-206. FISK, M. R., UPTON, B. G. J., FORD, C. E. & WHITE,

57

W. M. 1988. Geochemical and experimental study of the genesis of magmas of Reunion Island, Indian Ocean. Journal of Geophysical Research, 93, 4933-4950. FITrON, J. G., JAMES,D. & LEEMAN,W. P. 1991. Basic magmatism associated with the late Cenozoic extension in the western United States: compositional variations in space and time. Journal of Geophysical Research, 96, 13693-13711. FRASER, K. J., HAWKESWORTH,C. J., ERLANK,A. J., MITCHELL, R. H. & SCOTT-SMITH, B. H. 1985. Sr, Nd and Pb isotope and minor element geochemistry of lamproites and kimberlites. Earth and Planetary Science Letters, 76, 57-70. FUROMOTO, A. S., WEBB, J. P., ODEGAARD,M. E. & HUSSONG,D. M. 1970. Seismic studies on the Ontong Java Plateau. Tectonophysics, 34, 71-90. GILL, R. C. O., NIELSEN,T. F. D., BROOKS,C. K. & INGRAM, G. A. 1988. Tertiary volcanism in the Kangerdlugssuaq region, E Greenland: traceelement geochemistry of the Lower Basaits and tholeiitic dyke swarms. In: MORTON, A. C. & PARSON, L. M. (eds) Early Tertiary Volcanism and the Opening of the NE Atlantic. Geological Society, London, Special Publication, 39, 161-179. GREEN, D. H. 1971. Composition of basaltic magmas as indicators of conditions of origin: application to oceanic volcanism. Philiosophical Transactions of the Royal Society of London, A268, 707-725. GRIFFrrHs, R. W. 1991. Interaction of mantle plume heads with Earth's surface and onset of smallscale convection. Journal of Geophysical Research, 96, 18295-18310. & CAMPBELL,I. H. 1990. Stirring and structure in mantle starting plumes. Earth and Planetary Science Letters, 99, 66-78. HALLIDAY, A. N., DICKIN, A. P., FALLICK, A. E. & FITTON, J. G. 1988. Mantle dynamics: a Nd, Sr, Pb and O isotopic study of the Cameroon Line volcanic chain. Journal of Petrology, 29, 181-211. HART, S. R., GERLACH,D. C. & WHITE, W. M. 1986. A possible new Sr-Nd-Pb mantle array and consequences for mantle mixing. Geochimica et Cosmochimica Acta, 50, 1551-1557. HAWKESWORTH, C. J., GALLAGHER, K., KELLEY,S., MANTOVANI, M., PEATE, D. W., REGELOUS, M. & ROGERS, N. W. 1992. Paranfi magmatism and the opening of the South Atlantic. This Volume. ~, KEMPTON,P. D., ROGERS,N. W., ELLAM,R. W. VAN CALSTEREN, P. W. 1990. Continental mantle lithosphere, and shallow level enrichment processes in the Earth's mantle. Earth and Planetary Science Letters, 96, 256-268. ~ . , MANTOVANI,M. ~: PEATE,D. 1988. Lithosphere remobilization during Paranfi CFB magmatism. In: MENZIES,M. A. & Cox, K. G. (eds) Oceanic

and Continental Lithosphere: Similarities and Differences. Journal of Petrology Special Litho~,

sphere Issue, Oxford University Press, 205-223. MANTOVANi, M. S. M., TAYLOR, P. N. & PALACZ, Z. 1986. Evidence from the Paranfi of south Brazil for a continental contribution to

58

A.D. SAUNDERS E T A L .

, MACDOUGALL, J. D., LUGMAIR, G. W., MURDupal basalts. Nature, 322, 356-359. HEROT, J. M. & PEATE,D. W. & HAWKF_SWOR'm,C. J. ALl, A. V., SANKER DAS, M. & GOPALAN, K. 1982. Origin of the Deccan Trap flows at 1991. The petrogenesis of Mesozoic Gondwana Mahabaleshwar inferred from Nd and Sr isotopic low-Ti flood basalts. Earth and Planetary Science and chemical evidence. Earth and Planetary SciLetters, 10S, 134-148. HmL, R. I. 1991. Starting plumes and continental ence Letters, 60, 47-60. , NATLAND,J. H., WHITE, W. M., POREDA, R., break-up. Earth and Planetary Science Letters, BLOOMER,S. H., FISHER, R. L. & BAXTER,A. N. 104, 398-416. HOGG, A. J., FAWCETT,J. J., GrI'rlNS, J. & GORTON, 1989. Isotopic and geochemical provinces of the M. P. 1989. Cyclical variation in composition in western Indian Ocean spreading centres. Journal continental tholeiites of East Greenland. Canad/an of Geophysical Research, 94, 4033-4052. 1ournal of Earth Science, 26, 534-543. ~, NICOLLET, C. & DuPuY, C. 1991. MadaHOLM, P. M. 1988. Nd, Sr and Pb isotope geochemgascar basalts: tracking oceanic and continental istry of the Lower Lavas, E Greenland Tertiary sources. Earth and Planetary Science Letters, 104, Igneous Province. In: MORTON,A. C. & PARSON, 350-363. L. M. (eds). Early Tertiary Volcanism and the & SPENCER,K. J. 1991. Isotopic evidence for the Opening of the NE Atlantic. Geological Society, origin of the Manihiki and Ontong Java oceanic London, Special Publication, 39, 181-195. plateaus. Earth and Planetary Science Letters, HOOPER, P. R. 1988. The Columbia River Basalts. 104, 196-210. In: MACDOUOALL,J. D. (ed.) Continental Flood STOREY,M., DUNCAN,R. A., SPENCER,K. J. & Basalt. Kluwer Academic Publishers, Dordrecht, I~INGLE, M. In press (a). Geochemistry and geoNetherlands, 1-34. chronology of Leg 130 basement lavas: insights JORDON, T. H. 1981. Continents as a chemical boundinto the nature of the Ontong Java Plateau. In: ary layer. Philosophical Transactions of the Royal KROENKE, L., BERGER, W., JANACEK,T. et al. Society of London, A301, 359-373. (eds). Proceedings of the Ocean Drilling ProKENT, R. W. 1991. Lithospheric uplift in eastern gram, Scientific Results 130. Ocean Drilling Gondwana: evidence for a long-lived mantle Program, College Station, Texas. plume system? Geology, 19, 19-23. McDOUGALL, I. & DUNCAN,R. A. 1980. Linear vol~, STOREY,M. & SAUNDERS,A. D. In press. Large canic chains---recording plate motions. Tecigneous provinces: sites of plume impact or tonophysics, 63, 275-295. plume incubation? Geology. MCKENZIE, D. 1985. The extraction of magma from KHUTORSKOY,M. D. & YARMOLUK,V. V. 1989. Heat the crust and mantle. Earth and Planetary Science flow, structure and evolution of the lithosphere of Letters, 74, 81-91. Mongolia. Tectonophysics, 164, 315-322. 1989. Some remarks on the movement of small KLEIN, E. M. & LANGMUIR,C. H. 1987. Global cormelt fractions in the mantle. Earth and Planetary relation of ocean ridge basalt chemistry with axial Science Letters, 95, 53-72. depth and crustal thickness. Journal of Geophys& BICKLE,M. J. 1988. The volume and composiical Research, 92, 8089-8115. tion of melt generated by extension of the lithoKUSZNIR, N. J. & PARK, R. G. 1987. The extensional sphere. Journal of Petrology, 29, 625-679. strength of the continental lithosphere: its de& O'NIONS, R. K. 1983. Mantle reservoirs and pendence on geothermal gradient, and crustal ocean island basalts.Nature, 301,229-231. composition and thickness. In: COWARD,M. P., MENZIES, M. A. 1990. Petrology and geochemistry of DEWEY,J. F. & HANCOCK,P. L. (eds) Continealal the continental mantle: an historical perspective. Extensional Tectonics. Geological Society, LonIn: MENZIES, M. A. (ed.) Continental Mantle. don, Special Publication, 39, 35-52. Oxford Science Publications, Oxford, 31-54. LEITCH, A. M., YUEN, D. A. & SEWELL, G. 1991. & MURTHY, V. R. 1980. Nd and Sr isotope Mantle convection with internal heating and presgeochemistry of hydrous mantle nodules and sure-dependent thermal expansivity. Earth and their host alkali basalts, implications for local Planetary Science Letters, 102, 213-232. heterogeneities in metasomatically veined LIGHTFOOT, P. & HAWKESWORTH,C. 1988. Origin of mantle. Earth and Planetary Science Letters, 46, Deccan Trap lavas: evidence from combined 323-334. trace element and Sr-, Nd- and Pb-isotope ~, ROGERS,N., TINDLE,A. & HAWKESWORTH,C. J. studies. Earth and Planetary Science Letters, 91, 1987. Metasomatic and enrichment processes in 89-104. the lithospheric peridotites, an effect of asthenoMAHONEY, J. J. 1988. Deccan Traps. In: MACsphere-lithosphere interaction. In: HAWKESDOUGALL,J. D. (ed.) Continental Flood Basalts. WORTH,C. J. & MENZmS,M. A. (eds) Mantle MetaKluwer Academic Publishers, Dordrecht, somatism. Academic Press, London, 313-361. Netherlands, 151-194. MORGAN, W. J. 1971. Convection plumes in the lower , LE ROEX, A. P., PENG, Z., FISHER,R. L. & NATmantle. Nature, 230, 42-43. LAND,J. H. In press (b). Western limits of Indian 1972. Plate Motions and Deep Mantle ConvecMORB mantle and the origin of low Z°rPb/2°4Pb tion. Geological Society of America Memoir, MORB: isotope systematics of the central South132, 7-22. west Indian Ridge (17-500E). Journal of 1978. Rodriguez, Darwin, Amsterdam . . . . . a Geophysical Research. second type of hotspot Island. Journal of Geo,

PLUME-LITHOSPHERE INTERACTIONS

physical Research, 83, 5355-5360. - - 1981. Hotspo~ Tracks and the Opening of the Atlantic and Indian Oceans. Wiley, New York, 443-487. 1982. Hotspot tracks and the early rifting of the Atlantic. Tectonophysics, 94, 123-139. NEGY, J. G., PANDEY,O. P. & AGRAWAL,P. K. 1986. Super-mobility of hot Indian lithosphere. Tectonophysics, 131,147-156. OLSON, P., SCHUBERT, G., ANDERSON, C. & GOLDMAN, P. 1988. Plume formation and lithosphere erosion: a comparison of laboratory and numerical experiments. Journal of Geophysical Research, 93, 15065-15084. PLATE, D. W. 1989. Stratigraphy and petrogenesis of the Parand continental flood basalts, southern Brazil. Unpublished PhD thesis, Open University. RENNE, P. R. & BASU, A. R. 1991. Rapid eruption of the Siberian Traps flood basalts at the PermoTriassic boundary. Science, 253, 176-179. RICHARDS,M. A., DUNCAN,R. A. & COURTILLOT,V. E. 1989. Flood basalts and hot-spot tracks: plume heads and tails. Science, 246, 103-107. SAUNDERS,A. D. 1985. Geochemistry of basalts from the Nauru Basin, Deep Sea Drilling Project Legs 61 and 89: implications for the origin of oceanic flood basalts. In: SCHLANGER,S. O., MOBERLY, R. et al. (eds) Initial Reports of the Deep Sea Drilling Project 89. US Government Printing Office, Washington DC, 499-517. , NORRY,M. J. & TARNEY,J. 1991. Fluid influence on the trace element compositions of subduction zone magmas. Philosophical Transactions of the Royal Society of London, A335, 377-392. , STOREY, M., KENT, R. W. & GIBSON, I. L. In press. Magmatic activity associated with the Kerguelen-Heard plume: implications for plume dynamics. Bulletin G(,ologique Societie du France. SCHILLING,J.-G. 1991. Fluxes and excess temperatures of mantle plumes inferred from their interaction with migrating mid-ocean ridges. Nature, 352, 397-403. SHIREY,S. B., BENDER,J. F. & LANGMU1R,C. H. 1987. Three-component isotopic heterogeneity near the Oceanographer transform, Mid-Atlantic Ridge. Nature, 325,217-223. SLEEP, N. H. 1987. Lithospheric heating by mantle plumes. Geophysical Journal Royal Astronomical Society, 91, 1-11. 1990. Hotspots and mantle plumes: some phenomenology. Journal of Geophysical Research, 95, 6715-6736. SPOHN,T. & SCHUBERT,G. 1982. Convective thinning of the lithosphere: a mechanism for the initiation of continental rifting. Journal of Geophysical Research, 87, 4669-4681. ~gSCHUBERT,G. 1983. Convective thinning of the lithosphere: a mechanism for rifting and midplate volcanism on Earth, Venus and Mars. Tectonophysics, 94, 67-90. STOREY, M., MAHONEY, J. J., KROENKE, L. W. & SAUNDERS, A. D. 1991. Are oceanic plateaus sites of komatiite formation? Geology, 19,

59

376-379. KENT, R., SAUNDERS,A. D., SALTERS,V. J., HERGT, J., WHITECHURCH,H., SEVIGNY,J. H., THIRLWALL,M. F., LEAT, P., GHOSE, N. C. & GIFFORD, M. 1992. Lower Cretaceous volcanic rocks on continental margins and their relationships to the Kerguelen Plateau. In: WISE, S. W. & SCHLICH,R. et al. (eds) Proceedings o/the Ocean Drilling Program, Scientific Results 120. Ocean Drilling Program, College Station, Texas, 000-000. , SAUNDERS,A. D., TARNEY, J., GIBSON, I. L., NORRY, M. J., THIRWALL, M. F., LEAT, P., THOMPSON,R. N. & MENZIES,M. A. 1989. Contamination of Indian Ocean asthenosphere by the Kerguelen-Heard mantle plume. Nature, 338, 574-576. ~, , , LEAT, P., THIRWALL, M. F., THOMPSON,R. N., MENZIES,M. A. & MARRINER, G. F. 1988. Geochemical evidence for plumemantle interactions beneath Kerguelen-Heard Islands, Indian Ocean. Nature, 336, 371-374. SUN, S.-S. & MCDONOUGH,W. F. 1989. Chemical and isotope systematics of oceanic basalts: implications for mantle composition and processes. In: SAUNDERS, A. D. & NORRY, M. J. (eds) Magrealism in the Ocean Basins. Geological Society, London, Special Publication, 42, 313-345. TARDUNO, J. A., SLITER, W. V., KROENKE, L., LECtIE, M., MAYER, H., MAHONEY, J. J., MUSGRAVE,R., STOREY,M. ~gWINTERER,E. L. 1991. Rapid formation of the Ontong Java Plateau by Aptian mantle plume volcanism. Science, 254, 399-403. THOMPSON, R. N. & GIBSON, S. A. 1991. Subcontinental mantle plumes, hotspots and pre-existing thinspots. Journal o/the Geological Society, London, 147, 973-977. ~, LEAT, P. T., DICKIN, A. P., MORRISON,M. A., HENDRY, G. L. & GIBSON, S. A. 1989. Strongly potassic mafic magmas from iithospheric mantle sources during continental extension and heating: evidence from Miocene minettes of northwest Colorado. Earth and Planetary Science Letters, 98, 139-153. WATSON, S. & MCKENZIE, D. 1991. Melt generation by plumes: a study of Hawaiian volcanism. Journal o/Petrology, 32, 501-537. WEAVER, B. L. & TARNEY, J. 1983. Chemistry of the subcontinental mantle: inferences from Archaean and Proterozoic dykes and continental flood basalts. In: HAWKESWORTH,C. J. & NORRY, M. J. (eds) Continental Basalts and Mantle Xenoliths. Shiva, Nantwich, Cheshire, UK, 158-185. • , WOOD, D. A. & TARNEY,J. 1987. Geochemistry of ocean island basalts from the South Atlantic: Ascension, Bouvet, St. Helena, Gough and Tristan da Cunha. In: FIx-rON, J. H. & UPTON, B. G. J. (eds) Alkaline Igneous Rocks. Geological Society, London, Special Publication, 30, 253-267. WHITE, R. S. & MCKENZlE,D. P. 1989. Magmatism at rift zones: the generation of volcanic continental

--,

60

A. D. SAUNDERS ETAL. margins and flood basalts. Journal of Geophysical Research, 94, 7685-7730.

WrrHJACK, M. 1979. A convective heat transfer model for lithospheric thinning and crustal uplift. Journal of Geophysical Research, 84, 3008-3022. WOOD, D. A. 1978. Major and trace element variation in Tertiary iavas of Eastern Iceland and their significance with respect to the Iceland geochemical

anomaly. Journal of Petrology, 19, 393-436. YUEN, D. A. & FLEITOUT, L. 1985. Thinning of the lithosphere by small-scale convective destabilization. Nature, 313, 125-128. ZI~DLER, A. & HART, S. R. 1986. Chemical Geodynamics. Annual Reviews in Earth and Planetary Sciences, 14, 493-571.

Asthenosphere-derived magmatism in the Rio Grande rift, western USA: implications for continental break-up S. A . G I B S O N 1, R . N. T H O M P S O N 1, P. T. L E A T 1'5, A . P. D I C K I N 2, M. A . M O R R I S O N 3, G . L. H E N D R Y 3 & J. G . M I T C H E L L 4

1Department of Geological Sciences, University of Durham, South Road, Durham DH1 3LE, UK 2Department of Geology, McMaster University, 1280 Main Street West, Hamilton, Ontario L8S 4M1, Canada 3School of Earth Sciences, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK 4Department of Physics, The University, Newcastle upon Tyne NE1 7RU, UK 5Present address: British Antarctic Survey, High Cross, Madingley Road, Cambridge CB3 OET, UK Abstract: Magmas that are generated at continental rift zones provide an insight into the processes operating during the early stages of continental break-up. Our detailed study of mafic volcanism along the axis of the Rio Grande rift shows that, throughout both phases (30-17 and < 13 Ma) of its evolution, magmas with compositions interpreted as melts from the asthenospheric mantle have reached the surface. This recognition of early phase (26 Ma) magmas with incompatible trace element concentrations and radiogenic isotope ratios resembling those normally associated with ocean-island basalts and small seamounts (OIB) is significant because: (1) magmas dominated by the composition of asthenosphere-derived melts are not usually thought to be characteristic of early-phase continental rifting; (2) Tertiary mafic magmatism of an age greater than late Miocene in Colorado and New Mexico was hitherto regarded as subduction-related. Previous studies have shown that the final erupted composition of asthenosphere-derived melts is determined by the potential temperature of the convecting mantle, the amount and rate of lithosphere extension, fractional crystallization and crustal contamination. However, in the Rio Grande rift and elsewhere, such as the Basin and Range province, Eifel, NW Sardinia and the Cameroon Line, the final composition of these melts is also significantly influenced by earlier magmatic episodes. During the initial stages of asthenosphere melt generation the magma batches that first penetrate may heat a previously undisturbed segment of lithosphere and mix with strongly potassic, low temperature melt fractions. When these segments have been subsequently temporan.'ly purged of such fusible potassic fractions the asthenosphere-derived melts can rise unimpeded through the sub-continental lithosphere.

When continents break-up, the constructional plate margin between the two continental fragments must pass through an intracontinental rift stage. It follows that continental rifts can provide information on the early stages of continental separation. Importantly, continental rifts provide continuous exposure across the rift axes whereas rocks at passive margins around ocean basins, that formerly occupied rift axes, are buried beneath sediment and water. In this paper we describe mafic igneous rocks exposed in the axis of one of the world's classic continental rift systems, the Rio Grande rift, western USA, and examine the temporal and spatial variation of melt contributions from the asthenospheric

and lithospheric mantle to magmatism. We focus on the Oligo-Miocene volcanism in the Espanola Basin, New Mexico, which is located in the central portion of the rift. Most of these magmas originated from asthenospheric mantle with the same elemental and isotopic compositions as the source of the magmas of Northern Hemispheric ocean-islands and seamounts (NHOIB). These early rift-related magmas are geochemically similar to the much younger Pliocene Cerros del Rio lavas of the same region. In other parts of the rift, in southern New Mexico and NW Colorado, magmas derived predominantly from an asthenospheric mantle source are not apparent before approximately 5

From STOREY,B. C., ALABASTER,T. & PANKHURST,R. J. (eds), 1992, Magmatismand the Causes of ContinentalBreak-up, Geological Society Special Publication No. 68, pp. 61-89.

61

62

S.A. GIBSON E T A L .

Ma. We have used the combined data to construct a geochemical model in which the incompatible trace element concentrations and radiogenic isotope ratios of these rift-related magmas are a result of: (1) the amount and rate of extension; (2) composition of the asthenosphere; (3) the presence or absence of preexisting large-scale volcanismi (4) crustal contamination. We have compared our geochemical model for the Rio Grand rift with other examples of continental extension-related volcanism in order to establish its global significance.

Model of continental extension-related volcanism The importance of both asthenospheric and lithospheric portions of the mantle as source reservoirs during continental extension-related magmatism is well documented (for example, Fitton & Dunlop 1985; Fitton et al. 1988, 1991; Leat et al. 1988; Thompson et al. 1990). It has been shown recently that the volume and composition of magmatism generated during continental extension is directly related to the amount of stretching of the lithosphere and the potential temperature of the underlying anhydrous asthenosphere (McKenzie & Bickle 1988). This is because extension of the lithosphere causes asthenospheric upwelling and leads to decompression melting. Volatile-rich, low viscosity melts are believed to leak from the asthenosphere into the overlying sub-continental lithospheric mantle (SCLM: Haggerty 1989; McKenzie 1989). The volume of each upward moving melt increment would be extremely small and advect negligible heat and therefore leave the thermal structure of the overlying lithosphere undisturbed (McKenzie 1989). Each increment of this melt would freeze immediately it reached a level within the lithosphere where the ambient temperature was less than its solidus. The continuation of this process over geologically long periods of time could permit the accumulation of substantial volumes of frozen melt, possibly as complexes of veins, dykes and sills in a vertically thin zone. McKenzie (1989) pointed out that this 'sensitive zone' would remelt extensively, if subjected to even small amounts of decompression during continental extension or heating above a later mantle plume. Such a zone might provide a source of copious contamination to later upwelling magmas from asthenospheric sources because the heat advected by the dccpcr source melts would locally raise temperaturc of vein complexes above their solidus (Thompson et al. 1990).

In the Rio Grande rift and surrounding region, it is the Cenozoic strongly potassic mafic magmas (lamproites and minettes) that are generally supposed to approximate to refusion products of small melt fractions leaked from the asthenosphere in to the SCLM. These occur temporally and spatially at the fringes of rift-related igneous activity; i.e. along the flanks of the structure and early or sporadically in local volcanic sequences (Leat et al. 1990; Thompson et al. 1990, in press; Gibson et al. 1991, in press). The isotope systematics of these potassic rocks suggest that K-rich small melt fractions may have been accumulating in the lithosphere from as long ago as the Proterozoic to as recently as the Mesozoic. Inversion modelling or rare-earth element abundances in rift-related potassic rocks from NW Colorado suggests that their distinctive compositions are probably ultimately related to incipient melt leakage from hydrated asthenosphere above a subduction zone (McKenzie & O'Nions 1991). This is consistent with tectonic reconstructions of the region which show that subduction was probably repetitive beneath the Western USA, during its accretionary growth between 1.8 and 1.1 Ga (Condie 1982) and again during the Mesozoic (Sveringhaus & Atwater 1990). The type of magmatic model that we have outlined here, involving mixing between asthenospheric-source basaltic and SCLM-source ultrapotassic melts, contrasts with the many published petrogenetic schemes for continental magmatism that invoke wholesale rather than selective SCLM fusion. The extent to which the small melt fractions react with the surrounding essentially dry SCLM as they freeze (Menzies this volume; Hawkesworth et al. this volume) will have a significant effect on the composition of any subsequent melt. If this interaction occurred on a massive scale, large portions of the SCLM could become progressively 'wet' and subsequent melting might be wholesale (Hawkesworth et al. this volume).

The Rio Grande rift S u m m a r y o f tectonic e v o l u t i o n a n d p r e s e n t - d a y structure

The Rio Grande rift is superimposed on a region with a long and complex history of deformation and heating, which affects the location and composition of the subsequent rift-related magmatism (see below). The development of the Rio Grande rift formed part of a series of Cenozoic extensional events that affected the

RIO GRANDE RIFT MAGMATISM

110 ° I

42"~- - ' -

108 ° I

u

63 102 °

104"

1 0 6 7 Ma) basalts of New Mexico and SE Arizona and correlated this with a corresponding southv~ard increase in the amount of lithospheric extension. They identified three end-members within the young basalts; a 'depleted' asthenospheric mantle (DM),

70

S.A. GIBSON E T A L .

II

-

I

I

II

~e,~

~v~ ~

~c ~g

~ ~__.~. ~,..~~ t'~~ ~ ~ ~ ~

~

¢~. ~,-~v'~~

.~.

~I~

~I

'~I I~"

~c

"~'~'~~'8°..

~~vv

IB~'~"~.. ~

.~. oo

~.~''~-~-~.~, ~ , ~ a

i~-,--~

~

~'~ •-

i~ i~ II

,~.-°®:'~"°S-"

~ - ~ ~ - ~ i ~

~i~ ~I'~"

~~-~-~-~2

~-~~-~

~r, . ~ - ~ -

....,

~

~ ~

~

II

~

RIO G R A N D E RIFT MAGMATISM

71

II

O0

e~

I"-

•

@0

.

.

.

.

*o ~..

II II

II

If e.,.

i

S.A. GIBSON ETAL.

72

II

II

II

II

II

II

m

k,g t'~l

~

~

|

I

r..,

!

1

°-

RIO GRANDE RIFT MAGMATISM

73 r~

°~

0

~1. r. o

~'~

~

o

.¢

-

~

74

S.A. GIBSON E T A L . centrations and Sr and Nd isotopes, suggest that they are predominantly derived from an asthenospheric mantle source as outlined above. In order to assess the relative contributions of asthenospheric and lithospheric mantle sources to the compositions of magmas from the axis of

an 'enriched' lithospheric mantle (EM; Fig. 8) and a crustal component. According to this model the Oligo-Miocene Espanola Basin mafic magmas would be derived from enriched lithospheric mantle. However, our interpretations, based on both incompatible trace element con0.5132 "0)

/

\

/

0.513

30-17 Ma

\

/'

OIB

\ \ \

\

0.5128 \

I

•

"

0.5126

0.5124

F

t

"''"""

0.5122

0.512 0.702

I

I

I

t

0.~

0.~

0.~

0.~

I 0.707

!

I

0.~

0.~

87Sr/e6Sr

0.5132

b)

17 Ma and open squares < 13 Ma volcanism. The effects of adding 10% of these dominantly lithosphere-derived magmas to 90% of an asthenosphere-derived magma from the Potrilio-Palomas volcanic field (6LT99) are also shown. The K-rich marie magmas are as follows: A, Navajo minette (89SB141); B, Two Buttes minette (89SB234); C, Middle Park minette (6LT56) and D, Leucite Hills average madupite. The EM (enriched lithospheric mantle) and DM (depleted asthenospheric mantle) of Perry et a/. (1987) are shown for comparison. Data sources are as follows: 1, Riley dyke swarm (Table 1 and our unpubl. data); 2, Navajo Province and Dulce dyke swarm (Table 1, Gibson etal. in press and our unpubl, data); 3, Two Buttes (Gibson et al. in press); 4, Wasatch Plateau (Tingey et al. 1991); 5, Middle Park minette (Leat et al. 1991); 6, Elkhead Mountains (Thompson et al. 1990); 7, Leucite Hills (Fraser 1987); 8, Independence volcanic suite (Meen & Eggler 1987); 9, Crazy Mountain (Dudas et al. 1987); 10, Smoky Butte (Fraser 1987).

the Rio Grande rift it is necessary to evaluate the contributions of other processes, such as crustal contamination, magma mixing and replenishment fractional crystallization. Detailed studies have shown that open-system processes have significantly affected the concentrations of incompatible trace elements and radiogenic isotope ratios in the large volume magmatic systems in the Rio Grande rift, e.g. Taos (McMillan & Dungan 1988), Flat Tops (Gibson et al. 1991). However, the small-volume magmas that are common in the Rio Grande rift tend to lack a significant crustal component in their compositions. This is because they are volatile charged and do not possess the thermal capacity to fuse appreciably the continental crust as they traverse it. They also characteristically have relatively high contents of incompatible trace elements, including Sr, Nd and Pb, so that the geochemical impact of any crustal contamination is minimal. In the cases where local crustal contamination did occur, it is relatively easy to detect

this process and quantify it. This is because small-volume melts did not develop sufficient magmatic throughput to establish long-lived, open-system chambers in which individual magma batches mix and homogenize, and can also develop subtle shifts in the ratios of incompatible trace elements during polycyclic behao viour ( O ' H a r a & Matthews 1981). Southern

New Mexico

The vast outpouring of magma associated with Eocene-Oligocene volcanic activity in southern New Mexico, forming the Mogollon-Datil volcanic field (36-24 Ma, McIntosh 1991) was followed by an episode of further calc-alkaline volcanism that was associated with the early development of the southern part of the Rio Grande rift. The lavas of Sierra de las Uvas (Fig. 1) are exposed on the faulted margins of the rift and fill horst and graben structures. They represent some of the oldest (28 Ma Seager et al. 1984)

76

S.A. GIBSON E T A L .

and most voluminous rift-related magmatism in the south of the rift and were contemporaneous with the earliest magmatism in the Espanola Basin. The Uvas lavas have relatively low 143Nd/ 144Nd ratios, in comparison with the OligoMiocene Espanola Basin magmas (Fig. 7a) but are similar to the Oligo-Miocene San Luis Hills iavas in southern Colorado (see below). In the southern part of the Rio Grande rift a lull in magmatism between 16 and 13 Ma (Seager et al. 1984) was followed by a further intense episode of basic volcanism that occurred < 5 Ma, after renewed rifting between 9.6 and 7.1 Ma (Seager et al. 1984). Alkali-olivine basalts were accompanied by subordinate tholeiitic basalts in the Palomas and Potrillo volcanic fields along the Mexico border (Fig. 1). These late-phase magmas have some of the hightest ]43Nd/l'~Nd ratios observed in the volcanic activity associated with the Rio Grande rift (Fig. 7b). These ratios overlap the fields of Atlantic and Pacific M O R B ( A P M O R B ) and Northern Hemisphere OIB

( N H O I B , Fig. 9). Large individual flows of Recent mafic lavas were also erupted in the Jornado Basin, east of the Rio Grande at Black Mesa (1 Ma, Table 2) and the Malpais flow (0.2 Ma, Table 2) near Bosque del Apache, and in the Tularosa Basin near Carrizozo. These are tholeiitic basalts and have lower 143Nd/144Nd and higher S7Sr/a6Sr ratios than the Palomas and Potrillo magmas. On Fig. 7b they plot in a similar position to the olivine tholeiite (OT) lavas from the Albuquerque Basin, further north. Preliminary data suggest that the whole-rock chemistry of these lavas may have been affected by assimilation of upper crustal material.

Central New Mexico

Early-rift-related volcanism on the shoulders of the Albuquerque Basin, e.g. at Cerro Colorado, sugggests that contemporaneous magmatism may have occurred along the rift axis at this

0.5132

DM 0.5130

0.5126 1

D 0

•

I:~0

I

•

0

I

°

0.5126 co

:

°

EM

,

5

m

0.5124

'

I

OOm

I

e

]

O

•

I

9 10

• •

0.5122

7

New Mexico 0.5120

~

I

I

I

I

I

I

32

33

34

35

36

37

t)11 rl

Colorado

"

I

I

I

I

3B

39

40

41

Degrees Latitude

Fig. 9. North-south variation of Nd isotope ratios in mafic volcanics along the rift axis from NW Colorado to southern New Mexico. Closed squares represent early-phase (30-17 Ma) and open squares late-phase (< 13 Ma) rift-related volcanism. The ranges of APMORB (Atlantic and Pacific mid-ocean ridge basalts), NHOIB (Northern Hemispheric ocean-island basalts), WWOIB (world-wide ocean island basalts) are from Wilson (1989). The fields of EM and DM and the effects of mixing 10% K-rich mafic magma are taken from Fig. 8. Localities and data sources for Nd isotope ratios are as follows: 1, Potrillo-Palomas volcanic field (Table I and our unpubl, data); 2, Sierra de Las Uvas (Table 1 and our unpubl, data); 3, Valley of Fire lava, Tularosa Basin and Malpais flow, Jornada Basin (Table 1); 4, Albuquerque Basin (Perry et al. 1987); 5, Southern Espanola Basin (Dunckeretal. 1991 and Gibson etal. in press); 6, Northern Espanola Basin (Table 1, Gibson et al. in press & our unpubl, data); San Luis Hills lavas (Thompson et al. 1991); 8, Triangle Peak (Leat et al. 1989); 9, Flat Tops volcanic field (Gibson et al. 1991 and Table 1), Yarmony Mountain (Lear et al. 1990); McCoy, Dotsero and Willow Peak iavas (Leat et al. 1989); 10, Yampa province (Leat et al. 1991); 11, Walton Peak (Thompson etal. in press).

RIO G R A N D E RIFT M A G M A T I S M

77

+1 +1 +1

+1 +1

+1 +1

+1 +1

+1 +1

+1 +1

8 i#1 0

ZZZZ

I

.~.

+1 +1

~ .

•

•

Q

+1

.

+1 +1

~

+1 ~

.

+1 +1

+1 +1

+1 +1

+1 +1

o +1

o +1

.

~

+1 ~m

d

+1 ~

..

.=. 0,.t

.=. "13

.1=

¢II

.,,.

8

el

~a

[..

r~

78

S.A. GIBSON E T A L .

latitude but has been obscured by subsequent deposition of clastic sediments. Mafic dykes were emplaced at 13 Ma (Aldrich et al. 1986) but the most intense volcanic activity in the Albuquerque Basin appears to have occurred within the last 5 Ma, e.g. Cat Hills, Santa Ana Mesa, Lucero, Albuquerque volcano (Baldridge 1979). During this volcanic activity both alkaliolivine basalt and olivine tholeiite lavas were erupted. The tholeiites from this area unlike those in the Espanola Basin, have similar or lower 143Nd/144Nd ratios and similar or higher STSr/86Sr ratios than the contemporaneous alkali basalts (Fig. 7b). This feature has been attributed to crustal contamination by Baldridge (1979). The alkali basalts from the Albuquerque Basin have 143Nd/~Nd ratios that are similar to, or slightly lower than, the lavas of similar age from the Palomas and Potrillo volcanic fields further south. On the basis of radiogenic isotope ratios Perry et al. (1987) have proposed that the lavas in Albuquerque Basin are derived from an enriched lithospheric mantle (EM) rather than a depleted asthenospheric mantle (DM), although these ratios are not dissimilar to those of NHOIB (Fig. 9).

Northern New Mexico and southern Colorado

At approximately 26 Ma magmatism occurred in the axis of the rift in the San Luis basin (Fig. 2; Thompson et al. 1991). This is roughly at the same time as the early magmatism 50 km to the south in the Espanola Basin. The lavas consist of a series of alkali and tholeiitic basalts, many of which have undergone AFC-style processes (Thompson et al. 1991). The olivine tholeiites are the least contaminated lavas. These have lower 143Nd/~44Nd ratios than the contemporaneous Espanola Basin magmas (Fig. 7a) and have slightly higher concentrations of LIL and La/ Ta ratios (Fig. 5a). The San Luis Hills lavas are thought to be a result of mixing between asthenosphere- and lithosphere-derived components, together with limited crustal contamination (Johnson & Thompson 1991). During the Pliocene, some of the most voluminous volcanic activity occurred in the central part of the rift with the eruption of the Taos volcanic field (Fig. 1). These lavas overlie those of the San Luis Hills and range in composition from basalts and andesite to dacites. They evolved within an open magmatic system in which the most fractionated

lavas assimilated considerable quantities of granulite-facies crust (McMillan & Dungan 1986, 1988). The less-evolved Servilleta basalts of the Taos field have low concentrations of K20 and La/Ta ratios (Fig. 5b) but have relatively low t43Nd/144Nd isotope ratios (Fig. 7b) These have been interpreted as depleted MORB-type magmas that have been contaminated to a small extent by lower crust (Dungan et al. 1986). NW Colorado

In the northernmost part of the Rio Grande rift, in NW Colorado, the amount of lithospheric extension has not yet been quantified, but was clearly less than further south (Cordell 1982). The low degree of extension is reflected in the composition of the associated volcanism. Detailed studies of early-phase mafic magmas in the northern area, e.g. Fiat Tops (Gibson et al. 1991), Yarmony Mountain (Leat et al. 1990), Walton Peak (Thompson et al. in press), suggest that they all represent, to varying extents, mixtures of melts derived from asthenospheric and lithospheric mantle sources. At many of these localities the lithospheric component is manifested in anomalously K-rich mafic flows that appear sporadically within the lava successions. All of these early-phase magmas from NW Colorado are more enriched in K20 and have higher La/Ta ratios than the contemporaneous magmas in the Espanola Basin (Fig. 5a). They also have lower 143Nd/ 144Nd and STSr/S6Srisotope ratios that are higher than, or similar to, their temporal equivalents in the rift axis further south (Fig. 7a). A lull in volcanic and tectonic activity occurred in NW Colorado between 20 and 13 Ma (Leat et al. 1988). Renewed volcanism occurred at Flat Tops, the Glenwood Springs area and in the Elkhead Mountains and was followed by localized crustal uplift. This volcanic activity in NW Colorado may be a consequence of heating of the SCLM by the approaching Yellowstone plume, together with renewed extension (Thompson et al. 1990). Following the removal of the most fusible melt fractions from the lithospheric mantle beneath NW Colorado, magmas with similar incompatible element concentrations to OIB were emplaced at Yampa (Leat et al. 1991). These magmas have similar 87Sr/ 86 Sr to, but lower 143 Nd/ 144Nd isotope ratios than the OligoMiocene Espanola Basin magmas (Fig. 7b). They also have slightly elevated concentrations of some incompatible trace elements relative to true OIB (Fig. 5b) and even the most OIB-like have probably mixed with a few percent of lithosphere-derived melts.

RIO GRANDE RIFT MAGMATISM The latest episode of basic volcanism in the northern segment of the Rio Grande rift occurred during the Quaternary in the rift axis south of Yampa, at Dotsero, Triangle Peak, McCoy and Willow Peak (Leat et al. 1989), in areas that had not locally experienced any significant previous Phanerozoic mafic magmatism. These lavas have similar concentrations of incompatible trace elements and Sr and Nd-isotope ratios to the early-Miocene volcanics in the area (Fig. 7b) and are thought to represent mixing between asthenospheric and lithospheric mantle sources (Leat et al. 1989).

Discussion o f variations in asthenospheric and lithospheric mantle sources during rifting Using concentrations of incompatible trace elements and radiogenic isotope ratios, it is possible to establish the concentrations of various mantle sources to the Oligocene-Recent magmas that were erupted in the axis of the Rio Grande rift. The compositions of these different mantle sources may have changed in space and time during the development of the rift.

Along axis isotopic variations. At first sight it is tempting to make the simple assumption that magmas with lower 143Nd/l~Nd ratios than the Espanola Basin basanites were contaminated to a greater degree by low 143Nd/144Nttmelts from SCLM sources. However, Fig. 9 shows that the situation is more complicated. K-rich mafic magmas are useful indicators of the compositional and isotopic nature of the underlying SCLM. Although they themselves are probably mixtures of melts from asthenospheric and lithospheric mantle sources (Thompson et al. 1990), their chemistry is typically so dominated by the composition of the lithospheric component that, in many cases, it is difficult to define any asthenospheric component. Figure 8 shows the variation of 14aNd/144Nd isotope ratios in K-rich mafic magmas along a north-south profile from Montana to New Mexico. It is apparent from this diagram that the lithosphere giving rise to the potassic liquids in the north of the region has lower 143Nd/ 144Nd ratios than that in the south. If the lithospheric source of the K-rich mafic magmas had similar values of Sm/Nd throughout this region, the values of 143Nd/144Nd may indicate a varying age of the SCLM along the traverse in Fig. 8. There is an abrupt change in 143Nd/144Nd of the K-rich rocks in N. Colorado at approximately 39°N, somewhat south of the Cheyenne Belt (41°N) where the Wyoming Archean craton

79

meets lithosphere of Proterozoic age. Eggler et al. (1988) have used evidence from mantle xenoliths in nearby kimberlite pipes to suggest that at depth (up to 200 km) the Wyoming craton extends further south than the surface location of the Cheyenne Belt and underlies Proterozoic crust. The data of Fig. 8 support this idea. Figure 9 shows the corresponding north-south variation in 14aNd/144Nd ratios for magmas from the rift axis. The effect of adding as little as 10% of 'local' lithospheric melt to an asthenospheric melt of constant composition, typified by a sample from the Potrillo lavas, is also shown on this diagram. From this we conclude that, although the OIB-type magmas in the north of the rift have lower 143Nd/144Ndratios than the OIBtype magmas further south in the rift, they may have actually interacted with a similar or smaller amount of lithospheric melt. It seems likely that the magmas from the Albuquerque and Espanola basins and the Yampa province were derived from melts of upwelling asthenospheric mantle but have mixed with a very small amount (< 10%) of melt from the lithosphere.

Composition of the asthenosphere. We have shown that OIB-like rift-related magmas were erupted along the length of the axis of the Rio Grande rift throughout the whole of its development. Leat et al. (1991) have argued that OIBtype magmas in NW Colorado are derived from mantle advected by the Yellowstone plume that has approached the area during the last few Ma. They further suggested that the mantle of this plume had the composition of NHOIB, similar to that of the Hawaii and Iceland plumes. Brandon & Goles (1988) and Draper (1991) came to similar conclusions. If melts from such mantle differ significantly, in trace element and isotopic composition, from melts of asthenosphere that existed below the early Rio Grande rift, current knowledge does not allow them to be distinguished. Future data, e.g. 3He/4He isotope ratios, may permit us to distinguish or correlate these asthenospheric sources. It is nevertheless apparent that, at least in the Oligo-Miocene, the asthenosphere underlying the central section of the rift was heterogeneous with respect to its volatile content. We have proposed that the spatial and temporal coexistence of silica-saturated and undersaturated OIB-type magmas in the Espanola Basin is best explained by the presence of hydrous 'patches' in the underlying athenospheric mantle (Gibson et al. in press). This may be a result of volatiles escaping from the subducting FaraUon Plate, that has been imaged beneath the area by seismic tomography (Grand 1987).

80

S.A. GIBSON E T A L .

The importance o f pre-existing magmatism. In both the northern and southern sectors of the Rio Grande rift, the eruption of magmas whose compositions are dominated by melts from an asthenospheric mantle source only occurs in areas where there is evidence of a previous episode of large-scale volcanism, for example, the Yampa province is adjacent to the voluminous Flat Tops volcanic field, and the Potrillo Palomas volcanic field overlies the fringes of the Sierra de las Uvas lavas. In both cases, the earlier phase of magmatic activity appears to have been responsible for melting all of the more-fusible hydrous and halogen-rich components from the lithosphere so that later melts could traverse the lithospheric mantle with relatively little contamination. Such a melting event would effectively raise the solidus temperature of the 'depleted' lithosphere and, in order for subsequent melting of the lithosphere to occur, it would have to be stretched to a greater extent, (i.e. increase/3) or heated by raising the potential temperature (Tp) of the underlying asthenosphere. (Tp is the temperature the mantle would have if it was able to rise adiabatically to the surface without melting.) The absence of any significant lithospheric contribution to the compositions of the OligoMiocene magmas in the Espanola Basin is anomalous considering that an appreciable lithospheric component was involved in contemporaneous magmatism, only 50 km to the north, in the San Luis Hills. In the case of the Espanola Basin, evidence of pre-rift volcanic activity is preserved in volcanic clasts in early to late Oligocene sediments (Ingersoll et al. 1990). These suggest that intermediate to silicic activity started soon after 38 Ma, at the Cerrillos, Ortiz and Espinaso magmatic centres, in the south of the Espanola Basin, and reached a peak at 34 Ma (Fig. 3; Kautz et al. 1981). Shortly after rifting had been initiated in the Espanola Basin, the Servilleta Plaza volcanic centre developed to the north of Ojo Caliente (23-22 Ma; Ingersoll et al. 1990; Fig. 3). This volcanic activity may have removed the more fusible components from the underlying lithospheric mantle and permitted the upward migration of 'uncontaminated' asthenosphere-derived melts during the Oligo-Miocene. Relationship to extension. McKenzie & Bickle

(1988) have shown that the volume and composition of magmatism generated during continental extension is directly related to the amount of lithospheric extension and the potential temperature of the anhydrous asthenospheric mantle. A complicating factor in applying the model of

McKenzie & Bickle (1988) to the Rio Grande rift is that we suspect that the asthenosphere was partially hydrous in the Oligo-Miocene (see above), whereas they only consider the anhydrous system. On Fig. 2 we have calculated the apparent amount of stretching of the lithosphere (/3~) and the crust (/3,c) along the rift axis. The values are calculated using the same method as McKenzie (1984) in which/3, represents the local change in thickness of the crust or lithosphere, beneath a given point. In this case the melt thickness, if any, generated during extension is assumed to be added to the crust thickness, i.e. by underplating. At the rift axis, this value is usually greater than true/3 as the latter represents the amount of stretching over an area and is the ratio of the final to the initial surface area (as originally defined by McKenzie 1978). CordeU (1982) has shown that lithospheric extension, associated with the development of the Rio Grande rift, has affected a broad zone (as much as 750 km wide) and is centred on the rift axis. The amount of extension decreases symmetrically from the rift axis to the rift flanks and shows a correlation with magma composition (Gibson et al. in press). A value for true ~ would integrate the amount of lithospheric stretching at both the rift axis and the rift flanks. Since, in this study we are only concerned with magmatism that was generated at the rift axis, it seems more appropriate to use/3,. Due to the difference in present-day lithospheric thickness beneath the Colorado Plateau (150 km; Davis et al. 1984) and the Great Plains (200 kin; Davis et al. 1984) we have calculated a range of/3a~ values based on these end-members. Our values for present-day lithospheric thicknesses along the rift axis are taken from a variety of sources (see Fig. 2) and in most cases are slightly greater than the thicknesses calculated in the recent gravity study of Cordell et al. (1991). This is probably because the values of Cordell et al. (1991) are sensitive to accurate crustal thicknesses. For example at the Colorado/New Mexico state line /3~a according to Cordell et al. (1991) would be 3.5-4.0 instead of our calculated value of 2-2.5. Using the calculations of McKenzie & Bickle (1988) such high values of/3al would predict the local generation of much greater volumes of magmatism than we see in this part of the rift. This absence of voluminous surface magmatism may in theory be explained by crustal underplating, but such a proposal conflicts with the presence of basanitic magmas in the Espanola Basin. The occurrence of such magmas suggests that initially only a small amount of partial melting of the asthenospheric mantle occurred in this area.

RIO GRANDE RIFT MAGMATISM A difference in crustal thicknesses is apparent between the Colorado Plateau (40 km; Keller et al. 1979) and Great Plains (50 kin; Stewart & Pakiser 1962) and so we have calculated a range for in/3~c which also takes into account some of the pre-rift thinning beneath the Rio Grande rift. Figure 2 shows that there is a discrepancy in /3a values for the crust and lithosphere at any given locality along the rift axis. In all cases the value of fl~ is greater than//~c. This difference may be due to underplating of the crust (see above) or replacement of the lithosphere by the asthenosphere, by a mechanism other than stretching. We shall now consider how these variations in amounts of crustal and lithospheric thinning compare with the variations in magma composition that we have discussed earlier. Figure 2 shows that the amount of lithospheric and crustal thinning increases from north to south along the rift axis. According to the model of McKenzie & Bickle (1988), this increase should correlate with an increasing degree of partial melting of the asthenosphere, if the latter has a constant potential temperature. In northern New Mexico the values of/3a are intermediate, for both the crust and lithosphere, between those we calculated in NW Colorado and southern New Mexico (Fig. 2) and we might also expect the composition and volume of magmatism in this area to be intermediate. However, we have already shown (Fig. 6) that asthenospheresource magma compositions all along the rift axis are very similar and that their occurrence is predominantly controlled by pre-existing magmatism. At first sight it appears that the magmatism in northern New Mexico was much more voluminous than in NW Colorado (Fig. 2). But many of the volcanic fields in the former are associated with the Jemez lineament and are only partially related to extension in the Rio Grande rift. The presence of OIB-type magmas in NW Colorado may reflect a northward rise in the potential temperature of the asthenosphere along the rift axis, acting in the opposite sense to the changes in/3. The asthenosphere is probably at normal potential temperatures in at least the southern segment of the rift and rises towards the northern segment, which is only 500 km from Yellowstone Park and the mantle plume widely supposed to underlie it (Fitton etal. 1991; Lear et al. 1991; Thompson & Gibson 1991). Implications f o r rifting models. The develop-

ment of continental rift systems has been described in terms of 'active' and 'passive' end members (Sengor & Burke 1978). These terms

81

relate to the mechanisms that cause the upward movement of the asthenosphere-lithosphere boundary. The difficulties in applying these simplified terms to the modern Rio Grande rift have been discussed by Golombek et al. (1983), Baldridge et al. (1984) and Olsen et al. (1987). Many of the distinguishing characteristics that can be applied to other major continental rift zones, such as the relative timing of lithospheric and crustal thinning, are difficult to determine in the Rio Grande rift due to its long and complex pre-rift history. Olsen et al. (1987) have pointed out that the volume of magmatism associated with the rift is relatively small and that there is no compelling evidence for a major thermal event in the convecting mantle that is uniquely associated with rifting. They suggest that the rift developed as a mixture of active and passive processes. The identification of asthenosphere-derived melts in magmatism associated with the development of the Rio Grande rift has been used to constrain geophysical models (e.g. Davies 1991). The identification of such melts in the late phase of rifting (e.g. Reid & Woods 1978) has been interpreted as evidence of the progressive replacement of the lithospheric mantle during asthenospheric upwelling; as extension progresses from the onset of rifting (Perry et al. 1987). Our data from the Espanola Basin do not conflict with this model but suggest that significant asthenospheric upwelling occurred much earlier, at the beginning of rifting. In many cases, such as the San Luis Hills (Johnson & Thompson 1991), this predominantly asthenospheric mantle source signature is not obvious in the resulting magmatism, due to mixing with lithosphere-derived melts. This early upwelling of the asthenosphere coincides with the large magnitude extension between 30 and 18 Ma and may contribute to the high geotherm and ductile nature of the lithosphere at this time (Morgan et al. 1988). Bussod & Williams (1991) have noted that a 'thermal bulge' occurred at 22 Ma in the uppermost lithospheric mantle beneath the south of the rift. They suggest that this is due to dyke injection within the uppermost mantle, rather than lithospheric thinning and asthenospheric upweiling. Davis (1991) has incorporated this interpretation into a passive model for rifting beneath the Rio Grande rift in which he suggests that extensional stresses in an initially hot lithosphere caused abyssal fractures that were subsequently intruded by melts from the asthenosphere. This injection of magma into the lithosphere caused widespread uplift and triggered small scale asthenospheric convection together with crustal faulting and thinning.

82

S.A. GIBSON E T A L .

Changing magma sources during continental rifting and break-up The temporal and spatial patterns of magmatic composition we have described from the Rio Grande rift show several features that may be useful to studies of magmatism during continental fragmentation. Although it is an oversimplification to make a binary subdivision of a range of possibilities that may have existed at passive (former rifted) continental margins, it is convenient to distinguish between those that overlay the heated asthenosphere near mantle plumes and those that did not. The only major exception to this subdivision would occur if continental rifting and subsequent break-up took place close enough behind a subduction zone for water from the latter to enter and enhance decompression fusion- and hence magma production- in asthenosphere of otherwise normal potential temperature (Tp = 1280°C). We suspect that this may have been the case locally beneath the Rio Grande rift in the Oligo-Miocene (Leat et al. 1988; Gibson et al. in press). In contrast, we envisage an asthenosphere beneath the rift, during the last few Ma, that was/is essentially dry with traces of CO2 and H20 etc. Southwestern

USA

As we have mentioned earlier, the development of the Rio Grande rift was part of a much wider series of events that affected the whole of the western USA. At the present time, the causes of extension and magmatism in the western USA remain controversial between those who favour a subduction-related origin (e.g. Best & Christiansen 1991; James & Henry 1991; O'Brien et al. 1991) and those who suggest that there is no slab involvement (e.g. Gans et al. 1989; Dudas 1991; Norman & Mertzman 1991). During the Cenozoic, magmatism and extension were widespread in the western USA and their initiation has traditionally been linked to changes in plate motions, i.e. passive rifting processes. During the early Tertiary the magmatic arc that had developed in the Mesozoic at the continental margin swept northeastwards for about 1000 km. The Rio Grande lies at the eastern margin of this magmatic arc. This phenomenon has been related to a shallowing in the angle of the subducted slab (Lipman et al. 1972; Coney & Reynolds 1977). During the mid-Tertiary a dramatic shift in magma composition, from silicic and intermediate to basaltic, occurred in the western USA. Christiansen & Lipman (1972) and Snyder et al. (1976) have linked this to the change from eastward dipping subduction to a

strike-slip margin off the Pacific coast between 30 and 20 Ma (Atwater 1970). It is clear from recent data, however, that there are discrepancies with the subduction-related origin for this early Tertiary magmatism. The region of calc-alkaline magmatism is much broader than corresponding ones at present-day subduction zones and also the temporal and spatial migration of magmatism and extension shows inconsistencies. For example, in the Basin and Range Province the northern and southern segments appear to have evolved separately (Glazner & Bartley 1984; Taylor et al. 1991). In some areas the onset of volcanism and extension predates the passage of the trailing edge of the subducted slab. In fact Gans et al. (1989) have challenged the subduction-related origin for magmatism and extension throughout the Basin and Range province and Rio Grande rift. In their study of the temporal and spatial variation of magmatism and extension in the eastern Great Basin, they have shown that striking similarities exist in eruptive and extensional histories, with other highly extended parts of the Basin and Range province, despite regional variations in absolute timing. As a result of this study Gans et al. (1989) suggest that primitive basalts were erupted late in the extensional/ magmatic history of the region, due to a change in composition and thinning of the crust. They propose that the earlier voluminous silicic-tointermediate magmatism represent the 'thermal refining' of the most fusible melt fractions from the crust. This is very similar to the model that we are proposing for the Rio Grande flit, except that we believe that the composition of the lithospheric mantle, rather than the crust, is most significant in determining the magma composition in this region. Gans et al. (1989) have proposed an active model for the Basin and Range province, involving asthenospheric upwelling, although they do not postulate the cause of the initial melting of the asthenosphere. A review of large-scale trends within Cenozoic magmatism in the southwestern USA has shown an overall change from magmatism dominated by lithospheric mantle components (between 25 and 5 Ma) to magmatism dominated by an asthenospheric component (< 5 Ma) in areas of greatest overall extension, i.e. Basin and Range (Fitton et al. 1988, 1991; Kempton et al. 1991). From this Fitton et al. (1991) suggest that the lithosphere beneath the Basin and Range has either been stripped of its capacity to contribute towards magmatism or has been removed completely during the last 5 Ma. However, our evidence from the Rio Grande rift suggests that the lithosphere was loc-

RIO GRANDE RIFT MAGMATISM ally devoid of its more fusible melt fractions as early as 26 Ma. The Trans Pecos magmatic province, Texas, adjoins the southern end of the Rio Grande rift and is part of the Basin and Range province, although there is no evidence for crustal thinning (Barker 1987). James & Henry (1991) have noticed a change from calc-alkaline (48-30 Ma) to OIB-type magmas (28-17 Ma) in the Trans Pecos province. Their data shows that OIB-type magmas were also sporadically associated with the calc-alkaline magmas and they have interpreted these as being derived from the mantle wedge. The origin of the calc-alkaline magmatism has been interpreted as either subduction-related (Lipman et al. 1972; Henry & Price 1984; Price et al. 1987) or as a product of backarc spreading over an asthenospheric diapir (Barker 1979). But the origin of the Trans Pecos magmatism remains controversial. Nelson & Nelson (1987) have suggested that the Cenozoic Trans Pecos magmas are the result of mixing between lithosphere- and asthenosphere-derived melts and that their compositions are independent of tectonic setting whilst James & Henry (1991) maintain a subduction-related origin. In many respects, the incompatible trace element concentrations and isotope ratios of the Trans Pecos magmatic province are similar to those from the Rio Grande rift and it may ultimately be established that they have a similar origin. Whilst we believe that the physical presence of a subducted slab beneath the Rio Grande (and also the Western USA) was important for magma generation and extension, we suspect that its influence on the chemistry of these magmas was relatively minor in comparison with lithospheric melts and crustal contamination. The dominant chemistry of these lithosphere-derived melts, with additional local effects of crustal contamination, would probably overprint any subductive component within the early voluminous magmas. E i f e l a n d Sardinia

A comparison of Pliocene-Quaternary mafic alkali magmatism at two sites in Europe, Eifel (Germany) and NW Sardinia, illustrates the significance of pre-existing magmatism to the composition of subsequent magmas. In both places, small volumes of silica-undersaturated magmas have been erupted during the last few Ma in regions that have been undergoing modest but widely distributed lithospheric extension during much of the Tertiary (Cherchi & Montadert 1982; Meissner 1986). Crustal thicknesses are approximately 30 km in both areas (Peruzza et

83

al. 1990; Prodehl & Giese 1990); somewhat thinner than most continental values but not obviously attenuated. Total lithospheric thicknesses are approximately 50 km beneath Eifel (Prodehl & Giese 1990) and approximately 75 km beneath NW Sardinia (Panza 1985). Both these values are low enough for underlying convecting mantle at normal potential temperatures (1280°C; McKenzie & Bickle 1988) to approach its anhydrous solidus and certainly to begin melting in the presence of any H20 or CO 2. In the east and west Eifel volcanic fields the sodic alkali-olivine basalts, basanites and nephelinites are accompanied by leucitites. Their Sr and Nd isotope systematics have been interpreted by Worrier et al. (1986) as mixing between sources with respectively a relative time-integrated enrichment and depletion in Rb/Sr and Sm/Nd. They showed that plausible models of the magma genesis required garnet in a lherzolite source rock. With a 50 km lithosphere thickness, this places the principle source of magma genesis within the asthenosphere. If the convecting mantle beneath the Eifel is taken to have a similar composition to that of NHOIB, the most probable source of the contrasting isotopic comp.onent with relatively low 143Nd/I44Nd and high O'Sr/86Sr is in the SCLM. Such an appraisal of the Eifel geophysical framework and isotopic data leads to the conclusion that the asthenospheresource Eifel magmas incorporated a substantialvariable chemical input from the SCLM during their uprise. In contrast, Rutter (1987) showed that, after allowance for assimilation of a few percent of continental crust, the alkali-olivine basalt and basanite lavas of NW Sardinia had elemental and isotopic compositions that did not require any chemical input from a SCLM source. Unlike the Eifel, the NW Sardinia Plio-Pleistocene mafic alkalic lavas were preceded by widespread calc-alkaline magmatism between the late Oligocene and Miocene. Presumably these earlier magmas incorporated any available fusible exotica during their uprise through the SCLM, so that the latter behaved in a refractory way during the subsequent magmatism. C a m e r o o n line

The Tertiary alkaline magmatism of the Cameroon line, west Africa is an excellent example of the effect of local magmatic pre-history in determining whether or not a substantial melt fraction from lithospheric sources enters continental extension-related magmas. Fitton & Dunlop (1985) showed that, when averaged, the basic magmas along the continental and oceanic seg-

84

S.A. GIBSON E T A L .

ments of this volcanic zone were identical in both elemental and isotopic composition. Halliday et al. (1988) subsequently found subtle trends in the radiogenic isotope systematics of the continental segment lavas and attributed them to a contribution from a lithospheric mantle source in the magmas. Nevertheless, they also emphasize the remarkably 'oceanic' overall chemistry of the Cameroon line continental lavas; for example only the Potrillo-Palomas and Espanola Basin lavas of the Rio Grande rift have values of 143Nd/l'~Nd in the same range (Fig. 7). The crust in this part of west Africa was extensively re-worked during the Pan-African tectonic event (500-600 Ma) but contains remnants of earlier crust. Bea et al. (1988) reported extensive Siluro-Devonian basaltic magmatism associated with rifting in northern Cameroon. Incompatible element ratios in these olivine tholeiites contrast'with those of subsequent Tertiary volcanics and show similar features (e.g. relatively high Ba/Nb and La/Nb) to the low 143Nd/ 144Nd ratios of the Rio Grande rift magmas. During the Cretaceous, the major rift of the Benue Trough formed at an average distance of only 300 km to the northwest of the Cameroon Line. Extensive magmatism, much of it basaltic, accompanied the formation of the Benue Trough (Baudin et al. 1988) and more may be hidden beneath its thick sedimentary infill. If lithospheric extension across the major Benue Trough rift was on the same scale as across other major currently active continental rifts, it will have thinned the lower part of the African plate across a zone at least 700 km wide approximately centred on the rift axis. In this context, it is interesting to note that Tertiary magmatism has continued around the Benue Trough, in a zone approximately 750 km wide. The physical model we outlined above would be sensitive to lithospheric thinning. The SCLM overlying the attenuated outer zone might be expected to become purged of its fusible potassic fraction that accumulated between the Devonian and Cretaceous, thus permitting the OIB-like asthenospheric-source, Cameroon Line magmas to reach the surface. Our model resembles that of Halliday et al. (1988, Fig. 8) but extends it back to include the Palaeozoic tectonomagmatic history of the area.

Conclusions Like most continental basalts, rift-related basalts are dominated by lithospheric and asthenospheric components. In the Rio Grande rift, asthenosphere-dominated marie magmatism is restricted to the central rift axis, but it does

not occur everywhere, and its presence is not simply related to the amount and rate of continental extension. One of the major controlling factors that determines mantle source contributions to magma composition is earlier magmatism. We have shown that in the Rio Grande rift asthenosphere-derived magmas are present throughout the development of the rift but they are only evident in areas that had previously experienced either pre-rifl or rift-related voluminous volcanic activity. Continuous exposure across the axis of the Rio Grande rift has revealed that asthenosphere-derived melts are present during the early stages of continental break-up but at continental margins similar magmas may have been obscured by the deposition of later sediments. Our model for the evolution of the rift, based on geochemical data, agrees with recent geophysical studies that invoke large amounts of lithospheric extension and asthenospheric upwelling at the beginning of rifting. Previous geochemical studies have proposed that large scale asthenospheric upwelling did not take place beneath the rift axis until < 10 Ma. The rather sparse magmatism of the Rio Grande rift is probably a good sample of the extent and composition of the igneous products that lie deeply buried beneath later sediments on the subsided passive margins of continents, that split outside the immediate vicinities of mantle plumes. The pattern of temporal and spatial variations in magma sources that we observed in the rift may potentially apply quite well to the low intensity magmatism of other rifts developed over asthenosphere with normal potential temperatures. Nevertheless, this application may be rather more complex in cases where continental extension has occurred above the head of a mantle plume, leading to high-volume magmatism. This project was funded by the NERC (UK) (grant GR3/5299) and additional financial support from the University of Durham. We would like to express our gratitude to G. A. Izett, P. W. Lipman, D. L. Peck and R. A. Thompson (US Geological Survey) and E. E. Larson (University of Colorado) for their continuous support during this project. I. W. Sinclair, B. Bennet, R. Hardy and J. Wiodarczyk provided assistance with chemical analyses. We wish to thank J. G. Fitton and R. Ellam for reviewing this manuscript.

References ALDRICH, M. J., JR., CHAPIN,C. E. & LAUGHLIN,A. W. 1986. Stress history and tectonic development of the Rio Grande rift, New Mexico. Journal of Geophysical Research, 91, 6199-6211. ATWA~R, T. 1970. Implications of plate tectonics for

RIO GRANDE RIFT MAGMATISM the Cenozoic tectonic evolution of western North America. Geological Society of America Bulletin, 81, 3513-3536. BAt.DnIDOE,W. S. 1979. Petrology and petrogenesis of Plio-Pleistocene basaltic rocks from the central Rio Grande rift, New Mexico, and their relation to rift structure. In: RIEKER, R. E. (ed.) Rio Grande rift: Tectonics and Magrna~m. American Geophysical Union, Washington, 323-353. , DAMON, P. E., SHAFIQULLAHM. & BRIDWELL, R. J. 1980. Evolution of the central Rio Grande rift, New Mexico: New potassium-argon ages. Earth and Planetary Science Letters, $1,309-321. --, OLSEN, K. H. & CALLENDER, J. F. 1984. Rio Grande rift, problems and perspectives. New

Mexico Geological Society Field Conference Guidebook, 35, 1-12. , PERRY, F. V. & SHAFIQULLAH,M. 1987. Late Cenozoic volcanism of the southeastern Colorado Plateau. 1. Volcanic geology of the Lucero area, New Mexico. Geological Society of America Bulletin, 99, 463-470. BALTZ, E. H. 1978. Resume of Rio Grande depression in north-central New Mexico. Circular of the New

Mexico Bureau of Mines and Mineral Resources, 163, 210-228. BARKER, D. S. 1979. Cenozoic magmatism in the Trans pecos province: Relation to the Rio Grand rift. In: RIEKER,R. E. (ed.) Rio Grande rift: Tectonics and Magmatism. American Geophysical Union, Washington, 382-392. 1987. Tertiary alkaline magmatism in Trans Peeos Texas. In: FITI'ON,J. G. & UPTON,B. G. J. (eds) Alkaline Igneous Rocks. Geological Society, London, Special Publication, 30, 415-431. BAUDIN, P., PoPoFF, M. & COULON, C. 1988. Mesozoic magmatism of the southern part of the Benue Trough (Nigeria). Proceedings oi: the In-

ternational Geological Correlation Project, 227, Giens, France. Mesozoic to present-da~ magmatism of the African plate and its structi~ral setting, 34-38. BEA, A., COCHEME,A. B., KAMPUNZU,A. B. & TROMPETTE, R. 1988. The Paleozoic lavas from northern Cameroon and the Tertiary magmatism of the Cameroon line: comparative geochemistry and source implications. Proceedings of the In-

ternational Geological Correlation Project, 227, Giens, France. Mesozoic to present-day magmatism of the African plate and its structural setting, 45-48. BEST, M. G. & CHRISTIANSEN,E. H. 1991. Limited extension during peak Tertiary volcanism, Great Basin of Nevada and Utah. Journal of Geophysical Research, 96, 13509-13528. BRANDON, A. D. & GOLES, G. G. 1988. A Miocene subcontinental plume in the Pacific Northwest: geochemical evidence. Earth and Planetary Science Letters, 88, 273-283. BmDWELL, R. J. 1976. Lithospheric thinning and the late Cenozoic thermal and tectonic regime of the northern Rio Grande rift, Vermejo Park.

New Mexico Geological Society Guidebook, 27, 283-228. BussoD, G. Y. A. & WXLLIAMS,D. R. 1991. Thermal

85

and kinematic model of the southern Rio Grande rift: interferences from crustal and mantle xenoliths from Kilbourne Hole, New Mexico, Tectonophysics, 197, 373-389. CHAPIN,C. E. & SEAOER,W. R. 1975. Evolution of the Rio Grande rift in the Socorro and Las Cruces areas. Field Conference Guidebook, New Mexico Geological Society, 26, 297-321. Ch~RCm, A. & MO~ADERT, L. 1982. Oligo-Miocene rift of Sardinia and the early history of the Western Mediterranean Basin. Nature, 298, 736-739. CHRIS'nANSEN,R. L. & LIPMAN,P. W. 1972. Cenozoic volcanism and plate-tectonic evolution of the western United States. II. Late Cenozoic.

Philosophical Transactions of the Royal Society of London, A271, 249-284. CONDm, K. 1982. Plate tectonic model for Proterozoic continental accretion in the southwestern United States. Geology, 10, 37-42. CONEY, P. J. & REYNOLDS, S. J. 1977. Cordilleran Benioff z~ones. Nature, 270, 403-406. COlU:~ELL, L. 1982. Extension in the Rio Grande rift. Journal of Geophysical Research, 87, 8561-8569. , ZORIN, Y. A. & KELLER, G. R. 1991. The decompensative gravity anomaly and deep structure of the region of the Rio Grande rift. Journal of Geophysical Research, 96, 6557-6568. DALHEIM, H. A. DAVIS, P. & ACHAUER, U. 1989. Teleseismic investigation of the East African rift - Kenya. Journal of African Earth Sciences, 8, 461-470. DAvis, P. M. 1991. Continental rift structures and dynamics with reference to teleseismic studies of the Rio Grande and East African rifts. Tectonophysics, 197, 309-325. , PARKER, E. C., EVANS, J. R., IYER, H. M. & OLSEN, K. H. 1984. Teleseismic deep sounding of the velocity structure beneath the Rio Grande

rift. New Mexico Geological Society Guidebook, 35, 29-38. DECKER, E. R., BUCHER, G. J., BUELOW, K. L. HEASLER, H. P. 1984. Preliminary interpretation of heat flow and radioactivity in the Rio Grande rift zone in central and northern Colorado.

New Mexico Geological Society Guidebook, 35, 45-50. DRAPER, D. S. 1991. Late Cenozoic bimodal magmatism in the northern Basin and Range province of Southeastern Oregon. Journal of Volcanology and Geothermal Research, 45, 299-328. DUDAS, F. O. 1991. Geochemistry of igneous rocks from the Crazy Mountains, Montana, and tectonic models for the Montana alkalic province. Journal of Geophysical Research, 96, 13261-13277. , CARLSON, R. W. & EOOLER, D. H. 1987. Regional middle Proterozoic enrichment of the subcontinental mantle source of igneous rocks from central Montana. Geology, 15, 22-25. DUNKER, K. E., WOLFF, J. A., HARMON,R. S., LEAT, P. T., DICKIN, A. P. & THOMPSON, R. N. 1991. Diverse mantle and crustal components of the NW Cerros del Rio volcanic field, Rio Grande rift, New Mexico. Contributions to Mineralogy and Petrology, 108, 331-345. DUNOAN, M. A., LINDSTROM,M. M., MCMmLAN, N.

86

S . A . GIBSON ETAL.

J., MOORBATH, S., HOEFS, J. & HASKIN, L. A. 1986. Open system magmatic evolution of the TaGs Plateau volcanic field, northern New Mexico 1. The petrology and geochemistry of the Servilleta basalt. Journal of Geophysical Research, 91, 5999-6028. EATON, G. P. 1986. A tectonic redefinition of the southern Rocky Mountains. Tectonophysics, 132, 163-193. 1987. Topography and origin of the southern Rocky Mountains and Alvarado Ridge. In: COWARD, M. P. DEWEY, J. F. & HANCOCK,P. L. (eds) Continental Extensional Tectonics. Geological Society, London, Special Publication, 28, 355-369. EGGLER, D. H., MEEN, J., WELT, F., DUDAS, F. O., FURLONG, K. P., MCCALLUM,M. E. & CARLSON, R. W. 1988. Tectonomagmatism of the Wyoming Province. Colorado School of Mines Quarterly, 83, 25-40. FrrroN, J. G. & DUNLOP, H. M. 1985. The Cameroon line, West Africa, and its bearing on the origin of oceanic and continental alkali basalt. Earth and Planetary Science Letters, 72, 23-38. ~, JAMES, D., KEMPTON,P. D., ORMEROD, D. S. & LEEMAN, W. P. 1988. The role of lithospheric mantle in the generation of late Cenozoic basic magmas in the western United States. Journal of Petrology, Special Lithosphere Issue, 331-349. , - - & LEEMAN, W. P. 1991. Basic magmatism associated with late Cenozoic extension in the western United States: compositional variations in space and time. Journal of Geophysical Research, 96, 13693-13711. FRASER, K. J. 1987. Petrogenesis of kimberlites from

South Africa and lamproites from western Australia and North America. PhD thesis, Open University. G^NS, P. B., MAHOOD, G. A. & SC.ERMER, E. 1989.

Synextensional magmatism in the Basin and Range Province; A case study from the eastern Great Basin. Geological Society of America Special Paper, 233. GIBSON, S. A., THOMPSON,R. N., LEAT, P. T., MORRISON, M. A., HENDRY, G. L. & DICKEN, A. P. 1991. The Flat Tops volcanic field NW Colorado 1: Lower Miocene open system multi-source magmatism at Flander, Trapers Lake. Journal of Geophysical Research, 96, 13609-13627. , --, --, , , - - & MITCHELL,J. G., in press. Ultrapotassic magmas along the flanks of the Oiigo-Miocene Rio Grande rift, USA: monitors of the zone of lithosphere extension beneath a continental rift. Journal of Petrology. GLAZNER, A. F. & BARTLEY, M. J. 1984. Timing and tectonic setting of Tertiary low angle normal faulting and associated magmatism in the southwestern United States. Tectonics, 3,385-396. GOLOMBEK, M. P., McGmL, G. E. & BROWN, L. 1983. Tectonic and geologic evolution of the Espanola Basin, Rio Grande rift: structure, rate of extension, and relation to the state of stress in the western United States. Tectonophysics, 94, 483-507. GRAND, S. P. 1987. Tomographic inversion for

shear velocity beneath the North American Plate. Journal of Geophysical Research, 92, 14065-14090. I-D,OGERTV, S. E. 1989. Upper mantle opaque mineral stratigraphy and the genesis of metasomatites and alkali-rich melts. In: DANCmN, R. V. (ed.)

Proceedings of the Fourth International Kimberlite conference, Perth 2. Geological Society Australia, Special Publication, 14, 687-699. HALLIDAY, A. N., DICKIN, A. P., FALLICK, A. E. & Frta'ON, J. G. 1988. Mantle dynamics: A N d , Sr, Pb and O isotope study of the Cameroon line volcanic chain. Journal of Petrology, 29, 181-211. HAWKESWORTH, C. J., GALLAGHER, K., KELLY, S., MANTOVANI, M., PEATE, D. W., REGELOUS, M. & ROGERS, N. W. 1992. Parand magmatism and the opening of the south Atlantic. This volume. HENRY, C. D. & PRICE, J. G. 1984. Variations in caldera development in the Tertiary volcanic field of the Trans Pecos Texas. Journal of Geophysical Research, 71, 6119-6126. INGERSOLL, R. V., CAVAZZA, W., BALDRIDGE,W. S. & SHAFIQULLAH,M. 1990. Cenozoic sedimentation and paleotectonics of north-central New Mexico: Implications for initiation and evolution of the Rio Grande rift. Geological Society of America Bulletin, 102, 1280-1296. IYER, H. M. & HITCHCOCK, T. 1989. Upper mantle velocity structure in the continental U.S. and Canada. Geological Society of America Memoir, 172, 681-710. IZE'I-r, G. A. 1975. Late Cenozoic sedimentation and deformation in northern Colorado and adjoining areas. Geological Society of America Memoir, 144, 179-209. JAMES, E. W. & HENRY, C. D. 1991. Compositional changes in the southern Cordillera, Texas and Mexico: transition from continental arc to intraplate extension. Journal of Geological Research, 96, 13561-13575. JOHNSON, C. M. & THOMPSON, R. A. 1991. Isotopic composition of Oligocene mafic volcanic rocks in the northern Rio Grande rift: evidence for contributions of ancient intraplate and subduction mago matism to evolution of the lithosphere. Journal of Geophysical Research, 96, 13593-13608. KALrrz, P. F., INGERSOLL,R. V., BALDRIDGE,W. S., DAMON, P. E. & SHAFIOULLAH,M. 1981. Geology of the Espinaso Formation (Oligocene), north-central New Mexico. GeologicalSociety of America Bulletin, 92, 980-983. KEMPTON, P. D., FITTON, J. G., HAWKESWORTH,C. F. & ORMEROD, D. S. 1991. Isotopic and trace element constraints on the composition and evolution of the lithosphere beneath the southwestern United States. Journal of Geophysical Research, 96, 13713-13735. KELLER, G. R., BRAmE, L. W. & MORGAN, P. 1979. Crustal structure, geophysical models and contemporary tectonism of the Colorado Plateau. Tectonophysics, 61,131-147. ~ . , KHAN, M. A., MORGAN, P., WENDLANDT,R. F., BALDRIDGE, W. S., OLSEN, K. H., PRODEHL, C. & BRA1LE, L. W. 1991. A comparative study of

RIO GRANDE RIFT MAGMATISM the Rio Grande and Kenya rifts. Tectonophysics, 197, 355-371. KELLEY, V. C. 1978. Geology of the Espanola Basin, New Mexico. New Mexico Bureau of Mines and Mineral Resources. LARSON, E. E., OZIMA, M. & BRADLEY,W. C. 1975. Late Cenozoic basin volcanism in northwestern Colorado and its implications concerning tectonism and the origin of the Colorado river system. Geological Society of America Memoir, 144, 155-178. LEAT, P. T., THOMPSON,R. N., DICKIN,A. P., MOmUSON, M. A. & HENDRY, G. L. 1989. Quaternary volcanism in northwestern Colorado: Implications for the roles of asthenosphere and lithosphere in the genesis of continental basalts.

Journal of Volcanology and Geothermal Research, 37, 291-310. - - , MORRISON,M . A . , H E N D R Y , G . L . & DIcKIN, A. P. 1988. Compositionally diverse Miocene-Recent rift-related magmatism in northwest Colorado: Partial melting and mixing of magmas from 3 different asthenospheric and lithospheric mantle sources. Journal of Petrolog),, Special Lithosphere Issue, 351-377. , , , - &~ 1990. Geochemistry of mafic lavas in the early Rio Grande rift, Yarmony Mountain, Colorado, U.S.A. Chemical Geology, 81, 23-43. , , , -& 1991. Alkaline hybrid mafic magmas of the Yampa area, NW Colorado, and their relationship to the Yellowstone mantle plume and lithospheric mantle domains. Contributions of Mineralogy and Petrology, 107,310-327. Ln~tAN, P. W. 1980. Cenozoic volcanism in the western United States: implications for continental tectonics. In: Studies in Geophysics, Continental Tectonics. National Academy of Science, Washington, D.C. 161-174. & MEtmERT, H. H. 1975. Late Cenozoic volcanism and development of the Rio Grande depression in the southern Rocky Mountains. ,

Geological Society of America Memoir, 144, 119-154. , PROSTr~, H. J. & CRms'nANSEN, R. L. 1972. Cenozoic volcanism and plate-tectonic evolution of the western United States. 1. Early and middle Cenozoic. Philosophical Transactions of the Royal Society of London, A217, 217-248. LUEDKF, R. G. & SMn'H, R. L. 1978. Map showing distribution, composition and age of late Cenozoic volcanic centers in Arizona and New Mexico. Geological Society of America Bulletin, 83, 1335-1344. MANLEY, K. 1979. Stratigraphy and structure of the Espanola Basin, Rio Grande rift, New Mexico. In: RIEKER,R. E. (ed.) Rio Grande rift: Tectonics and Magmatism. American Geophysical Union, Washington, 71-86. & MEHNERT, H. H. 1981. New K-Ar ages for Miocene and Pliocene volcanic rocks in the northwestern Espanola Basin and their relationship to the history of the Rio Grande rift. Isochrord west, 30, 5-8.

87

MAY, S. J. 1984, Exhumed mid-Miocene volcanic field in the Tesque Formation, northern Espanola Basin, New Mexico. New Mexico Geological Society Guidebook, 35, 171-178. MClNTOSH, W. C. 1991. Evaluation of paleomagnetism as a correlation criterion for Mogolion-Datil ignimbrites, southwestern New Mexico. Journal of Geophysical Research, 96, 13459-13483. MCKENZJE, D. P. 1978. Some remarks on the development of sedimentary basins. Earth and Planetary Science Letters, 40, 25. 1984. A possible mechanism for epeirogenic uplift. Nature, 307, 616-618. 1989. Some remarks on the movement of small melt fractions in the mantle. Earth and Planetary Science Letters, 95, 53-72. & BiCKLE,M. J. 1988. The volume and composition of melt generated by extension of the lithosphere. Journal of Petrology, 29, 625-679. & O'NIONS, R. K. 1991. Partial melt distributions from inversion of rare-earth element concentrations. Journal of Petrology, 32, 1021-1091. MCMILLAN, N. J. & DUNGAN, M. A. 1986. Magma mixing as a petrogenetic process in the development of the Taos Plateau volcanic field, New Mexico. Journal of Geophysical Research, 91, 6029-6045. & 1988. Open system magmatic evolution of the Taos Plateau volcanic field, northern New Mexico: Andesites and dacites. Journal of Petrology, 29, 527-557. Mr~sN, J. K. & EC,GLER, D. H. 1987. Petrology and geochemistry of the Cretaceous Independence volcanic suite, Absaroka Mountains, Montana: clues to the composition of the Archean subMontanan mantle. Geological Society of America Bulletin, 98, 238-247. MFXssNER, R. 1986. The continental crust. International Geophysics series, 34. Academic Press, London. MFJ,izI~, M. A. 1992. The lower lithosphere as a major source for continental flood basalts: a reappraisal. This volume. MOROAN, P. & GOSNOLD, W. D. 1989. Heat flow and thermal regimes in the continental United States. Geological Society of America Memoirs, 172, 493-522. , SEAGER, W. R. & GOLOMBEK, M. 1986. Cenozoic thermal, mechanical and tectonic evolution of the Rio Grande rift. Journal of Geophysical Research, 91, 6263-6276. MORmsoN, M. A., HENDRY,G. L. & LEAT, P. T. 1987. Regional and tectonic implications of parallel Caledonian and Permo-Carboniferous lamprophyre dyke swarms from Lismore, Ardgour.

Transactions of the Royal Society of Edinburgh: Earth Sciences, 77, 279-288. NELSON, D. O. & NELSON, K. L. 1987. Geochemical comparison of alkaline volcanism in oceanic and continental settings; Clarion Island versus the eastern Trans Pecos magmatic province. In: MORR,S, E. M. & PASTERlS,J. D. (eds). Mantle Metasomatism and Alkaline Magmatism. Geological Society of America Special Publication, 215, 317-333.

S . A . GIBSON ETAL.

88

NORMAN, M. D. & MERTZMAN, S. A. 1991. Petrogenesis of Challis volcanics from central and southwestern Idaho: trace element and Pb isotope evidence. Journal of Geophysical Research, 96, 13279-13293. O'BRIEN, H. E., IRVING,I. J. & MCCALLUM,I. S. 1991. Eocene potassic magmatism in the Highwood Mountains, Montana: Petrology, geochemistry and tectonic implications. Journal of Geophysical Research, 96, 13237-13260. O'HARA, M. J. & MATHEWS, R. E. 1981. Geological evolution in an advancing, periodically tapped, continuously fractionating magma chamber.

Journal of Geological Society, London, 138, 237-277. OLSEN, K. H., BALDRIDGE, W. S. & CALLENDER, J. F. 1987. Rio Grande rift: an overview. Tectonophysics, 143, 119-139. PANZA, G. F. 1985. Lateral variations in the lithosphere in correspondence of the southern segment of EGT. Proceedings of the 2nd workshop on the European Geotraverse Project. European Science Foundation, Strassbourg, 47-51. PARKER,E. C., DAVIS,P. M., EVANS,J. R., IYER, H. M. & OLSEN, K. H. 1984. Upwarps of anomalous asthenosphere beneath the Rio Grande rift. Nature, 312, 354-356. PERRY, F. V., BALDRIDGE, W. S. ,g' DEPAOLO, D. J. 1987. Role of asthenosphere and lithosphere in the genesis of late Cenozoic basaltic rocks from the Rio Grande rift and adjacent regions of the southwestern United States. Journal of Geophysical Research, 92, 9193-9213. PERUZZA,L., FISCHER,G., SCHEGG,P. A. & RANIERI, G. 1990. Analysis of the Sardinian north-south magnotelluric profile. Results from the 5th study

centre, the European Geotraverse integrative studies. European Science Foundation, Strassbourg, 329-346.

PRICE,J. G., HENRY,C. D., BARKER,D. S. & PARKER, D. F. 1987. Alkalic rocks of contrasting tectonic settings in Trans Pecos Texas. In" MORRIS, E. M. & PASTERIS, J. D. (eds) Mantle Metasomatism and Alkaline Magmatism. Geological Society of America Special Publication, 215, 333-346. PRODEHL, C. & LIPMAN, P. W. 1989. Crustal structure of the Rocky Mountain Region. Geological Society of America Memoirs, 172, 249-284. & GIESE, P. 1990. Seismic investigations around the EGT in central Europe. Results from tha 5th

study centre, the European Geotraverse integrative studies. European Science Foundation, Strassbourg, 77-97. & PAKISER,L. C. 1980. Crustal structure of the southern Rocky Mountains from seismic measurements. Bulletin of the Seismological Society of America, Part. 1, 91,147-155. REID, J. R. & WOODS, G. A. 1978. Oceanic mantle beneath the southern Rio Grande rift. Earth and Planetary Science Letters, 41,303-317. RUTrER, M. J. 1987. Evidence for crustal assimilation by turbulently convecting, marie alkaline magmas: Geochemistry of mantle xenolith-bearing lavas from northern Sardinia. Journal of Volcanology and Geothermal Research, 32,343-354.

SEAGER, W. R., SHAFIOULLAH,M., HAWLEY,J. W. & MARVIN, R. F. 1984. New K/Ar dates from basalts and the evolution of the southern Rio Grande rift. Geological Society of America Bulletin, 95, 87-99. SENGOR, A. H. C. • BURKE, K. 1978. Relative timing of rifting and volcanism on earth and its tectonic implications. Geophysical Research Letters, 5, 419-421. SINNO, Y. A., DAGGET, P. H., KELLER, G. R., MORGAN, P. & HARDER, S. H. 1986. Crustal structure of the southern Rio Grande rift determined from seismic refraction profiling. Journal of Geophysical Research, 91, 6143-6156. SNYDER, W. S., DICKINSON,W. R. & SILBERMAN,M. L. 1976. Tectonic implications of space-time patterns of Cenozoic magmatism in the western United States. Earth and Planetary Science Letters, 32, 91-106. STEIGER, R. H. & JAGER, E. 1977. Convention on the use of decay constants in geo- and cosmochronology. Earth and Planetary Science Letters, 86, 359-362. STEWART, S. W. & PAKISER, L. C. 1962. Crustal structure in eastern New Mexico interpreted from the Gnome explosion. Bull,.tin of the Seismological Society of America, 52, :017-1030. SVERINGHAUS, J. & ATWATER, T. 1990. Cenozoic geometry and thermal state of the subducting slabs beneath western North America. In: WERNICKE, B. P. (ed.) Basin and Range extensional

tectonics near the latitude of Las Vegas, Nevada. Geological Society of America Memoir, 176, 1-22. TAYLOR,W. J., BARTLEY,J.M., Lux, D. R. & AXEN, G. J. 1989. Timing to Tertiary extension in the Railroad valle~-Pioche transect, Nevada: constraints from AR/ 39Ar ages of volcanic rocks. Journal of Geophysical Research, 94, 7757-7774. THOMPSON,R. A., JOHNSON,C. M. & MEHNERT,H. ,H. 1991. Oligocene Basaltic volcanism of the northern Rio Grande rift: San Luis Hills, Colorado. Journal of Geophysical Research, 96, 13577-13592. THOMPSON, R. N. & GIBSON, S. A. 1991. Subcontinental mantle plumes, hotspots and pre-existing thinspots. Journal of the Geological Society, London, 148, 973-977. - - , LEAT, P. T., MORRISON, M. A., HENDRY, G. L., DiCKIN, A. P. & MITCHELL,J. G., in press. Lamproitic drip-feed into early Miocene continental extension-related mafic magmatism: Walton Peak, northwest Colorado. Journal of the

Geological Society, London. , LEAT,P. T., MORRISON,M. A., HENDRY,G. L.,

~,

DICKIN, A. P. & GIBSON, S. A. 1990. Strongly potassic mafic magmas from lithosphere mantle sources during continental extension and heating: evidence from Miocene minettes of northwest Colorado, U.S.A. Earth and Planetary Science Letters, 38, 139-153. MORRISON, M. A., DICKIN,A. P. & HENDRY,G. L. 1984. An assessment of the relative roles of crust and mantle in magma genesis: an elemental approach, Philosophical Transactions of the

RIO GRANDE RIFT MAGMATISM

Royal Society of London, A310, 549-590. TINGEY, D. G., CHRISTIANSEN,E. H., BEST, M. G., Rulz, J. & Lux, D. R. 1991. Tertiary minette and melanephelinite dikes, Wasatch Plateau, Utah: records of mantle heterogeneities and changing tectonics. Journal of Geophysical Research, 96, 13529-13544. ~ , O. 1979. The Rio Grande rift system in Colorado. In: RIEKER, R. E. (ed.) Rio Grande rift: Tectonics and Magmatism. American Geophysical Union, Washington, 33-56. WERNICKE,B. P., CHRISTIANSEN,R. L., ENGLAND,P. C. & SONDER, L. J. 1987. Tectonomagmatic evolution of Cenozoic extension in the North American Cordillera. In: COWARD, M. P.,

89

DEWEY, J. F. & HANCOCK,P. L. (eds). Continental Extension Tectonics. Geological Society, London, Special Publication, 28, 203-221. WILKINSON,P., MITCHELL,J. G., CATYERMOLE,P. J. & DOWNIE, C. 1986. Volcanic chronology of the Meru-Kilimanjaro region, northern Tanzania. Journal of the Geological Society, London, 143, 601-605. WILSON, M. 1989. Igneous Petrogencsis. Unwin Hyman, London. WORNER, G., ZINDLER, A., STAUDIGEL, H. & SCtrMINCKE, H. U. 1986. Sr, Nd and Pb isotope geochemistry of Tertiary and Quaternary alkaline volcanics from West Germany. Earth and Planetary Science Letters, 79, 107-119.

Episodic alkaline igneous activity across Africa: implications for the causes of continental break-up D. K. B A I L E Y

Department o f Geology, University o f Bristol, Bristol BS8 1R J, UK Except for the northwestern and southern extremities, the African plate has been anorogenic throughout the Phanerozoic, yet the interior has been subject to spectacular episodes of uplift, rifting and magmatism from Mesozoic to present. Interior Africa thus offers an excellent test of intraplate dynamics as prototypes for the initiation of continental break-up. Crucial to the test are the spatial and temporal controls on the activity. Spatially, there is unequivocal evidence that the disposition of the activity is controlled by pre-exist!ng lithosphere structures. From Cretaceous to Recent times, alkaline magmatism (including carbonatites and kimberlites) has been the norm, typically in uplifted and rifted areas. Four major periods of activity emerge in the igneous record. The classic Early Cretaceous rifting and magmatism of central and southern Africa (135-110 Ma) is only one part of an igneous episode widely recorded across the continent. A Late Cretaceous episode (90-80 Ma), best known for kimberlite activity in the old cratons, is also widely registered elsewhere. Two major Tertiary episodes characterize the modern East African Rift zone, an initial EoceneOligocene (c. 40 Ma) and a Miocene-Recent episode (starting around 23 Ma), and corresponding peaks are recorded in many other areas, especially in the northern part of the continent. In most areas there is evidence for more than one episode, indicating that since the Cretaceous the same sites have been subject to repeated alkaline magmatism. Contemporaneous plate-wide activity, repetition at individual sites, and localization by existing anisotropies in the plate, are all indicative of responses to external events capable of affecting the whole plate. Such events would include continental collision, stages in Gondwana break-up, and marked changes in global plate motions. Spatial and temporal relations thus show that rifting and alkaline magmatism are permissive, and may themselves be symptoms of more fundamental processes that cause the break-up of continents. If so, pre-Cretaceous episodes of intra-plate alkaline magmatism may offer valuable time-markers, recording similar events earlier in the Earth's history. One possible example may be the global episode of kimberlite activity recorded around 1200 Ma. Abstract:

When continents break-up they are manifestly rifting apart and in many places it is clear that the n e w continental margin follows an ancient lineament in the lithosphere, e.g. the Kenya-Somalia margin of Africa (Bailey 1961). Where there is associated magmatism, it invites the question whether break-up is permissive or driven by hot mantle rising below the lithosphere. For the continental flood basalts of south Africa, Cox (this volume) argues for a large scale mantle plume operating during break-up, while Bertrand (1991) has concluded that the contemporaneous tholeiites in west Africa (and N. America) are expressions of permissive magmatism associated with opening of the Atlantic. These are cases linked with continental separation, but the same alternatives, active or permissive (passive), are also applicable to intra-continental rifting and magmatism as a possible prototype for conditions leading to continental break-up (summaries in Condie 1982; Park 1988). In this regard, interior Africa may provide a crucial record because it offers a wide display of uplift, rifting and magmatism during the Cretaceous, and again

during the T e r t i a r y - R e c e n t , across a region devoid of orogeny for the whole of the Phanerozoic (and over wide areas for a much longer period). The magmatism in African rift and extension zones is essentially alkaline, (following the criteria recommended by Serensen 1974) with either feldspathoidal or peralkaline characteristics. In some cases, transitional basalts may be involved and then the association includes trachytes a n d peralkaline rhyolites. Three instances of rift tholeiite are noted by Kampunzu & Mohr (1991), but the volumes are small, and they are evidently not characteristic as they are not reported elsewhere. On the other hand, carbonatites and their ultra-alkaline associates find their most spectacular development in Africa. Together with a persistent peralkaline character across the silicate magma range ( n e p h e l i n i t e rhyolite) they form one of the hallmarks of the activity. Hence, although major outpourings of tholeiite in south and west Africa may mark the initiation of G o n d w a n a break:up, since that time alkaline magmas have been erupted, in several episodes, at many areas scattered widely

From STOREY,B. C., ALABASTER,T. & PANKHURST,R. J. (eds), 1992, Magmatism and the Causes of Continental Break-up, Geological Society Special Publication No. 68, pp. 91-98.

91

92

D.K. BAILEY

across the African interior. Whether such magmatism is permissive, or results from deep mantle plumes impinging on the base of stable lithosphere may be judged against this pattern of activity. Essentially, what are at issue are the control and timing of movement and magmatism within an anorogenic plate. The prime concern in this paper is timing. Control of location has been covered previously, and is widely accepted, and is therefore summarized first.

Spatial controls It has long been recognized that continental rifts tend to follow the old structural fabric of the plate (McConnell 1951). McConnell's observations on the rift and shield structure of East Africa were widened in scope and extent by Black & Girod (1970) and Thorpe & Smith (1974) who pointed out that all Phanerozoic anorogenic alkaline magmatism in Africa was restricted to regions affected by the Pan-African tectono-thermal event (650-550 Ma). Subsequent observers have endorsed this view, notably Cahen et al. (1984) covering African geochronology, and Kampunzu & Popoff (1991) in the introductory Chapter to a recent text on extensional magmatism in Africa (Kampunzu & Lubala 1991). Their structural map of Africa (Fig. 1) is an excellent graphic realization of the relationship. In every other contribution to that volume, in which tectonic framework is considered, the magmatism is connected to earlier structural and/or thermal (Pan-African) anisotropics in the African plate. To this may be added, the latest seismic study of the Kenya Rift (Prodehl et al. 1991) which identifies the western margin as coinciding with the pre-existing boundary between the Archean and the PanAfrican belt. If all this anorogenic igneous activity across Africa is located by lithosphere anisotropies, it follows that the magmatism must have its site controlled by the lithosphere (Bailey 1961, 1977). Essentially, old zones of weakness provide channels for the release of fluids and melts from the lithosphere and sub-lithosphere mantle (Bailey 1983). Repeated episodes of activity (in some areas extending to the Precambrian) record periodic re-opening of the same lesions in the lithosphere, and the timing may offer the best clue as to the causes of lithosphere re-activation.

Time relations If the lithosphere acts as a template for the periodic release of mass and heat, then the old

lesions must be yielding under membrane stresses applied to the plate. Igneous chronology may therefore be vital in identifying the causes of the membrane stresses, especially if any correlation can be found with external events. Two clearly defined episodes need consideration in the first instance: (1) Cenozoic activity across the northern half of Africa, and stretching through East Africa, and (2) Cretaceous magmatism in central and southern Africa. Information on all areas is updated in Kampunzu & Lubala (1991) and many are reviewed in detail: they are located in Fig. 1 and listed in Fig. 2. Cenozoic

From Fig. 2 it may be seen that Africa was magmatically 'quiet' between Late Cretaceous and Late Eocene. Relatively few dates are recorded over this time, as shown in the compilation of Cahen et al. (1984) covering the whole continent. Along the Cameroon Line, an older series of ring complexes yield dates in this period and this is the best-documented exception, but some areas in Chad and Libya are still poorly known. Even in the Cameroon, however, a renewed spate of volcanism started in Oligocene times (D6rueUe et al. 1991) and this accords with

2

¥

Fig. 1. Location map showing the main areas of alkaline igneous, carbonatite and kimberlite activity from Cretaceous-Recent. Numbers correspond with the listing in Fig. 2. Cratonic regions unaffected by the Pan-African event (650-550 Ma) are outlined with thin broken lines: rifting and fracture patterns (shown by Kampunzu & Popoff 1991) are here omitted for clarity. Locations 22-25 indicate important kimberlite areas within the cratons.

EPISODIC ALKALINE MAGMATISM IN AFRICA lax.

93 JURASSIC

14.too CANARY IS. MADEm& SFa~BOAL C. VERDE LOS HOOOAit AIR TIBSSTi UWINAT NUBIA RED SEA HILLS

1.2

l 3

ell

m

m

3.4

U

i

6 ?

?.6 ?

,=

8.7.20 ,.,

CAMEROON E. A. RIFT I~D~YA ETHIOPIA Y E I d~ W. RIFT RU~ CIIILWA LUPATA s. LEBOMBO

9 I0 I0.I I I0 7

?

ii

ii|

ta,n t___

m

1,4 14.'/

i

i

?

I$.16 ANOO(.~ DAMARA

I? ? 7.1t. 19

7,18 22 ?3

25

26

KALAHARI CONC_~ W. AFRICA TANZANIA ELUOAHI)A

m

i

tim

•.

20.21 ~.22 20,23 7.0,23

24

---.

~

•

aT . . . . . . . . . . . . . . . . . . .

.

n

~,=

='=

el

T n

m

T---T

m elm

Fig. 2. Comparative chronology of alkaline igneous, carbonatite and kimberlite activity in Africa from Cretaceous-Recent. Named areas are from those shown on the map of Kampunzu & Popoff (1991). Heavy lines indicate areas where radiometric dates are sufficient to suggest a period of continual activity. Triangles indicate peaks of activity, where suggested in the reference. Activity peaks in the Cretaceous, for a given area, have so far been specified only for Kalahari kimberlites, in other areas age ranges are broadly concordant. Thin dashes indicate ages inferred from stratigraphic evidence. The major continental flood tholeiite episodes of the Jurassic are shown for comparison: depicted by fine broken lines ending with T (between areas 21 and 22; area 24). Carbonatite/nephelinite/melilitite, or kimberlite, are so far reported from all areas except 3, 6, 10, 17 (possibly due to paucity of data). References: 1, Araria & Ortiz (1991); 2, Le B as et al. (1986); 3, Bellion & Crevola (1991); 4, Stillman et al. (1982); 5, Dautria & Girod (1991); 6, Moreau et al. (1991); 7, Cahen et al. (1984); 8, Kampunzu & Popoff (1991); 9, Drruelle et al. (1991); 10, Kampunzu & Mohr (1991); 11, Baker (1987); 12, Harkin (1960); 13, Ebinger etal. (1989); 14, Woolley & Garson (1970); 15, Turner & Rex (1991); 16, Lee (1974); 17, Liyungu & Bailey (unpubl. data); 18, Marsh (1973); 19, Prins (1981); 20, Nixon (1987); 21, Gurney et al. (1991); 22, Demaiffe et al. (1991); 23, Lubala (1991), 24, Bell & Blenkinsop (1989). the marked outbreak of'activity across the continent in Late Eocene to Early Oligocene times (c. 40 Ma). This initiation of activity, at widely scattered centres across the plate, after a relatively long period of quiessence, would seem to signal an external event affecting the whole continent, and especially its northern part (Fig. 1). In the Kenya rift the earlier activity, possibly associated with pre-rift warping, is partly obscured by the major outpourings of the Miocene (Baker 1987, Fig. 2) and Miocene peaks or changes in activity are recorded in other areas as far away as Hoggar (Dautria & Girod 1991) and the Canary Islands (Arafia & Ortiz 1991). These episodes around 20 Ma are also matched by spectacular carbonatite-nephelinite activity off-rift, in E Uganda, and may signal another plate-wide disturbance. A possibly more localized signal is in Miocene initiation of activity at Rungwe, Tanzania, which correlates with a significant change in activity in the Kenya

rift around 7 Ma (Baker 1987). Although most of the Tertiary activity is associated with rifts and extension zones it also occurs in other environments. In eastern Africa, the two phases seen in the rift are nicely registered in the carbonatite/ nephelinite magmatism in eastern Uganda (along a N - S belt to the west of the Kenya rift). A n older group of sub-volcanic complexes in south eastern Uganda give ages c. 40 Ma, while the volcanoes further north were active from c. 20 Ma (Bell & Blenkinsopp 1989). Diamondiferous kimberlite perforated the Tanzania craton c. 40 Ma (Nixon 1987) signalling the release of craton stresses coinciding with the onset of rift magmatism.

Cretaceous In many of the areas affected by Tertiary volcanism, there are records of earlier Cretaceous magmatism (Fig. 2). Repeated episodes of

94

D.K. BAILEY

alkaline magmatism are a typical feature of this activity worldwide (Barker 1974) but of particular interest in the present context, is that two peaks in Cretaceous activity appear in many places across the African plate. Early Cretaceous alkaline/carbonatite magmatism forms the classic Chilwa Igneous Province in S Malawi, (135-110 Ma, Woollev & Garson 1970) and there was magmatism of similar type and age in other parts of the postKaroo/Cretaceous rift system in Zambia (Bailey 1960, 1989; Turner & Rex 1990), Moqambique (Woolley & Garson 1970), and Angola (Allsopp & Hargraves 1985) and southwestern Africa (Marsh 1973). Of special note is the Rungwe rift intersection in S Tanzania where Chilwa age carbonatites (Snelling 1965) were followed by both Miocene and Quaternary alkaline activity in the Rungwe volcanics (Harkin 1960). Here the eastern and western branches of the modern rift cross, and the link with the older southern rift system is highlighted. Repeated carbonatite/ alkaline magmatism stretching back to the Precambrian marks this rift intersection as an outstanding case of intermittent re-opening of the lithosphere (Bailey 1961, 1977). In west and north Africa, similar activity of Early Cretaceous age has so far been recorded in Benue and the Red Sea Hills (Fig. 2). In a discussion of alkaline complexes in the continental margins of south-western Africa and Brazil/Uruguay, Marsh (1973) noted a bimodal Cretaceous age distribution in alkaline magmatism in Brazil and foresaw that late Cretaceous activity would come to light in south western Africa. Subsequently Dawson (1980) detected such a broad pattern in African kimberlite ages, but felt that there might be two peaks in the Early Cretaceous (140 and 120 Ma). More data was available for Nixon's listing (1987) where the bimodal Cretaceous distribution is clear for the Kaapvaal craton kimberlites. A compilation of the best available kimberlite ages for the Kalahari craton shows one marked peak for Group II kimberlites between 115 and 130 Ma and another, for Group I kimberlites, at 8 5 " 95 Ma (Smith et al. 1985): Group I and Group II are primarily distinguished by geochemical and isotopic criteria, and the two age peaks are indicated in Fig. 2. The first peak matches the Early Cretaceous alkaline/carbonatite activity registered elsewhere in Africa. The second peak matches the ages in Brazil, pointed out by Marsh (1973), and now known to be registered in kimberlite activity in the Congo and W. Africa cratons (Nixon 1987). As in the case of the Early Cretaceous, this Late Cretaceous kimberlite activity also correlates with alkaline/carbonatite

activity, in areas as far away as Hoggar and the Red Sea Hills (Fig. 2). The compilation of ages by Smith et al. (1985) also shows a further peak, around 70 Ma, for off-craton Group I kimberlites and melilitites. Although some dates elsewhere in Africa are close to this, more data would be needed to confirm a separate peak around 70 Ma, and it seems prudent to retain the more general view of one Late Cretaceous peak (70-90 Ma). Apart from the Kalahari craton two Cretaceous peaks of activity are registered in the Congo and W. Africa cratons, Angola, the Red Sea Hills, and the Mid-Zambezi rift. In the last area both Early and Late Cretaceous activity is recorded in the same complex (Liyungu & Bailey, unpublished data), as appears to be the case in the Red Sea Hills (Cahen et al. 1984). S u m m a r y o f the Cretaceous-Recent time relations Two important features emerge, as may be. seen in Fig. 2. (1) Alkaline magmas, carbonatites and kimberlites were erupted across Africa in bursts of activity, with two periods peaking in the Early and Late Cretaceous, followed by a marked lull, which ended in resurgences in Late EoceneEarly Oligocene and in Miocene times. (2) Nearly all areas have yielded evidence of repetition, with most showing both Cretaceous and Tertiary episodes of activity. Even on the histogram of all igneous rock ages, from 400 Ma to present, for all of Africa (Cahen et al. 1984, Fig. 22.12) the above episodes clearly emerge. It would be unavoidable that most Cretaceous-Recent samples in the total were alkaline rocks, carbonatites or kimberlites, so the high points in the histogram are another confirmation, of the punctuated nature of this magmatism in Africa. The platewide aspect implicit in the histogram, is made explicit in Fig. 2. Plate-wide synchroneity, and the repetitions of activity point to periodic reactivation across the whole plate, presumably in response to external events. Possible causes of plate-wide reactivation

Periodicity and repetition, in the generation of similar magmas at widely dispersed points across the continent, must rule out random sub-lithosphere plumes as the generation mechanism. If the magmas were the result of a sub-lithosphere convection system this would have to affect the whole of Africa. Any such hypothesis would entrain its own additional problems, not least

EPISODIC ALKALINE MAGMATISM IN AFRICA the requirement for a new convection system for each magmatic period, to take account of the intervening horizontal movement of the lithosphere. A more plausible interpretation is that periodic reactivation signals lateral forces affecting the whole plate, for which the strongest candidate would be distinct, or abrupt, changes in plate motion or configuration. This view is strengthened by the observation that after the Pan-African tectono-thermal event, the African plate was relatively quiessent for about 350 Ma until the Mesozoic, and the subsequent dramatic change in magmatic character is broadly coincident with the break-up and dispersion of Pangea. At a local, and more restricted level, three different kinds of event have been suggested as causitive mechanisms in specific igneous provinces. (1) Continental collisions. After an exhaustive consideration of the Canary Islands magmatism, Arafia & Ortiz (1991, Fig. 15) conclude that the main Tertiary volcanic phases (Oligocene; Miocene; Pliocene) closely followed the main compressional events in NW Africa (AfricaIberia collision). A similar conclusion is reached by BeUion & Crevola (1991) for the Tertiary and Quaternary activity in Senegal and Cape Verde. (2) Changes in plate motion patterns. Bellion & Crevola (1991) suggest that the Late Cretaceous magmatism registered in Cape Verde, Senegal and Los, are related to changes in plate motion directions in the 'mid-Cretaceous'. (3) Gondwana break-up. In considering the alkaline complexes of south western Africa, Marsh (1973) proposed that the Early Cretaceous eruptions of Damaraland, Luderitz and Angola (120-135 Ma) were controlled by transform lineaments activated by the initial opening of the S Atlantic. When all of Africa is considered (Fig. 2) it may be seen that each of the above cases is echoed in four period of magmatism across the continent (Early Cretaceous; Late Cretaceous; EoceneOligocene; Miocene), and it is appropriate to consider whether similar events could be responsible. Cenozoic

Obviously the collision of Africa and Europe is complex, and diachronous in different places along its length, sufficiently perhaps to deter detailed comparisons with wide magmatic episodes in Africa. But the correlation suggested by Arafia & Ortiz (1991) involving major episodes in Oligocene and Miocene, may not be

95

solely restricted to north western Africa. Park (1988) in presenting a synoptic view of Alpine orogenesis, indicates that most subsequent authors have followed Dewey et al. (1973) in identifying two major Alpine compressional phases during Tertiary times, EoceneOligocene, and Miocene. Certainly these two periods would correspond with widely recurring magmatic pulses in Africa, and a further pointer to a connection with continental collision is the marked concentration of Cenozoic volcanic activity in the provinces across the northern part of the continent. Early Cretaceous

Marsh's suggestion (1973) that the activity in south western Africa was connected with the opening of the S Atlantic, needs modification when it is appreciated that contemporaneous activity broke out across the plate, to as far away as the Red Sea Hills. Possibly all this Early Cretaceous activity is connected with some fundamental change in Gondwana break-up, but some observers have deduced that the Africa/ Europe collision was also initiated at this period (Olivet et al. 1987), in which case, braking forces on the African plate could be responsible. Of course, the two events, Atlantic opening and initiation of collision, may not be unconnected, and certainly an event large enough to affect the whole plate is needed in the Early Cretaceous. L a t e Cretaceous Current information does not show any well documented collision or break-up event that might correlate with the Late Cretaceous igneous period. As noted above, BeUion & Crevola (1991) related this to plate motion change, and Marsh (1973) referring to the Late Cretaceous activity in Brazil, pointed to a major shift in the pole of rotation of Atlantic opening at this period. Again, the plate-wide distribution of the activity seems to indicate wider changes than those along the Atlantic margin. Olivet et al. (1987) in their account of the dispersion of Pangaea, draw particular attention to a sudden change in the movement patterns of the southern continents. From 190 to 90 Ma, Australia and India were moving southeast relative to Africa, but between 90 and 80 Ma these two plates changed direction, since when they have moved 'northwards' with respect to Africa. The time of this switch in direction, called by Olivet et al. (1987) the 'hinge', coincides with the younger peak in igneous Cretaceous ages across the African continent.

96

D.K. BAILEY

One magmatic characteristic of the Late Cretaceous activity may be distinctive, namely, the wide eruption of Group I kimberlite, especially in the Kalahari and Congo cratons. This magmatism might be symptomatic of a differem type of external event in the Late Cretaceous, which further research may clarify. l n t r a - c o n t i n e n t a l rifting and mantle flow As a general mechanism for intraplate rifting and igneous activity, mantle upwelling (even permissive upwelling) is ruled out by the nature and chemical variability of the magmatism, and by its vagarious distribution (including total absence in some of the most spectacular rift sections such as Lake Tanganyika). Of course, permissive upwelling must take place when there is significant mechanical thinning of the lithosphere, as in the transition to complete separation and ocean opening, with corresponding changes in the volume and character of the volcanism: such a development may be envisaged for Afar and Ethiopia (Bailey 1983). An earlier stage may be registered in the Kenya dome, but this is part of the rift zone that has experienced sustained alkaline igneous activity for more than 23 Ma: the present thermal structure might then be partly a time-integrated product of small scale melt diapirism, not just mantle inflow. Whatever the cause, shallower asthenosphere in such regions may have a similar seismic tomography signature to that anticipated for a mantle plume, as suggested from the most recent study of the Kenya rift (Green etal. 1991), where the data are difficult to reconcile entirely with either the passive or active models. Further discussion of tomography and plumes is provided by Anderson et al. (this volume). If the mantle upwelling were active, i.e. a deep mantle plume impinging on the base of the plate, there is an additional problem in explaining the spatial control by the lithosphere. Any notion that the rift activity results from lithosphere weaknesses being exploited by a plume centred elsewhere begs the question, leaves the magmatism and rifting as permissive, and still fails to confront the forementioned problems (Bailey 1983). But the coup de grace for the idea of plume generated rift magmatism is the repetition of alkaline/ carbonatite activity, especially around rift intersections, from Precambrian through to the present (Bailey 1977).

Implications for the causes of continental break-up in alkaline activity patterns Magmas rich in alkalies and CO2 characterize intraplate extensional magmatism, and in Africa

they display three features beating directly on the question of the causes of continental breakup. The magmas are: (1) localized in zones of old weakness in the lithosphere; (2) erupted at widely dispersed areas across the continent, in specific time periods; (3) repeated in the different areas. All three factors are consistent with permissive magmatism, and the localization by old weaknesses in the lithosphere indicates that in most instances, intra-plate rifting is passive, not active. Continental rifts can therefore offer patterns of continental break-up geometry, but not a prototype for the break-up mechanism. The evidence from Africa indicates that the extension, and extensional magmatism, are responses to global events, in some instances possibly the same fundamental processes that are responsible for break-up. One of the great potentials of the ancient continents may lie in their history of alkaline, carbonatite and kimberlite magmatism, preserving long term records of global events .'The two Cretaceous episodes recognized in Africa were first noted for Brazil (Marsh 1973) and it opens up the exciting prospect that earlier episodes may be similarly marked. For instance, Skinner et al. (1985) report a worldwide record of Proterozoic kimberlite activity around 1200 Ma, which must surely signal some global changes (possibly in plate motions) over that eruptive period. Revision of the final manuscript was much assistedby constructive discussion from B. Storey.

References ALLSOPP,H. L. & HARGRAVES,R. B. 1985. Rb-Sr ages and palaeomagnetic data for some Angolan alkaline intrusives. Transactions Geological Society South Africa, 88, 295-299. ANDERSON, D. L., ZHANG, Y.-S. d~. TANIMOTO, T. 1992. Plume heads, continental lithosphere, flood basalts and tomography. This volume. AR~,t~A, V. & ORTtZ, R. 1991. The Canary Islands: Tectonics, magmatism, and geodynamic framework. In: KAMPUNZU& LUBALA(eds) op. tit., 109-249. BARLEY, D. K. 1960. Carbonatites of the Rufunsa valley, Feira District. Geological Survey of Northern Rhodesia, Bulletin 5. ..... 1961. The mid-Zambezi-Luangwa rift and related carbonatite activity. Geological Magazine, 98, 277-284. • 1977. Lithosphere control of continental rift magmatism. Journal of the Geological Society, London, 133, 103-106. 1983. The chemical and thermal evolution of rifts. Tectonophysics, 94, 585-597.

EPISODIC ALKALINE MAGMATISM IN AFRICA 1989. Carbonate melt from the mantle in the volcanoes of south-east Zambia. Nature, 388, 415-418 (and 374). BAKER, B. H. 1987. Outline of the petrology of the Kenya rift alkaline province. In: FrrroN, J. G. & UPTON, B. G. J. (eds) Alkaline Igneous Rocks. Geological Society, London, Special Publication, 30, 293-311. BARKER, D. S. 1974. Alkaline rocks of North America. In: SORENSEN, H. (ed.) The Alkaline Rocks. Wiley, London, 160-171. BELL, K. & BLENKINSOP, J. 1989. Neodymium and strontium isotope geochemistry of carbonatites. In: BELL, K. (ed.) Carbonatites. Pergamon, Oxford, 278-300. BELLION, Y. & CREVOLA, G. 1991. Cretaceous and Cainozoic magmatism of the Senegal Basin (West Africa): a review. In: KAMPUNZU& LUBULA(eds) op. cit., 189-208. BERTRAND, H. 1991. The Mesozoic tholeiitic province of northwest Africa: a volcanotectonic record of the early opening of the Central Atlantic. In: KAMPUNZU& LUBALA(eds) op. cit., 147-188. BLACK, R. & GIROD, M. M. 1970. Late Palaeozoic to Recent igneous activity in West Africa and its relationships to basement structure, In: CLIFFORD, T. N. & GASS, I. G. (eds) African magmatism and tectonics. Oliver and Boyd, Edinburgh, 185-210. CAHEN, L., SNELLING,N. J., DELHAL,J. (~ VAIL,J. R. 1984. The geochronology and evolutionof Africa. Clarendon, Oxford. CONDm, K. C. 1982. Plate tectonics and crustal evolution. Pergamon, Oxford. Cox, K. G. 1992. Karoo igneous activity and the early stages of the break-up of Gondwana. This volume. DAUTRIA, J. M. & GIROD, M. M. 1991. Relationships between Cainozoic magmatism and upper mantle heterogeneities in the Hoggar (Central Sahara). In: KAMPUNZU& LUBALA(eds) op. cit., 250-268. DAWSON, J. B. 1980. Kimberlites and their xenoliths. Springer-Verlag, Berlin. DEMAIFFE,D., FIEREMANS,M. & FIEREMANS,C. 1991. The kimberlites of central Africa: a review. In: KAMPUNZU& LUBALA(eds) op. cit., 537-559. DERUELLE, B., MOREAU, C., NKOUMBOU, C., KAMBOU,R., LISSOM,J., NJONFANG,E. & NONO, A. 1991. The Cameroon Line: a review. In: KAMPUNZU& LUBALA(eds) op. cit., 274-327. DEWEY, J. F., PITMAN,W. C. III., RYAN, W. B. F. & BONNIN,J. 1973. Plate tectonics and the evolution of the Alpine system. Bulletin Geological Society America, 84, 3137-3180. EBINGER,C. J., DEINO,A. L., DRAKE,R. E. & TESHA, A. L. 1989. Chronology of volcanism and rift basin production: Rungwe Volcanic Province, East Africa. Journal Geophysical Research, 94, 14785-15803. GREEN, W. V., ACHAUER,U. & MEYER,R. P. 1991. A three-dimensional seismic image of the crust and upper mantle beneath the Kenya Rift. Nature, 354, 199-203. GURNEY, J. J., MOORE, R. O., OTIER, M. L., KIRKLEY, M. B., HOps, J. J. & MCCANDLESS,T.

97

E. 1991. Southern Africa kimbedites and their xenoliths. In: KAMPUNZU• LUBALA(eds.) op. tit., 495-536. HARKIN, D. A. 1960. The Rungwe volcanics at the northern end of Lake Nyasa. Geological Survey of Tanganyika, Memoir II. KAMPUNZU, A. B. & LUBALA, R. T. (eds) 1991 Magmatism in extensional structural settings. Springer-Veflag, Berlin. & MOHR, P. 1991. Magmatic evolution and petrogenesis in the East African Rift System. In: KAMPUNZU,& LUBALA(eds) op. cit., 85-136. & POPOFF, M. 1991. Distribution of the main Phanerozoic African rifts and associated magmatism; introductory notes. In: KAMPUNZU& LUBALA(eds) op. cit., 2-10. LE BAS, M. J., REX, D. C. & STILLMAN,C. J. 1986. The early magmatic chronology of Fuerteventura, Canary Islands. Geological Magazine, 123, 287-298. LEE, C. A. 1974. The geology of the Katete carbonatite, Rhodesia. Geological Magazine, 111, 133-142. LUBALA, R. T. 1991. African kimberlites: introduction. In: KAMPUNZU& LUBALA(eds) op. cit., 488-494. McCoNNELL, R. B. 1951. Rift and shield structure in East Africa. International Geological Congress, 14, 199-207. MARSH, J. S. 1973. Relationships between transform directions and alkaline igneous rock lineaments in Africa and South America. Earth and Planetary Science Letters, 18, 317-323. MOREAU, C., ROCCI, G., BROWN,W. L., DEMAIFFE, D. & PEREZ,J.-B. 1991. Palaeozoic magmatism in the Air Massif, Niger. In: KAMPUNZU& LUBALA (eds) op. cit., 328-352. NIXON,P. H. 1987. African-Arabian plate. In: NIXON, P. H. (ed.) Mantle Xenoliths. John Wiley & Sons, New York, 187-194. OLIVET, J.-L., GOSLIN,J.) BEUZART,P., UNTERNEHR, P., BONNIN,J. d~ CARRE, D. 1987. The break-up and dispersion of Pangea. Coedition Eli Aquitaine (Pau) and IFREMER (Brest). PARK, R. G. 1988. Geological structures and moving plates. Blackie, Glasgow. PRINS, P. 1981. The geochemical evolution of the alkaline and carbonatite complexes of the Damaraland igneous province, South West Africa. Annale Universiteit von SteUenbosch, Series A1 (Geology), 3, 145-278. PRODEHL,C. etal. (KRISP Working Party) 1991 Large scale variation in lithospheric structure along and across the Kenya rift. Nature, 354, 223-227. SKINNER, E. M. W., SMITH, C. B., BRISTOW,J. W., SCOTI"SMITH, B. H. & DAWSON,J. B. 1985. Proterozoic kimberlites and lamproites and a preliminary age for the Argyle lamproite pipe, western Australia. Transactions of the Geological Society of South Africa, 88 335-340. SMITH, C. B., ALLSOPP, H. L., KRAMERS, J. D., HUTCHINSON, G. & Roomcg, J. C. 1985. Emplacement ages of Jurassic-Cretaceous kimberlites by the Rb-Sr method on phlogopite

98

D. K. BAILEY

and whole rock samples. Transactions of the Geological Society of South Africa, 88, 249-266. SNELLINO, N. J. 1965. Age determinations on three African carbonatites. Nature, 205, 491. SORENSEN, H. 1974. Definitions. In: S~aRENSEN, H. (ed.) The Alkaline Rocks. John Wiley & Sons, New York, 3-11. STILLMAN, C. J., FURNESS, H., LE BAS, M. J., ROnER'rSON, A. H. F. & ZXELONKA,J. 1982. The geological history of Maio, Cape Verde Islands. Journal of the Geological Society, London, 139, 347-361. THORPE, R. S. & SMITH, K. 1974. Distribution of

Cenozoic volcanism in Africa. Earth and Planetary Science Letters, 22, 91-95. TURNER, D. C. & REX, D. C. 1991. Volcaniclastic carbonatite at Kaluwe, Zambia: age and relations to sedimentary rocks in the Zambezi rift valley.

Journal of the Geological Society, London, 148, 13-15. WOOLLEY,A. R. & GARSON,M. S. 1970. Petrochemical and tectonic relationship of the Malawi carbonatite-alkaline province and the LupataLebombo volcanics. In: CLIFFORD,T. N. & GASS, I. G. (eds) African Magmatism & Tectonics. Oliver & Boyd, Edinburgh, 237-262.

Plume heads, continental lithosphere, flood basalts and tomography D O N L. A N D E R S O N 1, Y U - S H E N Z H A N G 2 & T O S H I R O T A N I M O T O 1

~Seismological Laboratory, California Institute of Technology, Pasadena, California 91125, USA 21nstitute of Tectonics, University of California at Santa Cruz, Santa Cruz, California 95064, USA Abstract: High-resolution uppermantle tomographic models are interpreted in terms of plate tectonics, hotspots and plume theories. Ridges correlate with very low velocity areas to a depth of 100 km, probably a result of passively induced upweiling and partial melting. Past positions of ridges also exhibit very low seismic velocities in the uppermantle. At depths greater than 100 km, some low velocity anomalies (LVA) may record past positions of migrating ridges. Buoyant upwellings induced by spreading do not track the migration of surface ridges; they lag behind. At depths greater than about 150 km many LVA (Atlantic and Indian oceans) are more closely related to hotspots, and past positions of ridges than to current ridge locations. In the upper 200 km of the mantle, back-arc and continental extension areas are generally slower than hotspot mantle, possibly reflecting partially molten and/or hydrous mantle. The Pacific ocean ridges tend to be LVA, and probably hot, to about 400 km depth. The surface locations of hotspots, ridges and continental basaltic magmatism seem to require a combination of hot uppermantle and suitable lithospheric conditions, presumably the existence of tensile stresses. The high-velocityregions of the upper 200 km of the mantle correlate with Archaean cratons. Below 300 km the regions of generally fast seismic velocity, and therefore cold mantle, correlate with regions probably underlain by ancient slabs, where the uppermantle may be cooled from below. A moving plate, overriding a hot region, and being put into tension, will behave as K it were being impacted from below by a giant plume head. At sublithospheric depths there are very large LVA (VLVA) in the Pacific and Indian oceans and in the North and South Atlantics. The large continental and oceanic flood basalt provinces seem to have formed over these large, presumably hot, regions. These VLVA do not appear to be plume heads nor is there any obvious damage to the lithosphere under the present locations of flood and plateau basalt provinces. The uppermantle does not appear to be isothermal; the LVA are not restricted to hotspot locations. We suggest that LVA are hotcells in the uppermantle which reflect, in part, the absence of subduction cooling. Plate tectonic induced rifting causes massive magmatism if the break occurs over hotcells, i.e. low-seismic velocity regions. Flood basalts (CFB) may result from the upwellings of already hot, even partially molten, mantle. In contrast to plume heads and plume tails, hotcells are robust features which are fixed relative to one another. They are most pronounced in parts of the mantle that have not been cooled by subduction. There is a close relationship between CFB initiation sites, LVA and ridges and, we believe, hotcells.

The driving mechanism of plate tectonics is now generally attributed to boundary and plate forces such as slab pull and ridge push ( F o r s ~ & Uyeda 1975) with the asthenosphere being r~sistive or lubricative rather than forcing. Ridges are viewed as passive reactions to plate forces rather than the tops of convection cells (Lachenbruch 1976). Ridges are regarded as mobile but, even if passive, they may induce upwelling from some great depth, particularly at fast spreading ridges such as the East Pacific Rise (EPR). The existence of deep low-seismic velocity regions, such as at the East African Rift ( E A R ) , may indicate a component of active upwelling. Ridges

can migrate freely but the upwellings they induce will lag behind (Houseman 1983). Subduction zones migrate at a much slower rate than ridges. Although plate forces are important the mantle is not completely passive. Lateral temperature and pressure gradients may exert an important influence on plate motions (Officer & Drake 1983). These forces are generally ignored and this introduces certain paradoxes such as the rapid motion of the Indian plate in spite of the paucity of suitably located 'slab pull' regions and the presence of dragging continents. The plume hypothesis originally put forward

From STOREY,B. C., ALABASTER,T. & PANKHURST,R. J. (eds), 1992, Magmatismand the Causes of ContinentalBreak-up, Geological Society Special Publication No. 68, pp. 99-124.

99

100

D.L. ANDERSON E T A L .

by Morgan (1971) assigns a major role to deep mantle plumes. These plumes, as conceived by Morgan and his co-workers, account for most of the mass transport out of the mantle and more than 5.0% of the heat flow. In this theory, plumes drive the plate and are, essentially, the main source of buoyancy and mass flow in the mantle. In contrast, more recent plume theories (Campbell et al. 1989; Richards et al. 1989; Sleep 1990) consider plumes to be a secondary mode of convection. Hotspots have been shown to account for only about 6% of the terrestrial heat flow and to be a minor source of material (Davies 1990). These types of hotspots and plumes are clearly quite different than the robust plumes treated by Morgan (1971). White & McKenzie (1989) also consider plumes to be part of the large scale convective pattern of the mantle even though they are less voluminous than those treated by Griffiths & Campbell (1990). White & McKenzie (1989) attribute continental flood basalts (CFB) to lithospheric extension over hot upwellings while others assign a more active role to plumes in upliRing and breaking the lithosphere (Richards et al. 1989). The small scale and perceived fixity of hotspots motivated the hypothesis of deep, narrow plumes originating below the effects of moving plates. The anomalous geochemistry of hotspot magmas gave additional impetus for siting their source deep (Schilling 1973; Anderson 1975). It is generally assumed that chemically depleted midocean ridge basalts (MORB) come from the asthenosphere, or shallow sublithospheric mantle, even though enriched basalts erupt at most times and places where one expects the shallowest mantle to be sampled. The most depleted basalts are associated with long-lived and rapid spreading, not the initial stages of lithospheric disruption. This suggests that the traceelement and isotopic characteristics of hotspot magmas are acquired at shallow-depth. The shallow mantle is easily contaminated or metasomatized by a variety of mechanisms, (subduction, trapped melts) and may also house basalt-depleted (i.e. infertile) buoyant, harzburgites (Anderson 1981, 1983a, b, 1987a). Technically, the 'asthenosphere' is a weak layer in the uppermantle but it is now generally assumed by geochemists and petrologists to be the depleted reservoir. Furthermore, the perceived attributes of the 'depleted asthenosphere' have been transferred to the whole uppermantle. Thus, the words 'asthenosphere', 'upper mantle', 'convecting mantle' and 'depleted mantle' are all used interchangeably in the current geochemical literature. We use 'uppermantle' to highlight the fact that semantics may

be behind some of the current controversies. 'Uppermantle' means the mantle above the 650 km discontinuity. It contains the asthenosphere, the mesosphere (transition region) and the oceanic and continental mantle lithospheres. If the 'uppermantle' is defined as depleted, homogeneous and the MORB-source then, of course, it cannot provide enriched magmas (OIB, CFB) or be the source of plumes. Recent theories of giant plume heads, and remobilized continental lithosphere (CL) filling up the asthenosphere have considerably weakened the arguments for a depleted isothermal homogeneous asthenosphere. The distinctive geochemistry of hotspot basalts may be acquired in the shallow metasomatized mantle even of the heat source is deep and the bulk of the basalt comes from a deep, fertile reservoir (Anderson 1983b, 1985). If the uppermantle is inhomogeneous in composition and temperature, then the deep mantle plume hypothesis may not be necessary, or even viable. In particular, if continental insulation (Anderson 1982a, b; Gurnis 1988), recycling, metasomatism and subduction can account for observed geochemical and temperature variations, all available data could be explained by an uppermantle origin of hotspot geochemistry. Temperature variations of about 200°C exist in the shallow mantle and there are hot regions unrelated to hotspots (Lago et al. 1990). Relative fixity of hotspots is a problem for all plume theories. Swell heights, geoid and heat flow can be satisfied with lateral variations confined to the upper few hundred kilometres (Moriceau et al. 1991). There is as yet no evidence from seismology, or any other source, that plumes originate deeper than about 300 km or that the broad low-velocity regions in the uppermantle are connected by narrow tails extending deep into the lower mantle. The I = 2 component of lower mantle tomography correlates well with the l = 2 component of the hotspot distribution (Hager & Richards 1989) but shorter wavelengths correlate better with uppermantle tomography (Kedar et al. 1992). High-resolution 3D images of the seismic velocity variation in the uppermantle have recently become available (Zhang & Tanimoto 1991a, b, 1992). Some of the current issues of plate tectonics and hotspots can now be addressed with data of high resolution from the third dimensions. We discuss these recent high-resolution tomographic results. The new images have a lateral resolution of about 1000 km to depths of about 500 and 100-200 km depth resolution, depending on depth. Preliminary interpretations of these results are given in recent short

PLUMES, PLATE AND TOMOGRAPHY papers (Zhang & Tanimoto 1992; Anderson et al. 1992).

Mid-uppermantle (290 kin) (Fig. 1) This part of the mantle is well below the plate and should give us an idea of the plan form of mantle convection, including the locations of hot upweUings and plume heads. The mantle does not appear to be isothermal. It is characterized by large domains of both slow and fast seismic velocity. There is a general tendency for the seismic velocities in the mantle above about 300 km to be fast under continents and slow under oceans but this does not tell us the thickness of the plates. Many non-cratonic areas have fast velocities and some cratons have relatively slow velocities at depths below about 220 km. This is consistent with the moving plate being thinner than about 200 km, the approximate thickness of the thickest elastic plate as determined from flexural studies (Anderson 1990). Note the extremely broad LVA in the Pacific and Indian oceans. Clearly, these residual (or exterior) oceans differ from the newly opened Atlantic and Arctic oceans (interior oceans). The mantle beneath the Pacific plate is particularly slow from the fastest spreading part of the East Pacific Rise (EPR) toward the NW, generally in the spreading direction. The EPR itself is not located in a particularly central position in this large thermal anomaly. If hot mantle is a necessary condition for the location of ridges and hotspots, these could be almost anywhere in the North Pacific. At this depth (290 km) hotspots (black squares) appear to be randomly distributed in broad low-velocity regions. Hawaii is downstream from the EPR and is remote from plate boundaries and other hotspots. The Pacific plate is, of course, shrinking and there may be material converging, in the deeper mantle, toward the central Pacific. Hotspots are not characterized by large circular or ovoid LVA. The lowest velocity regions are beneath SW North America, the Southern Marianna Arc, New Zealand (NZ), SE Asia, Indo China, the northern Indian ocean and near the Gulf of Aden. The deep LVA near New Zealand, and the Afar are well known (Dziewonski & Anderson 1981; Nataf et al. 1984, 1986). Many of the above areas exhibit lithospheric tension or are regions of recent spreading. The island arc regions may be affected by volatile,fluxing due to slab dehydration. Regions of current subduction tend to have fast seismic velocities below the shallow mantle (Nataf et al. 1986; Zhang & Tanimoto 1991a; Zhou & Anderson 1989). The slow velocities

101

under NZ-Tonga-Fiji and Mariannas extend throughout the uppermantle (Figs 1-3) and could be the result of recent arc-rise collision or back-arc spreading. At the resolution considered, there is no evidence for cold mantle in these regions although, undoubtedly, cold slab extends to the deepest earthquakes, about 680 km. Since other covergence regions give the expected fast velocity signal it appears that resolution is not the problem. The above anomalous regions also do not have the high geoid signals associated with some other subduction zones. The results around NZ-Tonga-Fiji, indicating a broad region of presumably hot and low density mantle are particularly important since there is no depression of the 650 km discontinuity (Richards & Wicks 1990). A chemical boundary will be elevated by hot mantle which will also elevate phase boundaries with negative Clapeyron slopes (i.e. cold material transforms at greater depth). The top of the lower mantle has high temperatures in this region (Tanimoto 1990a) which would uplift chemical and phase boundaries and heat the base of the uppermantle, partially offsetting the cold slab effect. The tomographic results show that most of the uppermantle at convergence regions with backarc spreading has low seismic velocities and, presumably, low density. The question is, does the geoid reflect the thin dense slab (Hager & Richards 1989) or the broad buoyant surrounding area? The effects on depression and elevation of boundaries will probably be opposite for these two extremes. Many hotspots also exhibit geoid highs and LVA to about 200 km depth. There are many very low-velocity regions (VLVA) that are not near hotspots. It is likely that a combination of hot mantle plus appropriate lithospheric and stress conditions are required for hotspot upweUings to occur. The absence of hotspots above the VLVA in the northern Indian ocean is likely due to lithospheric compression in this region. Although this region is in a geoid low, suggesting perhaps some deeper cause for the lack of hotspots, it is little different from the Mt Erebus region in Antarctica, which has low seismic velocities, a geoid low, lithospheric extension (Davey 1981) and a hotspot (or at least, active volcanism). Other non-hotspot regions in geoid lows with high inferred uppermantle temperatures are the Pacific off North America, NE South America and the adjacent ocean, and the SW Atlantic. We suggest that these areas also have thick lithosphere or lithosphere under compression. The NE Pacific is a geoid low with hotspots, suggesting that lithospheric extension occurs in this region. Note that there are LVA in the wakes of

102

D.L. ANDERSON E T A L .

the Americas, India, Australia and Greenland. These regions have recently been vacated by thick continental lithosphere. The Mid-Atlantic Ridge (MAR) has migrated west and apparently decoupled from some of the Atlantic hotspots at about 70 Ma, stranding them on the African plate (O'Connor & Duncan 1990). At 290 km depth and deeper the uppermantle LVA does not follow the northern MAR but is offset to the east, following the central Atlantic hotspots. Ridges can move more rapidly than their associated upwellings (Houseman 1983), so this eastern Atlantic LVA may represent a previous location of the MAR or the position of continental break-up. There is also a discontinuous LVZ extending NW from Bermuda to Iceland and Jan Mayan. This may be due to the continental edge effect, discussed by Vogt (1991). He pointed out that the Bermuda swell paralleled the coast, as does the LVA, which is inconsistent with the hotspot hypothesis. Note also the LVA paralleling the Central Indian Ridge to the west and the E - W LVA south of the southern ridges, i.e. the former positions of these ridges. Spreading induced upwelling can extend to great depth and can still feed ridges which have migrated away. These passive upweUings may also leak material to the overlying hotspot. Theoretically, columnar type upwellings (plumes) are most stable near ridges (Cserepes & Christensen 1990). Note the gross global asymmetry in Fig. 1 between the slow Pacific hemisphere and the fast Eurasian, North Polar region and North American hemisphere. There is a broad LVA in the south central Pacific. Anderson (1987b) suggested that high temperatures in this region were responsible for extensive igneous activity in the Cretaceous. The plateaus in the western Pacific formed during a period of global plate reorganization, continental break-up and rapid spreading from centres in the Central Pacific. These spreading centres have since migrated toward the east. Other massive basalt provinces originated over what are now LVA in the Indian ocean, the N Atlantic and the S Atlantic. The VLVA in the S Indian ocean, which contains the Kerguelen-Heard hotspot, may represent the source of Gondwana magmatism and uplift, prior to rifting between India and Antarctica (Kent 1991). Eurasia has not moved far since its assembly as the northern part of the supercontinent of Pangaea and it is therefore still underlain by foundered oceanic lithosphere. The high seismic velocities found throughout the uppermantle under Eurasia suggest that the mantle here has been cooled by cold subducted oceanic slabs.

There is a remarkable absence of hotspots in this vast region.

The Mesosphere (430 and 490 km depth F i g s 2 & 3) The mesosphere, or transition region, is the region of the mantle between the two major seismic discontinuities at 400 and 650 km depth. The mesosphere is part of the uppermantle and is usually lumped with the asthenosphere or shallow mantle by petrologists (the 'depleted mantle', the 'convecting mantle'). However, it is mineralogically and rheologically distinct and may be chemically distinct as well. It may, for example, be the depleted reservoir and not well mixed with the shallow mantle. Much of the basaltic fraction of the Earth may be in this layer (Anderson 1983a, 1987a). This may be where slabs accumulate (Anderson 1979b, 1981,1982c; Anderson & Bass 1986). There is little correlation of the seismic velocities (Fig. 2) with present surface tectonics but good correlation with post Pangaeatic subduction (Anderson 1982b, 1989). In some respects the velocities in the mesosphere are the negative images of shallow velocities; regions beneath convergence zones, cratons, young oceans and old oceans all tend to change sign. In particular, mid-ocean ridges are generally underlain by fast velocities and, presumably, cold mantle; old oceans, with thick lithospheres, are often above hot mesosphere. Subduction zones generally mark sharp lateral boundaries between slow and fast mantle, e.g. see Tonga-Fiji, Marianas, Peru, Cascadia and the Aleutians. The slowest regions are under central Africa (surface hotspots around the edges but not in the centre), the N Indian ocean (no hotspots), W Australia (craton), New Zealand-Kermadec (island arc), W Pacific (stable ocean, no hotspots), SW USA (tectonic) and parts of eastern USA and northern Canada (stable). The fast regions generally correlate with expected locations of subducted oceanic lithosphere overridden by continents since the break-up of Pangaea. This is consistent with slab confinement to the uppermantle. The high-velocity anomaly (HVA) beneath NE Australia is probably due to the subduction of the North New Guinea plate associated with the joining of North and South New Guinea. The high veiocitiesin E, SE and S Asia may be the result of long continued subduction of the Pacific, Kula, and various other oceanic plates over the past 230 Ma. The high velocities under Eurasia may be due to subduction related to

PLUMES, PLATE AND TOMOGRAPHY

103

closures of oceans as microcontinents assembled which are on old, thick lithosphere. If the linear to form northern Pangaea (Russian and Siberian LVA, offset from the MAR, is a passive upweUplatforms, Kazakhstan block, Tarim craton, ing, then we might expect that the overlying hotspots will die out as the ridge migrates further Sino-Korean craton). Fast velocities are also prominent in areas west. The African plate is nearly stationary and where subduction has been long-lived (Kam- it is no surprise that eastern Atlantic hotspots are chatka through Japan, Borneo, northern Aus- nearly fixed. Note that the Kerguelen plateau is tralia through the New Hebrides, the Alpine at one extremity of a VLVA in the southern belt, the Andes, the Bering Sea, and central Indian ocean. Asia north of India). A belt of high velocities paralleling the Cordillera and extending from the Canadian Arctic through central North The outer shell (Fig. 4) America, the Caribbean and down west central Figure 4 shows the shear velocities at a depth of South America probably represents older sub- 66.1 km near the top of the mantle. Note the duction, due to the overriding of Pacific ocean 'ring of fire' around the Pacific, and the slow plates by the Americas. The locations of these velocities associated with ridges and back-arc fast bands are where the active margins of the basins (Zhang & Tanimoto 1992). Old crust is Americas were some 40 to 70 million years ago. generally underlain by fast mantle. Tectonically The fast belt extending across the Alps and into stable regions tend to be fast, a combination of Saudi Arabia is possible Tethyan lithosphere low temperatures and a refractory mineralogy (due to closure of an old ocean). (Jordan, 1975, 1978; Anderson & Bass 1984). LVA in this depth interval again occur in the Mantle temperatures were much hotter when wakes of drifting continents. These are par- ancient shield mantle formed, depleting it so ticularly evident for North America, India, and effectively in low melting components, that Antarctica but are also seen east of South cratonic mantle is well below the solidus. AlAmerica, south of Australia and east of Green- though cold, it is probably buoyant because of its land. The former positions of migrating ridges chemistry (low Fe, Al) and mineralogy (low garalso are evident, particularly the central Atlantic net). Archaean komatiites apparently melted at ridge, which has moved west, and the Indian temperatures well in excess of those inferred for hot present day mantle. If cratonic mantle lithoocean ridges, which have moved east and north. The mid-Pacific at one time had ridges and triple sphere is the residue of such high-temperature melting then it will be immune to damage by junctions and, at that time, passive rifting caused present-day hotspot magmas. upwellings and partial melting. Houseman (1983) showed that such upwellings will not Figure 5 shows velocities at 130 kin. The fastfollow the ridge if the ridge migrates too fast. est regions, without exception, are associated Absence of subductive cooling and presence of with Archaean shields. The Indosinia block heating from below are other mechanisms for (Cambodia) and Arabia are the only shields that generating hot Pacific uppermantle. There are are not HVA. Evidence for the former is weak. LVA, and presumably hot upwellings, in the Even the small Tarim shield in China shows up lower mantle under the central Pacific and as a HVA. Note that most of the Atlantic and Africa (Tanimoto 1990a). Even if there is no Indian ocean ridges follow the lowest velocity transfer of material there will certainly be trans- regions but are now offset in some regions comfer of heat from the lower to the uppermantle. pared with the shallower mantle zone in Fig. 4. Hot buoyant upwellings (and LVA) in the trans- This suggests when other depths are also conition region (mesosphere) can be caused by heat- sidered, that most ridges are passive, with hot ing from below, continental extension, ridge material upweiling into the space made available migration and intrinsic instability caused by par- by plate spreading (Lachenbruch 1976). Melting tial melting of phase changes. Contrary to pop- occurs during adiabatic decompression and this, ular belief all upwellings do not have to originate plus the high temperature, causes very low seisin a thermal boundary layer. mic velocities. This also induces, by continuity We can speculate that the linear LVA, with arguments, upward flow from greater depth, a three hotspots, NW of Africa may mark the form of convection triggered from above. Note mantle source for Jurassic pre-dri~ volcanism that most of the Atlantic and Indian ocean along the east coast of North America (Newark hotspots are embedded in hot, slow mantle at group, White Mountains, and eastern North this depth. These hotspots are related to present America dyke swarms) and that the material or past positions of the ridge. If the migration of from this region is now feeding the MAR and the ridges is taken into account, the close association Azores in preference to the overlying hotspots of hotspots with ridges, present and recent, is

104

D.L. ANDERSON E T A L .

quite remarkable. There are also many hotspots close to continental edges. Discontinuities in lithospheric thickness can also set up small scale convection (Vogt 1991). The Pacific ridge systems are embedded in very broad low velocity areas (LVA), with the Nazca and Cocos plates being entirely underlain by slow mantle which forms an asymmetric LVA around the EPR. Parts of the Australian-Antarctic and SW Indian ridges are not LVA at this depth, suggesting slow spreading or cold underlying mantle or very shallow melting. Note the discontinuity near the Australian-Antarctic Discordance (AAD). This is a deep ridge, low geoid region and the boundary between geochemical domains (Klein et al. 1988). It appears that the eastern part of the ridge south of Australia is being fed from the Pacific side. These types of ridges may be completely passive, existing only because of geometric constraints. Many hotspots are located in very large lowvelocity anomalies (Iceland, Hawaii, Azores) but again the sizes, shapes and trends of these VLVA, at present resolution, are not related to plate motions. There are other similar VLVA and LVA which are not obviously related to surface hotspots. As postulated earlier, these may also provide massive outpourings of basalt, such as CFB or oceanic plateau basalts (OPB), if plate reorganization puts the overlying lithosphere into extension. Hotspots may be the focused effect of upwelling triggered from above by extensional strains or discontinuities in the lithosphere. The extent of the volcanism would depend on the temperature and history of both the lithosphere and asthenosphere. There may also be instabilities at the top of a partially molten layer. These can also be triggered from above. Spreading centres can migrate away from upwellings they induced, upwellings which may still be manifest as volcanism on the overlying plate as well as by lateral transport to the ridge. The maximum depth extent of VLVA is unknown. This will only be evident when highresolution methods are developed for greater depths. At the moment all we know is that the large scale patterns of velocity anomalies are not everywhere continuous across the 650 km discontinuity (Tanimoto 1990b). The backarc basins in the western Pacific, and sites of subduction or recent subduction, are among the slowest regions at shallow depths. The slowest regions include W North America, New Zealand, SE Asia, the Philippine Sea plate, E Pacific and NE Australia-Coral Sea. Low seismic velocities can be explained by high temperatures and partial melting. The latter sets in at lower temperatures for volatile-rich mantle.

The extremely low shallow mantle velocities in convergent regions, particularly those with active backarc spreading, may be due to the high volatile content caused by dehydration of subducted oceanic crust and volatile-fluxed melting of the mantle wedge. The volatile fluxed or metasomatized material may constitute an important part of the shallow mantle, even away from subduction zones and continental lithosphere (Anderson 1982c). The volatiles released by subducting slabs are not entirely accounted for by island arc volcanism and may be stored in the shallow mantle. The buoyant material may explain why ridges, surface hotspots and trenches are mostly in geoid highs and have low seismic velocities in the shallow mantle. The dense slab will also contribute to the geoid but note that many ridges and hotspots also have underlying fast (cold) regions in the mesosphere. Note that the ridges are not underlain uniformly by slow material in the uppermantle. There are particularly slow regions in the north and south Atlantics, around triple junctions and near some hotspots. Ridges are also not uniform in depth or chemistry. Regions of continental extension (e.g. Rhinegraben, Kenya Rift, SW USA) have slow seismic velocities. Note the low velocities associated with lithospheric extension near Lake Baikal and the East African Rift. The LVA associated with the latter extends to great depth but the Lake Baikal region is underlain by cold mantle and will probably never be an active magmatic centre. Note that the older parts of the Atlantic and Pacific oceans are fast, indicating, in part, thickening of the lithosphere. Some of the material represented in this depth slice is entrained, sublithospheric material and thermal boundary layer (TBL). In contrast to White (1988) we do not equate the lithosphere with the TBL, which can be twice the thickness of the strong layer. The lower half of the TBL constitutes part of the perisphere (see below). In most models of plate tectonics and hotspots, bathymetry is due to lithospheric thickening over a homogeneous, isothermal mantle with perhaps a few interruptions by hotspot swells which are thought to represent the tops of deep mantle plumes. There is supposed to be no deep contribution to oceanic bathymetry (e.g. White 1988). The tomography, and recent geoid studies, indicate that this is much too simple a view. Ridges and hotspots are embedded in larger scale warm uppermantle provinces. Hotspots are not centrally located above the hottest mantle nor is hottest mantle always associated with hotspots. Recent geoid studies (Baudry & Kroenke 1991; Maia & Diament

PLUMES, PLATE AND T O M O G R A P H Y

105

Fig. 1. Shear velocities at a depth of 290 km (after Zhang & Tanimoto 1991a). Continental and plate boundaries are shown. The dots are hotspots. Blue regions have fast seismic velocities and are probably cold. Orange is slow and hot. Note that the total range of velocity variations changes (decreases) with depth.

Fig. 2. Shear velocities at a depth of 430 km. See Fig. 1 for details. Note the change of the scale of velocity variation.

106

D.L. ANDERSON ETAL.

Fig. 3. Shear velocities at a depth of 490 km. See Fig. i for details.

Fig. 4. Shear velocities at a depth of 66.1 km. See Fig. 1 for details.

PLUMES, PLATE AND T O M O G R A P H Y

Fig. 5. Shear velocities at a depth of ~30 km. See Fig. 1 for details.

Fig. 6. Shear velocities at a depth of 230 km. See Fig. 1 for details.

107

108

D.L. ANDERSON E T A L .

1991) show that Pacific hotspots are embedded in long geoid highs but are not at one end. One gets the impression that it is a combination of hot mantle and suitable lithospheric conditions that determines where hotspots emerge. CFB provinces are also generally on the edges of Archaean cratons and, when corrected for continental drift, originated over regions that are now LVA. Adiabatic ascent from c. 150 km under rifted cratons may result in more extensive melting than upwellings under thinner lithosphere.

Asthenosphere (230 km Fig. 6) The close association of the E Atlantic sinuous LVA with past positions of the westwardly migrating MAR, and with hotspots, is particularly well displayed by this image. At even greater depth, this LVA breaks up and loses its continuity (Fig. 1). The Iceland and Azores LVA are not centred, or most prominent, under the surface expression of the hotspots but the islands are near the ridge, the most convenient access of material from the LVA from depth to the surface.

Discussion If long linear LVA are passive upwellings associated with spreading ridges, then the association of deep LVA with hotspots must be explained. Hotspots, in most current theories, are active upwellings, possibly from the lower mantle, which control the locations of continental break-up (Morgan 1983). On the other hand, continents tend to break-up along pre-existing lines of weakness, generally following suture zones and mobile belts and avoiding cratons until the final separation. Hotspots and ridges may also be induced by plate spreading over a non-homogeneous, non-isothermal mantle, with the geometric and chemical characteristics controlled by conditions in the shallow mantle and asthenosphere as well as by the fluid dynamics of the deeper mantle. Plate forces (slab pull, plate thickening) may ultimately be responsible for rifting (the passive case) but the actual location and timing of the rift or midocean ridge may be controlled by sublithospheric conditions (perhaps, hot mantle). Upwelling of this hot mantle may be facilitated and made more intense by the spreading suction. Thus, the distinction between active and passive spreading and upwellings is not necessarily very clear. Plate motions have a large effect on convective motions in the mantle. Plume-like upwellings occur only near diverging plates; midplate

plumes are not steady features (Cserepes & Christensen 1990). The plume-like structures near ridges extend throughout the system while the sheet-like upwellings are shallow features. An important question is, how long can a plumelike upwelling exist after a ridge has migrated away; can Hawaii be the result of a c. 100 Ma triple junction that has since migrated away? According to recent ideas (Campbell & Griffiths 1990; Griffiths 1986) continental flood basalts (CFB) represent plume heads associated with the initiation of deep mantle plumes. Others suggest that the continental lithosphere is 'remobilized', 'delaminated', 'pre-weakened' or otherwise damaged or removed in the processes of flood-basalt magmatism (Hawkesworth et al. 1986; Mahoney et al. 1989). If these scenarios are true, there should be low seismic velocities in the upper 100-200 km of the mantle under continents where continental flood basalt provinces occur. Places to look for this effect include the North Atlantic Tertiary Province (NATP; Greenland, Scotland, Norway), the ParamiEtendeka (Brazil, SW Africa), the Deccan (W India), the Karoo (SE Africa) and the Siberian Traps. None of these areas show any evidence for entrained plume heads or removed lithosphere. Conductive cooling extends to about 200 km in about 200 million years, so there is little chance for conductive cooling of a hot plume head emplaced at about 200 km. Continental flood basalts may rather be associated with lithospheric extension associated with a continent moving over hot areas (LVA) of the mantle which may, in fact, be partially molten prior to extension. Note the VLVA in the North Atlantic (source of NATP, ?Siberia), W. Indian ocean (Deccan) and S Indian ocean (Karoo) where these CFB are presumed to have formed. These are all regions which have not experienced postPangaeatic subductive cooling. The hot mantle inferred to exist under hotspots and CFB may not have experienced cooling by subduction. Note that inferred backtracked locations of massive Cretaceous volcanism (Pacific plateaus, Kerguelen) and other flood basalts are VLVA (SE Pacific, S Indian ocean, N Atlantic, etc.). Hot areas of the uppermantle may be due to the absence of cooling rather than the importation of plume heads from great depth in the mantle. In addition to cooling by subduction of old oceanic lithosphere, instabilities in old continental TBL can also deliver cold material to the uppermantle. Most of the Pacific ocean has not experienced any of these kinds of cooling for 200-500 Ma and can be expected to contain hotter than average convection cells. Intense magmatism in the Pacific may be related to plate reorganiza-

PLUMES, PLATE AND TOMOGRAPHY tion over hot mantle and unrelated to giant plumes or plume heads. There are large lateral variations in seismic velocity at all depths in the uppermantle. These are most pronounced in the upper 300 km where a given change in temperature has the greatest effect because of phase changes and partial melting (Anderson 1989). Dehydration reactions in downgoing slabs also contribute to shallow seismic velocity variations. The smaller variations at greater depth cannot be used to argue that most of the uppermantle is isothermal. The relatively small velocity variations below 400 km are significant and do not imply smaller temperature variations than at shallow depths. Relatively small density variations at this greater depth are implied. The Pacific and Indian oceans are clearly distinct from the Atlantic ocean. This may be related to the location of Pangaea, previous ocean closures, and post-Pangaeatic subduction. Hotspots are generally embedded in broad low velocity regions of the uppermantle but lithospheric conditions and present and past locations of plate boundaries may also influence the positions of low velocities and hotspots. Previous positions of the mid-Atlantic ridges and SE Indian ocean show up as 'ghost' low-velocity zones in the uppermantle. Some of the older parts of the Atlantic, Pacific and Indian oceans have slower than average velocities below some 300 km depth. These appear to be generally unrelated to hotspots or to deep mantle effects. Although their source is unknown, these are undoubtedly buoyant regions which should contribute to the bathymetry of old ocean basins. In fact, the departure of old oceanic lithosphere from the square-root-time law is well known. It has been attributed to small scale convection, lithospheric delamination and dynamic support (Parsons & Sclater 1977; Colin & Fleitout 1990; Cazenave & Lago 1991). The tomography favours a sublithospheric explanation. The absence of surface hotspot expressions under most old oceanic lithosphere suggests strong lithospheric, perhaps under compression, rather than absence of hot mantle. Most hotspots on old oceanic lithospheric are near fracture zones or continental edges and, in fact, probably started under continents (Wilson 1990). The large number of hotspots on the African plate may be related to its central position in the supercontinent of Pangaea, or to the fragmentation of Pangaea.

Lithospheric extension v. deep plumes It is likely that the Rhinegraben and Lake Baikal are in regions of externally induced tension, pos-

109

sibly due to collision of Africa and India, respectively, with Eurasia, and that their volcanism is entirely passive rather than related to hotspots or sublithospheric LVA. They have low seismic velocities between 100 and 200 km. However, below 300 km both regions appear to be very cold. The Eiffel 'hotspot' does not have the time progressive track originally attributed to it (Cantarel & Lippolt 1977) and may not be a deep seated thermal anomaly. The rift related quaternary volcanism in E China lies above deep, cold regions which surrounds a deep hot region. The East African Rift (EAR), the Gulf of California, the Gulf of Aden and the Rio Grande Rift are above deep LVA. The surface expression of hotspots is associated with shallow LVA, spreading centres (past and present), continental edges, and absence of slabs. However, the relation between hotspots and LVA changes with depth. The above observations are all consistent with lithospheric control on the surface locations of some, if not most, hotspots and a general control by post-Pangaeatic Subduction on high velocities in the transition region (and the absence of hotspots). Old slab probably underlies most of Australia, the Americas and S and E Asia. These areas have few hotspots. Even older slab probably underlies much of N Asia, another hotspot free area. Many proposed hotspots are underlain by colder than average mantle below 200-400 km depth (e.g. St Helena, Tristan da Cunha, Iceland, Easter, Nunivak, San Felix, Juan Fernandez, Hoggar, Tibesti, E. China, Eiffel, Azores, Ascension, Galapagos, Crozet and Amsterdam). Almost all of these hotspots are near fracture zones, rift zones and lithospheric discontinuities, or the effects discussed earlier, and some have been questioned as hotspots based on other considerations. Some of these may be plate boundary and triple junction phenomena and some may be due to plate reorganization. Many hotspots are actually hot lines, or erupt for long periods of time after leaving the point of initiation, or do not exhibit the predicted age progression, or are up to 1000 km away from the conjectured plume (Veevers 1984; Okal & Batiza 1987; Phipps 1988; O'Connor & Duncan 1990). If hot mantle actually has the dimensions and shapes of the seismic LVA then these observations, combined with lithospheric control of exit points, can be understood. The similarity in isotope and trace-element chemistry between continental and oceanic magmas along some hotspot tracks argues for an extended sublithospheric source rather than a CL or plume head source. These kinds of observations are often attrib-

110

D.L. ANDERSON E T A L .

uted to shallow channels or tilted plume conduits or .attachment of the plume head to the moving lithosphere. Alternatively, there may be a global enriched (metasomatized) sub-lithospheric layer, which could explain the catholic nature of enriched basalts, and hotter than average convection cells in the uppermantle. The uppermantle has multiple convection cells and these are unlikely to all have the same temperature, particularly considering their different histories. The hotspots which are embedded in broad and deep LVA include Samoa, Mt Erebus and East African Rift volcanoes. These seem to be regions where the lithosphere is breaking due to plate tectonic boundary forces, over extended regions of particularly hot mantle. These surface hotspots may be fed by a large high-temperature uppermantle region, or hot convection cel~, rather than by a narrow, deep, feeder tube, as in the plume theories. Other hotspots over extensive and deep LVA (hotcells?) include Hawaii, Rrunion, Canaries, Cape Verde, Kerguelen and Tasmania. These are all, currently, non-ridge hotspots. Few of the on-ridge or near-ridge hotspots occur over slow (hot) mantle that extends as deep as 300 or 400 kin. Most ridges, however, are migrating. Hawaii occurs near the boundary of hot and cold, or less hot, uppermantle and is related to an extensive NS trending LVA. Other regions, such as eastern China, which are generally not considered hotspots, in spite of the hotspot chemistry of the magmas, are associated with deep LVA, which, however, are often offset from the surface volcanism. The concepts of large lateral transport distances, and large regions of high temperature (VLVA, LVA), feeding localized surface volcanoes, are at variance with the concept of narrow (100200 km) plumes which are absolutely fixed with respect to one another. However, common amendments to plume theories involve large plume heads, easy capture of plumes by ridges, long distance lateral plume flow through shallow channels, large geochemical radius of influence around plumes and tilting of plume conduits by mantle flow. These modifications to the 'hotspot' models are unnecessary if the mantle is not isothermal, and if the uppermantle is not homogeneous, i.e. there may be a global metasomatized shallow layer which provides the geochemical signature often attributed to deep mantle sources. Although it is generally assumed that the Parami basalts formed over a fixed Tristan hotspot there are difficulties with this interpretation (Morgan 1983). Plate reconstructions apparently require motion of this hotspot (Molnar &

Stock 1987; Peate etal. 1990). It appears that the mantle source for the Paran~ is displaced from the predicted position of Tristan and that it is laterally extensive. We suggest that the Paranfi CFB formed over one of the South Atlantic LVA south of Tristan de Cunha. These LVA may represent hotcells. We also suggest that the Gondwana flood basalt province extending across Africa, Antarctica, India, Australia (Karoo, Ferrar, Tasman) formed when this region was above the LVA (66 to > 290 km depth) in the southern Indian ocean, perhaps centred on Kerguelen. This region has apparently had anomalously high temperatures since at least 270 Ma (Kent 1991). However, we see no reason for attributing all uppermantle VLVA, or regions of extensive volcanism, to plume heads. The Indian ocean was insulated by Pangaea, was isolated from subductive cooling, has been swept by migrating ridges and triple junctions which undoubtedly initiated upwelling and melting, and is underlain by hot lower mantle. The 150 Ma pre-heating time between plume head injection and magmatic activity, as proposed by Kent (1991), should have done significant damage to the S Gondwana continental lithosphere. As we discuss later, there is little evidence for such damage.

A modest proposal The mantle is certainly convecting and there are therefore hot buoyant upwelling regions and cold, dense descending regions. What role this convection plays in driving the plates or (vice versa) and controlling their speed and direction is uncertain. In the Morgan (1971) plume theory hot upwelling plumes play a dominant role. Most current investigators assign a dominant role to the plates (Forsyth & Uyeda 1975; Hager 1978); body forces on the thickening plate and sinking slabs control plate velocities and directions, and perhaps, the planform of uppermantle convection. Plate divergencies and convergences probably control, to a large extent, convection cell boundaries in the mantle. Lateral temperature gradients can also drive convection, and in this case, there is no critical Rayleigh number. Currents flow from hot toward cold areas. Froidevaux & Natal (1981) discuss this mechanism using cold vertical slabs as the source of cold and obtain upwellings near the centres of large continents. Cold slabs at the base of the system can also set up lateral temperature gradients. Officer & Drake (1983) conclude that lateral density gradients beneath the plates are important in driving mantle convection. Upwellings associated with these mechan-

PLUMES, PLATES AND TOMOGRAPHY isms may be confused with deep mantle thermal boundary layer instabilities, or plumes. Partial melt zones in the uppermantle can also generate upwellings unrelated to TBL. The tomography shows that the mesosphere under Africa, Antarctica and the Indian ocean has low seismic velocities and these regions are therefore probably hot. A long wavelength geoid high is centred over Africa and it has been proposed that this region has been hot for several hundred million years (Anderson 1982b). The S Indian ocean region may have been as well (Kent 1991). The central Pacific, is also a long wavelength geoid high and has low uppermantle seismic velocities. The large plateaus in the W Pacific (Ontong Java, Shatsky Rise, etc.) may have formed over the S Pacific, suggesting hot mantle, or hotcells, has existed there since at least the early Cretaceous (Anderson 1989). This region has not experienced slab cooling for a comparable period of time and has been a region of constantly changing ridge and triple junction configurations. We suggest that since the break-up of Pangaea the circum-African continents have been moving from hot mantle toward cold mantle, driven both by the uppermantlemesospheric lateral temperature gradients and slab-plate body forces. If so, most of the continents have reached the end of their journey since, to proceed further, they must intrude into hot Pacific mantle. North America, in particular, seems to be boxed in by hot mantle. Continents and their associated subduction zones also appear to have been moving toward cold, downwelling lower mantle. The continents moving toward continentdipping trenches and, therefore, overriding cold oceanic lithosphere, will insert cold slab into the mantle beneath them and in their trail, until they get close to a ridge. Thus, the high-velocity regions under parts of the Atlantic and Indian oceans probably reflect mainly subduction cooling but they may also have been relatively cold prior to the break-up of Pangaea. On the other hand the high velocity regions in the mesosphere under northern Europe and central Asia, the Arctic and parts of the Pacific have not experienced recent subduction and these may be the regions toward which uppermantle flow was directed during the initial breakup of Pangaea, at least that component controlled by lateral temperature gradients. These cold regions may reflect slabs subducted during the assembly of Eurasia. If this analysis of the present tomography is taken as a guide we predict deep uppermantle currents directed north from Antarctica and the southern oceans and from the Pacific toward •

I

111

Eurasia and the Arctic and from Africa toward the surrounding areas. Smaller scale diverging flows in the transition region may originate under the north Atlantic, western Australia, and the Indian geoid and velocity low. Horizontal temperature gradients can drive mantle flow and continental drift (Elder 1967; Officer & Drake 1983). The mantle currents discussed above may also represent the 'return flow' which balances the plate motions. The very low seismic velocities and, presumably low viscosity, under the Indian-Australian plate may explain why this plate is moving rapidly in spite of having several continents and little slab. Superposed on the thermal currents in the sublithospheric mantle are plate induced motions, generally directed from ridges to trenches. The scale of mantle convection is unknown. The scales of swells and uplifted regions have recently been taken as the scale size of plume heads. However, they could instead be telling us the size of uppermantle convection cells• There are probably many hotcells under the Pacific plate, perhaps elongated in the spreading direction by plate drag (Maia & Diament 1991). It is of interest that the most active current hotspot, Hawaii, is downstream from the most rapidly spreading segment of the EPR. There is no obvious plume head associated with the Hawaiian hotspot, in the sense that CFB provinces are attributed to the Tristan, Rrunion, Heard and Iceland hotspots. Tomography shows that there are vast LVA in the uppermantle and that ridges, hotspots and CFB initiation sites tend to occur in these regions. These regions have also not been cooled by subduction. Uppermantle convection cells will be embedded in these VLVA and some may be hotter, or convecting more vigorously, than others. Convection may even be intermittent, causing periodic overturn. What we call 'hotspots' may be related to these types of phenomena, plus lithospheric extension, rather than to narrow deep plume tails extending to the core. Nevertheless, plume tails and heads, and whole mantle convection schemes have become popular and we will next test these with the tomographic results.

Plume heads, continental lithosphere and flood basalts Recent models for continental flood basalts include (1) the initial impact of a giant mushroom plume head (the 'starting plume' hypothesis; Richards et al. 1989; Hill et al. 1991) (2) melting caused by lithospheric thinning over a steadystate plume (White & McKenzie 1989) or (3)

112

D.L. ANDERSON E T A L .

delamination, melting or remobilization of the continental lithosphere (CL; Hawkesworth et al. 1986). In contrast to these plume-related hypotheses there is the possibility that plate reorganizations, extensional stresses, zones of weakness in the lithosphere and the prior existence of hot or partially molten regions of the mantle (hotcells) are involved in massive basaltic outpourings. There may be extensive hot regions of the uppermantle that are unrelated to the recent arrival of a plume head, and which can provide massive amounts of basalt if the overlying lithosphere is put into tension. High-resolution surface wave tomography can be used to test the plume head and lithospheric damage (thinning, stretching, remobilization, heating or delamination) hypotheses. We find no evidence, at our resolution, for the involvement of plume heads or lithospheric modification at any of the sites of Triassic to Tertiary continental flood basalts. These massive outpourings of basalt seem rather to be related to plate tectonic reorganization, over broad deep regions of hot, probably partially molten, mantle. Pieces of thick continental lithosphere apparently can separate along lines of weakness, such as suture zones, without substantial rifting or thinning, allowing egress to already buoyant, or even partially molten, mantle, which melts further upon ascent and decompression. Cratonic lithosphere probably formed at high temperature and appears to be immune to significant thermal damage with current mantle temperatures, both because it is cold and because it is refractory.

Plume theories In the plume theories the mantle is assumed to be nearly isothermal except in the vicinity of a plume. In multicell convection, with variable surface (lithosphere) conditions and high-viscosity slabs there can be large lateral temperature gradients (Anderson 1982b; Froidevaux & Nataf 1981; Gurnis 1988). The tomography shows this as well. In each convection cell, of course, there will be an upwelling or 'plume' but all convection cells are not the same temperature. The 'cold' downwellings of hotcells may be hotter than the 'hot' upwelling parts of coldcells. The point is that in a realistic mantle there are lateral temperature gradients (and variations in volatile content). Individual cells may be relatively isothermal, homogeneous and well-mixed. Plume theories are based on numerical, laboratory or conceptual experiments that involve uniform boundary conditions and, at most, a few cells. Generally, the system is heated from below

and has constant properties and no phase changes. Such a world has a few hot plumes to remove the heat from the bottom of the system to the top; plumes arise from instabilities in the lower thermal boundary layer. All convection cells are about the same. For a real mantle, there is non-uniform heating from below (for uppermantle convection), non-uniform cooling from above, internal heating, moving plates and induced upwelling from above. Therefore, regions that are hotter than average, or even partially molten, are not necessarily plumes. Continental flood basalt (CFB) volcanism is characterized by rapid eruption of large volumes of basalt, typically 106 km 3 in 106 years. Often, CFB are associated with rifting and continental breakup although sometimes the magmatism precedes extensive rifting. Richards et al. (1989) argue that major CFB provinces are consequences of the initial arrival of deep-seated mantle plumes which promote continental breakup. White & McKenzie (1989) argue for an anomalously hot, asthenospheric mantle, associated with steadystate plumes, in order to generate large volumes of basaltic magma at relatively shallow levels. Lithospheric stretching and thinning is required to allow ascent and melting of the anomalous asthenosphere, i.e. the asthenosphere is normally sub-solidus. The ascent and melting of the large head of a new plume has been suggested as the cause of the anomalously large rate of eruption of modern flood basalts over short periods of time near the beginning of many hotspot tracks (Morgan 1981; Courtillot et al. 1986; Richards et al. 1989; Hill et al. 1991). Griffiths & Campbell (1990) argue that flood basalts are the result of large spherical masses from the core-mantle boundary which collapse to 2000 km diameter pancakes when they encounter the lithosphere. The plume head consists of hot core-mantle boundary (CMB) material and, mostly, entrained material from the lowermost mantle above D " (the layer at the base of the mantle). As the giant sphere enters the uppermantle it no longer entrains ambient mantle but simply pushes it aside. Magmatism resulting from plume heads is therefore a combination of lowermost and lower mantle material but no uppermantle or asthenospheric material. Plumes are assumed to consist of a large, composite buoyant spherical vortex head followed by a narrow, hot, uncontaminated tail containing material from just above the molten iron core. One would expect the highest temperature mantle magmas (picrites, komatiites) to be representative of plume tails and the enriched source region in the plume head-tail theories, and to be depleted in Ander-

PLUMES, PLATES AND TOMOGRAPHY son's (1981, 1982d, 1983a)depleted mesosphere theory. In Anderson's model the meosphere was depleted by the removal of a small melt fraction to the continental crust and the shallow mantle (perisphere). It is maintained as a depleted reservoir by de-metasomatization, at shallow depth, of the downgoing slab. A metasomatized shallow mantle is consistent with mass balance calculations (Anderson 1983a). White & McKenzie's (1989) explanation for flood basalts is quite different than Griffiths & Campbell (1990). This involves externally imposed lithospheric extension coincident with a hot steady plume. In their model a steady flow ascends and cools by conduction as it spreads laterally below the lithosphere. It melts and causes hotspots only if the overlying lithosphere is coincidently stretching and thinning. The asthenosphere is depleted, as it is in most theories, and fuels the midocean ridges. Griffiths & Campbell (1990) criticize the steady plume hypothesis because they do not believe it explains CFB coincidence with the initiation of hotspot tracks, the absence of volcanism prior to the CFB and the very short duration of anomalous rates of eruption. On the other hand, Griffiths & Campbell (1990) ignore externally imposed, or plate tectonic, extension and rifting, the role of melting, the interaction of plumes with mantle convection and plates, the spacing of plumes, their initiation in a convecting, internally heated mantle and the expected geochemistry of source material in contact with the core. Neither theory addresses the source of MORB after the near-ridge asthenosphere is filled with plume heads. White & McKenzie (1989) do not address the depth of origin of plumes nor how they form. However, their hypothesis and the original plume hypothesis of Morgan (1971) view plumes as the main sources of mantle upwelling, which are therefore simply the rising parts of mantle convection cells albeit, very narrow. In the plume head hypotheses, plumes are viewed as small-scale thermals superimposed on the main scale of mantle convection. In the uppermantle version of the starting plume hypothesis plume heads, which originate in the uppermantle, are much smaller and little entrainment occur, i.e. they consist almost entirely of source material, which would be the subasthenospheric transition region, or mesosphere (Griffiths & Campbell 1990). In this model, uppermantle plume heads, upon spreading, cannot achieve the lateral dimensions observed for flood volcancism supporting 'the hypothesis that plumes responsible for flood basalts originate at the core-mantle boundary' (Griffiths & Campbell 1990).

113

The lithosphere and asthenosphere are not involved in the melts that constitute CFB in the starting plume hypothesis. The melts are entirely from a mechanical mixture of CMB material and entrained lowermost mantle. In the White-McKenzie version of the plume hypothesis, plumes consist mainly of uppermantle material which partially melts to provide CFB when the lithosphere thins adequately. In the lithospheric delamination or remobilization schemes CFB are derived from the continental lithosphere. The asthenosphere is generally thought to be depleted and the source of midocean ridge basalts. Plume material is almost invariably thought to arise from an enriched or primitive reservoir, either shallow (CL) or deep mantle. Depleted MORB-like picrites and komatiites, the best candidates for plume material, are paradoxes in these schemes. Lithospheric remobilization is attributed to continental collision, or stretching, or the impact of a plume of unspecified composition. If the hot, bouyant upwellings are depleted then they may pick up their enriched geochemical signature in the metasomatized shallow mantle (Anderson 1983b, 1985). In the White-McKenzie theory, lithospheric heating and thinning is an intrinsic part of the CFB story. Extensive melting only occurs when the lithosphere becomes acceptably thin. The continental lithosphere originally adjacent to hotspots and overlying the plume heads should differ from the lithosphere elsewhere along the break which may develop into a new ocean. In addition, the plume head is entrained by the plate and dragged along by the diverging lithospheres. Many authors would like to erode, delaminate or remobilize the continental lithosphere by plume heads or processes associated with continental collision or breakup (Hawkesworth et al. 1986; Storey etal. 1989; Mahoneyetal. 1989, 1991). Specifically, lithospheric removal has been proposed for Brazil, India and Madagascar and implied for other CFB sites. Delaminated continental lithosphere has been proposed as a general source of enriched basalts by Allegre & Turcotte (1985) and McKenzie & O'Nions (1983). In all cases the motivation was to insert enriched material into the 'asthenosphere' or 'convecting mantle' which, it is widely felt, would otherwise be depleted and unable to provide CFB or ocean island basalts (OIB). The alternative is that there is a global shallow, inhomogeneous enriched or metasomatized layer (refractory peridotite) which overlies the depleted reservoir (Anderson 1983b, 1989) and which contaminates or mixes with basalts from

114

D.L. ANDERSON E T A L .

the deeper more fertile reservoir. Hot regions of the mesosphere provide the bulk of the material (basalts) to spreading ridges, hotspots and continental flood basalt provinces and these basalts are variably contaminated by the easy melting components of the shallow mantle, which is melted by adiabatically decompressing and melting plume or diapir material. Continental tholeiites are often attributed to asthenospheric upwelling and melting of enriched heterogeneous subcontinental lithospheric mantle. Although this process is unlikely because of the low temperature and probable infertility of the continental lithosphere (as opposed to the shallow mantle under the lithosphere) the predictions for seismic tomography would be similar to the plume head and lithospheric delamination or reactivation scenarios. In the Anderson (1989) scenario, the shallow mantle (perisphere) is enriched but not necessarily fertile and this contaminates depleted basalts from slightly greater depth. The shallow mantle can be contaminated (enriched) by subduction (dehydration, sediments) and by trapped, small volume melts (kimberlites, lamproites). When plumes reach the surface they are mixtures and can range from depleted to enriched. This is similar to the continental lithosphere contamination models but the enriched layer is viewed as global and thick and weak, or low viscosity, rather than lithospheric. It is isolated from the rest of the mantle by its buoyancy and low viscosity. We call this layer the 'perisphere'. Chemically, it plays the role that many attribute to the continental lithosphere but physically it behaves as asthenosphere. The perisphere is constantly refreshed by recycling, slab dehydration and residual LILrich melts. It is constantly being depleted by eruption of enriched magmas. A minority opinion holds that within-plate magmatism is the result of lithospheric dynamics and plate reorganizations rather than rising deep mantle plumes (Giret & Lameyre 1985; Phipps 1988; Zehnder et al. 1990; Sykes 1978). In the deep mantle plume hypotheses hotspots appear at random times and places since the source layer is remote and decoupled from plate tectonics and the locations of extending lithosphere, continental boundaries, lithospheric discontinuities and zones of weakness. The association (i) in space, of hotspot initiation sites with zones of crustal weakness, lithospheric discontinuities, cratons and ridges, and (ii) in time, with episodes of rapid plate motion and of plate reorganization are coincidental in the hydrodynamic thermal boundary layer instability models. These models also do not address the relative fixity of hotspots. The thermal boundary

layer at the base of the mantle is expected to be as mobile as the surface boundary layer, and more mobile than a mesospheric boundary layer since the core is essentially invicid. In layered convection models (Anderson 1979a, 1991) the high viscosity of the lower mantle and the very high viscosity of bottomed out slabs can establish a fixed reference system at the bottom of the uppermantle. In uppermantle convection schemes one can view the individual convection cells as relatively fixed to one another and the hot upwelling parts are therefore relatively fixed; they may be the main upwellings rather than secondary thermals. Convection cells in the uppermantle may have dimensions of 500-1000 km and clusters of hotcells may remain hot for long periods of time. The complete plume hypothesis is untestable. The narrow plume tails, 10-200 km in diameter, and extending deep into the mantle are below the resolution of geophysical techniques and cannot be resolved by numerical or theoretical computation. They give no signal and have no measurable effect. The geophysical effects (bathymetry, geoid, tomography) of the large flattened plume heads are little different from alternative models of uppermantle structure involving large hot regions associated with larger scales of mantle convection, uppermantle convection cells, continental insulation, absence of subduction, or passive upwellings associated with migrating ridges, or lithospheric discontinuities and other plate induced phenomena. Large scale hot regions in the uppermantle can be generated by a variety of mechanisms. Passively induced upwellings, such as by ridge spreading, sample both the shallow mantle and, as time goes on, or spreading rate increases, the deeper mantle. This is not the place to discuss whole-mantle v. layered-mantle convection but phase changes, viscosity jumps and changes in chemistry can all serve to isolate the uppermantle from the lower mantle. Recent plume models envisage 1000-2000 km diameter circular pancake shaped features of order 100-200 km thick underneath the lithosphere. These hot circular regions either surround current hotspots (White & McKenzie 1989) or are split and carried along by the lithosphere underneath flood basalt provinces (Phipps 1988; Hill et al. 1991).

Tomography Although no geophysical or computational technique can resolve the narrow plume tails conjectured to link plume heads with the deep mantle there are other predictions of the above theories

PLUMES, PLATES AND TOMOGRAPHY that can be tested with available tomographic resolution. We present and interpret a series of tomographic maps; in particular, we look in detail at a number of CFB provinces. Figure 6 shows the shear-velocity at 230 km depth. In the plume models we expect to see large low velocity circles surrounding all hotspots, or their conjectured initiation sites (CFB), or elliptical patches elongated in the plate direction. We expect to see shallow linear bands of low-velocity material under ridges, updrafts caused by spreading. However, we see broad areas of both low and high seismic velocity. Very low seismic velocities faithfully follow ridges in the upper 100 km (Fig. 4). By 300 km depth, however, the band of low-velocities in the Atlantic extends from the Azores along the north central M A R and then to Cape Verde, Ascension, St Helena, and Tristan (the reader unfamiliar with hotspot names should consult a globe or an atlas). This brings up the question of whether the hotspots which are embedded in this 'fossil ridge' feature are active or passive. Certainly the hotspots which are outside this feature (Jan Mayen, Great Meteor, Fernando, Arnold, and Trinidade) are relatively small and not particularly active. It should be pointed out that some of the slowest regions at depths greater than 200 km are not associated with hotspots at all. They occur along some continental edges and near some subduction zones. If the 'midplate' mantle is isothermal, with all intermediate wavelength bathymetry and geoid anomalies caused by lithospheric thickening, except around hotspots (White & McKenzie 1989; Davies 1988, 1990) then seismic velocities beneath the lithosphere will be characterized by about 40 circular or elliptical patches, 2000 km in dimension. Figures 5 and 6 show that this is not the situation; there are extensive hot regions in the Pacific and Indian oceans with smaller hot patches and bands elsewhere. There appears to be buoyant sublithospheric support under old oceanic plate.

North Atlantic Tertiary Province, NATP (60 Ma) We now discuss the continental flood basalt (CFB) provinces and their relation to conjectured plume heads. The 60 Ma old basalts of E Greenland and NW Europe are attributed to the arrival of a plume head (Richards et al. 1989) or extensive stretching of the Greenland-NW Norway-Scotland lithosphere (White & McKenzie 1989). The conjectured plume tail is now under Iceland and the plume head effects should underlie Greenland, Norway, and Scotland, i.e.

115

damaged lithosphere or entrained plume head. There has been little time (60 Ma) since this event for the plume head to cool or for the damaged lithosphere to repair itself so the plume head initiation event, or the lithospheric thinning event in the passive plume model, should be readily apparent as thin and warm lithosphere, and shallow and hot asthenosphere. Figures 5 and 6 show the seismic velocities at 130 and 230 km depth and we see that in the lands surrounding Iceland, where the conjectured plume head impacted the North Atlantic, the seismic velocities are all higher than average, indicating cold mantle. The slow velocities are confined to the young oceanic regions which did not exist at the time of the conjectured plume head or lithospheric thinning event. The tomography shows a L V A below 400 km depth north of Jan Mayen, extending from NE Greenland to Spitzbergen. This appears to have fed Iceland, and the northernmost M A R and may have provided basalt to the NATP and, possibly, to Siberia in the Triassic. The VLVA in the N Atlantic could be due to a hotcell, or cells, or a plume head. It is about the size of Greenland.

Deccan traps (65 Ma) The Deccan traps formed at 65 Ma, presumably over the Reunion hotspot. The responsible plume head would underlie most of W and S India. Storey et al. (1989) proposed that the Indian subcontinental lithosphere slid off and contaminated the Indian ocean asthenosphere during this event. Mahoney et al. (1989) argued that the distinctive character of Indian ocean basalts was the result of continental lithospheric remobilization preceding the break-up of Gondwana, particularly from the portion that would become greater India. We therefore expect to see a large area of thin lithosphere and very slow seismic velocities under most of India. The tomography, however, show fast velocities under the Indian subcontinent and indicates thick and cold lithosphere, and cold asthenosphere (say 110-200 km depth). The seismic velocities in the Indian lithosphere may be slightly lower than for other cratons but it is also a small craton. Heat flow and flexural studies (Gupta & Gaur 1984) are consistent with a thick, cold lithosphere today and during emplacement of the Deccan traps. The Rajmahal basalts in the NE corner of the subcontinent, a possible result of magmatism from the Kerguelen hotspot, also are associated with fast velocities in the shallow mantle. There is therefore little evidence for hot plume heads or thinned lithosphere under these two continental flood-basalt provinces, which should both

116

D.L. ANDERSON E T A L .

have traumatized the uppermantle under most of greater India. One might entertain instead the opposite hypothesis; splitting of cratonic lithosphere (rather than thinning) may allow adiabatic ascent from great depth ( > 150 km) and, therefore, extensive melting. The Indian ocean ridge, which was near India at the time of the Deccan magmatism, will also have induced upwelling. Gupta & Gaur (1984) showed that low heat flow is associated with the Deccan trap terrane. There is no sign of transient thermal perturbations. These are signs of thick lithosphere and a deep CFB source and one that did not significantly perturb the lithosphere. The thickness of the thermal lithosphere just south of the Deccans traps is 200 km and this region has a normal cratonic heat flow, 37.7 mW m -2, similar to that in Siberia. The thickness of the thermal lithosphere, or thermal boundary layer, is generally twice that of the elastic lithosphere. The present elastic thickness of the Archaean Slave craton is 100 km, about the same as inferred for the Deccan lithosphere (Grotzinger & Royden 1990). Curiously, despite being erupted onto Archaean Craton, only a very small fraction of the Deccan magmas reflect much contribution from material thought to comprise the CL. It was therefore suggested that there was not much lithosphere left under India by the time of the Deccan activity, owing to prolonged convection thinning preceding and during the breakup of Gondwana (Mahoney et al. 1989). These gyrations, regarding the fate of the Indian lithosphere are entirely the result of the perception that CFB should arise from enriched CL and that all basalts from shallow mantle and spreading ridges should be depleted, unless they have been contaminated by CL or plume material. The alternative is that plumes are initially depleted, or that migrating or slow spreading ridges can tap shallow enriched mantle (the perisphere). It is therefore of interest that India was near a spreading ridge at the time of the Deccan traps. The traps, in turn, are near a major ancient lithosphere break that crosses the subcontinent. The Indian ocean mantle appears to be hotter than average and spreading, of course, induced upweUing and further melting. Thus, it appears that lithospheric conditions (but not chemistry) are involved in CFB and the initiation of hotspots. We suggest that the shallow enriched perisphere can be isolated from the underlying depleted mantle because of its intrinsic buoyancy and low viscosity. It has low viscosity because of high-temperature and probably, high volatile content. This layer is, in part, the lower

part of the thermal boundary layer. The TBL, in turn, is White's (1988) 'lithosphere'.

Paran~i flood basalts (130 Ma) The extensive Paran~i basalts in Brazil (130 Ma) and the smaller Etendeka basalt province in W Africa have been attributed to the arrival of a massive plume head under the South Atlantic, or to a period of preheating and lithospheric extension by a plume head prior to opening of a new ocean. In addition, several authors have suggested that the Brazilian lithosphere delaminated, or slid-off, to contaminate the Atlantic with enriched materials. Hawkesworth et at. (1986) proposed that large portions of ancient Brazilian lithosphere contaminated a belt of South Atlantic asthenosphere which is now erupting at hotspot islands (Tristan da Cunha and Gough) which, in turn, contaminate nearby sections of the Mid-Atlantic Ridge. Bonatti (1990) proposed that part of the CL of equatorial South America slid off and remained behind to form the St Peter-Paul islets. These suggestions were made because of the widespread occurrence of enriched basalts and peridotites in the South Atlantic. If the oceanic uppermantle is depleted, as assumed by most authors, then enriched material must be imported from great distance or great depth. The alternative is the shallow mantle, almost everywhere, is metasomatized or enriched (i.e. in LREE, LIL, 87Sr/86Sr) even if infertile. In Anderson's (1981, 1983a, b) model the shallow mantle everywhere is enriched, unless shoved aside by long sustained spreading or deep upwelling. This is the perisphere. Taken together the plume head and CL scenarios fill up the entire Indian and South Atlantic oceanic mantle with plume heads and enriched continental lithosphere, leaving little room for the 'undepleted asthenosphere' which is supposed to exist in the shallow mantle everywhere to provide fuel for the mid-oceanic ridge system. The depleted MORB reservoir must therefore be below the enriched layer as in the perisphere model. Figures 5 and 6 show no evidence for the low seismic velocities which should underlie much of Brazil and SW Africa if the plume head hypothesis is valid. Instead we find faster than average seismic velocities in these regions, with the slower velocities confined to the young ocean and the younger parts of Walvis Ridge and the Rio Grande Rise, i.e. the newer rather than the older manifestations of the Tristan de Cunha hotspot. Southern African lithosphere should also have been affected by the earlier Karoo and Madagascar events, in

PLUMES, PLATES AND TOMOGRAPHY either the plume head or lithospheric stretching models of CFB. The time since the conjectured plume head, or stretching event, about 130 Ma, is too short for the lower lithosphere or asthenosphere to cool off substantially, since the major part of the effect is much deeper than the thermal diffusion distance. The regions of South America and Africa which should have been affected by the plume head do not differ from the surrounding continental areas. The seismic evidence is therefore not favourable to the plume head, lithospheric stretching or lithosphere delamination hypotheses. It is, however, consistent with Pangaea starting to break-up after having moved over mantle which is hotter in some places than in others. The last places of break-up coincide both with present hotspots and ancient cratons. The Atlantic ocean did not open initially at Tristan da Cunha, St Helena and Iceland. CFB and OPB may be the result of rapid draining of an already partially molten shallow mantle, aided by passively induced upwelling from great depth and adiabatic decompression and melting.

The Karoo (180 Ma) and Siberian (250 Ma) events These are two of the more massive continental flood basalt provinces and they have both been attributed to the arrival of a gigantic plume head, or extensive stretching of the lithosphere over a plume head (White & McKenzie 1989). These CFB provinces are old enough so that some cooling of the affected lithosphere, and top of the asthenosphere, may have occurred but they should still differ from adjacent areas that did not experience these thermomechanical traumas. Figures 5 and 6 show, however, that these regions are no different from other areas in their vicinity. If they were uniformly faster one might argue that the depletion of the plume head in melt offsets the higher temperatures but there is little to distinguish these areas, which were expected to be damaged by stretching and plume head insertion, from other areas which were not. It has also been proposed that the Madagascar lithosphere was removed during late Cretaceous magmatism by the Marion plume (Mahoney et al. 1991). The lithosphere in this part of the world therefore received double damage in the plume head hypotheses. Actually, southern Africa received triple damage because the later Paranfi-Etendeka event also would have affected this area if it were due to a 2000 km wide plume head.

117

The inferred lack of damage to the Siberian lithosphere is consistent with an analysis of lithospheric thickness associated with the Siberian trap event. The Siberian lithosphere is calculated to have been 180 km thick, as measured by deformation accompanying the loading (Zorin & Vladimirov 1989). There are several other studies showing normal or thick lithosphere under the sites of continental flood basalt magmatism (Watts & Cox 1989). The thickness of the Siberian platform lithosphere in the late Permian and Triassic, when intensive trap magmatism occurred, differed insignificantly from the present thickness, 180-200 km (Zorin & Vladimirov 1989). The heat flow 30-40 mW m -2 is close to the minimum values for continents. Normal lithosphere and a sub-lithosphere source for the CFB are indicated. The Siberian flood basalts formed during a period of rapid continental drift during the final assembly of Pangaea and just after Siberia docked. It was preceded by Ural-Taimyr subduction and back-arc extension in Siberia. Pangaea had just drifted rapidly north by 30°, possibly placing Siberia over the North Atlantic -Greenland tomographic slow region in the uppermantle. Other events of note at this time include plutons in the Kara massif and USA, massive accretion of terranes in the circumPacific including Wrangellia, Sonoma and Hercynian orogenies, Oslo Graben activation, rifts around Greenland, flood basalts in Morocco, Greece, Iran and the Caucases, an increase in seawater 87Sr/86Sr, and rapid true polar wander. This may have been a time of significant global plate reorganization, and, therefore, extensive magmatism at new plate boundaries. The rapid continental drift and the massive plate reorganization and microplate assembly taking place at the time of the Siberian flood basalts suggests that back-arc basin activity or drift over hot mantle, combined with plate extension, rather than a plume head, may have been responsible for the Siberian CFB. The tomography shows that the polar North AtlanticGreenland area has slow seismic velocities below 400 km depth. This polar high temperature region is the likely location of Siberia, just after a rapid northward movement of Pangaea. The North Atlantic Tertiary Province occupied approximately this same region 175 Ma later. A large region of the mantle that remains thermally distinctive for a long period of time may represent an uppermantle convection cell or several adjacent cells. The area now represented by the N Atlantic was covered for a long period of time by thick cratonic lithosphere and has not recently been cooled by subduction.

118

D.L. ANDERSON E T A L .

Discussion What are we to make of these tomographic results which seem to contradict the premises of all currently popular plume and flood basalt scenarios? Certainly, there were massive basalt flood events in all the areas discussed. Yet there is no evidence of lithosphere damage or the presence of a massive plume head which has been conjectured to be dragged away from the scene by subsequent plate motions. The hotspots associated with all of these events (except possibly for Siberia which has left no track) are embedded in slower than average mantle and these presumably hot regions can be traced down to 200 to 300 km. At these depths, however, the hotspots are not spots but are simply part of an extensive meandering low-velocity region of the uppermantle. These LVA are often parallel to ridges and coastlines rather than to spreading directions. There are many regions at depths between 200 and 400 km where the velocities are even lower than beneath the surface hotspots; backarc basins, NE Pacific and the Indian ocean, to name a few. Most of the Pacific ocean has low seismic velocities at most depths in the uppermantle. The rapid onset and short duration of basaltic flooding events suggests to some (Griffiths & Campbell 1990; Richards et al. 1990; Hill et al. 1991) that this marks the arrival of a plume head. The steady-state plume model has been discounted by these authors because of the presumed time constraints associated with lithospheric stretching and passive upwelling. However, in all of the currently popular plume hypotheses, it is assumed that the mantle is isothermal (based on selective bathymetry) except near plume heads and slabs, and that melting only occurs when a hot plume can rise to depths shallower than the ordinary thickness of the lithosphere. It has also been pointed out that CFB often precedes extensive rifting (White & McKenzie 1989; Hill et al. 1991) and that stretching and rifting are often unaccompanied by CFB or that rifting often produces small volume alkalics rather than large volume tholeiitic floods. We might point out that CFB tholeiites are actually similar to mid-ocean ridge tholeiites and can be considered as contaminated, fractionated MORB, or a precursory picrite, instead of being basalts from an independent, fertile reservoir (Anderson 1982d, 1989). Many hotspots (OIB) are very close in trace element and isotopic chemistry to MORB and some are more depleted than the CFB which are thought to be their initial expression. It may be that the differences between MORB and CFB and other 'enriched' magmas have more to do with shallow

mantle (perisphere) interactions than with the fertile reservoir, which they all might share. Depleted picrites and komatiites may be the deep and hot parent magmas that service both ridges and hotspots. Off-ridge hotspots tend to be the most enriched. In these cases there is more opportunity for sub-lithospheric cooling, fractionation and contamination with the shallow metasomatized mantle, prior to eruption. The ultimate melts are more likely to be evolved, small melt residues, that indicate deeper crystal-melt equilibration than basalts erupted under thin lithosphere spreading centres. Continental extension or rifting, however, allows massive melting upon adiabetic ascent, or even spilling of accumulated magmas. Thus, eruption through thick lithosphere, or from below thick lithosphere, introduces complications other than the temperature of the sub-lithospheric mantle. That is to say, it is not only the temperature of the uppermantle that controls the chemistry, quantity and style of eruption. The thickness of the lithosphere can, in fact, affect the temperature of the uppermantle. Anderson & Bass (1984) showed that temperatures below some 150-200 km even under cold cratons are close to the dry solidus and if material below this depth is allowed to rise adiabatically it will melt. In hotter parts of the mantle, or if the mantle contains volatiles, the melting point is probably generally exceeded at sublithospheric depths. In high heat flow and tectonically active regions the seismic velocities are very slow and imply the presence of melt to depths of order 390 km. Very small melt fractions can cause large velocity reductions. If cratonic mantle rifts, or drifts over hotter than average mantle (hotcells), there can be extensive sublithospheric melting without the need for a deep plume to cause melting or to trigger extensive volcanism. If the overlying lithosphere is under compression there is little likelihood of melt extraction and melts may pond and be available for rapid extrusion, without extensive precursory rifting, if the stress regime turns to tensile or extensional. The sublithospheric mantle may be contaminated by subduction and metasomatism and, therefore, have chemical properties often attributed to CL. A mixing fractionation scheme for converting deep depleted plumes to enriched OIB-type magmas by shallow mantle processes satisfactorily explains the Sr-Nd-Pb isotope systems (Anderson 1983a, b, 1985), as well as the trace-element chemistry of hotspots. The small-melt fraction enriched magmas may be kimberlitic or lamproitic and may also contain recycled sediments and hydrous

PLUMES, PLATES AND TOMOGRAPHY fluids. The perisphere/asthenosphere is probably laterally and vertically inhomogeneous. There is an interesting correlation of hotspot initiation sites and the presence of ancient thick cratonic lithosphere. The cratonic plate is probably about 180-220 km thick (Anderson & Bass 1984; Anderson 1990). The Archaean cratons are particularly evident in the tomographic maps of Zhang & Tanimoto (19914, b, 1992). Surface wave studies may have less radial resolution than body-wave studies and the shallow high-velocities may be smeared to greater depth but, in any case, the velocities above 200 km under the oldest cratons are faster than elsewhere, including other ancient platforms. Precambrian shields, in general, are themselves faster than younger areas, and the high-velocities extend to greater depth. The seismic plus the flexural studies therefore imply thick (100-200 km), cold, strong lithosphere under ancient cratons. Part of this is due to temperature but there is probably also a mineralogical or chemical component to the high velocities (Anderson & Bass 1984). We have already mentioned that many of the CFB provinces, including those that occurred prior to local sea-floor spreading, are adjacent to thick continental lithosphere or Precambrian shield crust. These include Parami, Karoo, Siberia, Greenland and Deccan. Interestingly, the last places to fail as the Atlantic ocean opened up are just those places which have been attributed to the weakening effects of plumes (Morgan 1981). These places are also regions of thick cratonic lithosphere. These age and lithospheric thickness relationships are inconsistent with plume theories and add yet another element of coincidence to the timings and locations of the plumes conjectured to be related to hotspots. The evolution of hot mantle under thick continental lithosphere, and its subsequent adiabatic ascent, are expected to be quite different than under thinner lithosphere. It has often been remarked upon that alkali basalts and CFB tend to occur along ancient suture zones. This is particularly evident for the Deccan traps, which occur near an ancient rift-horst suture that splits the subcontinent of India. The Deccan traps were emplaced on thick cratonic lithosphere, along an ancient suture, near a new Indian ocean spreading centre, on a small rapidly moving continent. This is another series of coincidences in the deep mantle plume hypotheses.

Fixity of hotspots The relative fixity of hotspots was one of the initial motivations for anchoring them in the deep

119

mantle which, at the time, was thought to be non-convecting and rigid (Wilson & Burke 1972; Burke & Wilson 1976). Associating hotspots with propagating fractures, plate reorganizations or lithospheric discontinuities has been criticized for not providing a fixed reference system. Contrary to common belief, however, there is no rationale for assuming that deep mantle plumes, particularly from D " , should be fixed relative to one another or stationary relative to the plates e.g. 'It should also be pointed out that steady deep plumes also do not provide a convincing explanation of stationarity of hotspots since plumes in high Rayleigh number convection tend to move horizontally on times comparable to the turn over time scale of the mantle -- 108 years' (Griffiths 1986); 'The fixity of hotspots with respect to plate motions is difficult to explain with simple mantle convection models. In fact, no numerical experiment has ever demonstrated such behaviour, that is, plumes that are independent of upper boundary layer motions' (Duncan & Richards 1991). When a relative motion between hotspots was inferred by Molnar & Stock (1987), Olson (1987) asserted that such motion was to be expected for entities arising from such a deep source. Actually, if hotspots are fixed to one another for periods as long as 120 million years, as argued by Duncan & Richards (1991) then this is evidence for robust plumes, rather than frail plume tails. One should favour the association of strong hotspots with the main buoyant mantle upwellings or particularly hot convection cells. Hotspot fixity also suggests a layered mantle style of convection where high-viscosity bottomed-out slabs, hot regions of the high-viscosity lower mantle, convectively maintained bumps on chemical interfaces, and stagnation points of uppermantle return flow can provide a stable uppermantle reference system. Individual uppermantle convection cells may also maintain their position relative to other cells. 'Hotspots' may actually be hotcells. In multicellular convection distant cells do not communicate readily or rapidly. Because of the shearing action of plates and their ability to entrain shallow uppermantle material one would want to have the hot upwellings to originate at, or to extend to, depths of at least 300 km and, possibly, into the transition region. These hot regions of the uppermantle probably preferentially occur in areas where subduction has not been extensive for the past 100 Ma or so. It is not necessary, as generally assumed, that upwellings originate in thermal boundary layers since phase changes and partial melting can also trigger instabilities. Rifting and spreading can also initiate upwellings.

120

D.L. ANDERSON E T A L .

Once an upwelling initiates, it generates buoyancy and low-viscosity and may maintain itself in one place, just as has been proposed for lower mantle plumes. It bears emphasizing that there is no chemical or physical characteristic of hotspots that demands a lower mantle or CMB origin, or even an origin in a thermal boundary layer. We still have no convincing evidence that hotspots are the result of active upwellings or that they are due to lower thermal boundary instabilities. If hotspots are the tops of active upwellings, they could represent hot uppermantle convection cells, in a steady-state scenerio, or an overturning uppermantle convection cell, in an unsteady convection scenerio. They do not need to be the tops of deep mantle thermals. The location of the enriched, or hotspot, reservoir is not agreed upon. Many authors feel that either a strong continental lithosphere or the lower mantle are the only possible locations of ancient isolated reservoirs. What is needed is a mechanism that allows about 1-2 Ga of isolation between the depleted and enriched reservoirs. The enriched reservoir may consist, in part, of ancient recycled continental material so it needn't be completely isolated for this amount of time. Also, the depleted reservoir may have been even more depleted in the Archaean so it also need not be completely isolated. The best location for an ancient depleted reservoir is below the depth of slab dehydration and sediment removal, and the depth of melt trapping by rising diapirs or sublithospheric magma ponding. The mesosphere is one such location. A strong buoyant layer, restricted to continental mantle is unstable (Kincaid 1990). A low-viscosity buoyant (possibly volatile-rich) layer can probably avoid entrainment into the deeper mantle (the oil slick effect) and is a suitable source and sink of enriched material. Considering the extensive regions that Zhang & Tanimoto (1991a, b, 1992) have found in the uppermantle having extremely low seismic velocities, and apparently high temperature, we see no reason why lithospheric extension over particularly hot regions cannot lead to extensive basaltic volcanism, whether these extended regions occur in continental or oceanic regions, or whether or not stretching is extensive enough actually to cause melting in unmelted mantle. It is not even necessary that a local upwelling was occurring prior to lithospheric extension. As a check on this hypothesis we might look for correlations between uppermantle low velocity regions, plate stresses and the presence or absence of extensive magmatism. The association of magmatic provinces with low-velocity re-

gions beneath extending plates is fairly obvious. We now ask, what about those hot regions which are amagmatic and what about those regions which have experienced tension without massive volcanism? There are extensive low-velocity regions under the Indian ocean, Central America, western equatorial Pacific, and the western Atlantic. These are all areas of regional convergence or compression. They lack midplate volcanism confirming that hot uppermantle is not a sufficient condition for magmatism. On the other hand many areas of Europe and Asia, such as Lake Baikal and the Rhine graben, are obviously under extension but lack extensive volcanism, presumably because the uppermantle is cold, as judged from the high seismic velocities. Some of these regions experienced secondary extension associated with convergence, a condition that can cool off the uppermantle if the lithosphere is thick. Rifting in China, back-arc basins, island arcs and high plateaus are examples of secondary extension caused by regional compression. Extensive volcanism and the existence of nondepleted basalts are often taken as prima facie evidence for the presence of plumes or plume heads. With equal logic they could be taken as evidence for the existence of lithospheric extension, hot uppermantle and enriched or metasomatized uppermost mantle. The presence of extensive volcanism and depleted basalts could also be taken as evidence for long sustained extension and the tapping of larger and deeper volumes. A moving plate, overriding a hot region of the mantle, and being put into tension, will behave, in many respects, as if it were being impacted from below by a giant plume head. An internal contradiction in the plume head theories is the displacement of 'depleted asthenosphere' by plume material along the mid-Atlantic and other oceanic ridges. The geochemical rational for importing enriched or non-depleted material from depth, in a plume, is that the shallow mantle (asthenosphere) under midocean ridges is 'known' to be depleted. A given convection cell may tend toward homogeneity and a relatively limited range of temperature. However, there are probably many convection cells in the mantle and distant ones do not readily equilibrate with each other, in either chemistry or temperature. Particularly hot cells may even be delivering melt to the base of the lithosphere. Hotcells are an alternative to plume heads. Plume theories presuppose a more-or-less isothermal mantle and require that anomalously hot mantle be imported from great depth. The tomography shows that the upper-

PLUMES, PLATES AND TOMOGRAPHY mantle is far from isothermal and that V L V A do not appear to be circular plume heads or to be restricted to hotspots. Perisphere

Anderson (1987a, 1989) suggested that basalts represent blends of melts from a deep depleted fertile source and a shallow refractory enriched layer. The shallow layer is enriched by subduction, primarily by dehydration of the upper part of the slab, and trapped small melt fractions such as lamproites or kimberlites. For example, CFB and OIB could be deeply derived depleted tholeiites or picrites which have been contaminated by lamproite melts as they evolve and cool in the shallow mantle. Because the shallow enriched layer could be, mechanically, either lithosphere or asthenosphere, we prefer to refer to it as the 'perisphere'. Since mantle has little strength at temperatures above 550-650°C most of the enriched shallow mantle is formally the asthenosphere. A weak, low-density shallow layer can be isolated from the underlying mantle because it is difficult to entrain. On the other hand it may be pushed aside or depleted by the earliest stages of passive or active upwelling of the deeper mantle. The new word is necessary because the asthenosphere is often referred to as 'depleted' and the continental lithosphere is usually referred to as 'enriched'. The word 'asthenosphere' is also used interchangeably with 'uppermantle', 'convecting mantle' and 'well mixed mantle' as well as 'depleted mantle'. The uppermantle is not necessarily homogeneous or entirely depleted (Anderson 1983a). Hotspot magmas could obtain their distinctive geochemical signatures by small melt fractions from the metasomatized shallow layer, both under continents and oceans. A prediction of this model is that picrites and komatiites, i.e. high temperature or large-volume melts, are depleted in the large-ion lithophile elements and low in 87Sr/86Sr and that enriched magmas are replaced by depleted magmas as continental rifting evolves to sea-floor spreading. Mature and rapidly spreading ridges should be the most depleted. Migrating ridges should tap more of the shallow mantle and, therefore, be more enriched (e.g. Indian ocean ridges). The small-volume melts from the perisphere (LIL- and isotopically enriched) are probably triggered by ascent of deeper, more fertile plumes or diapirs. We have used 'uppermantle' throughout this paper instead of 'upper mantle'. In the current literature the following equalities are assumed: upper mantle = convecting mantle = depleted mantle -- MORB-source = well mixed mantle.

121

We wish to refer to 'uppermantle' simply as the mantle above the 650 km discontinuity. It is not necessarily homogeneous, well mixed or unlayered or convecting as a unit. We feel that much unnecessary sematic confusion exists simply because of unproved attributes which have been assigned to conventional terms such as 'lithosphere', 'asthenosphere' and 'upper mantle'. The 'lower mantle' itself has also been variously described as 'primitive', 'primordial', 'undegassed', 'OIB-source' and so-on. Some geochemical attributes of OIB are automatically assigned to a lower mantle source because it is 'known' that the upper mantle is homogeneous and depleted. The lower mantle may, in fact, be depleted instead of primitive or enriched, and barren instead of fertile (Anderson 1983a). We appreciate the advice of several anonymous reviewers and thank B. Storey for his hospitality during the meeting and his patience regarding the manuscript. This work was supported by National Science Foundation Grants EAR-90-02947, EAR-91-08246, EAR-91-03526, and OCE-91-16213. Contribution No. 5117, Division of Geological and Planetary Sciences, California Institute of Technology, Pasadena, C A 91125.

References

ALLEGRE,C. J. & TURCOTTE,D. L. 1985. Geodynamic mixing in the mesosphere boundary layer and the origin of oceanic islands. Geophysical Research Letters, 12, 207-210. ANDERSON, D. L. 1975. Chemical plumes in the mantle. Geological Society of America Bulletin, 86, 1593-1600. - 1979a. Chemical stratification of the mantle. Journal of Geophysical Research, 84, Bll, 6297-6298. 1979b. The upper mantle transition region: Eclogite? Geophysical Research Letters, 6, 433-436. - 1981. Hospots, basalts and the evolution of the mantle. Science, 213, 82-89. - - 1982a. The chemical composition and evolution of the mantle. In: AKIMOTO,S. d~ MANGHNANI, M. H. (eds) High-Pressure Research in Geophysics. Advances in Earth and Planetary Science, 12,301-318. 1982b. Hotspots, polar wander, Mesozoic convection, and the geoid. Nature, 297, 5865, 391-393. 1982c. Isotopic evolution of the mantle; a model. Earth and Planetary Science Letters, 57, 13-24. - 1982d. Isotopic evolution of the mantle; the role of magma mixing. Earth and Planetary Science Letters, 57, 1-12. - 1983a. Chemical composition of the mantle. Journal of Geophysical Research, 8 8 , supplement, B41-B52. - 1983b. Kimberlites and the evolution of the

122

D . L . ANDERSON ETAL. mantle. In: KORNPROBST,J. (ed.) Developments in Petrology. Proceedings of the Third Kimberlite

Conference Vol. 1,395-403. 1985. Hotspot magmas can form by fractionation and contamination of MORB. Nature, 318, 145-149. 1987a. The Depths of Mantle Reservoirs. In: MYSEN, B. O. (ed.) Magmatic Processes: Physiochemical Principles. Special Publication No. 1 edition, Geochemical Society. 1987b. Global mapping of the uppermantle by surface wave tomography, in composition, structure and dynamics of the lithosphere asthenosphere-system. In: Fuct4s, K. & FROIDEVAUX,C. (eds) Geodynamics Series. American Geophysical Union, Washington DC, 16, 327. 1989. Theory of the Earth. Blackwell Scientific Publications, Boston, 1-366. 1990. Geophysics of the continental mantle: an historical perspective. In: MENZIES, M. (ed.) Continental Mantle. Clarendon Press, Oxford. 1991. Chemical boundaries in the mantle. In: SABADINI, R. & LAMBECK, K. (eds) Glacial Isostacy, Sea-Level and Mantle Rheology. Kluwer Academic Publishers, Dordrecht, 379-401. - - - & BAss, J. D. 1984. Mineralogy and composition of the upper mantle. Geophysical Research Letters, 11,637-640. & BASS, J. O. 1986. The transition region of the Earth's upper mantle. Nature, 320, 321-328. ~, TANIMOTO, T. & ZHANG, Y.-S. 1992. Plate tectonics and hotspots: The third dimension. Science, 256, 1645-1651. BAUDRV, N. & KROENKE, L. 1991. Intermediatewavelength (400-600 km) South Pacific geoidal underlations: their relationship to linear volcanic chains. Earth Planetary Science Letters, 102, 430-443. BONATI'I, E. 1990. Subcontinental mantle exposed in the Atlantic-ocean on St-Peter-Paul Islets. Nature, 345, 800-802. BURKE, K. C. & WILSON,J. T. 1976. Hot spots on the Earth's surface. Scientific American, 235, 46-47. CAMPBELL, I. H. t~ GRIFFITH, R. W. 1990. Implications of mantle plume structure for the origin of flood basalts. Earth Planetary Science Letters, 99, 79-93. CAMPBELL, P., GRIFFITHS, R. W. & HILL, R. I. 1989. Melting in an Archaean mantle plume: Heads its basalts, tails its komatiites. Nature, 339,697-699. CANTAREL, P. ~; LIPPOLT, H. J. 1977. Alter und Abfolge des Vulkanismus der Hucheifel, Neues Jahrbuch.fuer Mineralagie. Geologie und Palaeontolgie Monetshefte, 600-612. CAZENAVE, A. & LAGO, B. 1991. Long wavelength topography, seafloor subsidence and flattening. Geophysical Research Letters, 18, 1257-1260. COLIN, P. & FLEITOUT, L. 1990. Topography of the ocean floor: Thermal evolution of the lithosphere and interaction of deep mantle heterogeneities with the lithosphere. Geophysical Research Letters, 17, 1961-1964. COURTILLOT, V., BESSE, J., VANDAMME, D., MON-

-

-

-

-

-

TIGNY, R., JAEGER,J. J. & COPPETI'A,H. 1986. Deccan flood basalts at the Cretaceous/Tertiary boundary? Earth and Planetary Science Letters, 80, 361-374. CSEREPES, L. • CHRISTENSEN,J. 1990. Three-dimensional convection under drifting plates. Geophysical Research Letterg, 17, 1497-1500. DAVIES, G. F. 1988. Ocean bathymetry and mantle convection 2. Small-scale flow. Journal of Geophysical Research, 93, 10481-10488. 1990. Mantle plumes, mantle stirring and hotspot chemistry. Earth and Planetary Science Letters, 99, 99-109. DAVE¥, F. J. 1981. Geophysical studies in the Ross Sea region. Journal of the Royal Society of New Zealand, l l , 465-479. DUNCAN, R. A. & RICHARDS, M. A. 1991. Hotspots, mantle plumes, flood basalts and true polar wander. Review of Geophysics, 29, 31-50. DZIEWONSKI,A. M. & ANDLERSON, D. L. 1984. Seismic tomography of the Earth's interior. American Scientist, 72, 483-494. ELDER, J. 1967. Convective self-propulsion of continents. Nature, 214, 657-660. FORSYTH, D. & UYEDA, S. 1975. On the relative importance of the driving forces of plate motion.

Geophysical Journal. Royal Astronomical Society, 43, 163-200. FROIDEVAUX, C. ~; NATAF, H.-C. 1981. Continental drift: What driving mechanism? Geologische Rundschau, 166-176. GARFUNKEL, Z. 1975. Growth, shrinking and longterm evolution of plates. Journal of Geophysical Research, 80, 4425-4432. GIRET, A. & LAMEYRE, J. 1975. Inverted alkalinetholeiitic sequences related to lithospheric thickness in the evolution of continental rifts and oceanic islands. Journal of African Earth Science, 3, 261-268. GRIFFITHS, R. W. 1986. Dynamics of mantle thermals with constant buoyancy or anomalous internal heating. Earth Planetary Science Letters, 78, 435-446. & Campbell, I. H. 1990. Stirring and structure in mantle plumes. Earth Planetary Science Letters, 99, 66-78. GROTZINGER, V. & ROYDEN, L. 1990. Elastic strength of the Slave craton at 1.9 Gyr and implications for the thermal evolution of the continents. Nature, 347, 64-66. GUPTA, M. L. & GAUR, V. K. 1984. Surface heat flow and probable evolution of Deccan volcanism. Tectonophysics, 105, 309-318. GURNIS, M. 1988. Large-scale mantle convection and the aggregation and dispersal of continents. Nature, 332, 695-699. HAGER, B. H. 1978. Oceanic plate motions driven by lithospheric thickening and subducted slabs. Nature, 276, 156-159. & RICHARDS, M. A. 1989. Long-wavelength variations in Earth's geoid. Philosophical Transactions of the Royal Society of London, A328, 309-327. HAWKESWORTH, C. J., MANTOVANI, M. S. M., -

-

PLUMES, PLATES AND T O M O G R A P H Y TAYLOR, P. N. & PALACZ, Z. 1986. Evidence from the Paranfi of South Brazil for a continental contribution to D U P A L basalts. Nature, 322, 356-359. HILL, R. I., CAMPBELL,I. H. & GRIFFITHS,R. W. 1991. Plume tectonics and the development of stable continental crust. Exploration Geophysics, 22, 185-188. HOUSEMAN, G. A. 1983. The deep structure of ocean ridges in a convecting mantle. Earth and Planetary Science Letters, 64, 283-294. JACKSON, E. D. & SHAW, H. R. 1975. Stress fields in central portions of the Pacific plate. Journal of Geophysical Research, 80, 1861-1874. JORDAN, T. H. 1975. The continental tectosphere. Reviews of Geophysics and Space Physics, 13, 1-12. - 1978. Composition and development of the continental tectosphere. Nature, 257, 745-750. KEDAR, S., ANDERSON, D. L. & STEVENSON, D. J. 1992. Relationship between hotspots and mantle structure: Correlation with whole mantle seismic tomography. Proceedings of the NATO-ASI Meeting, Dynamic Modeling and Flow in the Earth and Planets. (In press). KENT, R. 1991. Lithospheric uplift in eastern Gondwana. Geology, 19, 19-23. KINCAID, C. 1990. The dynamical interaction between tectosphere and large scale mantle convection. LOS, 71, 1626. KLEIN, E. M., LANGMUIR, C. A., ZINDLER, A., STAUDIGEL, H. 8¢ HAMELIN, B. 1988. Isotope evidence of a mantle convective boundary at the Australian-Antarctic discordance. Nature, 3 3 3 , 623 -629. LACHENBRUCH, A. H. 1976. Dynamics of a passive spreading center. Journal of Geophysical Research, 81, 1883-1902. LAGO, B., CAZENAVE,A. tf¢ MARLY, J.-C. 1990. Regional variations in subsidence rate of lithospheric plates. Physics of the Earth and Planetary Interiors, 61,253-259. MAHONEY, J. J., NATLAND, J. H., WHITE, W. M., POREDA, R., BLOOMER, S. M., FISHER, R. L. & BAXTER, A. N. 1989. Isotopic and geochemical provinces of the western Indian Ocean spreading centers. Journal of Geophysical Research, 94, 4033-4052. - - N I C O T r E T , C. & DuPUY, C. 1991. Madagascar basalts: tracking oceanic and continental sources.

Earth and Planetary Science Letters, 104, 350-363. MAIA, M. & DIAMENT,M. 1991. An analysis of the altimetric geoid in various wavebands in the central Pacific Ocean: Constraints on the origin of intraplate features. Tectonophysics, 190, 133-153. MCKENzIE, D. & O'NIONS, R. K. 1983. Mantle reservoirs and ocean island basalts. Nature, 3 0 1 , 229-231. MOLNAR, P. & STOCK, J. 1987. Relative motions of hotspots in the Pacific, Atlantic and Indian Oceans since late Cretaceous time. Nature, 327, 587-591. MORGAN, W. J. 1971. Convective plumes in the lower mantle. Nature, 230, 42-43.

123

1981. Hotspot tracks and the opening of the Indian and Atlantic Oceans. In: EMILIANI,C. (ed.) The Sea. Wiley Interscience, New York, vol. 7, 443 -487. - 1983. Hotspot tracks and the early rifting of the Atlantic. Tectonophysics, 94, 123-139. MORICEAU, C., CHRISTENSEN,U. ~1;FLEITOUT,L. 1991. Geoid and topography associated with sublithospheric convection. Earth and Planetary Science Letters, 1 0 3 , 395-408. NATAF, H.-C., NAKANISHI, I. & ANDERSON, D. L. 1984. Anisotropy and shear velocity heterogeneities in the upper mantle. Geophysical Research Letters, l l , 109-112. --, & ~ 1986. Measurements of mantle wave velocities and inversion for lateral heterogeneity and anisotropy, Part III, Inversion. Journal of Geophysical Research, 9 1 , B7, 7261-7307. O'CONNOR, J. M. & DUNCAN, R. A. 1990. Evolution of the Walvis Ridge-Rio Grande Rise hot spot system: Implications for African and South American plate motions over plumes. Journal of Geophysical Research, 95, B l l , 17,475-17,502. OFFICER, C. B. d~; DRAKE, C. L. 1983. Plate dynamics and isostasy in a dynamic system. Journal of Geophysics, 54, 1-19. OKAL, E. & BATIZA, R. 1987. Hotspots: The first 25 years. In: KEATING,B H., FRYER,P., BATIZA, R. BOEHLERT, G. W. (eds) Seamounts, Islands and Atolls. Geophysical Monograph 43, American Geophysical Union, Washington D C, 1-11. OLSON, P. 1987. Drifting mantle hotspots. Nature, 327, 559-560. PARSONS,B. & SCLATER,J. G. 1977. An analysis of the variation of the ocean floor bathymetry and heat flow with age. Journal of Geophysical Research, 82, 803-827. PLATE, D. S., HAWKESWORTH,C. J., MANTOVANI,M. & SHOKOWSKY,W. 1990. Mantle plumes and flood basalts stratigraphy in the Paran~, South America. Geology, 18, 1223-1226. PHIPPS, S. P. 1988. Deep rifts as sources for alkaline intraplate magmatism in eastern North America. Nature, 334, 27-31. RICHARDS, M. A. & WICKS, C. W., Jr. 1990. S-P conversion from the transition zone beneath Tonga and the nature of the 670 km discontinuity. Geophysical Journal International, 1 0 1 , 1-35. --, DUNCAN, R. A. & COURTILLOT, V. E. 1989. Flood basalts and hot-spot tracks: Plume heads and tails. Science, 246, 103-107. SCHILLING, J.-G. 1973. Iceland mantle plume: geochemical study of Reykjanes Ridge. Nature, 242, 565-571. SLEEP, N. H. 1990. Hotspots and mantle plumes: Some phenomenology. Journal of Geophysical Research, 95, No. B5, 6715-6736. STOREY, M., SAUNDERS,A. D., TARNEY,J., LEAT, P., THIRLWALL, M. J., THOMPSON, R. N., MENZIES, M. A. 8¢ MARRINER, G. F. 1988. Geochemical evidence for plume-mantle interactions beneath Kerguelen and Heard islands, Indian Ocean. Nature, 336, 371-374.

- -

124

D. L. ANDERSON ETAL.

SYKES,L. R. 1978. Intraplate seismicity, reactivation of preexisting zones of weakness, alkaline magmatism, and other tectonism postdating continental fragmentation. Reviews of Geophysics and Space Physics, 16, 621-688. TANIMOTO, T. 1990a. Long wavelength S-wave velocity structure throughout the mantle. Geophysical Journal International, 100, 327-336. 1990b. Modeling curved surface wave paths: membrane surface wave synthetics. Geophysical Journal International, 1023, 89-100. VEEVERS, J. J. 1984. Phanerozoic earth history of Australia. Clarendon Press, Oxford. VOGT, P. R. 1991. Bermuda and Appalachian-Labrador rises. Geology, 19, 41-44. WATTS, A. B. & Cox, K. G. 1989. The Deccan Traps: an interpretation in terms of progressive lithospheric flexure in response to a migrating load. Earth and Planetary Science Letters, 93, 109-112. WHITE, R. S. 1988. The Earth's crust and lithosphere. Journal of Petrology, Special volume, 1-10. & MCKENZIE, n . 1989. Magmatism at rift zones: The generation of volcanic continental margins and flood basalts. Journal of Geophysical Research, 94, 7685-7729. WILSON, J. T. 1990. On the building and classification of mountains. Journal of Geophysical Research, -

-

-

-

-

95, 6611-6628. & BURKE, K. 1972. Plate tectonics and plume mechanics. EOS, 54, 238-239. ZEHNDER, C. M., MUTTER, J. C. & BUHL, P. 1990. Deep seismic and geochemical constraints on the nature of rift-induced magmatism during breakup of the North'Atlantic. Tectonophysics, 173, 545-565. ZHANG, Y. S. & TANIMOTO, T. 1991a. Global Love wave phase velocity variations and its significance to plate tectonics. Physics of the Earth and Planetary Interiors, 66, 160-202. & TAN~MOTO, T. 1991b. Three-dimensional global upper mantle structure and its tectonic significance. Journal of Geophysical Research, (In press). & TANIMOTO,T. 1992. Ridges, hotspots and their interaction as observed in seismic velocity maps. Nature, 335, 435-439. ZHOU, H.-W. & ANDERSON, D. L. 1989. Search for Deep Slabs in the Northwest Pacific Mantle. Pro-

ceedings of the National Academy of Science, USA, 86, 8602-8608. ZORIN, Y. A. & VLADIMIROV, B. M. 1989. On the genesis of trap magmatism of the Siberian platform. Earth and Planetary Letters, 93, 109-112.

The stress regime associated with continental break-up MARTIN

H . P. B O T T

Department o f Geological Sciences, University o f Durham, South Road, Durham DH1 3LE, UK

Abstract: Extensional syn-rift structures at passive margins indicate that continental breakup occurs in response to horizontal deviatoric tension in the continental lithosphere. An abrupt change in the state of stress occurs at the onset of the post-rift (break-up) unconformity when the large tension ceases. The stress regime at the time of continental break-up has been modelled by finite element analysis with a view to simulating the observed features. A local compression is associated with thinned crust and an opposing local tension is produced by hot, low density upwelted asthenosphere. The resulting stress associated with stretching and thinning of the lithosphere is relatively insignificant and is likely to be compressive unless the normal continental crust is very thin. One possible source of break-up tension is the large tensional loading stress associated with an underlying hot, low density upper mantle, such as now occurs in the present day uplifted continental rift systems. It is uncertain whether such local tension could readily cause break-up within the present compressional stress regime of normal continental regions. Another possible source of widespread continental tension would be the occurrence of subduction on opposite sides of a large continental mass such as Pangaea, giving rise to instability. At such times, splitting could more readily propagate outwards from a domed plume region leading to break-up, thus involving both active and passive factors. The radical change in stress regime at the onset of seafloor spreading is readily modelled as the change in stress associated with the development of a new weak plate boundary.

The occurrence of continental break-up is most conspicuously recorded in the deep geological structure at passive margins (e.g. Klitgord et al. 1988). At typical inter-plume margins, a syn-rift stage of variable timespan marks the inital stretching of the continental lithosphere. This is followed by a post-rift stage starting at the onset of seafloor spreading when thermal subsidence becomes the predominant tectonic process. Large extensional stresses dominate the syn-rift stage, but these abruptly give way to a much more subdued stress regime at the onset of the post-rift stage. U n d e r the present-day tectonic framework, normal low-lying continental regions are observed to be subjected to horizontal deviatoric compression (Zoback et al. 1989). These regions are not now in a state of tensional instability ready for break-up. On the other hand, uplifted regions such as continental rift systems and mountain ranges are subjected to local tension, and it has been suggested often that continental rift systems may be incipient foci of break-up. This paper investigates the factors which may be important in producing the extensional stress at the time of continental break-up and in accounting for the abrupt change in stress regime at the onset of seafloor spreading, using finite element stress and deformation analysis. The

primary Cause of break-up is also briefly discussed. Other recent mechanical studies of the process of continental break-up have emphasized the necking process (e.g. Braun & Beaumont 1989; Lin & Parmentier 1990).

Principles of modelling The types of stress regime which might affect a passive margin just prior to final break-up have been modelled by elastic-viscoelastic finite element analysis, using the D u r h a m package described by Waghorn (1984) and Bott et al. (1989). Quadrilateral and triangular isoparametric elements with eight and six nodes respectively have been used. The finite element grid (Fig. la) extends for 1050 km horizontally and 400 km vertically. The nodes down the right edge are constrained to zero horizontal displacement so that the grid can be taken to represent half of a symmetrical structure 2100 km wide •with the zone of incipient break-up at the centre. The nodes along the bottom and left edge are free except when stated to the contrary. Isostatic restoring forces are applied to nodes along the top of the model and at the Moho, with a density change of 2980 kg m -3 at the surface under air, 1950 kg m -3 under water and 350 kg m -3 at the normal Moho depth.

From STOREY,B. C., ALABASTER,T. & PANKHURST,R. J. (eds), 1992, Magmatism and the Causes of Continental Break-up, Geological Society Special Publication No. 68, pp. 125-136.

125

126

M . H . P . BOT£

(a)

Grid J

2 E v ¢O..

tm

400 0

Distance (km)

1050

Grid over thinned lithosphere, with anomalous densities

(b)

-

-I~5

0.0

__

-1~1 -3~3 +3~.0 +28~.5

+291.0

0.0

E

11"0

v tt3..

-290 7 -42.0

-42.0

0.0

c

E)

-16.7

-16.7

-16.7

0.0 -

100 850

.

Distance (km)

1050

Fig. 1. (a) The basic finite element grid used for the models, which is mirror-imaged about the right edge to represent a symmetrical model 2100 km wide and 400 km in depth extent. (b) An enlargement of the region of lithospheric thinning near the right edge of the grid, showing anomalous densities in kg m -3 assigned to the shallow elements in all the models except BK1, BK2 and B K 4 A as a result of crustal thinning and thermal anomalies.

The unthinned crust is assumed to be 40 km thick and the unthinned lithosphere to be 100 km thick. The uppermost 20 km of the normal crust is elastic. Young's modulus has been taken as 0.9×1011 Pa for the crust and 1.75x1011 Pa for the upper mantle, and Poisson's ratio is 0.27 throughout the models. The lithosphere below the elastic upper crust has been assigned a viscosity of 1023 Pa s and the underlying asthenosphere 1021 Pa s. The models have all been run for 1000 time steps of 500 years each so as to approach viscoelastic equilibrium. The approach to equilibrium is not of geological significance as the time span of evolution is much longer than

5 x 105 years. Although the viscosity distribution is rather simple, nevertheless the displacements and stresses in the elastic layer after full upward stress concentration has occurred are almost independent of the details of the underlying viscosity distribution. At the modelled passive margin representing the zone of break-up, the elastic layer, crust and lithosphere all thin to 20% of their normal continental thickness over a horizontal distance of 150 km. Anomalous densities in the model are referenced to the density-depth profile beneath normal continental lithosphere. The mean density of the crust is assumed to be 350 kg/m 3 below

STRESS REGIME AT CONTINENTAL BREAK-UP that of the upper mantle in the break-up zone. The anomalously low densities in the region of thinned lithosphere and upwelled asthenosphere have been computed using a continental thermal model of Morgan & Sass (1984) with a temperature of 1150°C at 100 km depth. If partial fusion occurred at this depth the temperature would be about 180 K higher. For a shield geotherm it would be much lower. The lithospheric temperature at a given location such as the Moho is maintained adiabatically during thinning and the temperature in the asthenosphere is assumed to be adiabatic. The anomalous densities associated with the lithospheric thinning have been computed from the anomalous temperature field assuming a coefficent of volume thermal expansion of 2.5x 10 -5 K -1 for the crust and 3x10 -5 K -~ for the upper mantle, and these are shown on the enlarged portion of the grid in Fig. lb. The stretched and thinned lithosphere is assumed to be in local isostatic equilibrium vAth the normal continental lithosphere, and this is accomplished by applying a distribution of boundary pressure representing the deficient load of the seawater relative to upper continental crust. This is shown in the models at the relevant nodes, using the same scale as that for the deviatoric stress. Zero horizontal out-ofplane deviatoric stress has been assumed. The two in-plane deviatoric stresses are thus equal and opposite so that only one of them needs to be displayed.

Local stresses associated with lithospheric thinning As a preliminary, the local stresses resulting from thinning of the lithosphere are studied in isolation. The model used is as shown in Fig. 1. The results apply in principle to sedimentary basin evolution as well as to continental breakup. Stresses are produced in the strong nearsurface elastic layer as a result of anomalous subsurface loading and the associated isostatic surface loading. The stresses are produced by the combined effect of subsurface and surface loading of opposite polarity which isostatic equilibrium requires. Two distinguishable types of stress occur. Loading stress (Bott 1991) is produced as a result of the opposite subsurface and surface forces and bending stress additionally occurs where the loading is not symmetrical so that the lithosphere bends. It is the loading stress, rather than bending stress, which drives most extensional and compressional tectonics. In the models presented here, except in model

127

BK4 and its variants, bending stress is minimal because of the imposed local isostatic equilibrium. The loading stress concentrates upwards into the strong near-surface elastic layer as the deviatoric stresses relax to small values in the viscoelastic material over the timespan of the modelling (Kusznir & Bott 1977). There are two opposing contributions to the loading stress produced by the thinned lithosphere. First, the positive anomalous densities produced by crustal thinning, with its associated isostatic surface depression, gives rise to a horizontal deviatoric compression which reaches a maximum value of 196 MPa in the elastic layer above the centre of the symmetrical model where the thinning is maximum (model BK1, Fig. 2a). Second, the negative subsurface load associated with the anomalous high temperatures, with its associated isostatic uplift, gives rise to horizontal deviatoric tension reaching a maximum of 98 MPa (model BK2, Fig. 2b). These two isolated effects are superimposed in model BK3 (Fig. 2c), which represents lithospheric thinning with both crustal thinning and thermal effects. These opposing effects combine to yield a maximum deviatoric compression of 98 MPa. Thus the effect of lithospheric stretching in this model is to produce a local compressional stress which will act to oppose further extension. As the thermal anomaly and the crustal thinning produce opposing stress fields, the actual resulting stress depends on the relative thickness of crust and lithosphere. The thicker the crust relative to the lithosphere, the stronger the compression. This variation has been tested by varying the crustal thickness while the other parameters remain identical to those of model BK3 (Fig. 2c). If the crust is reduced to 30 km thickness, the resulting maximum stress is a small deviatoric tension of 17 MPa, and if it is further reduced to 20 km, this tension increases to 44 MPa. The stress resulting from lithospheric thinning also depends on the assumed geothermal profile. If the temperature were to reach 1330°C at 100 km depth, appropriate to partial fusion (McKenzie & Bickle 1988), then the slightly increased thermal effect would increase the maximum tension in model BK2 by about 8%, and the maximum compression in model BK3 would be reduced by a similar amount. The difference would be almost insignificant. However, if the temperature perturbation extended deeper into the asthenosphere, the thermally originating tension might be substantially greater, significantly reducing the compression in model BK3 or even resulting in a small tension. For the most realistic scenario, with a crust

128

M . H . P . BOTT

Model BKI- Crustal thinning only DeviatoricStress

(a)

,

100 MPa t

Elastic crust

T

l

~-

Ductile crust

E

t

0

Moho

~

°

. .~. . . . . . . . . .". . . . . . . . . . . .

I I

--"----

+350

,~--.'"

--

+350

"~ .'7"_..-"". . . . . . . . . . . . . .". . . . . . . . . . . . . . . . . .

v

t-

Q.

# s SS

LithospherJc mantle

C3

0

,/•

0

#,• •

•

•

•

# •

~,#A,

•

.........

Asthenosphere

110

Model BK2: Thermal anomaly only Deviatoric Stress . .- . . . . . . .

(b)

--.

100 MPa ~-z__- -..-~-'- : -.5;.....

.2

......

E v

t-

f s s ~, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Q. (1) #

C3

#f

#f

•

•

#s if

•

110 Model BK3: Combined effect Deviatoric Stress -

(c)

100 MPa t

t

t

t

1

-t ,In

E

- .........................

"o;

...............................

v

t-" •

## S sS

C3

## r S •

,/

##

•

•

•

•

s ° s. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . #

110 800

Distance (km)

1050

STRESS REGIME AT CONTINENTAL BREAK-UP rather thicker than 30 km, the local stress produced by lithospheric thinning is likely to be a small compression which increases as the thinning proceeds. If the thermal perturbation is much larger than in m o d e l BK3 or the crust is very thin, there may possibly be a small but hardly significant tension. This demonstrates that the source of the extensional stress leading to break-up must originate elsewhere. The two main possibilities, as discussed qualitatively by Bott (1982), are (1) tensional stress produced by an underlying upper mantle hot spot below the region of lithospheric thinning, and/or (2) tension associated with subduction plate boundary forces. These are further modelled in the following sections.

Tension from underlying hot spot As indicated above, one possible source of strong lithospheric tension is an underlying hot, low density region in the mantle (hot spot), possibly arising from a deep mantle plume as envisaged by Loper (1985). The low density region gives rise to flexural isostatic uplift and to quite large associated loading stresses (Bott & Kusznir 1979; Bott in press). Since loading stresses increase linearly with the depth of a wide load (Bott 1991), quite small density anomalies at large depth give rise to large tensions in the strong upper lithosphere above. Model BK4A (Fig. 3d) shows the effect of such a hot spot on lithosphere of uniform thickness, with the elastic layer 20 km thick. The hot spot beneath the symmetrical model is 500 km wide, with an anomalously high temperature of 120 K at 100 km depth falling stepwise to zero at 400 km depth (as shown for model BK4, Fig. 3b). The maximum deviatoric tension of 75 MPa occurs in the elastic layer above the centre of the 500 km-wide anomalous region near the left edge of the model BK4A in Fig. 3d.

129

In model BK4 (Fig. 3a-c), such a hot spot has been superimposed on the lithospheric thinning model BK3. Additionally the asthenospheric temperatures above 100 km depth in the region of thinned lithosphere have been raised by 140 K above those in model BK3. The isostatic effect of the low densities associated with the hot spot is to produce a submarine uplift of about 1.2 km (Fig. 3a). In contrast to the small compression in model BK3 and the small tension in model BK4A, there is a large tension reaching a maximum of 255 MPa in the thin elastic layer. The effect of the hot spot is to produce tension which greatly outweighs the small compression resulting from the lithospheric thinning. As the lithosphere thins in response to the hot spot tension, the tension in the thinning elastic layer progressively increases during the necking process.

Modelling the break-up event It was pointed out that a radical modification of the stress regime from strongly tensional to more subdued occurs at the onset of seafloor spreading. The main new mechanical factor at this stage is a new zone of weakness introduced across the whole lithosphere at the ridge axis. Such a weak zone is introduced into the hot spot model BK4 in this section. The zone of weakness at the axis at the right edge of the grid has been modelled by extending the low viscosity asthenosphere up to the surface. The grid has been appropriately modified, and now includes a single low viscosity element 4 km wide replacing the elastic layer adjacent to the symmetrical axis. The effect of this modification is shown in Fig. 4, where two models with a weak axial region are compared with model BK4 which has a thin but strong crest extending across the axial region. The two new models BK5 and BK6 represent the extremes of weak and strong resistance to plate motion.

Fig. 2. (a) Model BK1 (enlarged axial region down to 110 km depth) showing horizontal deviatoric compressional stress (thick black lines) in the upper elastic layer produced by the anomalous density distribution shown in kg m -3 resulting from crustal thinning and the associated surface topographical loading (shown by arrows on the same scale as the stresses). In this and other figures, the continuous line denotes the Moho, the widely spaced broken line denotes the base of the upper crustal elastic layer, the medium spaced broken line marks the base of the lithosphere and the dotted lines separate regions of differing anomalous density; compressions are black and tensions are red. (b) Model BK2 (enlarged axial region down to 110 km depth), showing the horizontal deviatoric tension (thick red lines) resulting from the anomalous low densities caused by the anomalous temperature field. The density distribution in this model is identical to that in Fig. lb, except for the values displayed which are marked by asterisk. These occur where the high densities related to crustal thinning have been excluded. (c) Model BK3 (enlarged axial region down to 110 km depth), showing the horizontal deviatoric stress (compressions black, tensions red) resulting from both crustal thinning and thermal effects. The anomalous density distribution is identical to that shown in Fig. lb.

0901

(w~I) eouels!CI

OOg 0i7

(p) edlAI O01

OSOI

sseJlS o!Jo},E!AaCi eJeqdsoql!l peuu!qlun '},ods 1OH :V17~!I3 lepolAI

(w~I) eouels!cI

009 OLl

,/ •

L~-

/

sS s

E) -o

js

j~ s

s

I J

s

B

(o) sseJ},S o!Jo},e!AeCl

e d ~ O0 L

OcjO l

0

(w~I) eoue},s!(]

0017 EJ .9-

"(3 i,,-i. :3"

v

/

s

3

°

s

- ;

"

o

(q)

sseJIS O!JOIE[AeC]

e d N O0 L

¢'1

r-']

"13

L 3 (e)

lods loll :17NI3 lePOlAi JA~Ofl "d "H "IAI

O~I

STRESS REGIME AT CONTINENTAL BREAK-UP Model BK5 (Fig. 4b) has a weakened axis with free boundary conditions at the left edge as in model BK4. The effect of weakening the axis is to reduce the deviatoric tension in the elastic layer adjacent to the crest by about 40%. The maximum tension decreases from 255 MPa in model BK4 to 153 MPa in model BK5. This radical change in stress regime occurs because of the increased pressure exerted by the weak element on the edge of the elastic layer, with the exact amount of the reduction depending on the rheology and width of the weak zone. The increased compressional boundary traction is in equilibrium with the viscous drag on the base of the elastic plate as it moves away from the axis at a modelled velocity of 8 mm a -1. In model BK6 (Fig. 4c), the left edge of the elastic layer has been constrained to zero horizontal displacement, representing strong resistance to plate motion. The reduction in the tension in the elastic layer near the axis is more dramatic, falling to only 20 MPa which is about 8% of the tension in model BK4. In this model, the change in pressure exerted on the right edge of the elastic layer produces a supplementary compression which extends uniformly to the fixed edge of the plate. In the region above the hot spot this has the effect of almost annihilating the tension but away from the hot spot towards the left edge a significant horizontal deviatoric compression occurs, as can be seen on the whole-width plot shown in Fig. 4d. These two models with weak crest represent the extremes of weak and strong resistance to plate motion. In both models there is a significant reduction in tension in the break-up region, but the effect is more spectacular when the resistance to plate motion is stronger. Reality is probably somewhere between the two extreme models. These models predict an abrupt change in stress regime at the time of break-up.

Plate interior tension from subduction pull Another possible source of plate interior tension which may be of importance in continental break-up comes from plate boundary forces.

131

Present day unelevated continental regions are characterized by compression (Zoback et al. 1989), reflecting the dominance of ridge push in the present plate regime. In certain past times, however, subduction may have occurred on opposite sides of a large predominantly continental plate such as Pangaea. Subduction pull acting on both overriding and subducting plates has been modelled by Bott et al. (1989) and Whittaker et al. (in press). The subduction pull (trench suction) acting on opposite sides of such a plate would give rise to pervasive tension within the plate in a way which does not apply today (Bott 1982). A possible difficulty with this scenario comes from an inference of Price & Audley-Charles (1987), with reference to the Australia-Banda Arc collision zone, that the downgoing slab is likely to be fractured into slices by the bending stresses. They suggested that the °subduction plate boundary forces may be rendered ineffective as a result. However, the models presented in this paper and by Bott (1991) demonstrate that upper lithospheric stresses associated with sub-lithospheric loading do not require elastic body connection but are equally effectively developed if weak non-elastic material intervenes. Model BK7 (Fig. 5a, b) represents the simple lithospheric thinning model BK3 subjected to a tension of 200 MPa (deviatoric tension of 100 MPa) applied to the left edge of the 20 km thick elastic layer. This is equivalent to a tectonic force of 4.0× 1012N m -1. Because of the horizontally fixed right edge, this tension extends right along the elastic layer, being amplified where the layer thins. Tension of this magnitude, or possibly higher, might be expected from subduction on both sides of the plate. The resulting stress distribution (Fig. 5a, b) shows a maximum tension of 183 MPa at the axis. This represents the superimposition of a compression of 98 MPa as in model BK3 and the applied tension which is significantly increased as a result of the thinning of the elastic layer. The result of applying such a tension to the lithosphere as it thins would be to increase the deviatoric tension from an initial unthinned value of 100 MPa to about double this

Fig. 3. (a) Surface displacement profile for model BK4, showing the isostatic response to the deep low density region. (b) The near-horizontal deviatoric stresses for model BK4. The anomalous density of the deep, hot region is shown in kg m -3 but elsewhere anomalous densities are as in Fig. lb. Note the very large deviatoric tensions which occur where the lithosphere is thinned, in contrast to model BK4A shown in (d) below in which the lithosphere is unthinned. (e) Enlarged central region of model BK4, as outlined in (b) above, with stress at the same scale. (d) The near-horizontal deviatoric stresses shown at the same scale for the enlarged central region of model BK4A, which differs from model BK4 above in having an unthinned lithosphere.

132

M . H . P . BOTT

Model BK4: Hot spot Deviatoric Stress

100 MPa .

(a)

,

•

t

t

t

s

~

t

,,..- , " "

•

40 Model BK5" Weak crest, edge free Deviatoric Stress •

(b)

t

~

t

100 MPa

t..L.

t

-

t

1~

rl..

40 Model BK6: Weak crest, edge fixed Deviatoric Stress

(c)

40 800

•

~

t

-

- .......

100 MPa ~

t

- .....

t

lit

----"I

Distance (km)

1050

Model BK6: Full length plot Deviatoric Stress

1O0 MPa

(d) iii,

E

•

-12"

•

•

~,

o

v

t-

.........................

C1. C) CI

•

-8"

o

400 0

Distance (km)

1050

Fig. 4. (a) The deviatoric stresses in the enlarged central region of model BK4 (Fig. 3c) down to 40 km depth, for comparison with models BK5 and BK6 shown in (b) and (c) below, where the stress is at the same scale. (b) The near-horizontal deviatoric stresses for model BK5, which differs from model BK4 in having a weak crestal region extending to the surface• Note that the deviatoric tensions are smaller than in model BK4. (e) The near-horizontal deviatoric stresses for model BK6, which differs from model BK5 in having the nodes at the left edge of the lithosphere constrained to zero horizontal displacement. Note that the tensions are much smaller than those of models BK4 and BK5. (d) The full length model BK6, showing the horizontal deviatoric compressions which develop away from the hot spot and extend to the distant edge of the model. The stress scale is a factor of two larger than that of (a) to (c) above.

STRESS REGIME AT CONTINENTAL BREAK-UP value, as the lithosphere, crust and elastic layer all thin by a factor of five. This is to be expected for the necking process. Model BK8 (Fig. 5c) demonstrates the effect of introducing a weak zone at the crest, as in model BK5. In this model, the maximum deviatoric tension is reduced by about 20%, with the amount of reduction depending on rheology and width of the weak zone. The combined effect of the excess pressure exerted by the viscoelastic element on the right side of the elastic layer and the applied tension on the left edge is to cause the plate to move away from the axis at a rate such that the edge forces are in equilibrium with the underside viscous drag as the plate moves away from the axis at about 8 mm a -1.

Discussion Two recognizable sources of lithospheric tension may be of importance in causing continental break-up, (1) local tension associated with an underlying upper mantle hot spot and (2) regional tension resulting from subduction on opposite sides. The most favourable scenario for break-up would be when both sources of tension occur together. Such a situaton may have applied to the break-up of Pangaea as suggested by Bott (1982) and shown in Fig. 6. Unlike the present-day continental masses, Pangaea would be unstable when subjected to regional tension and break-up would be triggered by the local tension produced by the hot spot. Under such a stress regime, the crack would readily be able to propagate laterally away from the hot spot region. It is not clear that major break-up could occur readily in response to just one of these two sources of tension. At the present time of predominantly compressional stress in continental regions, it is possible that break-up could be initiated in a hot spot region such as East Africa or Baikal, but it is not obvious whether the crack could propagate much beyond the uplifted rift region. This possibility needs further investigation but this is outside the scope of the present contribution. What is the mechanical role of magmatism in break-up? The low density magma can only be a marginal factor in producing the tensional stress because of its small volume. However, magma must certainly play an important role in weakening and thinning the lithosphere. Similarly, the grain of the continental basement is likely to be a controlling factor in the detailed geometry of break-up.

133

Conclusions The stresses which may cause, and which result from, continental break-up have been modelled by elastic-viscoelastic finite element analysis, with the following findings: (1) Thinning of the lithosphere by uniform extension gives rise to local stress in the strong upper layer as a result of subsurface loading by the anomalous density distribution and surface loading from the accompanying isostatic topography. Crustal thinning gives rise to horizontal deviatoric compression and the thermally-produced low densities give rise to tension. For a 100 km-thick lithosphere and 40 km-thick crust, the compression outweighs the tension, causing a resulting deviatoric compression of 98 MPa in the model when the lithosphere, crust and elastic layer have all thinned by a factor of five. The tension outweighs the compression when the original crust is thinner than about 30 km, other factors being the same. Most usually, then, lithospheric thinning should produce a small local compression which opposes the thinning process. Deviatoric tension originating externally is thus required to initiate and drive breakup, and to give rise to instability as necking takes place. (2) One possible source of the tension is a low density upper mantle of the type associated with a hot spot. Quite a small thermal anomaly in the upper mantle of 120 K at 100 km depth decreasing to zero at 400 km depth gives rise to a region of anomalously low density which produces a deviatoric tension of 75 MPa in the 20 km-thick unthinned elastic upper lithosphere in model BK4A. As the lithosphere thins by a factor of five (model BK4) the modelled tension increases to 255 MPa. This is because of the concentration of the tension into the 4 km-thick elastic layer which greatly outweighs the local compression produced by the thinning as summarized in (1) above. (3) Another possible source of plate interior tension which may be of importance in break-up would be the presence of subduction pull on opposite sides of a large continental plate such as Pangaea. A deviatoric tension of 100 MPa in the 20 km-thick elastic upper lithosphere (tectonic force of 4× 1012 N m -I) would increase to over 180 MPa as the modelled lithosphere thins by a factor of five. (4) Weakening the axis of the symmetrical region of lithospheric thinning has been simulated by allowing a weak element to extend up to the surface. This has the effect of reducing the tension in the region of thinned lithosphere in proportion to the resistance to plate motion, and

134

M . H . P . BOTT

Model BK7: Applied tension Deviatoric Stress (a)

B

.,-,-.,

J

100 MPa g

E ((~.

a 400

Distance (km)

0

Deviatoric Stress

(b)

,

0

1050

t .__.

100 MPa 1' T ~ -T_ - - ,--"~_~-. . . . . .

t

E v

~, si"

.

.C: ,41.,I Q.

.J

.

.

f SJ ~

f

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

•

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

,/

CI / s¢ s# # "

110 800

...............................

Distance (km)

1050

Model BK8: Applied tension, weak crest Deviatoric Stress , t t

(c)

E

".,

~...:

100 MPa T t 111'

............. :..;..::.'.';.'. .... : ................... oS

v

¢-

t

sf D

,.I-.I

:

Q.

.

• •

CI

##

/

#e ##

•

•

./:.-- .... - .....................................

"

•

•

•

## ##

:/

r..............................................................

110 800

Distance (km)

1050

STRESS REGIME AT CONTINENTAL BREAK-UP Continental

lithosphere

135

in tension

(a) Enhanced tension :.::..:.).!

(b) Continental Split Tension Relieved

(c) Fig. 6. Possible mechanism of continental break-up shown in three stages, illustrating the roles of the two complementary sources of tension. (a) Subduction on opposite sides of a continental plate gives rise to tension in the continental lithosphere. (b) Later development of a sub-continental hot spot gives rise to enhanced tension and lithospheric thinning. (c) Continental break-up with formation of new ocean ridge and ridge push force, relieving the tension. Fsu, trench suction; Frp, ridge push. After Bott (1982). explains the abrupt change in stress regime at the onset of seafloor spreading as seen in the postrift u n c o n f o r m i t y at passive margins. (5) It is suggested that the most favourable condition for continental break-up occurs w h e n a very large continental mass such as P a n g a e a is m a d e unstable ~by subduction pull acting on opposite sides. In such circumstances, the de-

v e l o p m e n t of a hot spot in the underlying u p p e r m a n t l e (from a p l u m e rising from the core-mantle b o u n d a r y ? ) w o u l d be e x p e c t e d to trigger break-up. It is n o t clear w h e t h e r the present-day uplifted continental rift systems could initiate break-up in the presence of the m o r e stable compressional stress regime of the continents.

Fig. 5. (a) Model BK7 (full length) which is identical to model BK3 (Fig. 2c) except that an extensional boundary stress of 200 MPa is applied to the left edge down to 20 km depth (base of elastic layer), as shown. A small tension occurs in the elastic layer over the whole length. (b) Enlarged axial region of model BK7, as outlined in (a) above, showing the increase of tension resulting from the thinning of the elastic layer. (e) Enlarged axial region of model BK8, which differs from model BK7 in having a weak axial region extending to the fffifface, with a consequent reduction in the tensions, which are shown at the same scale as that of (a) and (b) above.

136

M. H. P. BOTT

References

Boar, M. H. P. 1982. The mechanism of continental splitting. Tectonophysics, 81,301-309. - 1991. Sublithospheric loading and plate-boundary forces. Philosophical Transactions of the Royal Society of London, A337, 83-93. In press. Modelling the loading stresses associated with active continental rift systems. Tec-

tonophysics. & KUSZNIR, N. J. 1979. Stress distributions associated with compensated plateau uplift structures with application to the continental splitting mechanism. Geophysical Journal of the Royal Astronomical Society, 56, 451-459. , WAGHORN, G. D. & WHITrAKER, A. 1989. Plate boundary forces at subduction zones and trencharc compression. Tectonophysics, 170, 1-15. BRAUN, J. & BEAUMONT,C. 1989. A physical explanation of the relation between flank uplifts and the breakup unconformity at rifted continental margins. Geology, 17,760-764. KLITGORD, K. D., HUTCHINSON,D. R. & SCHOUTEN, H. 1988. U.S. Atlantic continental margin; structural and tectonic framework. In: SHERIDAN, R. E. & GROW, J. A. (eds) The Atlantic Continental Margin, U.S. Geological Society of America, The Geology of North America, I-2, 19-55. KuszNm, N. J. & Boar, M. H. P. 1977. Stress concent-

ration in the upper lithosphere caused by underlying visco-elastic creep. Tectonophysics, 43, 247-256. LIN, J. & PARME~CrlER,E. M. 1990. A finite amplitude necking model of rifting in brittle lithosphere. Journal of GeophysicalResearch, 95, 4909-4923. LOPER, D. E. 1985. A simple model of whole-mantle convection. Journal of Geophysical Research, 90, 1809-1836. MCKENZIE, D. & B1CKLE,M. J. 1988. The volume and composition of melt generated by extension of the lithosphere. Journal of Petrology, 29, 625 -679. MORGAN, P. & SASS,J. H. 1984. Thermal regime of the continental lithosphere. Journal of Geodynamics, 1, 143-166. PRICE, N. J. & AUDLEY-CHARLES, M. G. 1987. Tectonic collision processes after plate rupture. Tectonophysics, 140, 121-129. WAGHORN, G. D. 1984. Numerical modelling of the stress regime at subduction zones. PhD thesis, University of Durham. WHITTAKER, A., Boar, M. H. P. & WAGHORN, G. D. In press. Stresses and plate boundary forces associated with subduction plate margins. Journal

of Geophysical Research. ZOBACK, M. L., ZOBACR, M. D. AND 27 OTHERS1989. Global patterns of tectonic stress. Nature, 341, 291-298.

Karoo igneous activity, and the early stages of the break-up of Gondwanaland K. G. C O X

Department o f Earth Sciences, Parks Road, Oxford, OX1 3PR, UK

Abstract: The widespread early Jurassic Karoo vulcanism of southern Africa appears to have been generated from a large-scale mantle plume originating in the sub-lithospheric mantle, but with a considerable additional contribution in some areas of material from the lithosphere. Subduction-related vulcanism may also play a significant role, but the relationships of the two types of vulcanism, if indeed a clear distinction does exist, are at present uncertain. Continental break-up probably took place in two stages, the first of which, in the early Jurassic, was not accompanied by true seafloor spreading. In this stage, the reactivation of an ancient shear zone is thought to have resulted in the movement of Antarctica north-eastwards relative to Africa (in the present-day African reference frame), with the creation of a zone of thinned continental crust, or a mixed zone of oceanic and continental crust, in Mozambique. The second stage took place 10-30 Ma later as Antarctica and Madagascar moved southwards relative to Africa with the formation of the oldest-known oceanic crust of the Indian Ocean.

The generation of the Mesozoic continental flood basalt volcanic rocks of Brazil, southern Africa, and Antarctica poses a series of geological problems, principally involving the evaluation of the respective roles of mantle plumes, subduction, and continental break-up. Figure 1 shows the pre-break-up configuration of the Gondwanaland continents, the broad zone of continental flood basalts (CFB) lying parallel to the Pacific margin of the super-continent, the proposed positions of mantle plumes, and the Gondwanide orogen, the site of subduction during the earlier Phanerozoic. Two important lines of continental separation are also schematically shown: the eastern one represents the earliest phase of break-up and is the principal subject of the present work. The geometrical relationships alone make it clear why the three themes, plumes, subduction, and break-up, may be thought to be inter-related. The present work will draw largely on the evidence from the Karoo province of southern Africa, and will focus particular attention on the role of plumes in the vulcanism. The Karoo Province

General accounts of the geology of the province are given by Eales et al. (1984) and Cox (1988), and the petrology and geochemistry are extensively discussed in Erlank (1984). The existing radiometric dating of the Karoo (Fitch & Miller 1984) presents a number of puzzling problems (see discussion by Cox 1988), but there is no

doubt that substantial eruptive activity occurred within the period c. 198 Ma-173 Ma, which places the province firmly in the early Jurassic. For present purposes the rocks closest to the southeast coast of Africa are the most important (see Fig. 2). These are mainly represented by seaward-dipping extrusive sequences passing beneath a cover of Cretaceous and younger sedimentary rocks, but also extending inland along a faulted trough in the Limpopo valley. In some areas of northern Mozambique the volcanic rocks are hidden completely by the sedimentary cover, but elsewhere there are long stretches of continuous outcrop, the whole zone extending for more than 1500 km from Mozambique Island in the north to Natal in the south. This eastern zone of the Karoo contains a great variety of rock types compared with the rest of the province in the continental interior, where monotonous evolved tholeiitic basalts and dolerites are the rule. In the east the greatest diversity is found in the northern part of the Lebombo monocline, round Nuanetsi and Tuli, and in the Mateke-Sabi monocline (see Fig. 4). Here, after initial minor eruptions of nephelinite, a great thickness of highly-potassic picrite basalts and basalts appeared, followed by extensive rhyolites. The area is also the site of important plutonic complexes, mainly of gabbro and granite. Elsewhere in the east e.g. the southern Lebombo, basalts and rhyolites are dominant, or the sequences are mainly basaltic as in northern Mozambique. The volcanic sequences in the eastern zone are impressively thick (e.g. perhaps

From STOREY,B. C., ALABASTER,T. & PANKHURST,R. J. (eds), 1992, Magmatismand the Causes of ContinentalBreak-up, Geological Society Special Publication No. 68, pp. 137-148.

137

138

K.G. COX

0 I

2000 ,

i

I

km

\

/

.

\ X

\

\

,j

!!.Karoo

Fig. 1. Pre-drift configuration of Gondwanaland showing distribution of continental flood basalt provinces (stipple) and their relationship to the Gondwanide orogen. Circles represent the postulated heads of the Karoo and Paran~i plumes (White & McKenzie 1989). Heavy broken lines show sites of the future S Atlantic ahd the initial areas of the Indian ocean.

as much as 12.5 km in Swaziland. Within the basaltic sequences, because of poor exposure, the effect of strike faulting in repeating the sequence is difficult to estimate). In the continental interior, by contrast, the maximum preserved thickness of volcanics is probably only 1.5 km (e.g. Lesotho), though the dolerites of the Karoo basin must substantially augment the total amount of basaltic magma emplaced into the crust.

The Karoo Plume Burke & Dewey (1972) first proposed a connection between plumes and Karoo vulcanism, placing one at Nuanetsi, and a second in the lower Zambezi valley, to account for the somewhat younger vulcanism of the Lupata Gorge (see Fig. 2). They suggested that rifted triple junctions controlled by basement structure were generated by plume activity. The faulted areas of volcanics extending along the Limpopo valley to the west of Nuanetsi, for example, is the proposed failed arm of such a rift system. More recently Campbell & Griffiths (1990) have also

proposed the existence of a large plume, the centre of which passed precisely beneath the Nuanetsi area. In contrast, White & McKenzie (1989, see Figs 1 and 3) proposed a plume centred further to the southeast, with a very broad active top more than 2000 km in diameter and circular in plan. While the other authors had deafly been decisively influenced by the existence of the thick picrite basalt sequences of Nuanetsi and adjacent areas, White & McKenzie (1989) sought only to position the plume axis on the continental margin, and postulate a plume diameter which was large enough to encompass, in a continental reconstruction, all the known seaward-dipping reflector sequences off the Dronning Maud Land coast. Such a circle also does in fact enclose most of the occurrences of Karoo volcanics on the African continent, including the seaward dipping sequences referred to above as the eastern zone. With regard to the Karoo plume there are two issues: what size and shape of the area did the plume influence, and where was its centre? Two lines of evidence suggest a 'big-plume' model of the White & McKenzie (1989) and Campbell &

THE KAROO PROVINCE

25*

139

35*

30*

40*

,',Z A M B I

!!iii!,i'i;i:~:~ili~~ozambi ,:~, ueq - 15"

~ntonioEnes

\ Beira

- 20*

- 25*

~ , O r angeT/~ Free State

~)urban

~

Post-KarooCover

~

Karoo and later lavas

~

Karoo sediments

~

Pre-Karoorocks 500 km

0 |

.......................

!

I

I

I

/

30*

35 °

40 °

,

.

- 30*

I

Fig. 2. Geological sketch map of SE Africa (after Eales et al. 1984). Numbered stars show postulated plume locations or plume axis locations: 1, Nuanetsi (Burke & Dewey 1972; Campbell & Griffiths 1990); 2, Lower Zambesi near Lupata Gorge; (Burke & Dewey 1972); 3, White & McKenzie plume axis (1989). Griffiths (1990) type, but n o t specifically centre d at Nuanetsi, probably represents the most realistic option. It has been argued elsewhere (Cox 1989) that the drainage patterns of CFB provinces preserve a record of uplift, probably generated by magmatic thickening of the crust (underplating), over huge areas consitent in scale with a big-plume model. In the Karoo case (Fig. 3), although the original domical pattern has been much-modified by erosion originating from the Limpopo valley rift, it is still well-preserved in the north-western part of Zimbabwe and in the area of the Transvaal, Orange Free State, and Lesotho. The directions of flow of the preserved dome-flank drainage patterns are certainly broadly consistent with a dome-centre

close to Maputo, as suggested by White & McKenzie (1989). The second argument depends on the strong influence on Karoo volcanic geology of veryobviously heterogeneous basement. The Burke & Dewey (1972) hypothesis implies this on the largest scale (i.e. on the scale of large cratonic blocks v. younger mobile belts) but it has also been demonstrated on smaller scales (i.e. between specific structural units within individual orogenic belts), particularly in the Nuanetsi-Tuli area (Cox 1970). Here, different zones within the basement Limpopo orogen have clearly influenced the location of Karoo sedimentary troughs, and of faulting, the emplacement of plutons, and, although Karoo lavas were spread

140

K.G. COX

\

1 0 I

600 ;

,

I

I

km

Fig. 3. Main drainage pattern of southeast Africa (after Cox 1989). The bold arc is the edge of the proposed plume head of White & McKenzie (1989), the centre lying c. 450 km ENE of Maputo. Outlined areas west of Harare and south of Pretoria are remnants of the original dome-flank drainage system. Shaded ornament shows area where dome-flank drainage has been destroyed by subsequent erosion. over the whole area, the distribution of the earliest sequences of lavas. This also seems highlyconsistent with a big-plume model, in which it is easy to visualize a large area subjected to plumegenerated extensional stress, but within which, rifting, vulcanism, and plutonism, are controlled by local physical heterogeneities. An area such as Nuanetsi, where the most primitive (hottest) basaltic rocks are found, does not therefore need to be directly above the plume axis. It simply represents the place where, at the present level of exposure, the most profound rifting occurred, and the vulcanism owes its existence only to the fact that it was within the area influenced by the plume. Triple junctions are thus still 'plumegenerated', as in the Burke & Dewey model, but the problem implied by that hypothesis ('how did the plume know where to come up, under such a special basement configuration?') need no longer concern us. A similar case has recently been argued by Thompson & Gibson (1991). Geochemical studies have added interesting but as yet unresolved ideas to the plume hypothesis. The northern parts of both the Paran,'i and the Karoo provinces, although the former is of early Cretaceous age, are both characterized by basaltic rocks highly-enriched in incompatible trace and minor elements e.g. K, Ti, P, Sr, Rb, Ba (Cox et al. 1967; Bellieni et al. 1984).

In the Karoo it is reasonably certain that the pierite basalts of Nuanetsi and the north Lebombo, which represent the most extreme example of such enrichment, are derived by the mixing of magmas from two quite different source components, because despite their early Jurassic age they yield a Proterozoic 'erupted isochron' (or 'pseudo-isochron') in the Sm-Nd system, which has to be interpreted as a mixing line (Ellam & Cox 1989, 1991). The enriched source endmember is likely to have been derived from ancient metasomatized mantle lithosphere, and the enriched provinces, of both the Paran~i and the Karoo, have thus been interpreted as representing the hottest, axial, area of the plumes, where contamination of sub-lithospheric melts derived from the plumes was most extreme (Ellam & Cox 1991). However, the proposed plume component, in so far as it is possible to deduce its composition against the background of heavy lithospheric contamination, does not appear to be a typical, so-called, 'plume-related' magma, i.e. similar to an ocean island basalt. In fact, at least as far as its incompatible element contents go, it looks more like MORB. This apparent discrepancy is so far unresolved and is regarded as a serious flaw in the model by some geochemists (numerous pers. comms). Geochemical study of these rocks continues, but it is worth remarking that the plume concept is essentially physical not chemical. Eventually it may be necessary to accept that the Karoo plume is geochemically different from those so far known in the oceans. The Ellam & Cox (1991) model has also been used to account for the geochemistry of basaltic rocks of the southern Lebombo and the thenadjacent Dronning Maud Land basalts of Antarctica described by Harris et al. (1990, 1991). These areas are interpreted as peripheral to the plume head and the rocks thus contain a much smaller lithospheric contribution. The hypothesis has never purported to explain the geochemistry of the widespread Ferrar dolerites and associated basalts of the Transantarctic Mountains, nor of the southern part of the Karoo province outside the southern Lebombo.

Subduction: geochemical evidence It is not proposed to discuss the evidence for subduction along the Pacific margin of Gondwanaland in detail in the present work. Southern Africa is not the best place to see evidence of this, and other works cover it more fully (e.g. Storey & Alabaster 1991). However, it has become clear that many of the rocks of the southern part of the Karoo province, particularly the widespread basalts and dolerites of the Lesotho

THE KAROO PROVINCE magma-type (Marsh & Eales 1984) have a geochemical signature that can be called 'calcalkaline' or 'subduction-related' (Cox 1983). Although this also applies to the southern Lebombo rocks mentioned above, it is a very mild signature in that area, and it is at least permissible to interpret the rocks as generated from a MORB-source (i.e. the plume) plus a very small contribution from a highly-enriched lithospheric component. The Lesotho magma-type however is impossible to interpret in this fashion, most obviously because it contains substantially less Ti than MORB. The Ferrar dolerites of the Transantarctic Mountains show the most extreme examples of this sort of geochemistry, being particularly enriched in Si and poor in Fe, and enriched in LIL incompatible elements relative to MORB while being depleted in HFS elements, notably Ti. In the Karoo, similar rocks are found amongst the dolerites of Falkland Islands, which were located in the extreme south of the province in the early Jurassic (Mitchell 1988). Initial Sr-isotope ratios in the Ferrar are generally high (c. 0.706-0.710), and not suprisingly crustal contamination has been invoked as an explanation. Direct crustal contamination does not appear however to stand up to detailed scrutiny, and the alternative of the incorporation of subducted sediment into the mantle source has been proposed (Hergt et al. 1991). At present however, there is little consensus as to the mineralogy of the source, whether melting took place under dry or wet conditions, whether melting was directly related to subduction at the time of the Karoo, or whether it simply reworked mantle that had been geochemically modified earlier, or indeed whether the source was essentially lithospheric or asthenospheric.

Plumes and subduction: their interplay Accepting that a reasonable argument can be made for the existence of a Karoo plume somehow similar to that proposed by White & McKenzie (1989), a number of outstandingly difficult problems remain. The ground plan of the White & McKenzie plume within southern Africa (Fig. 1) looks convincing enough, in that the dome-flank drainage patterns (Fig. 3) extend over the appropriate area, and if the NW Zimbabwean river pattern is projected back to the southeast, and the Orange-Vaal pattern to the northeast, a dome-centre somewhere near the proposed plume centre no. 3 (Fig. 2) is certainly indicated. However, the Antarctic half of the plume is less amenable to interpretation. Initially, it is clearly correct to include the basalts of

141

Dronning Maud Land within the orbit of the same plume, because geochemical and age similarities to Lebombo volcanics have been well demonstrated (Harris et al. 1990). But the Ferrar dolerites and Kirkpatrick basalts, and the huge Dufek intrusion, extend the zone of vulcanism through the Transantarctic Mountains as an apparently linear belt for a further 3000 kin. Furthermore, we have seen above how the geochemical evidence appears to suggest that subduction may be the important influence in areas close to the Pacific margin of Gondwanaland, while further into the hinterland, plumes are perhaps the dominant force. This constitutes a series of substantial geological problems, towards the solution of which one can at present do little more than pose the following questions. (1) Is the idea of a plume with a head circular in plan, too constricting? (2) Might the Karoo plume discussed so far merely be the western end of a very much more elongated mantle-upwelling (= plume?) which stretched right across Antarctica and even to Australia? If so, the postulated convective rolls of Froidevaux & Nataf (1981), which were specifically associated with subduction, need serious further consideration. In terms of geochemistry, a single plume or roll might produce quite different products in different areas, e.g. close to the site of subduction as opposed to further into the hinterland. (3) Alternatively, are plumes, and convective upwellings associated with subduction, really quite different physical phenomena? If they are, the presence of plumes in the hinterland of the subducting margin of Gondwanaland may be pure coincidence, but on the other hand might subduction somehow have triggered plume formation? It is clear that the reverse is unlikely to be true. Subduction was in evidence long before the plume was, e.g. Dalziel et al. (1987), Storey & Alabaster (1991).

Continental break-up The connection between CFB vulcanism and continental break-up has long been a preoccupying problem. In the Atlantic there is a clear near-coincidence in time between the onset of sea-floor spreading and the break-out of CFB vulcanism in the adjacent land masses, be it the late Triassic-early Jurassic development of the central Atlantic, the early Cretaceous opening of the south Atlantic, or the onset of sea-floor spreading in the North Atlantic in the Eocene. Cause and effect are of course still matters for

142

K.G. COX

debate. The earliest stages in the break-up of Gondwanaland do however paint a different picture. The earliest Jurassic vulcanism of the Karoo exhibits a major peak of activity at c. 200190 Ma (Fitch & Miller 1984) which cannot so far be matched by new sea-floor of the same age. In the Karoo case, though in the past it was tempting to favour models in which plumes might have a direct causative connection with continental break-up (e.g. Cox 1970), the information subsequently accumulated suggests that the Karoo plume was in fact something of a failure. In the western part of the Indian Ocean, sea-floor older than Mid-Jurassic has yet to be discovered, the oldest so far known being in the Somalia basin (anomaly M22 at 152 Ma, Cochran 1988), though it has been argued that seafloor spreading might have started as long as 170 Ma ago (Lawyer et al. 1991b). Nevertheless, on the present evidence there seems to be a gap of at least 20 Ma between the onset of vulcanism and the onset of sea-floor spreading. However, the Atlantic case suggests that the connection can exist under some circumstances, so the question is: how is the connection more direct in some cases than in others? As an answer, it seems likely that even if a large plume generates a substantial continental volcanic province, under unfavourable tectonic conditions continental break-up may not take place. In the next section, the tectonic condition of southern Africa during the late Palaeozoic and early Mesozoic is discussed. T e c t o n i c e v e n t s in s o u t h e r n A f r i c a

The general geology of southern Africa is admirably described by Tankard et al. (1982), and more specific papers relevant to the following discussion are provided by Rust (1975), Daly et al. (1989, 1991), and Groenewald et al. (1991). Profound and wide-spread unconformities began to affect southern Africa and neighbouring continents during the Devonian, as evidenced by the deposition of the marine strata now seen in the Cape Fold Belt (e.g. the Table Mountain Group, part of the Cape Supergroup), in Antarctica, and in the older strata of the Paran~i basin in South America. These rocks rest almost invariably on Precambrian basement, both Proterozoic and Archaean. Karoo sedimentation commenced in the Carboniferous with the glacial beds of the Dwyka tillite, deposited with little unconformity" on the older rocks in the south, but spreading widely on to the Precambrian basement elsewhere. In thc southern part of southern Africa the Karoo basin developed and accumulated large thicknesses of sedimentary rocks during the Permian and the Trias,

as was also happening in the Paranfi basin in South America. Some of the sedimentary rocks of the Karoo basin were perhaps still marine, for example the turbidites of the Ecca Group (Permian) in southern South Africa, so it is permissible to think of a Karoo marine transgression in some areas, followed by a transition to fluviatile and sub-aerial aeolian conditions during the deposition of the Beaufort Group, the Elliot Formation (red beds), and, ultimately, the Upper Triassic Cave Sandstone (now known as the Clarens Formation). Further north within southern Africa, outside the Karoo basin, the sedimentary sequences are thinner and probably less complete, and tended to develop in more isolated sedimentary troughs, such as are preserved in the valleys of the Limpopo, Zambesi, and Luangwa (part of the latter is visible at the northern edge of Fig. 2, about 400 km west of Lake Malawi), though the aeolian sandstones of the Clarens Formation probably covered most of the area. The unconformity is so profound that, for example, over wide stretches of Zimbabwe there is no preserved geological record between the Archaean and the early Mesozoic. By the end of the Trias, southern Africa, and probably much of Brazil, was a vast low-lying area dominated by aeolian and fluviatile sedimentation. The Cape Fold ~elt, associated with subduction along the margin of Gondwanaland, reached maximum activity during the Permian, and perhaps into the Trias, and built the highlands from which much of the Karoo sedimentary material in southern Africa was derived. The origins of the unconformity discussed above, are still however debatable. While the Karoo basin itself would appear to represent a foredeep of the Cape Fold Belt, the more isolated troughs of the northern part of southern Africa have been interpreted recently as half-graben structures reactivating NE-trending basement shear zones (Daly et al. 1989, 1991) as a consequence of compression from the Cape Fold Belt. During compression, such lines o f basement weakness (which are a dominant feature of southern African geology) became sites of sinistral transcurrent movements, accompanied by sedimentation in linear troughs, having a NE-trend at the present day. Thus, the northern part of southern Africa underwent extension in an essentially E - W direction in the present reference frame. Previously, the evidence of extension, and sedimentation in isolated troughs during the Permian, for example in the Bubye Coalfield of the Limpopo valley (Cox et al. 1965), suggested that the ensuing Karoo vulcanism represented the climax of a long-drawn-out series of events

THE KAROO PROVINCE

spreading anomalies but is simply a cartoon to illustrate the remarkable match of certain features of Mesozoic geology on the juxtaposed continental margins (incidentally it also provides an excellent match for the various features of basement geology discussed by Groenewald et al. 1991). This is taken to be the initial configuration of Africa and Antarctica before break-up, i.e. up to about the Trias-Jurassic boundary. Details of the geological features will be discussed later. Figure 5 is modified from Daly et al. (1991), and illustrates the reactivated basement shear zones within southern and central Africa during the Permo-Trias, with their associated sedimentary troughs, and, where known, their sense of sinistral shear. I suggest that there was one more important example of such a zone, that is along the Limpopo belt (the Limpopo lineament), acting as a normal half graben to the west of Nuanetsi (the Bubye coalfield) but developing into a sinistral shear zone to the east, and so allowing Dronning Maud Land to slide out in a NE direction (see Fig. 6a). A second, minor, linea-

originating in mantle convection (e.g. Cox 1970). But now we have the opportunity to try to decide between that sort of model, and the alternative which supposes that extension was already taking place for other reasons (e.g. influences of subduction), and that break-up was then perhaps aided by a plume which struck under these circumstances. In the Karoo case, however, the circumstances were evidently not favourable to sea-floor spreading in the early Jurassic. In the next section I put forward some tentative ideas, in general agreement with the suggestion of Marsh (1987) and Lawver et al. (1991a), that break-up took place in two stages, the first of which did not result in the creation of new ocean floor.

Stage 1 o f break-up In Fig. 4 a rather tight fit of Dronning Maud Land and SE Africa is illustrated. This is even tighter than the recently-proposed reconstruction (Martin & Hartnady 1986), and is not based on any considerations of the fitting of sea-floor

Mozambique iii:i!iiiiii ii:i:i:i:i:ii iiiiiiiiii ::::::i

143

Is.

Antonio

Lupata Gorge

Sii!!!i!

!iiiiiii Enes

::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

, e e e e o e e o o e o o o . e o o

========================

Nuanetsi

i,EXPLO

RA::iiiiiiiiiiiiiii

i

l AST!iiiiiiiiiiiiiiiiiiiii ~ , , , , • , , , , , , , • • , , o , , , , • • • , . * * , ~ o o o o o o o e o o o o o o e o e o o o o o e e o e e o o o o

a

* • J

*

*

~1) ............................... ~:~i .i1 .................................. ~iii?tl "I ~/I/

7

•

.

•

•

.

•

.

o . . .

•

;;,,/ Kirwan

,

~.-/-:-0

::

........

•

-

•

*

.

-

*

I,

•

,

*

*

*

•

-

*

*

* * *

,

,

*

*

*

-

**

*

,

,

*

*

*

*

*

*

* -

iWestfjella .... Zululand

•

-

.

I

500 km

EXPLORA (WEST) . . . . . . . . . . . . . . . 111 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Fig. 4. Cartoon illustrating tight fit of southeast Africa and East Antarctica (striped ornament shows area of overlap of present coast lines). Volcanic sequences, including the offshore Explora Wedge in Dronning Maud Land, are shown in black. Arrows indicate direction of dip in Karoo volcanic monoclines.

144

K.O. COX

merit is postulated further south (the Maputo lineament) delineated by the coast line NE of Maputo, and the abrupt northern termination of the Rooi Rand dyke swarm in the Lebombo (Armstrong et al. 1984), which implies more extension to the south, hence movement on the proposed lineament east of the Lebombo is also sinistral. Further SW, the lineament is taken to coincide with the maximum concentration of dolerite sills within the Karoo basin (Winter & Venter 1970). Movements on such lineaments in the late Trias or early Jurassic would be in keeping with the suggestions made by Daly et aL (1991) for the Permo-Trias. 1]" new ocean floor were developed, on the African side it would now be hidden below the coastal plain sediments of southern Mozambique. Nevertheless, the gap created between southern Africa and Dronning Maud Land was prob-

ably a complex zone, consisting of more than just new seafioor and post-Karoo sediments. In Fig. 6a it is referred to only as the Mozambique Thinned Zone (MTZ), because whatever else it may be, it is low-lying, much of it is below sea level, and it must represent thin crust. It 'might be oceanic with a sedimentary cover, it might be stretched pre-existing continental crust, or it might represent a mixture of the two. Recent geophysical evidence from the WeddelI Sea (Kristofferson & Hinz 1991) has demonstrated the existence of a failed rift, probably including a small strip of new oceanic crust, which in the tight reconstruction used in the present work would lie immediately east of the southern part of the Lebombo and west of the Explora wedge. This rift effectively occupies the area of the speculative southern part of the MTZ shown south of the Maputo lineament in Fig. 6a. The

INDIA 0

AFRICA

ANTARCTICA J

SOUTH AMERICA

J #~,,,,#

\

\

Strike slip fault

KFB \

\

Normal fault

I

'~

Thrust fault

\

\~

G.O N DWA N I D E

\

" \ RO.GEN

Fig. $. Distribution of Permo-Triassic sedimentary troughs, faults and reactivated shear zones in southern and central Africa (aRer Daly etal. 1991). The Limpopo and Maputo lineamentspostulated in the present work are individuallylabelled. Open arrows indicate direction of extension, filled arrows show compressive stress. KFB, Karoo Foreland Basin.

THE KAROO PROVINCE

STAGE 1

145

STAGE 2

Fig. 6. (a) Stage 1 of break-up showing postulated NE movement of Dronning Maud Land relative to Africa along a sinistral strike-slip fault localized by the Limpopo lineament. Associated volcanic sequences are shown in black. MTZ, Mozambique thinned zone. Broad arrows indicate main compressive stress (after Daly etal. 1991). (b) Stage 2 of break-up showing the development of a spreading ridge (double line) and two N-S transform faults, allowing Antarctica to move southwards relative to Africa. Associated volcanics are shown in black. The circle marks the possible position of the plume axis at this time.

further possible significance of 'thinned zones' in the context of Gondwanaland break-up will be discussed later. Most of the very wide-spread Karoo vulcanism of southern Africa probably accompanied Stage I of the break-up. Volumetrically, probably the most important was represented by the opposing sequences of seaward-dipping volcanics flanking the MTZ, on the African side by the Lebombo, and on the Antarctic side by the western part of the Explora Wedge (Hinz 1981). Why the Lebombo, which in all respects appears to be completely analogous to the seawarddipping reflector sequences described by marine geophysicists, is exposed above sea-level is an outstanding problem. Gilchrist & Summertield (1990) have speculated that the isostatic response to erosion of a continental hinterland can lead to uplift of the coastal tract because of the flexural rigidity of the lithosphere. This is a question which demands further investigation. In the present case it may be that the huge erosional effect of the Limpopo river system on the topographic dome left after the cessation of vulcanism (see Cox 1989), was the controlling factor. The features described above were probably the main consequences of the impingement of the Karoo plume in the early Jurassic. The northeastern part Of southern Africa, where t h e plume struck, was at the time under N--S compression and this part of Gondwanaland had a limited capacity to extend in an E-W direction (directions expressed in present-day African

reference frame, see Daly et al. 1991). It appears that the amount of extension, and lithospheric thinning, in the MTZ was sufficient to generate a significant melting anomaly, but factors that could have initiated true sea-floor spreading (e.g. the gravitational forces derived from the associated uplift, or perhaps the viscous-drag effects of the mantle convection itself) were ineffectual (of. Hill 1991).

Stage 2 o f break-up This may have started as long as 170 Ma ago, judging by the evidence of seafloor magnetic anomalies of the Somalia Basin (see previous discussion), which show that Madagascar began to separate from Africa at about this time. The strong implication is that East Antarctica, probably belonging to the same plate as Madagascar, also began to move. Figure 6b shows this in cartoon form. Seaward-dipping volcanic reflector sequences were developed on the new continental margins, represented by the Karoo volcanic sequences of the coastal area of northern Mozambique (Ant6nio Enes, Mozambique Island, see Fig. 2), and the eastern part of the Explora Wedge in Antarctica. The implications of this model are that these sequences are younger than those of the Lebombo and the western part of the Explora Wedge. Further data are needed, but at least the existing dates on the volcanics of the Ant6nio Enes area of Mozambique support this concept, since ages obtained there are mainly in the range 177-157 Ma (Jaritz et al.

146

K.G. COX

1977, el. the main Lebombo dates in the approximate range 198-173 M~i, Fitch & Miller 1984). The possible position of the axis of the Karoo plume, is more-or-less that postulated by White & McKenzie (1989), is also shown in Fig. 6b. The _plume head may have retained a minor but wide influence on younger magmatism in southeast Africa e.g. the Bumbeni complex in Zululand (Allsop et al. 1984: c. 140 Ma) and the Lupata Gorge volcanics on the Lower Zambezi and the alkaline plutonic complexes in southern Malawi (WooUey & Garson 1970: c. 100-140 Ma). During the second stage of break-up it is possible that the southern part of the original Mozambique Thinned Zone was detached, perhaps along the Maputo lineament, and deformed in a highly complex fashion. The Falklands Plateau, which was wrapped around the southern edge of Africa until the early Cretaceous, when the opening of the South Atlantic carried it westward as part of the South American plate, may have been part of this zone. But originally (i.e. in the early Jurassic) the Falkland Islands microplate was located close to the present coast of southeast Africa in the general vicinity of Port Elizabeth (Mitchell et aL 1986), whereas by the early Cretaceous it seems not only to have moved several hundred kilometres to the southwest, but also to have undergone a very substantial rotation (Mitchell et al. 1986; Taylor & Shaw 1989). By implication; the Falklands Plateau is now a complex collection of coalesced microplates, assembled at some time during the Jurassic. Outstanding problems The suggestions made above are tentative, and depend for their further elucidation on a considerably improved chronology based on radiometric dating and palaeomagnetism, including magnetostratigraphy. For example, Karoo vulcanism appears to cover a long time-span and may show systematic geographical diachronism. However, stratigraphic studies in the Lebombo (Cox & Bristow 1984) suggest younging from north to south but the existing radiometric ages suggest the reverse (Fitch & Miller 1984). Furthermore, while the eruption of the widespread Lesotho magma-type of the southern part of the Karoo province (Marsh & Eales 1984) is regarded as one of the earliest phases of activity (c. 193 Ma, Fitch & Miller 1984), the Ferrar magmatism of the Transantarctic Mountains, from the Theron Mountains through to Victoria Land, appears to be significantly younger (c. 175 Ma, Kyle 1991). Yet these are the two areas where geochemical studies indicate the

possible involvement of subduction. There is thus no clear indication at present of the relative ages of supposed plume-related (e.g. N Lebombo) and supposed subduction-related magmatism, or indeed whether there is a consistent age-difference between the two. Nor are the ages of different parts of the Explora wedge known. At present then, although it is possible to argue in favour of a two-stage break-up process, and to suggest that the most important plumerelated vulcanism accompanied stage 1, the relationship between stage 2 and vulcanism remains uncertain. The idea that stage 2 was accompanied by subduction-related vulcanism is tempting, but the existing radiometric ages hardly support this, since even the Ferrar magmatism seems to be too old. Accurate knowledge of the stratigraphie development of the extrusive rocks, and the relative timing of the intrusive phases, will eventually provide some of the most important constraints on the tectonic models. Other contraints will emerge from further petrological and geochemical studies, particularly the resolution of the debate concerning the origins of low-Ti basalts.

Rderences Au.soe, H. L., MANTON,W. I., BmSTOW,J. W. & ERLA~, A. J. 1984. Rb-Sr geochronology of Karoo feJsic volcanic,. In: ERLANK,A. J. (ed.) Parogenesis of the volcanic rocks of the Karoo prov-

/me. Geological Society of South Africa Special Publication, 13, 273-280. ARMS'mONO,R. G., BRISTOW,J. W. & Cox, K. G. 1984. The Rooi Rand dyke swarm, southern Lebombo. In: ERt.ANK,A. J. (ed.) Petrogenesis of the volcanic rocks of the Karoo province.

Geological Societyof South Africa Special Publication, 13, 77-86. BmaLmm, G., COMm-CmAnTa~OWH,P., MARQUES,L. S., MELFI,A. J., NARDY,A. J. R., [hCCImLLO,E. M. & ROISEmmRO,A. 1984. High- and low-TiO2 flood basalts from the Paranfi plateau (Brazil): petrology and geochemical aspects bearing on their mantle origin. Neues Jahrbuch far Mineralogle, Abhang, 150, 273-306. BURKE, K. & DEWEY, J. F. 1972. Plume-generated triple junctions: key indicators in applying plate tectonics to old rocks. Journal of Geology, 81, 406-433. C~BELL, I. H. & GRIV'FITrlS,R. W. 1990. Implications of mantle plume structure for the evolution of flood basalts. Earth and Planetary Science Letters, 99, 79-93. ~ , J. R. 1988. The Somali basin, Chain ridge and the origin of the northern Somali basin gravity and geoid low. Journal of Geophysical Research, 93(B10), 1985-2008. Cox, K. G. 1970. Tectonics and volcanism of the Karoo period and their bearing on the postulated

THE KAROO PROVINCE fragmentation of Gondwanaland. In: CLIFFORD, T. N. & GASS, I. G. (eds) African magmatism and tectonics. Oliver & Boyd, Edinburgh 211-235. 1983. The Karoo province of southern Africa: origin of trace element enrichment patterns. In: HAWKESWORTH,C. J. & NGRRY,M. J. (eds) Continental basalts and mantle xenoliths. Shiva, Nantwich 139-157. 1988. The Karoo province. In: MACDOUOALL,J. D. (ed.) Continental flood basalts. Kluwer Academic Publishers, Dordrecht 239-271. 1989. The role of mantle plumes in the development of continental drainage patterns. Nature, 342, 873-877. & BRISTOW,J. W. 1984. The Sabie River Basalt Formation of the Lebombo monocline and south-east Zimbabwe. In: E ~ K , A. J. (ed.) Petrogenesis of the volcanic rocks of the Karoo province. Geological Society of South Africa Special Publication, 13,125-147. , JOHNSON,R. L., MONKMAN,L. J., STILLMAN,C. J., VAIL,J. R. & WOOD,D. N. 1965. The geology of the Nuanetsi igneous province. Philosophical Transactions of the Royal Society, London, A2~, 71-218. , MACDONALD, R. & HORNUNG, G. 1967. Geochemical and petrographic provinces in the Karroo basalts of southern Africa. American Mineralogist, 52, 1451-1474. DALY, M. C., CHOROVIZC,J. & FAIRHEAD,J. D. 1989. Rift basin evolution in Africa: the influence of reactivated steep basement shear zones. In: COOPER, M. A. & WILLIAMS,G. F. D. (eds) Inversion Tectonics. Geological Society, London, Special Publication, 44, 309-334. , LAWRENCE,S. R., KIMUN'A, D. & BINGA, M. 1991. Late Palaeozoic deformation in central Africa: a result of a distant collision7 Nature, 350, 605-607. DALZIEL, I. W. D., STOREY,B. C., GARRETr, S. W., GRtmOW, A. M. & I-I~RROD,L. D. B. 1987. Extensional tectonics and the fragmentation of Gondwanaland. In: DEWEY, J. F., COWARD,M. P. & HANCOCK,P. (eds) Continental Extensional Tectonics. Geological Society, London, Special Publication, 28, 433-441. EALES, H. V., MARSH, J. S. & Cox, K. G. 1984. The Karoo igneous province: an introduction. In: ERL~K, A. J. (ed.) Petrogenesis of the volcanic rocks of the Karoo province. Geological Society of South Africa Special Publication, 13, 1-26. ELLAM, R. M. & COX, K. G. 1989. A proterozoic lithospheric source for Karoo magmatism: evidence from the Nuanetsi picrites. Earth and Planetary Science Letters, 92, 207-218. & 1991. An interpretation of Karoo picrite basalts in terms of interaction between asthenospheric magmas and mantle lithosphere. Earth and Planetary Science Letters, 105, 330-342. EaLANK, A. J. (ed.) 1984. Petrogenesis of the volcanic rocks of the Karoo Province. Geological Society of South Africa, Special Publication, 13. FrrcH, F. J. & MILLER,J. A. 1984. Dating Karoo i~neous rocks by the conventional K-Ar and ~Ar/

147

~Ar age spectrum methods. In: ERLAmC, A. J. (ed.) Petrogenesis of the volcanic rocks of the Karoo province. Geological Society of South Africa Special Publication, 13, 247-266. FROIDEVAUX, C. & NATAF, H. C. 1981. Continental drift; what driving mechanism? Geologische Rundschau, 70, 166-176. Glt.CHmS'r, A. R. & SUMMERFmLD,M. A. 1990. Differential denudation and flexural isostasy in formation of rifted-margin upwarps. Nature, 346, 739-742. GROENEWALD,P. B., GRANTHAM,G. H. &WATr,J~YS, M. K. 1991. Geological evidence for a Proterozoic to Mesozoic link between southeastern Africa and Dronning Maud Land, Antarctica. Journal of the Geological Society, London, 148, 1115-1123. HARRIS,C., ERLANX,A. J., DUNCAN,A. R. & MARSH, J. S. 1991. The geochemistry of the Kirwan and other Jurassic basalts of Dronning Mand Land and their significance for Gondwana reconstruction. In: THOMVSON,M. R. A., CRAM~, J. A. & "I'HoMSON, J. W. (eds) Geological evolution of Antarctica. Cambridge, Cambridge University Press, 563-568. , MARSH,J. S., DUNCAN,A. R. & ERLANK,A. J. 1990. The petrogenesis of the Kirwan basalts of Dronning Maud Land. Journal of Petrology, 31, 341-369. I-IERGT, J. M., PF.ATE,D. W. & HAWKESWGRTH,C. J. 1991. The petrogenesis of Mesozoic Gondwana low-Ti flood basalts. Earth and Planetary Science Letters, 105, 134-148. HALL, R. I. 1991. Starting plumes and continental break-up. Earth and Planetary Science Letters, 104, 398-416. Hnqz, K. 1981. A hypothesis on terrestrial catastrophes. Wedges of very thick oceanward dipping layers beneath passive continental margins. Geologisches Iahrbuch, E22, 3-28. JARITZ, W., KREUTER,J., MOLLER, P. & HARRE, W. 1977. Die Vulcanitserien in Kiistengebiet yon Nord Moqambique. Geologisches Iahrbuch, B26, 147-165. KmSTOVe~RSON, Y. & HaNZ, K. 1991. Evolution of the Gondwana plate boundary in the WeddeU Sea. In: THOMSON,M. R. A., CRAME, J. A. & THOMPSON, J. W. (eds) Geological evolution of Antarctica. Cambridge, Cambridge University Press, 533-539. KYI.E, P. R. 1991. Timing and duration of Ferrar magmatism, Antarctica. IUGG Vienna, August 1991, Abstracts Volume. LAWVER, L., GAHAGAN, L. M. & DALZIEL, I. W. D. 1991a. Continental break-up and tectonic evolution. Nato Advanced Study Institute, 'Dynamic modelling and flow in the earth and planets', Alaska, June 1991, Abstracts volume. , ROYER, J.-Y., SANDWELL,D. T. & SCGTESE,C. R. 1991b. Evolution of the Antarctic continental margins. In: THOMPSON,M. R. A., CRAM~,J. A. & THOMSON,J. W. (eds) Geological evolution of Antarctica. Cambridge, Cambridge University Press, 533-539. MARStL J. S. 1987. Basalt geochemistry and tectonic

148

K. G. COX discrimination within continental flood basalt provinces. Journal of Volcanology and Geothermal Research, 32, 35-49. & EALES, H. V. 1984. The chemistry and petrogenesis of igneous rocks of the Karoo central area, southern Africa. In: ERLAr~K, A. J. (ed.)

Petrogenesis of the volcanic rocks of the Karoo Province. Geological Society of South Africa Special Publication, 13, 27-67. MARTn~, A. K. & HARTNADY,C. J. 1986. Plate tectonic development of the south west Indian Ocean; a revised reconstruction of east Antarctica and Africa. Journal of Geophysical Research, B91, 4767-4786. MrrCHELL, C. 1988. Petrology and geochemistry of

basaltic rocks of the Falkland Islands and Deccan Traps (India). D Phil. thesis, Oxford University. , TAYLOR, G. K., Cox, K. G. & SHAW,J. 1986. Are the Falkland Islands a rotated microplate? Nature, 319, 131-134. Rust, I. C. 1975. Tectonic and sedimentary framework of Gondwana basins in southern Africa. In: HARLAND,W. B. & HAMBREV,M. J. (eds). Gondwana Geology. Australian National University Press, Canberra, 537-564. STOREY, B. C. & ALABASTEn, T. 1991. Tectonomagmatic controls on Gondwanaland break-up models; evidence from the proto-Pacific margin of Antarctica. Tectonics, 10, 1274-1288. TANKARD,A. J., JACKSON,M. P. A., ERIKSON,K. A.,

HOBDAY,D. K., HtnCrER, D. R. & MINTER, W. E. L. 1982. Crustal evolution of southern Africa-3.8 billion years of earth history. Springer Verlag, New York. TAYLOR, G. K. & SHAW, J. 1989. Falkland Islands: new palaeomagneticdata and their origin as a displaced terrane from southern Africa. In: HILLHOUSE, J. W. (ed.) Deep structure and past kinematics of acreted terranes. Geophysical Monograph Series, AGU, Washington, D.C., 50, 59-72. THOMPSON, R. N. & GIBSON, S. A. 1991. Subcontinental mantle plumes, hotspots and thin spots. Journal of the Geological Society, London, 148, 973-977. WINTER, H. D. & VEm'ER, J. J. 1970. Lithostratigraphic correlation of recent deep boreholes in the Karoo-Cape sequence. Proceedings of the

Second Gondwana Symposium, South Africa, 395-408. WHrrE, R. S. & MeK£szm, D. P. 1989. Magmatism at rift zones: the generation of volcanic continental margins and flood basalts. Journal of Geophysical Research, 94, 7685-7729. WOOLLEY,A. R. & GARSON,M. S. 1970. Petrochemical and tectonic relationship of the Malawi carbonatite-alkaline province and the LupataLebombo volcanics. In: CLIFFORD,T. N. & GASS, I. G. (eds) African magmatism and tectonics. Oliver & Boyd, Edinburgh, 237-262.

Role of subduction-plate boundary forces during the initial stages of Gondwana break-up: evidence from the proto-Pacific margin of Antarctica B. C. S T O R E Y 1, T. A L A B A S T E R

2, M. J. H O L E 3, R . J. P A N K H U R S T

1&

H. E. WEVER 1

IBritish Antarctic Survey, High Cross, Madingley Road, Cambridge CB3 0ET, UK 2School of the Environment, University of Sunderland, Sunderland SR2 7BW, UK 3Department of Geology, University of Aberdeen, Aberdeen AB9 2UE, UK

Abstract: In the West Antarctic sector of Gondwana, early stages of break-up are associated with the large Antarctic-Karoo-Tasman basalt province. Formation of this within-plate province was synchronous with active margin tectonics and development of both a protoPacific margin magmatic suite along the Antarctic Peninsula and the extensive Tobifera volcanic suite associated with the Rocas Verdes marginal basin system of southern South America and South Georgia. Extension, concurrent with subduction and oceanward migration of the magmatic focus, resulted in a broad extensional province in a back-arc and intra-arc-setting. High geothermal gradients and basalt underplating caused crustal melting on the east coast of the Antarctic Peninsula and formation of bimodal basalt-rhyolite suites. Large-ion lithophile element enriched initial rifting magmas were succeeded, at least in part of the Rocas Verdes basin, by early drift magmas of transitional chemistry and then by entirely asthenospheric MORB magmas representing lithospheric rupture and sea-floor spreading. A plate interaction model is proposed for the initial stages of Gondwana break-up relating the broad zone of mantle melting to a reduction in subduction-plate boundary forces. The change from Gondwanide compression to lithospheric extension in the Jurassic may be linked to a change from shallow to steeply dipping subduction, and to a slowing of subduction rates caused by a change in plate boundary zone parameters. A possible reduction of compressive boundary stresses may have enabled unconfined, overthickened Permo-Triassic crust to extend because of gravitational instability, thus facilitating break-up. We suggest that break-up was not plume-related, but was due to variations in the regional stress field associated with changing plate-boundary forces. The continental crust was placed under tension with substantial lithospheric thinning and decompression melting of an enriched mantle source forming the broad linear zone of within-plate magmatism. The presence of a plume beneath the Karoo province may have thermally weakened the lithosphere and induced local rifting, contributing to, but not causing the eventual separation of East and West Gondwana.

Critical to differentiating between active and passive models for continental break-up is the origin of horizontal tensional stresses prior to rifting. In active plume models (e.g. Morgan 1971; Richards et al. 1989) tensional stresses produced by mantle plumes are thought to be sufficient to cause break-up. There appears little doubt that tensional effects imposed by a plume may lead to crustal extension (Houseman & England 1986) but it is less clear whether the gravitationally induced horizontal stresses are sufficient to cause break-up. In passive models, plumes are not the ultimate driving force and break-up is thought to be related to external forces that drive plate motions (e.g. White &

McKenzie 1989). Tensional stresses may, for example, be generated by active subducting zones on opposite tides of a large continent (Bott 1982; Dewey 1988a). In many hybrid models (e. g. Hill 1991), which combine elements of both active and passive models, the external forces that drive plate motions may place the lithosphere under tension but it is the arrival of a new plume that weakens the lithosphere causing it to split and form a new ocean. In the hybrid models, plumes play an important role in determining the exact position of break-up; and the models are compatible with the coincidence of large igneous provinces, mantle plumes and continental break-up but they do not require mantle plumes

From STOREY,B. C., ALAaASTER,T. & PANKHURST,R. J. (eds), 1992, Magmatismand the Causes of ContinentalBreak-up, Geological Society Special Publication No. 68, pp. 149-163.

149

150

B.C. STOREY E T A L .

w.

WS ?J ~

,

• ~/~'--~':

WSE

DML

EAST ANTARCTICA

VIBL 0 RSE

I

! 800 km I

--x-"i

Fig. 1. Map of West Antarctica showing the present day position of the crustal blocks (Dalziel & Elliot 1982; Drewry 1983; Storey et al. 1988a); AP, Antarctic Peninsula; TI, Thurston Island; MBL, Marie Byrd Land; EWM, Ellsworth-Whitmoremountains; HN, Haag Nunataks. Note the discordant trend of the Gondwanian fold belt in the EWM and Transantarctic Mountains. The Weddell Sea embayment (WSE) and Ross Sea embayment (RSE) are below sea-level and are most probably underlain by extended continental crust. The WSE is separated from the Weddell Sea (WS) by an ocean-continentboundary of Late Jurassic age (Bell et al. 1990) which together with the Andenes (AE) and Explora escarpments (EE) forms an almost linear feature that truncates a failed rift (FR) with dipping reflector units (DR; Kristoffersen & Hinz 1991) along the Dronning Maud Land (DML) margin. Fracture zone traces indicating the Late Jurassic and Cretaceous spreading directions in the WS, are based on the gravity map of Bell et al. (1990). An area of possible older sea-floor (?J) is indicated on the western side of the WS (LaBreque et al. 1989). Fig. 2. (Opposite) (a) Pre-break-up Late Triassic-Early Jurassic Gondwana reconstruction after Lawver& Scotese (1987), illustrating the Gondwanian fold belt, the plume head of White & McKenzie (1989), the Gastre Fault System (GFS) and its possible continuation between Africa and East Antarctica (Rapela & Pankhurst this volume), the Late Triassic to Early Jurassic magmatic belt of southern South America and Antarctic Peninsula. The Falkland Islands (FI) (Mitchell et al. 1986) and EWM (Grunow et al. 1991) have been rotated into possible pre-break-up positions. (b) Initial rifting stage: a Mid-Jurassic reconstruction of the proto-Pacific margin illustrating a subduction related magmatic belt, within-plate Karoo-Ferrar-Tasman province, a broad extensional province including a back-arc basin on the eastern margin of AP and the dipping reflector and failed rift system of the Explora wedge. The Gastre fault system and its continuation may have operated as a transfer fault system on the margin of the broad extensional province. During the initial drift stage (155 Ma onwards) and the final separation of East and West Gondwana late Jurassic sea-floor formed in the Weddell Sea and within the Rocas Verdes marginal basin.

SUBDUCTION AND GONDWANA BREAK-UP

151

PALEO-TETHYS

AFRICA

NEOTETHYS

SOUTH AMERICA EAST ANTARCTICA

Fold belt

, AUSTRALIA

Fold belt

X Magmaticarc

INDIA

AFRICA

head SOUTH AMERICA

EAST ANTARCTICA

Explore Wedge

Gastre Fault :System

~q FI

WSE , Ensialic J basin i

pRoViNCE /

/

/

/

/

f

Tasman

/ / MBL / N E W ZEALAND

/ Tobifera volcanism

Subductlon magmatism

Back-arc basin

W'dhin-plate magmatism

/

/

/

/

152

B.C. STOREY E T A L .

70eW

55°W T-

60°W

65°W

1

I

200 km

.-__-__-

65°S ~.,_~..,_-

65°S

+__

o~

75°W

o.¢ 70°S 70°S :-~

L- - _ - - . _--: (~ sk~_----,, vo_ ~.~_-----_-r~,, _.:.., ,, t oo.I~,C ,,, ~T.-_--_--_--_--: l I~. ,~-,"7 5 .to ~..~_~ _~. . ~ " - - - - - - - ' - - - - - - - - - - - - k P 7~xp ~ x ' ~ | - - , , I o : " ¢, _

75ow

70°W

:

"

65°W

60°W

55°W

0

Phase II magmatic rocks (180-160 Ma)

@

Phase I magmatic rocks (236-199 Ma)

oO

Peraluminous compositions Ensialic back-arc basin Terrestrial sedimentary rocks Fore-arc region

;r,-'l [Z~;;~,'-:: ",1

Palaeozoic basement

Fig. 3. Simplified geologicalmap of AP showing the distribution of magmatic rocks during Late Triassic to Early Jurassic times (Phase I, 236-199 Ma) and Mid-Jurassic times (Phase II, 180-160 Ma). Note the restriction of known phase I rocks to the eastern margin of AP. Phase II has a more widespread distribution; plutonic rocks intrude, and volcanic rocks unconformably over-lie the Permo-Triassic fore-arc region in the northern part of AP. Palaeozoic basement may be present beneath at least part of this fore-arc region.

SUBDUCTION AND GONDWANA BREAK-UP to be the fundamental force that drives plate motions. Although the association between the initial stages of Gondwana break-up, the presence of a mantle plume and the formation of the extensive within-plate Karoo-Ferrar-Tasman igneous province, is well known (e.g. Dalziel et al. 1987), their causal relationships remain speculative. Less well known, however, is that when the crustal blocks that make up West Antarctica (Fig. 1; Dalziel & Elliot 1982; Storey et al. 1988a) are restored to what we consider to be their prebreak-up position, the within-plate province is virtually continuous with a magmatic suite along the proto-Pacific margin of Antarctica and southern South America (Dalziel et al. 1987; Storey & Alabaster 1991). In the reconstruction of Fig. 2 the mantle plume postulated to lie beneath the Karoo province (Morgan 1981; White & McKenzie 1989) lies in a back-arc position. In fact, Cox (1988) suggested that the main melting anomaly responsible for the Karoo-Ferrar-Tasman basalt province was related to subduction and has drawn analogies between the province as a whole and back-arc spreading. Two major episodes of rift-related basaltic magmatism are recognized within the southern African Karoo province (193+5 and 178+5 Ma, Fitch & Miller 1984; Cox 1988). The main phase of Antarctic Ferrar and Tasman magmatism has been dated at 179+4 Ma (Kyle et al. 1981) and 175+8 Ma (Hergt et al. 1989a) respectively, although some magmatism in the Antarctic sector may be as old as 195 Ma (Brewer et al. 1991) and comparable in age to the Karoo province. The first phase of true sea-floor spreading (early drift phase) as indicated by anomaly M22 (155 Ma) from near the margins of the Mozambique and Somali basins (Martin & Hartnady 1986), followed some 40 Ma after the earliest rift-related magmatism. The age of Weddell Sea floor is also likely to be of Late Jurassic age based on the identified anomalies (M13-14 138 Ma) of Bell et al. (1990) and the location of the ocean-continent boundary (assuming normal spreading rates). However, the possibility cannot be excluded that older Jurassic sea floor exists on the eastern margin of Antarctic Peninsula (Fig. 1; LaBreque etal. 1989). The initial rifling stage may be associated with strikeslip movement along the Gastre Fault System, a major transcurrent boundary in southern South America (Fig. 2; Rapela & Pankhurst, this volume). Movement along this zone and a precursor of the Aghulas Fracture Zone may have resulted in translation and rotation of the Falkland Islands (Mitchell et al. 1986) and some of the crustal blocks of West Antarctica (Grunow et

153

al. 1991) during the early break-up stages.

In this paper we evaluate the evidence for subduction along the Antarctic sector of the protoPacific margin of Gondwana during the initial stages of Gondwana break-up by reviewing the magmatic record. We investigate temporal, spatial and compositional changes which may indicate critical changes in subduction zone parameters at the time of break-up and suggest a causal relationship between changes.io subduction forces and continental break-up We concentrate on the Antarctic Peninsula (Fig. 3) because the magmatic record is best known from this sector of Gondwana and make comparisons with the neighbouring, once contiguous, South America, South Georgia and Thurston Island parts of the margin. We consider a time frame of Late Triassic to Late Jurassic to include the magmatic record prior to sea-floor spreading and because Late Triassic sediment-filled grabens in South America (Gust et al. 1985) and Africa (Daly et al. 1991) are thought to represent the earliest rifting stage within the Gondwana supercontinent.

Antarctic Peninsula The Antarctic Peninsula (Fig. 3) is predominantly an ensialic Mesozoic magmatic arc related to subduction of proto-Pacific and Pacific ocean floor (Pankhurst 1982; Storey & Garrett 1985; Hole et al. 1991). Plutonic and volcanic rocks range in age from 236 to 10 Ma along the length of the 1500 km arc. Silurian and Carboniferous magmatic rocks occur within the basement and record an earlier subduction history for this margin (Milne & Millar 1989). Wide Mesozoic accretionary complexes contain blue-schist assemblages, ocean-floor and ocean-island basalts, and deformational structures typical of a plate margin setting (Storey & Garrett 1985) although they provide few constraints on the timing of subduction. AIlochthonous Triassic (Dalziel et al. 1981), Early Jurassic (Thomson & Tranter 1986), and Cretaceous (Holdsworth & Nell 1992) faunas within the prism are supportive of a Mesozoic age but do not constrain the timing of subduction. Autochthonous fore-arc basin sequences are Oxfordian to Albian in age (Butterworth et al. 1988) and may be as old as Middle Jurassic (Holdsworth & Nell 1992). Metamorphic ages from subduction complex rocks are Jurassic and Cretaceous in age (Tanner et al. 1982; Trouw et al. 1990; Herv6 et al. 1991) and may date the time of subduction. The.main time constraints on the history of subduction come from the magmatic record. In

154

B.C. STOREY ETAL.

the Antarctic Peninsula magmas emplaeed during the initial stages of Gondwana break-up can be divided into two tectono-magmatic regions; (1) arc magmatism, (2) back-arc magmatism. Arc magmatism Timing and distribution. Magmatism can be broadly divided into Late Triassic-Early Jurassic and Mid-Jurassic episodes (Pankhurst 1990). The former covers an interval of approx. 236 to 199 Ma and was restricted to the eastern side of the Mesozoic arc (Fig. 3) where it was emplaced into Palaeozoic basement as a series of granitoids (Pankhurst 1982; Hole 1986; Hole et al. 1991; Wever & Storey 1991). Magrnatism was contemporaneous with deposition of fluviatile sedimentary rocks in extensional grabens on Permo-Triassic accretionary complex in the northern part of the peninsula (Farquharson 1984). The second magrnatic episode (c. 180160 Ma) is represented by both volcanic and plutonic rocks and is more widely distributed than the earlier episode (Fig. 3; Pankhurst 1982, 1983; Thomson & Pankhurst 1983; Harrison & Piercy 1990; Millar et al. 1990; Hole et al. 1991). Both geochronological data and field relations indicate that the magmatic focus migrated from a Late Triassic back-arc position into the Triassic fore-arc region where volcanic rocks overlie Early Jurassic fluviatile sedimentary rocks and unconformably overlie uplifted accretionary complex (see Storey & Alabaster 1991). Composition. Both magmatic pulses comprise peraluminous and metaluminous compositions (Fig. 3). Although metaluminous and peraluminous rocks are present within each pulse of activity, the earlier one is dominated by peraluminous compositions, whereas the later one is dominated by metaluminous compositions. The peraluminous rocks are predominantly granitoids, have up to 5% normative corundum, a strongly fractionated REE profile, relatively high initial STSr/S6Srratios (0.712 to 0.720) and low eNdt values ( - 8 to - 9 ) whereas the metaluruinous rocks range in comp?sition from gabbro to granitoid with initial STSr/S6Sr ratios of 0.706 to 0.709 and eNdt values of - 2 to - 5 (Hole et al. 1991; Wever & Storey 1991). The extrusive equivalents vary from basaltic andesites through andesites, including geochemically unusual continental magnesian andesites (Alabaster & Storey 1990), to dacites and rhyolites. They plot as coherent and continuous trends on silica variation diagrams (Storey & Alabaster 1991). Petrogenesis. The geochemical and isotopic compositions of peraluminous granitoids and

rhyolites throughout the Antarctic Peninsula are compatible with an origin by partial melting of a pre-Mesozoic metasedimentary basement. Although the oldest reliably dated basement rocks in the region are early Palaeozoic (Milne & Millar 1989), Nd model ages for Late Triassic and Early Jurassic peraluminous granitoids are suggestive of derivation from a 1.4 Ga old crustal protolith. The metaluminous products of both episodes have trace element abundances more typical of volcanic arc magmas, and Sr- and Ndisotope characteristics intermediate between those of the peraluminous group and a presumed asthenospheric composition. Hole et al. (1991) thought that this was best explained by deep crustal interaction of mantle-derived primary magmas by assimilation-fractional crystallization (AFC) processes, although Pankhurst et aL (1988) pointed out that part of the range of isotopic variation could be equally due to a heterogeneous source region. Storey & Alabaster (1991) highlighted the geochemical and isotopic similarities between some mafic endmembers of the Pacific-margin suite and the contemporaneous Ferrar-Tasman within-plate suite, which is generally thought to have been derived from lithophile-element enriched subGondwana lithospheric mantle reflecting sediment subduction into a depleted mantle source during the Palaeozoic (Kyle et al. 1983; Hergt et al. 1989a, b). They therefore proposed that both were derived from an enriched mantle source, in a single broad zone of lithospheric mantle melting that extended from the Transantaretic Mountains to the Pacific margin: according to this model the main difference between the two suites would have arisen through their distinct fractionation trends (see Fig. 4). In view of the complex interplay of possible sources within an active subduction zone (Pearce 1983), further detailed geochemical data are required to test this hypothesis and the role of enriched mantle in controlling the composition of the arc magmas. However, it is clear that crustal anatexis and subduction were instrumental in generating contemporaneous magmas in the Antarctic Peninsula at this time. Together with the presence of extensional grabens within the arc, this seems to indicate that lithospheric attenuation in this part of the proto-Pacific margin of Gondwana commenced during Late Triassic to Early Jurassic times. Back-arc m a g m a t i s m

An extensional basin (Fig. 3) formed along the SE margin of the Antarctic Peninsula during a

SUBDUCTION AND GONDWANA BREAK-UP m Rb

synCOLG

ppm

~/Pirrlt

Hills

j

/L/ /-

f Trend 100

•

tj~ • : //_ .1~•

• ~"

VAG

2 *

,,,,

/

/

wpG

/

.rrsr-.s.msn O.,;,I,:. 10 10

/

ORG

,

,/,,I

100

,

, , , t,,l Y ppm

Nb+

Fig. 4. Rb-Nb+Y diagram illustrating a within-plate differentiation trend (trend 1) for the generation of the Pirrit Hills pluton (see Storey et al. 1988b) and a subduction-related volcanic arc differentiation trend (trend 2) for the generation of Antarctic Peninsula Middle Jurassic intermediate and acidic compositions. In both cases, the mafic precursor corresponds to the field of the Ferrar-Tasman dolerites (Hergt et al. 1989a, b). Discriminant boundaries are after Pearce et a/. (1984). Abbreviations are VAG, volcanic arc granite; synCOLG, syncoUisional granite; WPG, within-plate granite; and ORG, ocean-ridge granite.

period of Early Jurassic crustal reworking and bimodal magmatism, behind the active continental arc system (Wever & Storey 1992). The basin was infilled by several kilometres of arcderived sedimentary rocks containing locally abundant marine invertebrate fossils of Midand Late Jurassic age (Quilty 1977; Thomson 1983; Rowley et al. 1983). Bimodal volcanic and hypabyssal rocks are mainly exposed along the NW margin of the basin, where they intrude the • magmatic arc rocks as a series of dykes, and form flows and sills embedded within the sedimentary infiil. Silicic volcanic rocks have eNdt values of - 7 to - 9 , directly comparable to the peraluminous crustaUy derived S-type granitoids flanking the marginal basin system, and may represent their extrusive equivalents; although the possibility that they may be derived by selective fractionation from low eNdt mantle-derived marie rocks cannot be excluded. Basaltic end-members have been subdivided based on trace element abundances, into three geochemical groups (Wever & Storey 1992) with

155

island-arc tholeiite (group I), enriched midocean ridge basalt (group II) and calc-alkaline basalt (group III) characteristics. The basalts of group I and II reveal the least pronounced LILE/ HFSE enrichment (Th/Ta ~ 2) and have highest eNdt values (+3.7 to -1.2) whereas those of group III show the strongest LILE/I-IFSEenrichment (T1CI'a > 5) and lowest eNdt values ( - 2.3 to - 5), reminiscent of subduction-related basalts from active continental margins. It is uncertain to what extent this LILE enrichment is due to contemporaneous subduction, or is related to previous episodes of mantle enrichment. The group III basalts are the most comparable to those within the enriched Ferrar province and their trace element characteristics may reflect either sediment subduction into a depleted mantle source sometime during the Palaeozoic, as argued by Kyle et al. (1983) and Hergt et al. (1989a, b) for the Ferrar, or contemporaneous subduction processes. Wever & Storey (1991, 1992) concluded that crustal attenuation along the eastern margin of the Antarctic Peninsula commenced in Early Jurassic times and was contemporaneous with subduction-related magmatism. Decompressional melting of the underlying enriched mantle during extension was thought by them to have facilitated basaltic magma production. Emplacement of hot magmas into the crust resulted in crustal melting and formation of voluminous crustal derived granitoids and silicic volcanic rocks. The isotopic characteristics of the basalts preclude a purely asthenospheric origin and all contain at least a small contribution from an enriched mantle source. T h u r s t o n Island

The Jurassic igneous record of the adjoining Thurston Island crustal block of West Antarctica (Fig. 1) differs from that of the Antarctic Peninsula (Storey et al. 1991). Only one Early Jurassic pluton in known: a mildly peraluminous twomica granite with a relatively high initial S7Sr/a6Sr ratio (0.710) and low eNdt values of - 5 . 6 and - 6 . 8 (Pankhurst, unpublished data), indicating a significant lithospheric, probably crustal, contribution. In contrast, a major suite of Late Jurassic to Early Cretaceous metaluminous granitoids (150-110 Ma) was emplaced on Thurston Island, outboard of the Early Jurassic granite. These have uniform initial S7Sr/S6Sr ratios of 0.7052-0.7057 and eNdt values mostly in the range +0.8 to -1.8, indicating a significantly more LILE-depleted source than the Jurassic granitoids of the Antarctic Peninsula (Pankhurst, unpublished data).

156

B.C. STOREY E T A L .

South America and South Georgia Silicic magmatism was widespread in southern South America during Late Triassic and Jurassic times (Fig. 2). Calc-alkaline granitoids of c. 200 Ma with moderately elevated initial S7Sr/86Sr ratios ('modified I-types' of Pankhurst 1990, 'transitional' granitoids of Parada 1990) are known from the Coast Ranges of central Chile (Parada et al. 1988) and from the North Patagonian Massif (Rapela et al. 1992), but are apparently absent from the coastal Patagonian Batholith of southern Chile. These magmas were also derived either by crustal-mantle hybridization or from a LILE-enriched lithospheric source. The North Patagonian Massif magmatism is interpreted by Rapela & Pankhurst (this x~olume) as indicating emplacement along the strike-slip Gastre Fault System during Triassic rifting 6f this part of the Gondwana lithosphere, and the hack-arc region of western Argentina sho~vs further evidence of rift-related Triassic and early Jurassic sedimentary basins with calc-alkaline and alkaline volcanic rocks (Gust et al. 1985; Uliana et al. 1989; Ramos & Kay 1991). NNW-trending extensional grabens continued to be active throughout Patagonia during the Mid- to Late Jurassic burst of volcanic activity that formed the Chon-Aike/Tobifera formation (155-165 Ma, Gust et al. 1985). Accumulations are thickest within the grabens, whereas outside of the grabens the volcanic rocks rest with sharp lithological and structural discontinuity on Palaeozoic basement. Geological, isotope and geochemical data have previously been employed to support an origin for the silicic Tobifera province by crustal anatexis (Bruhn et al. 1978; Herv6 et al. 1981). Storey & Alabaster (1991) compared the scanty elemental and isotopic data available for the province (Kay et al. 1989) and the silicic volcanism on the Antarctic Peninsula. They argue that at least part of the South American province was a product of either differentiation of an enriched mantle source or crustal contamination of subductionrelated magmas rather than wholesale crustal anatexis. This is apparently supported by Rb-Sr data for the Chon-Aike formation giving an age of 161+5 Ma and a low initial 87Sr/a6Sr ratio of 0.706 (De Barrio, unpublished, quoted by Rapela & Pankhurst, this volume). In Late Jurassic times, coincident with the earliest known sea-floor spreading in the Mozambique Basin and the Weddell Sea and the beginning of Gondwana dispersal, the extensional silicic province of southern South America was split by the formation of a marginal

basin. The basin is now represented by uplifted ophiolitic rocks known as the Rocas Verdes (Dalziel et al. 1974; Mukasa et al. 1988) and their southerly continuation, the Larsen Harbour Complex of South Georgia (Storey & Mair 1982). An integrated elemental-isotopic study on basaltic lavas and dykes from South Georgia (Alabaster & Storey 1990) has revealed a change from lithospheric-derived to asthenosphere derived magmas, representing a change from lithospheric rupture to sea-floor spreading. This is illustrated in Fig. 5, in which the basalts show a marked time-controlled shift from initial lavas

20• • Initial Rift lO• • ee

Ferrar

/

Th/Ta ••

a

1

u

~Maln

,

1.01 • •

8

DrilteeEarly Drift

• , , 1 ''i 0.1 Ta/Sm

'

I 0.5

Main Drift

E-4-

8 Early Drift

2-

ENd 0--2--4-

Ferrar(~ nQ---8

Initial Rift By

b

. . . . . . . 0.01

I 0.1

'

1 0.5

Ta/Sm

Fig. 5. Plots of (a) Th/Ta-Ta/Sm and (b) ENd-Ta]Sm for stratigraphically controlled rocks with < 60% SiO2 from South Georgia and northern Antarctic Peninsula (see Storey & Alabaster 1991). Fields for Ferrar dolerites are from Kyle etal. (1983), P. R. Kyle, pers. comm. and RIP, unpublished data.

SUBDUCTION AND GONDWANA BREAK-UP having high Th/Ta (about 10) and Ta/Sm (about 0.15) ratios, through a series of early drift lavas (Th/Ta of about 5; Ta/Sm of about 0.06) to MORB-type lavas with low Th/Ta (about 1.5) and low Ta/Sm (about 0.02) ratios. The shift towards low Th/Ta and low Ta/Sm ratios with decreasing age is accompanied by an increase in eNdt from about - 5 (initial rift) through eNdt of about +2 (early drift) to eNdt of about +8 (main drift). These changes reflect contributions from different sources rather than simply changes in degrees of partial melting and/or fractionation processes of a common source. The initial rifting lavas have elemental and isotopic signatures similar to Ferrar province and consistent with their derivation from an early enriched subGondwana lithospheric mantle source. The main drift lavas have both N-type MORB and more depleted elemental and isotopic characteristics and represent the only known autochthonous magmatic rocks derived from a purely asthenospheric source sampled from the proto-Pacific margin province of Gondwana. The change from initial rifting to sea-floor spreading is represented by lavas showing geochemical and isotopic characteristics between the two end members.

Tectonic controls on Gondwana break-up models The relationship between Pacific margin magmatism, within-plate magmatism and lithospheric extension during the initial stages of Gondwana break-up suggest that three factors may be important in producing the extensional stress at the time of break-up: (1) subduction; (2) crustal thickening and (3) mantle plume.

Subduction The driving mechanism for plate tectonics is now generally accepted to be dependant on plate boundary forces such as slab pull and ridge push (Forsyth & Uyeda 1975). Bott (1982, this volume) has suggested that tension associated with subduction forces on both sides of a large continent may lead to continental break-up. The review of the magmatic record along the proto-Pacific margin of Antarctica prior to and during the initial stages of Gondwana break-up clearly shows that subduction was taking place and that the active margin was extensional during Late Triassic and through Jurassic times. The earliest recorded Mesozoic subduction-related magmatism in the Antarctic Peninsula sector is Middle Triassic in age and it marks the end of an

157

apparent hiatus following Carboniferous magmarism. Late Triassic-Early Jurassic and MidLate Jurassic magmatic pulses here and along the South American margin were contemporaneous with the two magmatic episodes in the Karoo-Ferrar-Tasman within-plate province. The active margin and Ferrar-Tasman magmas could have been derived from a similar enriched sub-Gondwana lithospheric mantle source, indicating a broad zone of lithospheric mantle melting during the initial break-up stages. During the Jurassic subduction-related magmatism migrated from the eastern margin towards the western margin of the Antarctic Peninsula and Thurston Island with emplacement of granitic plutons closer to the trench. The margin was clearly extensional; grabens formed in the forearc region in advance of the westward migrating magmatism, and basins were, in the back-arc region of Southern South America and Antarctia, infilled by Mid- and Late Jurassic sedimentary rocks and bimodal volcanic rocks. These may have been contiguous with basins now preserved in the Weddell Sea embayment area and the formarion of the 'failed rift' proposed by Kristoffersen & Hinz (1991) forming a broad extensional province (Fig. 2b). Structural analysis of dyke trends and associated normal faults in the Ferrar province of the Transantarctic Mountains, has shown that a uniform regional stress field existed within the Gondwana lithosphere during the initial rifting stages (Wilson, in press). The extension direction was normal to the proto-Pacific margin suggesting that the stress field was related to plate-boundary forces generated along the active margin during subduction. Although the tectonic and magrnatic changes recorded along the proto-Pacific margin are most probably related to changes in subduction zone parameters such as (1) convergence velocity (2) subduction rollback and absolute plate motions (3) age of the subducting slab, and (4) dip of the subducring slab, one can only speculate as to what these specific changes may have been. A decrease in convergence rates, for example, implies weak coupling between the subducting and overriding plate resulting in an extensional margin. The relationship between the rollback of the subduction trench and the motion of the overriding plate toward or away from the trench is also important (Uyeda & Kanamori 1979). Where the overriding plate retr~ats from the trench, or where it advances more slowly than~i~ollback, the margin is extensional. Molnar & Atwater (1978) have suggested that cold dense oceanic slab is easier to subduct and will result in extension. The discovery of

158

B.C. STOREY ETAL.

geochemically unusual continental magnesian andesites (dated 160 Ma) within Jurassic volcanic rocks on the Antarctic Peninsula and South Georgia led Alabaster & Storey (1990) to suggest, based on the model of Tatsumi & Maruyama (1989), that subduction of young hot oceanic lithosphere prior to ridge trench interaction was an important factor in Gondwana break-up models and drew comparisons with the Basin and Range province. Although it is generally accepted that young oceanic lithosphere is difficult to subduct and results in compressional tectonics (Molnar & Atwater 1978), Carlson et al. (1983) show that encroachment of a ridge upon a trench causes a decrease in convergence rate, thus enhancing the tendency of the overriding plate to extend by back-arc or intra-arc extensional processes. Of equal importance to the model of Bott (1982) is the presence of subduction on the opposite Tethyan margin of the large Gondwana supercontinent (Fig. 2a). During Permo-Triassic times much of the N & NE margin of Gondwana faced Palaeo-Tethys and was an active subducting margin ($eng6r et al. 1988). In Late Permian times rifting subparallei with the margin heralded the opening of Neo-Tethys and other smaller basins, initially as back-arc basins above Palaeo-Tethyan subduction zones. McKenzie & Weiss (1975) suggested that major ocean spreading centres may originate from back-arc basins behind subduction zones and ~jeng6r (1979) applied this model to the opening of Neo-Tethys as a major ocean basin during Jurassic times. Crustal thickening

The spatial relationship between lithospheric extension and pre-existing orogens involving thicker than normal continental crust is well known (Dewey 1988b). Wilson (1966) noted that the Central and North Atlantic oceans nucleated on the Appalachian/Caledonian orogen, and the Basin and Range province has been related, amongst other things, to extensional collapse of crust orogenically thickened during the compressional Laramide orogeny (Coney 1987). According to Hales (1969) the driving force for plate motions is the body weight of the lithosphere, and continental rifting has been attributed to gravity glide from areas of overthickened crust ~Price et al. 1988). The initial stages of Gondwana break-up followed a period of compression during which the Permo-Triassic Gondwanian fold belt (Du Toit 1937) formed inboard of the Antarctic Peninsula. The fold belt is represented by the PermoTriassic Cape fold belt in southern Africa, the

Sierra de la Ventana fold belt of South America and the fold belt in the Ellsworth and Pensacola mountains of West Antarctica. Interestingly, the location of earliest sea-floor spreading in the South Atlantic-Weddell Sea area corresponds geographically to the same general area of crustal thickening during the Permo-Triassic Gondwanian folding (Fig. 2a). Lock (1980) postulated that Gondwanian folding was related to fiat-plate subduction and a compressive margin regime. Changes in subduction zone parameters e.g. slab dip and a reduction in compressive boundary forces during the initial stages of break-up may have enabled unconfined, overthickened Permo-Triassic crust to extend by gravitational instability. Consequently the effect of crustal thickening and extensional collapse of continental crust combined with changes in plate convergence forces may have been important during Gondwana break-up. Mantle plume

The role of mantle plumes in continental breakup is a subject of debate that is discussed extensively in recent literature (e.g. Richards et al. 1989; White & McKenzie 1989; Campbell & Griffiths 1990; Hill 1991). In active models extension is driven by mantle plumes (Richards et al. 1989) whereas in passive models extension is due to other factors (White & McKenzie 1989). Although the likely presence of a mantle plume beneath the Karoo province an~ the neighbouring Antarctic province in DrOnning Maud Land (Fig. 2; White & McKenzie 1989; Cox 1989) is not questioned here, its role, either active or passive, in the disintegration of Gondwana, and the possible requirement of higher-than-normal mantle temperatures in the production of the Ferrar-Tasman province are clearly important issues. The large volumes and ~mpositions of some magmas in the Karoo province, including sea-ward dipping reflector sequences of the Explora Wedge, offshore Dronning Maud Land are typical features of mantle plume models (e.g. White & McKenzie 1989). The irregular dyke trends in neighbouring Dronning Maud Land province (Grantham & Hunter 1991) are also consistent with at least local tension in the vicinity of a plume head. However, the difference in timing of 40-50 Ma between the plume-related Karoo magmatism and final break-up makes it unlikely the plume played an active role in the initial separation of east and west Gondwana. As far as the Ferrar-Tasman province is concerned its linear extent for upwards of 3000 kin, subparallel to the proto-Pacific margin is not

SUBDUCTION AND GONDWANA BREAK-UP compatible with conventional models of mantle plumes of 2000 km diameter (White & McKenzie 1989; Campbell & Griffiths 1990) although it is possible that plumes could have a more linear shape. It is also conceivable that the plume head was deflected to the Transantarctic Mountains lineament by a thick lithospheric keel of the Precambrian cratonic areas (see model of Thompson & Gibson 1991); however, the distances involved and the uniform stress field (Wilson in press) make this unlikely. Furthermore, the lack of age progression within the KarooFerrar-Tasman province, rules out a simple relationship to migration across a plume head. It is also by no means certain that higher-thannormal mantle temperatures are required to produce the Ferrar-Tasman magmas. The volumes concerned are low compared to many continental flood basalt provinces and may be such that decompressional melting during lithospheric thinning above normal temperature mantle is sufficient to produce the required magmas. Our geochemical data from the magmatic rocks of West Antarctica show that the igneous rocks so far sampled do not contain a plume source signature. The entirely asthenospheric derived magmas sampled from the uplifted ophiolite fragment of South Georgia, formed PERMO-TRIASSIC

PRE-RIFTING

from a depleted MORB source and not a 'plume' source. If elevated temperatures are required to melt the lithospheric mantle source and form the Ferrar tholeiitic magmas, then processes other than mantle plumes may be involved. Higherthan-normal temperatures may, for example, be due to the insulating effect of the Gondwana supercontinent (Anderson 1982) or linked in someway to subduction/back-are processes (Cox 1978).

Break-up model and concluding remarks The magmatic record along the proto-Pacific margin of Gondwana records important changes in subduction zone parameters during the initial stages of Gondwana break-up consistent with extension in the overriding plate. This, together with the record of subduction along the Tethyan margin, suggests that subduction pull on the opposite sides of the supercontinent (Fig. 2) may have given rise to tension within the Gondwana plate leading eventually to break-up. The gravitational potential of crust overthickened during the Permo-Triassic compressional event may have been a contributing factor also. A mantle plume beneath the Karoo province, although not essential to the model, may have

STAGE

TPG . . . . . . . . . . .

Gondwanide fold belt

I

. . . . -,-.~.',-,~--, ,~. ,,. ,,, ~',,

I

, ,,,.-,,;~r~-',--":;':""-,,;"

~-...J

EARLY-MIDDLE

159

JURASSIC BBG

'

"""

High compressive stress Slow absolute motion Crustal thickening

INITIAL RIFTING STAGE AP ~>..____~,~

~-

PLB

EWM ~

WSE

TAM

!

[ ~

J ~JP~ "~ ~.~ ,/

,a,lc und.rp,atlng Lower compressive stress Slower subduction rates Oblique subduction Subducting younger crust Lithospheric melting

Fig. 6. Schematic sections illustrating the main tectonomagmatic features along the proto-Pacific margin of Antarctica prior to and during the earlystagesof Gondwana break-up (Storey& Alabaster 1991). Abbreviations are BBG, Botany Bay Group fluviatilesedimentaryrocks; TPG, TrinityPeninsula Group accretionarycomplex; AP, Antarctic Peninsula; EWM, Ellsworth-Whitmore mountains; TAM, Transantarctic Mountains; PLB, Palmer Land ensialic basin; WSE, Weddell Sea embayment.

160

B.C. STOREY E T A L .

thermally weakened the lithosphere, increased magma production rates and induced local rifting, contributing to, but not causing, the eventual separation of East and West Gondwana. Regional tensional forces may have been weak at this time such that sea-floor spreading and continental break-up did not, as far as we know, commence until approximately 155 Ma, 40 Ma after the main plume-related volcanic event of the Karoo (195+4 Ma). The change from Gondwanide compression to lithospheric extension may be linked in Early Jurassic times to a reduction in plate boundary forces and a change from a shallow to a steeply dipping subduction zone (Fig. 6). During the initial rifting stage a broad extensional province developed in southern South America and West Antarctica (Fig. 2) behind an oceanward migrating magmatic arc. Initial rifting magmas were formed by decompression melting of a mantle source previously enriched (as in the case of the Ferrar magma) or enriched by contemporaneous subduction-induced processes. The mantle may have risen passively in response to crustal extension associated with changing plate boundary forces or possibly by subduction induced convective flow (model of Sleep & Toksoz 1971; Jurdy & Stefanick 1983). Anatectic melting at the base of the crust and fractionation of the mafic magmas formed silicic magmas on the margin of the extending basin. A plume beneath the Karoo provinces may have increased magma production rates, formed the dipping reflector sequences and induced local rifting. The broad extensional province is separated from the stable cratonic areas by a major transfer fault zone represented in southern South America by the Gastre Fault System (Rapela & Pankhurst, this volume). Its eastorly continuation cuts across the axis of the plume head and may have been controlled by the plume's position. Movement (possibly transtensional) along this zone may have resulted in translation and rotation of the Falkland Islands (Mitchell et al. 1986) and the Ellsworth-Whitmore mountains crustal block of West Antarctica (Grunow et al. 1991) during the early break-up stages, and NE movement of East Antarctica relative to Africa (Fig. 2b). A major change in the orientation of the stress field in the late Mid-Jurassic accompanied the initiation of sea-floor spreading and the eventual separation of east and west Gondwana. In the Weddell Sea, the ocean-continent boundary truncated earlier rifts and sediment filled basins in the Weddell Sea embayment areas (Fig. 1). The coincidence of break-up with emplacement of high magnesian andesites (155-160 Ma) in-

dicative of atypical thermal conditions during subduction, suggest that geothermal gradients may have been high at this stage and influenced break-up. Although the true causes of Gondwana breakup may never be firmly established our data indicate the importance of a plate interaction model encompassing a reduction in subduction plate boundary forces. The combination of subduction pull on the opposite sides of Gondwana combined with the more local tensional effects of a mantle plume and collapse of overthickened lithosphere may have been sufficient to cause break-up and the final separation of east and west Gondwana. The delay of c. 30 Ma between the Karoo magmatism and final separation suggests the mantle plume did not provide the essential trigger for Gondwana break-up but it may have controlled the ultimate position of break-up and of major fault systems.

References ALABASTER, T. & STOREY, B. C. 1990. Antarctic Peninsula continental magnesian andesites: indicators of ridge-trench interaction during Gondwana break-up. Journal of the Geological Society, London, 147, 595-598. ANDERSON, D. L. 1982. Hotspots, polar wander, Mesozoic convection, and the geoid. Nature, 297, 391-393. BELL, R. E., BROZENA, J.M., HAXBY, W. F. & LABRECOUE,J. L. 1990. Continental margins of the western Weddell Sea: insights from airborne gravity and Geosat-derived gravity. Antarctic Research Series, 50, 91-102. BoTr, M. H. P. 1982. The mechanism of continental splitting. Tectonophysics, 81,301-309. 1992. The stress regime associated with continental break-up. This volume. BREWER,T. S., REX, D., GUISE,P. & STOREY,B. 1991. Ar40/Ar39 Age determinations from the Theron and Pensacola Mountains, Antarctica: implications for the age of Mesozoic magmatism in Antarctica. Programme and abstracts, Magmatism and the causes of continental break-up. Geological Society, London, P. 1. BRUHN,, R. L., STERN, C. R. & DEwrr, M. J. 1978. Field and geochemical data bearing on the development of a Mesozoic volcano-tectonic rift zone and back-arc basin in southernmost South America. Earth and Planetary Science Letters, 41, 32-46. BtFI'I~RWORTH,P. J., CRAME,J. A., HOWLETr,P. J. & MACDONALD,D. I. i . 1988. Lithostratigraphy of Upper Jurassic-Lower Cretaceous strata of eastern Alexander Island, Antarctica. Cretaceous Research, 9, 249-264. CAMPBELL,I. H. & GRIFFrrBs,R. W. 1990. Implications of mantle plume structure for the evolution of flood basalts. Earth and Planetary Science Letters, 99, 79-93.

SUBDUCTION AND GONDWANA BREAK-UP

161

CAm.soN, R. C., HmDE, T. W. C. & UYEDA, S. 1983. FORSYTH, D. & UYEDA, S. 1975. On the relative imThe driving mechanism of plate tectonics: reportance of the driving forces of plate motion. lation to age of the lithosphere at trenches. Geophysical Journal of the Royal Astronomical Geophysical Research Letters, 10, 297-300. Society, 43, 163-200. CONEY, P. J. 1987. The regional tectonic setting and GRANTHAM,G. H. & HUNTER,D. R. 1991. The timing and nature of faulting and jointing adjacent to the possible causes of Cenozoic extension in the North American cordillera. In: COWARD,M. P., Pencksokket, western Dronning Maud Land, Antarctica. In: THOMSON,M. R. A., CRAME, J. DEWEY, J. F. & H,(NCOCK, P. L. (eds) ContinA. & THOMSON,J. W. (eds) Geological Evolution ental Extensional Tectonics. Geological Society, London, Special Publication, 8, 15-39. of Antarctica. Cambridge University Press, Cox, K. G. 1978. Flood basalts, subduction, and the 47-51. break-up of Gondwanaland. Nature, 274, 47-49. GRUNOW, A. M., KErn', D. V. & DALZJEL, I. W. D. 1991. New Paleomagnetic data for Thurston 1988. The Karoo Province. In: MACDOOGALL, Island: Implications for the tectonics of West J. D. (ed.) Continental Flood Basalts. Kluwer Academic Publishers, Dordrecht, 239-271. Antarctica and Weddeil Sea opening. Journal of 1989. The role of mantle plumes in the developGeophysical Research, 96, BII, 17935-17954. ment of continental drainage patterns. Nature, • GUST, D. A., BIDDLE, K. T., PHELPS, D. W. & 342, 873-877. ULIANA, M. A. 1985. Associated Middle to Late DALY, M : C., LAWRENCE, S. R., KIMUNA'A, D. & Jurassic volcanism and extension in southern BINGA, M. 1991. Late Palaeozoic deformation in South America. Tectonophysics, 116, 223-253. central Africa: a result of a distant collision? HALES, A. L. 1969. Gravitational sliding and conNature, 350, 605-607. tinental drift. Earth and Planetary Science DAt.ZJEL, I. W. D. & ELLIOT, D. H. 1982. West Letters, 6, 31-34. Antarctica: problem child of Gondwanaland. HARRISON,S. M. & PIERCY,B. A. 1990. The evolution Tectonics, 1, 3-19. of the Antarctic Peninsular magmatic arc; Evid- ~ , DEwrr, M. J. & PALMER,K. F. 1974. Fossil marence from northwestern Palmer Land. In: KAY, ginal basin in the southern Andes. Nature, 250, S. M. & RAPELA, C. W. (eds) Plutonism from 291-294. Antarctica to Alaska. Geological Society of , ELLIOT, D. H., JONES, D. L., THOMSON,J. W., America, Special Paper, 241, 9-25. THOMSON, M. R. A., WELL, N. A. & ZINSMEIS- HERGT, J. M., CHAPPELL,B. W., MCCULLOCK,M. T., "mR, W. J. 1981. The geological significance of McDOUGALL, I. & CHIVAS, A. R. 1989a. Geosome Triassic microfossils from the South chemical and isotopic constraints on the origin Orkney Islands, Scotia Ridge. Geological of the Jurassic dolerites of Tasmania. Journal of Magazine, lie, 15-25. Petrology, 30, 841-883. , STOREY, B. C., GARRETr, S. W., GRUNOW,A. - - , FAURE, G. & MENSING,T. M. 1989b. The M., HERROD,L. D. B. & PANKHURST,R. J. 1987. geochemistry of Jurassic dolerites from Portal Extensional tectonics and fragmentation of Peak, Antarctica. Contributions to Mineralogy Gondwanaland. In: COWARD,M. P., DEWEY, J. and Petrology, 102, 298-305. F. & HANCOCK,P. L. (eds) Continental ExtenHERVI~, F., DAVIDSON,J., GODOY, E., MPODOZIS, C. sional Tectonics. Geological Society, London, & COVACEVlCH,V. 1981. The Late Paleozoic in Special Publication, 28, 433-441. Chile: stratigraphy, structure and possible tectonic framework. Anais da Academia Brasileira DEWEY, J. e. 1988a. Lithospheric stress, deformation de Ciencias, 53, 361-373. and tectonic cycles: the disruption of Pangaea , LOSKE, W., MILLER, H. & PANKHURST, R. J. and the closure of Tethys. In: AUDLEY-CHARLES, 1991. Chronology of provenance, deposition and i . G. & HALl.AM, A. (eds) Gondwana and metamorphism of deformed fore-arc sequences, Tethys. Geological Society, London, Special southern Scotia arc. In: THOMSON, M. R. A., Publication, 37, 23-40. CRAME, J. A. & THOMSON,J. W. (eds) Geological 1988b. Extensional collapse of orogens. TecEvolution of Antarctica. Cambridge University tonics, 7, 1123-1139. Press, 429-435. DREWRY, D. J. 1983. Antarctica: Glaciological and geophysical Folio 2. Scott Polar Research InHILL, R. I. 1991. Starting plumes and continental stitute, Cambridge. break-up. Earth and Planetary Science Letters, Du Torr, A. L. 1937. Our Wandering Continents. 104, 398-416. Oliver & Boyd, Edinburgh. HOLDSWORTH,D. K. & NELL, P. A. R. 1992. Mesozoic FARQUHARSON, G. W. 1984. Late Mesozoic, nonradiolarian faunas from the Antarctic Peninsula marine conglomeratic sequences of northern and their geological significance. Journal of Antarctic Peninsula (the Botany Bay Group). Geological Society, London, 149, 000-000. British Antarctic Survey Bulletin, 65, 1-32. HOLE, M. J. 1986. Time controlled geochemistry o1: FITCH, F. J. & MXLLER,J. A. 1984. Dating Karoo igneigneous rocks o1: the Antarctic Peninsula. PhD ous rocks by the continental K-Ar and 40Ar/39Ar thesis, Bedford College, University of London. age spectrum methods. In: ERLANK, A. J. (ed.) , PANKHURST,R. J. & SAUNDERS, A. D. 1991. Petrogenesis of the volcanic rocks o1: the Karoo Geochemical evolution of the Antarctic PeninProvince. Geological Society of South Africa, sula magmatic arc: the importance of mantleSpecial Publication, 13, 247-266. crust interactions during granitoid genesis. In:

B.C. STOREY E T A L .

162

THOMSON,M. R. A., ~ ,

J. A. & THOMSON, J. W. (eds) Geological Evolution of Antarctica. Cambridge University Press, Cambridge, 369-374. HOUSEMAN, G. & ENOLAND, P. 1986. A dynamical model of lithosphere extension and sedimentary basin formation. Journal of Geophysical Research, 91,719-729. JURDY, D. M. & SEFANICK,M. 1983. Flow models for back-arc spreading, convergence and subduetion. Tectonophysics, 99, 191-206. KAY, S. M., RAMOS,V. A., MPODOZlS,C. & SRUOGA, P. 1989. Late Paleozoic to Jurassic silicic magmatism at the Gondwana margin: Analogy to the Middle Proterozoic in North America? Geology, 17, 324-328. KmSTO~RSON, Y. & HINZ, K. 1991. Evolution of the Gondwana plate boundary in the Weddell Sea area. In: THOMSON,M. R. A., CRAME, J. A. & THOMSON, J. W. (eds) Geological Evolution of Antarctica. Cambridge University Press, Cambridge, 225-230. Kyt~, P. R., ELLIOT, D. H. & StyrmR, J. F. 1981. Jurassic Ferrar Supergroup tholeiites from the Transantarctic Mountains, Antarctica, and their relation to the initial fragmentation of Gondwana. In: CRESSWEta., M. M. & VELLA,P. (eds) Gondwana Five, Proceedings of the Fifth International Gondwana Symposium, Wellington, New Zealand. A. A. Balkema, Rotterdam, 283-287. , PANmtURST, R. J. & BOWMAN, J. R. 1983. Isotopic and chemical variations in Kirkpatrick Basalt Group rocks from southern Victoria Land. In: OLIVER,R. L., JAMES,P. R. & JAGO,J. B. (eds) Antarctic Earth Science. Australian Academy of Science, Canberra, 234-237. LA BRECOUE,J. L., BROZENA,J., PARRA,J. C., KELLER, M. A., HAXBY, W., RAYMOND,C. A., KOVAKS,L., BELL,R., YANEZ,G., PEERS, M., CANDE, S. & VALLADARES, J. 1989. USAC aerosurvey results for the Weddell Basin: Part 1. Abstracts, 28th International Geological Congress, Washington DC, USA, 2-250. LAWYER, L. A. & SCOESE, C. R. 1987. A revised reconstruction of Gondwanaland. In: McKENZm, G. D. (ed.) Gondwana Six: Structure, Tectonics & Geophysics. AGU Geophysical Monographs, 40, 17-23. LOCK, B. E. 1980. Flat plate subduction and the Gape Fold Belt of South Africa. Geology, 8, 35-39. MARTIN, A. K. & HARTNADY,C. J. H. 1986. Plate tectonic development of the South Indian Ocean: A revised reconstruction of East Antarctica and Africa. Journal Geophysical Research, 91, 4767-4786. MCKENZlE, D. P. & WEISS, N. O. 1975. Speculations on the thermal and tectonic history of the earth. Geophysical Journal of the Royal Astronomical Society, 42, 131. MILLAR, I. L., MtLNE, A. J. & WHrrHAM, A. G. 1990. Implications for Sm-Nd garnet ages for the stratigraphy of northern Graham Land, Antarctic Peninsula. Zentralblau fur Geologie und Pali~ontologie, 1, 97-104.

Mna~, A. J. & MILLAR, I. L. 1989. Short paper: The significance of mid-Paleozoic basement in Graham Land, Antarctic Peninsula. Journal of the Geological Society, London, 146, 207-210. MrrCHELL, C., TAYLOR,G. K., Cox, K. G. & SHAW,J. 1986. Are the Falkland Islands a rotated microplate? Nature, 319, 131-134. MOLNAR, P. & ATWAER, T. 1978. Interarc spreading and Cordilleran tectonics as alternates related to the age of subducted oceanic lithosphere. Earth and Planetary Science Letters, 41,330-340. MORGAN,W. J. 1971. Convection plumes in the lower mantle. Nature, 230, 42-43. 1981. Hotspot tracks and the opening of the Atlantic and Indian oceans, In: EMnaANI, C. (ed.) The Sea Vol. 7. Wiley Interscience, New York, 443--475. MUKASA,S., BRUECKNER,H. K. & DALZmL,I. W. D. 1988. Zircon U-Pb constraints on the kinematic evolution of the northern Scotia are. GSA Annual Meeting, Denver, Colorado, 1988, Abstracts with Programs, 20, A12. PAN~UnS'r, R. J. 1982. Rb-Sr geochronology of Graham Land, Antarctica. Journal of the Geological Society, London, 139, 701-711. 1983. Rb-Sr constraints on the ages of the basement rocks of the Antarctic Peninsula. In: OLIVER,R. L., JAMES,P. R. & JAGO, J. B. (eds) Antarctic Earth Sciences. Australian Academy of Sciences, Canberra, 367-371. 1990. The Paleozoic and Andean magmatic ares of West Antarctica and southern South America. In: KAY,S. M. & RAPELA,C. W. (eds) Plutonism from Antarctica to Alaska. Geological Society of America Special Paper, 241, 1-7. , HOLE, M. J. & BROOK, M. 1988. Isotope evidence for the origin of Andean granites. Transactions of the Royal Society of Edinburgh, Earth Sciences, 19, 123-133. PAn~LDA, M. A. 1990. Granitoid plutonism in central Chile and its geodynamicimplications; A review. In: KAY,S. M. & RAPELA,C. W. (eds) Plutonism from Antarctica to Alaska. Geophysical Society of America, Special Paper, 241, 51-66. , RIUANO, S., SEPULVEDA, P., HERVI~, M., HERVI~, F., PUIG, A., MUNIZAGA,F., BROOK, M., PANKHURST,R. J. & SNELLING,N. 1988. Mesozoic and Cenozoic plutonic development in the Andes of Central Chile (30°30'-32°30'S). Journal of South American Earth Sciences, 1 , 249-260. P~ARCE, J. A. 1983. Role of the sub-continental lithosphere in magma genesis at active continental margins. In: HAWKESWORTH,C. J. & NORRY,M. J. (eds) Continental Basalts and Mantle Xenoliths. Shiva, Nantwich, England, 230-249. , HARRIS,N. B. W. & TINI)LE, A. G. 1984. Trace element discrimination diagrams for the tectonic interpretation of granitic rocks. Journal of Petrology, 25, 956-983. PratE, N. J., PRICE, G. D. & PRICE, S..C. 1988. Gravity glide and plate tectonics. In: AUDLEYCHARLES,M. G. & HALLAM,A. (eds) Gondwana and Tethys. Geological Society, London, Special Publications, 37, 5-21.

SUBDUCTION AND GONDWANA BREAK-UP QUILTY, P. G. 1977. Jurassic ammonites from Ellsworth Land, Antarctica. Journal of Paleontology, 44, 110-116. RAMOS, V. A. & KAY, S. M. 1991. Triassic rifting and associated basalts in the Cayo basin, central Argentina. In: HARMON,R. S. d~.RAPELA,C. W. (eds) Andean magmatism and its tectonic setting. Geological Society of America Special Publication, 265, 79-91. RArELA, C. W. & PANK,URST, R. S. 1992. The granites of northern Patagonia and the Gastre Fault System in relation to the break-up of Gondwana, This volume. , - - & HARmSON,S. M. 1992. Triassic 'Gondwana' granites of the Gastra District, North Patagonian Massif. In: BROWN, P. E. & CHAPPELL, B. W. (eds) Second Hutton symposium on granites and related rocks. Transactions of the Edinburgh Royal Society: Earth Sciences, 83, 291-304. RICHARDS, M. A., DUNCAN,R. A. & COURTILLOT,V. E. 1989. Flood basalts and hotspot tracks: Plume heads and tails. Sciences, 246, 103-107. ROWLEY, P. D., VENNUM,W. R., KELLOGG, K. S., LAUDON,T. S., CARRARA,P. E., BOYLES,J. M. & THOMSON,M. R. A. 1983. Geology and plate tectonic setting of the Orville Coast and eastern Ellsworth Land, Antarctica. In: OLIVER, R. L., JAMES, P. R. dk JAGOJ. B. (eds) Antarctic Earth Science. Australian Academy of Science, Canberra, 245-250. SENGOR, A. i . C. 1979. Mid-Mesozoic closure of Permo-Triassic Tethys and its implications. Nature, 279, 590-593. , ALTMER, D., CIN, A., USTAOMER,T. & HSO, K. J. 1988. Origin and assembly of the Tethyside orogenic collage at the expense of Gondwana Land. In: AUDLEY-CHARLES,M. G. & HALLAM, A. (eds) Gondwana and Tethys. Geological Society, London, Special Publications, 37, 119-181. SLEEP, N. M. & TOKSOZ,'N. 1971. Evolution of marginal basins. Nature, 233, 548-550. STOREY, B. C. & ALABASTER,T. 1991. Tectonomagmatic controls on Gondwana break-up models: evidence from the proto-Pacific margin of Antarctica. Tectonics, 10, 1274-1288. & GARRETr, S. W. 1985. Crustal growth of the Antarctic Peninsula by accretion, magmatism and extension. Geological Magazine, 122, 5-14. & MAIR, B. F. 1982. The composite floor of the Cretaceous back-arc basin of South Georgia. Journal of the Geological Society, London, 139, 729-737. ~, DALZIEL,I. W. D., GARRETF, S. W., GRUNOW, A. m., PANKHURST, R. J. & VENNUM, W. R. 1988a. West Antarctica in Gondwanaland: Crustal blocks, reconstruction and break-up processes. Tectonophysics, 155, 381-390. , HOLE, M. J., PANKHURST,R. J., MILLAR,I. L. & VENNUM, W. R. 1988b. Middle Jurassic withinplate granites in West Antarctica and their bearing on the break-up of Gondwanaland. Journal of Geological Society, London, 145,999-1007. ~, PANKHURST,R. J., MILLAR,I. L., DALZIEL,I. W. D. & GRUNOW,A . M. 1991. A new look at

163

the geology of Thurston Island. In: THOMSON, M. R. A., CRAME,J. A. & THOMSOM,J. W. (eds) Geological Evolution of Antarctica. Cambridge University Press, 399-403. TANNER, P. W. G., PANKHURST, R. J. & HYDEN, G. :982. Radiometric evidence for the age of the subduction complex in the South Orkney and SQuth Shetland Islands, West Antarctica. Journal of the Geological Society, London, 139, 683-690. TATSUMI, Y. • MARUYAMA,S. 1989. Boninites and high-Mg andesites: tectonics and petrogenesis. In: CRAWFORD,A. J. (ed.) Boninites and related rocks. Unwin Hyman, London, 50-:71. THOMPSON, R. N. & GmSON, S. A. 1991. Subcontinental mantle plumes, hotspots and thin spots. Journal of the Geological Society, London, 148, 973-977. THOMSON, M. R. A. 1983. Late Jurassic ammonites from the Orville Coast, Antarctica. In: OLIVER, R. L., JAMES,P. R. & JAGO,J. B. (eds) Antarctic Earth Science. Australian Academy of Science, Canberra, 315-319. & PANKHURST, R. J. 1983. Age of post-Gondwanian calc-alkaline volcanism in the Antarctic Peninsula region. In: OUVER,R. L., JAMES,P. R. d~ JAGO, J. B. (eds) Antarctic Earth Science. Australian Academy of Science, Canberra, 328-333. & TRArCa'ER, T.H. 1986. Early Jurassic fossils from central Alexander Island and their geological setting. British Antarctic Survey Bulletin, 70, 23-39. TRouw, R. A. J., PANKHURST,R. J. & KAWASHITA,K. 1990. New radiometric age data from Elephant Island, South Shetland Islands. Zentralblau fiir Geologie und Paliiontologie, 1, 105-118. ULIANA, i . A., BIDDLE, K. T. & CERDAN, J. 1989. Mesozoic extension and the formation of Argentine sedimentary basins. Bulletin of the American Association of Petroleum Geologists, 46, 599-614. UYEOA, S. & KANAMORI,H. 1979. Back-arc opening and the mode of subduction. Journal of Geophysical Research, 84, 1049-1061. WEVER, H. E. & STOREY,B. C. 1991. Magmatic evolution of a rear-arc region on NE Palmer Land, Antarctic Peninsula: evidence for an early Jurassic marginal basin. Abstracts, Eighth International Symposium on Gondwana, Hobart 1991. &~ 1992. Bimodal magmatism in northeast Palmer Land, Antarctic Peninsula: Geochemical evidence for a Jurassic ensialic back-arc basin. Tectonophysics, 205, 239-260. WHITE, R. & MCKENZIE, D. 1989. Magmatism at rift zones: the generation of volcanic continental margins and flood basalts. Journal of Geophysical Research, 94, 7685-7729. WILSON, T. J. In press. Jurassic faulting and magmatism in the Transantarctic Mountains: implications for Gondwana break-up mechanisms. In: FINDLAY, R. n . (ed.) Gondwana 8 - Assembly, Evolution and Dispersal. A. A. Balkema, Rotterdam.

Jurassic magmatism and tectonism associated with Gondwanaland break-up: an Antarctic perspective D A V I D H. E L L I O T

Department o f Geological Sciences and Byrd Polar Research Center, The Ohio State University, Columbus, Ohio 43210, USA

Abstract: Magmatic and tectonic activity in Antarctica associated 'with the early stages of continental rifting and break-up of Gondwanaland culminated with tholeiitic magmatism at about 175-180 Ma. In the Ross Sea sector of the Transantarctic Mountains, Jurassic igneous rocks, comprising pyroclastic rocks and Ferrar Group tholeiitic basalts, overlie fluvial strata of the Permian-Upper Triassic Gondwana sequence. Petrological, structural and volcanological data suggest that the Jurassic pyroclastic rocks and overlying flood basalts were erupted into a volcano-tectonic rift system associated with lithospheric extension and decompression melting. Geochemically the Ferrar tholeiites form part of the Gondwana lowTi province, but they exhibit marked differences in initial STSr/~'Srratios and in high field strength element abundances in comparison with other parts of the province. Three Early to Middle Jurassic tectono-magmatic terrains were present in this part of Gondwanaland: a plate margin magmatic arc; a belt of silicic within-plate igneous rocks inboard of the arc; and a continental flood basalt province (the Gondwana low-Ti province). Silicic volcanism mainly preceeded emplacement of Ferrar tholeiites, whereas major silicic activity elsewhere either accompanied basaltic magmatism or, though spatially separated preceded it. Younger magmatic episodes have obscured much of the plate margin record. Lack of reliable age determinations makes details of relations between tectonism and magmatism difficult to assess.

Models for the generation of continental flood basalts explain the production of very large volumes of magma either by active plume-driven processes (Duncan & Richards 1991) or by stretching of the lithosphere with passive upwelling of the mantle, the process being enhanced by the presence of a plume (White & McKenzie 1989). Where stretching and passive upwelling (White & McKenzie 1989) has occurred, rifting should precede magmatism. Given normal temperature mantle, crustal thinning and subsidence should be followed by eruption of basaltic magmas formed by decompression melting; however, if by coincidence a sufficiently hot plume were uncovered at the site of rifting, uplift would precede magmatism. The active plume model (Richards et al. 1989; Campbell & Griffiths 1990; Duncan & Richards 1991), which incorporates aspects of the White & McKenzie (1989) model, was developed in part because of the requirement, in the case of continental flood basalt provinces, that large volumes of magma be produced over short periods. In this model, flood basalt volcanism should precede rifting of the continental crust. These alternaltive models can be evaluated by examination of the sequence of geological events and the magmas formed. Tholeiitic rocks of the Jurassic Ferrar Group

of Antarctica constitute a continental flood basalt province with a relatively small estimated volume of 5 x 105 km 3 (Kyle et al. 1981), but a unique geochemical signature (Kyle 1980). The Ferrar tholeiites form the major part of a larger Jurassic igneous province in Antarctica that includes Dronning Maud Land (Fig. 1). Basaltic intrusions of the Ferrar Group crop out at Horn Bluff west of north Victoria Land and along the Transantarctic Mountains to the Theron Mountains. Extrusive equivalents cap the Devonian?Lower Jurassic Beacon Supergroup, the Permian and younger part of which forms the Antarctic Gondwana sequence (Barrett et al. 1986; Elliot et al. 1986b). The Ferrar tholeiites differ from most other continental flood basalt provinces in that they form a linear belt. They were associated with crustal extension although, like the Columbia River Basalts, they were not followed by sea floor spreading. The field relations suggesting the tectonic setting in which the Ferrar tholeiites were emplaced, the geochemistry of the Ferrar tholeiites, and the spatial and temporal relations with other magmatic rocks both within Antarctica and on the formerly adjacent continents are discussed: The Late Jurassic to Early Cretaceous tholeiitic magmatism in the South Atlantic reg-

From STOREY,B. C., ALABASTER,T. & PANKHURST,R. J. (eds), 1992, Magmatism and the Causes of ContinentalBreak-up, Geological Society Special Publication No. 68, pp. 165-184.

165

166

D.H. ELLIOT

.~

I ,r ~ ' ~ ~.~_l~l

70o.s~-._::A "~"

KraulM t n s . / / ~ ~ _

~ .

t./

H o ~ . e u ~ n a ~ t ~ , ¢ Esmark'mUg.t[":~.TAnt~etiea[ "~] Weda~# ~ 7---" ~ ~ c \ ; /l 4sow ,~;~N~2~ [ ~ ,_._~ | Littlewood // Nunataks JU...~ "~ Theron M l n s . ~_~. ' . °. . . ,, i o ~m ~ ~ ~'~ Sh~ckleton~ W.hiehaway .. ,ange l["J~'Nunataks" × )~" ~ EAST D~sek# ~ e ~ a c o l a ~,,, ANTARCTICA .

.

.

.

EIIsworthMtns./~""r~M"~ m.~ ~ + + +~ "/',,x ined+ ~J- i o SOUTH+"¢~ s0os ~/%

Whltmore Mtns.

+9OoZ~00S.

POLE "O~ ~ M t . ~ MI. ~Bumstead s t , 0I , ~ ,

WEST

y

~

evolution of the major trunk drainage system which received sediment from a cratonic source on the present polar plateau flank and a magmatic source on the Gondwana plate margin in West Antarctica (Vavra 1984; Barrett et al. 1986; Collinson 1991; Isbell 1991). The floods of epiclastic volcanic debris and the sparse occurrence of ash suggest the magmatic source was a distal active arc. The Permian basins in Victoria Land, however, reflect a more cratonward location relative to the plate boundary, being intracratonic and isolated from input of volcanic detritus from the inferred arc (Woolfe 1991) Jurassic volcanic rocks that cap the Beacon sequence are exposed in three widely separated areas of the Transantarctic Mountains; tholeiitic intrusions, however, occur throughout the range.

~_~Marshal O~,,

VICTORIA :

lABS°w "-~'~

"~

-"---.,~'~Alanl-tk 1350E

-I- Jurassic granitic Dlutons

"~

v Kirkpatrick Basalt Ferrar intrusive rocks

~ ~

Middle

I o lt~-~-~ [O ~

Early Jurassic

~°/u~

Late

Plutonic, metavolcanic and metasedimentary rocks of Late Proterozoic to Early Palaeozoic age form most of the basement in the Transantarctic Mountains although older Proterozoic gneisses and schists, with Archean sources, occur on the cratonic margin of the range (Borg & DePaolo 1991; Stump 1992). The Beacon Supergroup was deposited on a pro-Devonian erosion surface cut across the basement. The Devonian part of the Beacon Supergroup is a passive margin sequence; in contrast, much of the PermoTriassic part (Fig. 2) constitutes the sedimentary fill of a foreland basin (Collinson 1991). The facies, palaeocurrents and petrology of these rocks record the subsidence of the basin and the

E~pos=e Hill Mawson I ZIALI 100+m

.

Las y

Fremouw YV:¥~ 700m ~

~~ o :'i~ ;~--~.~'.-~.z2::; "=

~

]~ ' x ~ ~ N ~

w... .

'~ o~O

L~ v ° o

'~f"::":':"

' ~.~ lOOm ' ~" :'i Feather :::i,::!;ii!~i:i'!i:-t50m

l m

Tectonic setting of the Jurassic volcanic rocks

~.

Permian I ~

l

7sam

l VT~lV

c.rapaco 120m t] ~ALl

~z~

......... ~

[

3BOrn I ~ [ ~ [

~ ~

.-..ol ..:_N:

Fig. 1. Locality of the Transantarctic Mountains showing the distribution of Jurassic tholeiitic rocks and granitoids.

ion is considered only with reference to other parts of the Gondwana geochemical province to which the Ferrar belongs,

~ ~

~ 6 a~,~; , ! 180-530m

LAND

~Undifarentiatedbedr°ckl.~_O;oo S

Prebble --^

I ®1~

'

Mesa Range

62em

..,-r=nir..... II~1~^~,, I:: I ' ~ ~

.

.

.

I~

r~kroona

.

ao-~,omK\\\\\\\\\"~. ~',.\\\\\\\\"~ ~'~ "l,'~m

C,0 0 ~

Motschel

0 O 0 glacial bed~ o o -0-650m

wr

.usa.

~

A 6 o X \ \\

Breccia/ Laharlc beds Tuff Diamicttfe Hiatus

--'-=-¢ Shale :& & Sandstone (,vOlCanic; proximal) i¥~/i Sandstone (volcanic: arc derived) ~::~,::~-iSandstone (quartzose/arkosic)

Con,

Fig. 2. Simplified stratigraphic columns for the Victorian Group, Beacon Supergroup (Antarctic Gondwana sequence) and overlying pyroclastic rocks and basalt lavas of the Ferrar Group. The central Transantarctic Mountains column is for the Beardmore Glacier region (Barrett et al. 1986), south Victoria Land column is a composite for that region (Barrett & Kohn 1975; Kyle 1979; Bradshaw 1987), and the north Victoria Land column is also a composite (Collinson etal. 1986; Elliot etal. 1986a, b). The principal arc-derived volcanic sandstones are marked by symbols; the quartzose sandstones in the lower Falla Formation include a minor volcanic component.

JURASSIC MAGMATISM & TECTONISM: ANTARCTICA

167

The principal structure and tectonic feature of was as much as 200-300 km closer to the Transthe Transantarctic Mountains is the abrupt thin- antarctic Mountains in the Early to Mid-Jurassic ning of the crust, from about 40 km under the than it is today. Crust of normal thickness (c. 40 range to about 20 km beneath the Ross Sea- km) probably existed prior to break-up. Ross Ice Shelf region (Behrendt & Cooper 1991; Behrendt et al. 1991). The crustal thinning and the range-parallel faults along the mountain Tectonostratigraphic development of the front may reflect several stages in the evolution Jurassic volcanic province of the region, starting with Jurassic rifting and followed by Cretaceous and Cenozoic uplift of Jurassic magmatism marks a major change in the the range (Stump & Fitzgerald 1992; Gleadow & tectonic development of the Transantarctic Fitzgerald 1987; Fitzgerald & Gleadow 1988). Mountains. Foreland basin deposition in the Repeated movements of the range and deepen- Beardmore Glacier region terminated with the ing of adjacent sedimentary basins, at least in the plant-bearing Upper Triassic beds of the lower Ross Sea region (Cooper et al. 1987), have FaUa Formation (Table 1, Fig. 2). Changes in obscured original Jurassic features. Little evi- thickness of the lower part of the Falla Formadence exists for any major structural boundary tion suggest a significant disconformity occurs at on the polar plateau flank of the Transantarctic the contact with the mainly volcanogenic upper Mountains, although that might simply reflect part of the formation. Lack of continuous secsparse geophysical data. Furthermore, the pre- tions precludes establishing whether such relasent day and former extent of the Ferrar tions exist in Victoria Land. tholeiites is uncertain. The width of the outcrop belt of the Ferrar Group ranges up to 160 km in the Beardmore Glacier region although Stratigraphy o f the volcanogenic sequences generally less than 100 km. Only in the sector The section between the Upper Triassic plantnear the South Pole is there evidence for extens- bearing beds (lower FaUa Formation), the upive sub-ice mesa topography that can be inter- permost beds of the foreland basin fill, and the preted as the Beacon Supergroup with intruded Middle Jurassic basalts is best dgveloped in the Ferrar sills (Drewry 1972), although Steed & Beardmore Glacier region where it is repDrewry (1982) suggested, on the basis of resented by 310 m of silicic and basaltic volgeophysical surveys, that Beacon beds might be canogenic strata (Table 1, Fig. 2). In the other present as much as 300-400 km to the west of two localities, both in Victoria Land, the section north Victoria land. On the Ross Sea flank, contains a major hiatus or is not exposed in conxenoliths of Ferrar rocks are known only from tinuous section. Cenozoic volcanic centres along the TransanIn the Beardmore Glacier region the voltarctic Mountains foothills (Berg 1991), suggest- canogenic section below the basalts consists of ing that (1) the Ferrar province either did not ex- about 235 m of strata dominated by fine-grained tend into the present Ross Sea, or (2) the Beacon silicic tuff (upper Falla Formation) and about 75 and Ferrar rocks were removed by erosion, or m of beds dominated by coarse pyroclastic (3) that fortuitously such rocks were not en- debris of basaltic composition (Prebble Formatrained by the Cenozoic magmas. The present tion). The change from the carbonaceous beds outcrops of the Ferrar Group may be the rem- of the lower Falla Formation to the volcanogenic nants of a linear belt, perhaps as much as two to strata of the upper FaUa Formation is abrupt. three times as wide as its present known extent, The silicic beds are divided into three informal but probably not the remnants of a vast out- units, the lower consisting of sandstone and pouring as in southern Africa (Marsh & Eales tuffaceous beds, the middle dominated by re1984). sistant tuff beds and the upper consisting of nonIn a broader context, Ferrar magmatism has a resistant tufts. Quartzose sandstones in the spatial relationship to the Early to Mid-Jurassic lower unit of the upper Falla grade up into tufGondwana plate margin (Elliot 1974; Cox 1978), faceous beds, whereas in higher units the occurring in a belt parallel to, but inboard of, sandstones consist almost entirely of reworked the margin and emplaced in the Gondwana fore- volcanic debris. The abundance of silicic glass land basin region. The foreland basin region in- shards in the tuffaceous beds and the thickness cluded what is now the Ross embayment in West of individual tuff beds (up to 30 cm) point to proxAntarcitica which is floored by thin (c. 20 km • imal sources, as do accretionary lapilli. thick) continental crust (summarized in The overlying coarse pyroclastic beds of the Behrendt et al. 1991) extended as a consequence Prebble Formation consist of laharic deposits together with minor breccias, tufts and volcanic of Gondwana break-up and subsequent events. The thinned crust suggests that the plate margin sandstones. The beds contain blocky siderome-

168

D.H. ELLIOT

Table 1. Upper Triassic to Mid-Jurassic stratigraphy at Mount Falla, central Transantarctic Mountains

Age Middle Jurassic .._9...

Early Jurassic

Formation

Thickness

Kirkpatrick basalt

Tholeiitie lava flowswith sparse lacustrine and tuffaceous inter beds

445 m

Prebble Fm.

Laharic deposits, pyroelastic breccias, tuff and 'tuffaceous sandstone

-75m

Upper FaUa Fro.

Unit 3. Tuff Unit 2. Tuff and tuffaceous sandstone Unit 1. Sandstone and tuffaceous sandstone

106 m 63 m 66 m

Lower Falla Fm.

Quartzose volcaniclastie sandstone, siltstone and carbonaceous shale

296 m

oo.~...

Late Triassic

Description

lane shards as well as accretionary lapilli and armored lapilli, all pointing toward the importance of water in the depositional setting. The coarse deposits contain clasts of basaltic rocks (many of which are medium grained dolerites), silicic tuff and sedimentary rocks. The basaltic clasts indicate emplacement of sills and dykes in the Beacon sequence before lava eruption, whereas the dasts of tuff and sedimentary rocks (most of which can be identified as Beacon Supergroup) point to either or both redeposition of underlying rocks exposed by erosion and incorporation of wall rock fragments. The capping lavas of the Kirkpatrick Basalt (Fig. 2), as much as 520 m thick in the Beardmore Glacier region (Barrett et al. 1986), comprise a small number (c. 10) of relatively thick flows (30-175 m thick) together with a variable number of thin flows (1-20 m thick). Continuing silicic volcanism is indicated by the sparse occurrence of silicic glass shards in interbeds (Elliot et al. 1991). Continued influence of water (Fleming & Elliot 1991) is suggested by lacustrine beds and in a number of flows by the presence of pillow basalt, hyaloclastite and thick glassy entablatures (accepting the mechanism for entablature formation of Long & Wood 1986). In south Victoria Land (Fig. 2) the Triassic Lashly Formation (Barrett & Kohn 1975) is disconformably overlain by as much as 450 m of basaltic pyroclastic rocks of the Mawson Formation (Bradshaw 1987). The Carapace Sandstone consists mainly of detrital grains reworked from the Mawson beds. The section is capped by up to 380 m of basalts (Kyle 1979). In north Victoria Land the plant-bearing Triassic Section Peak Formation (Coilinson et al. 1986) is overlain by pyroclastic rocks of the Exposure Hill Formation which itself is capped by as much as 780 m of Kirkpatrick Basalt (Elliot et al. 1986a, b). Comagmatic dolerite sills occur throughout

the Transantarctic Mountains, emplaced mainly into the Beacon Supergroup and only locally into the pre-Devonian basement. Individual sills are up to 400 m thick and where the Beacon sequence is fully developed the cumulative thickness may exceed 1200 m. Dykes are widely scattered and generally less than 10 m in width. No extensive dyke swarms have been found. Evidence of syn-volcanic tectonism

A variety of features in the Jurassic volcanic sequence (upper Falla Formation, Prebble Formation and Kirkpatrick Basalt) of the Beardmore Glacier region provide evidence for the tectonic setting during deposition, and complement results (Wilson in press a, b), from structural studies on mesoscopic faults and Ferrar dykes, that indicate faulting was coeval with dyke emplacement. Wilson (in press a, b) has identified normal fault arrays orientated NE and NW, and dyke arrays with NNW-NNE and E N E - E W trends. The pattern has been interpreted as a triaxial strain field with vertical shortening and two directions of horizontal extension. At a number of localities, the lower member of the volcaniclastic upper part of the Falla Formation includes arkosic grits (very coarse sand to granule size) containing angular to subrounded grains of quartz, plagioclase and Kfeldspar (microcline and orthoclase), and concentrations of coarse sand-sized garnet. Some angular K-feldspar clasts attain 2 cm in length. The grain size and sediment immaturity require a proximal igneous and metamorphic basement source. Large angular feldspars are absent from the lower Falla Formation and older parts of the Victoria Group; sandstones in these rocks are dominantly medium to fine grained although locally containing pebbly beds with rounded clasts mostly of vein quartz and silicic volcanic

JURASSIC MAGMATISM& TECTONISM: ANTARCTICA rocks (Barrett et al. 1986). A basement source component has been recognized for some of these sandstones. The Victoria Group is at least 1400 m thick along the cratonic flank of the basin, suggesting that the sequence probably extended onto the craton for a considerable distance (more than 100 km). The abrupt influx of arkosic grits with large angular K-feldspar clasts argues for a significant change in the transport distance from the source region and in the rate of source region erosion. Topographic relief must have been generated, exposing basement rocks to denudation; the most likely cause was normal faulting. Normal faulting involving basement rocks is also indicted by monoclinal flexures sub-parallel to the NNW-NNE dyke array identified by Wilson (in press a, b). The emplacement of basaltic breccia, identical to that in the Prebble Formation, in the hinge line of one monocline demonstrates an Early to Mid-Jurassic age for the flexuring (after deposition of unit 2 of the upper Falla Formation and before eruption of the Kirkpatrick Basalt). Monoclinal flexuring also affected Kirkpatrick Basalt flows in the southern Marshall Mountains but it can be dated only as younger than the lowest three flows. The thickness of many flows, up to 175 m for the lowest flow at Mt Bumstead, suggests ponding of the lavas and hence topography, and the capping flow, at least 75 m thick, indicates that topography was present throughout the time of lava eruption. At Allan Hills in south Victoria Land (Figs 1 & 2) the Mawson Formation (Borns & Hall 1969; Ballance & Watter 1971), the Prebble Formation equivalent, was deposited on an erosion surface cut 500 m down into the underlying Triassic and Permian strata; this unconformity points to vertical movements following termination of fluvial deposition. The Mawson Formation and the Exposure Hill Formation in north Victoria Land (Elliot et al. 1986a) both contain evidence for phreatomagmatic activity and a significant role for water. The Kirkpatrick Basalt in north Victoria Land includes several thick flows (> 100 In), indicating ponding by topography. An episode of silicic volcanism preceding the basaltic activity is known in south Victoria Land only from megaclasts in the basaltic breccias (Bradshaw 1987). In contrast, in north Victoria Land the Upper Triassic Section Peak Formation (Collinson et al. 1986) contains silicic glass shards as does the matrix of some of the coarse pyroclastic beds of the Exposure Hill Formation (Elliot et al. 1986a). The onset of silicic volcanism is poorly dated but the later stages were contemporaneous with Ferrar extrusive activity.

169

The evidence for extension from the fault and dyke arrays and the basement block faulting, the presence of topography, and the occurrence of water during eruption of the basaltic rocks, are all consistent with a volcano-tectonic rift. The orientation of extensional structures in the central Transantarctic Mountains can be inferred from the fault and dyke arrays (Wilson in press a, b) and the monoclinal flexures that affect the volcanic strata. The monoclinal flexures are orientated N to NNE, similar to part of the dyke array. The structural studies have shown two directions of extension, perpendicular and parallel to the trend of the mountains. On regional grounds, the principal extension direction is inferred to be perpendicular to the range as a whole. No unequivocal major faults of Jurassic age parallel to the range that might be principal bounding faults have been identified yet, although range-parallel faults that can be dated only as post-Beacon are known from a number of localities in the Transantarctic Mountains (Barrett 1965; Barrett & McKelvey 1981; Bentley & Clough 1972).

Geochemistry of the Ferrar tholeiites The striking isotopic and geochemical signature of the Ferrar tholeiites has been known for more than two decades (Compston et al. 1968). Since that time the contrasting signatures of the Mesozoic tholeiites in Dronning Maud Land (Fig. 1) have been established (Faure & Elliot 1971; Faure etal. 1979; Furnes etal. 1987; Harris et al. 1990) and Brewer (1989) has proposed that the boundary of the Ferrar geochemical province lies in the Theron Mountains region. The Dronning Maud Land tholeiites show close similarities to the Karoo tholeiites. Cox et al. (1967) recognized two geochemical provinces in the Karoo based on the relative abundances of potassium and high field "strength elements. Although lacking the distinction in K, the similar division in the Cretaceous Paran~i basalts (Bellieni et al. 1984) led Cox (1988) to propose subdivision of the Gondwana tholeiites as a whole into a low-Ti and a high-Ti province. Ferrar tholeiites are relatively high in SiO2 and low in TiO2 (Fig. 3) and P205; in those elements as well as certain trace elements (e.g., depletion in Nb and Sr), they have a distinct calealkaline aspect. Within the lavas, two compositional types have been recognized and have been referred to (Siders & Elliot 1985) as the low-Ti and high-Ti flows; these low-Ti and high-Ti compositions are now designated the Mount Fazio and Scarab Peak chemical types respectively (Fleming et al. in press b) in order to avoid con-

170

D.H. ELLIOT

fusion with the terminology applied to the Gondwana province as a whole and to which petrogenetic significance is attached. The Scarab Peak and Mt. Fazio chemical types both belong to the Gondwana low-Ti province. No basalt flow is known that has more than 7.2% MgO (greater MgO contents are found in cumulate zones of thick flows and are accompanied by relatively high Cr and Ni). Most flows in north Victoria Land range between 6 and 7% MgO (Siders & Elliot 1985), whereas those in south Victoria Land range between 4 and 7% MgO (Kyle et al. 1983). In the Beardmore Glacier region most flows are rather evolved with MgO between 2.3 and 4% and only a few have greater MgO (up to 7.2%, Faure et al. 1974). The capping flow here and in north Victoria Land (Fig. 3) has an iron-rich evolved composition 4.0 %TiO 2 3.0

46

50

54 %SiO 2

(EgO T ~ 15%). The sills include a number that are olivine-bearing and the chilled margins have up to 10% MgO (Gunn 1966; Kyle 1980); trace element data for these olivine-bearing rocks are, however, very sparse. The Australian (Tasmania, Kangaroo Island, western Victoria) tholeiites have the same isotopic and geochemical characteristics (e.g. high initial 87Sr/grSr ratios) as the Ferrar tholeiites and form part of the Ferrar geochemical province (Hergt et al. 1989b, 1991). The transition zone tholeiites from the Theron Mountains (Brewer & Clarkson 1991) and Whichaway Nunataks (Brewer 1989) occupy two fields on an eNd-eSr diagram (Fig. 4), one overlapping with the Ferrar province and the other the Karoo Central Area. These rocks (Brewer 1989; Brewer & Brook 1991) include compositions similar to the Ferrar province (Fig. 3) in their low TiO2 contents, as well as compositions relatively enriched in high field strength elements (TiO2, P205); except for the Whichaway Nunataks, full details are not yet available. One sub-group of their high titanium group (HTG), although having a lower range of SiO2, is similar to the Scarab Peak chemical type of north Victoria Land. The high-Ti group from this region, however, does not belong to the Gondwana high-Ti province according to the criteria (Ti/Y, Zr/Y) of Erlank et al. (1988). Dronning Maud Land tholeiites (Harris et al. 1990, 1991a) show

58

Fig. 3. Plot of TiO2 versus SiO2to show subdivisions of the Antarctic Jurassic tholeiites. All the Antarctic tholeiites except some dykes from Dronning Maud Land belong to the Gondwana low-Ti province according to the criteria of Erlank et al. (1988). The average of the Tasmanian dolerites is included because those rocks belong to the Ferrar geochemical province which is defined by high initial STSr/a6Srratios. The Ferrar province is represented by fields 4, 5 and 6, and some of the Theron Mountains and Whichaway Nunataks samples. Data source: 1, Theron Mountains high-Ti group (Brewer & Brook 1991); 2, Dronning Maud Land (Harris et al. 1990, 1991a; Brewer & Brook 1991); 3, Theron Mountains low-Ti group (Brewer & Brook 1991); 4, Central Transantarctic Mountains (Faure et al. 1974); 5, Mount Fazio chemical type, north Victoria Land (Siders & Elliot); 6, north Victoria Land and central Transantartic Mountains evolved iron-rich lavas (Scarab Peak chemical type of north Victoria Land) (Faure et al. 1974; Siders & Elliot 1985; Fleming et al. in press b); dots, Whichaway Nunataks dolerites (Brewer 1989); T, average Tasmanian dolerite (Hergt et al. 1989b).

0

S. Lebombo (1) / Kirwan Antarctica (2) ,"N" / ", ~l~// Ka'roo Central A r e a ( 1 ) EBAUR~~__X

~--.

/ - T .h. . . .

Antarctica (3)

eNd

-5 (low TI) 161 ~. . . . . I

-10 -50

0

l

='--~~-----~ i

+50

+100

+150

+200

eSr

Fig. 4. Plot of eNd versus eSr for rocks assigned to the Gondwana low-Ti province (Cox 1988) to show the variable isotopic characteristics, of the province as a whole but the similarity between the Ferrar province and the Paran~l-Etendeka province. The fields for the Antarctic tholeiites and Tasmanian dolerites are stippled. Data sources: 1, Hawkesworth et al. (1984); 2, Harris et al. (1991a); 3, Brewer & Clarkson (1991); 4, Fleming et al. (in press b), Hergt et aL (1989a), Kyle et al. (1987), Pankhurst et al. (1986); 5: Hergt et al. (1989b). 6, Hawkesworth et al. (1986). Although Jurassic magmatism is the focus of this paper, the Paran~i and Etendeka provinces are an integral part of the Gondwana low-Ti province and are included for completeness.

JURASSIC MAGMATISM & TECTONISM: ANTARCTICA clear similarities to the low-Ti province of the Karoo, in particular the Sabie River Formation in the case of the Kirwan Escarpment basalts. However, some dykes in the Ahlmann Ridge, Dronning Maud Land (Harris et al. 1991b) have affinities to the Karoo high-Ti province in their high Ti/Y and ZrfY ratios. The Ferrar geochemical province, the Theron Mountains and Whichaway Nunataks sills and dykes, and most of the Dronning Maud Land tholeiites belong to the Gondwana low-Ti province which stretches from Australia to southern Africa (Cox 1988; Harris et al 1990; Hergt et al. 1991) but also includes the Cretaceous tholeiites of the Etendeka and the southern part of the Paranfi. The Ferrar geochemical province is distinguished from the low-Ti province of the southern Karoo and Dronning Maud Land by two characteristics: at comparable MgO contents, higher minimum initial STSr/S7Sr ratios (Fig. 4) and greater relative depletion in certain high field strength elements (Fig. 5). Although differing in age, both the Paranfi low-Ti province and the Etendeka have a range of initial ratios of strontium that start at lower values than the Ferrar province, but extend to higher values. Furthermore, they do not show the depletions in TiOe and P205. The most mafic tholeiites in the Gondwana low-Ti province are not necessarily those with the lowest initial STSr/SrSrratios, and in the case of the Ferrar province it is the evolved 4.0

3.0

2.0

O 1.0 ,~ o ~ .~ •~ "~

171

Scarab Peak chemical type that has the lowest initial ratios (Mg Number ~ 24 and initial 87Sr/ 86Sr = 0.7083-7094; Fleming & Elliot 1988; Fleming et al. in press b). Ferrar tholeiites are more depleted (e.g. in Ti, P) than Lesotho and Dronning Maud Land basalt (Fig. 5). Olivinebearing sills (Gunn 1966; Kyle et al. 1983) show yet greater depletion in at least some of the high field strength elements (Ti, Zr). The high initial S7Sr/~r ratio ( _ 0.7085) of the Ferrar province as a whole has been used to argue against a significant component of crustal contamination by Kyle (1980) and Kyle et al. (1983). Hergt et al. (1989b) considered this further and used both isotopic and high field strength element data to provide a quantitative estimate of the amount of clay-rich terrigenous sediment incorporated through subduction pro• cesses. Further, although , 51 80 ns variable (range +5.5 to +9.3) and has been modelled by bulk assimilation and by AFC (Hoers et al. 1980; Mensing et al. 1984), mineral separates yield values of +5.5 to +6.0 which also have been used to argue against significant amounts of contamination (Menzies & Kyle 1990; Fleming et al. in press b). High 6180 values (> +6.0), at least for the capping flow of the north Victoria Land basalts, clearly reflect low temperature processes (Fleming et al. in press b). On the basis of Sr data, Hergt etal. (1989b) estimated that subduction of sediment into the subcontinental mantle occurred sometime between 900+200 Ma and 180 Ma. Neodymium model ages set an upper limit of 1.0-1.5 Ga for the time of enrichment (e.g. Storey et al. 1988; Brewer & Clarkson 1991).

Temporal and spatial relations of Jurassic magmatism in Gondwanaland

0.9 o.a 0.7 0.6 0.5 0.4 O.I 0.2

m

!

|

I

I

I

n

!

I

I

|

Rb

Be

K

Nb

Ce

Sr

p

Zr

1"1

Y

Ca"

Fig. 5. Element concentrations for basaltic rocks of the Gondwana low-Ti province normalized to a Lesotho Basalt (Marsh & Eales 1984). Other data: 1, Average Tasmanian dolerite chilled margin (Hergt et al. 1989b); 2, Basalt lava, Kirwan Escarpment, Dronning Maud Land (Harris et al. 1990); 3, Basalt lava, north Victoria Land (Siders & Elliot 1985); 4, Basalt lava, Beardmore Glacier region, Antarctica (T. H. Fleming & D. H. Elliot unpublished data). The choice of MgO concentration (c. 6.8%) for comparison is determined by the very limited range exhibited by the Tasmanian dolcrites.

Relations between magmatism and tectonism, within Antarctica and with the adjacent continents, in the initial rifting of the Antarctic sector of Gondwanaland can be assessed only by using reconstructions for the Jurassic (e.g. Lawver et al. 1985, 1991; Grunow et al. 1991). This topic has been considered by Storey & Alabaster (1991) for the Antarctic Peninsula and southern South America. In the following discussion, three time intervals are considered, based on the apparent principal episodes of magmatism in southern Africa (Fitch & Miller 1984), the inferred centre for a hotspot related to continental break-up (White & McK~nzie 1989). Although most of the rocks referred to in the following sections have been dated radiometrically, palaeontologically dated sequences are also discussed. Correlation of the

172

D.H. ELLIOT

two time scales has still to be resolved satisfactorily for the Jurassic, as shown by the discrepancies (as much as 10 Ma for the Oxfordian/Kimmeridgian boundary) between the l U G S scale (Cowie & Bassett 1989) and that of Harland et al. (1989). The Harland et al. (1989) timescale is used here because they discuss the most recent review of the age assignmentsof the M anomaly sequence (Kent & Gradstein 1986). E a r l y J u r a s s i c ( 1 9 0 - 1 9 5 M a , Fig. 6)

This time interval includes the first major peak in basaltic magmatism during which the Lesotho basalts were erupted and dolerite sills and dykes emplaced into the G o n d w a n a sequence in the Central Area of the Karoo province and the Sabie River Formation was erupted in the southern Lebombo belt (Fitch & Miller 1984). The Falkland Islands dolerites may have been emplaced at this time (a single K/Ar date of 192 Ma; Cingolani & Varela 1976), and according to

l/ill

South

190-195 Ida

It // ~Africa

"~

Falkland . . . . . ~,0/ / ~ 4 ~ . ~ _ ~ ~s "~-~.,~ EHsworth- / ' ~ W,itmor./~,~r~Bl°~ ~ ' ~ Ahh,enn Ridg, Kraul Mtn8.

~...

~\ 8 ~ - ' " " Theron Mtns. Whlchaway N. ~\~ East I Antarctica

•

Thurston Jones Mine.

?

Bounty

I.

-..,v/

~ ~,/Plaalataau _3.

~Campbell

km 1000

I,~'~ -"-'

Victoria

Land •

'~ ~ W. Victoria N • w Australia South Wales

Marine volcaniclastlc rocks • Rhyolltlo rocks Intermediate to eIItcic rocks Foltated granitolda

4" Granltolds V Basalt ~\~,i Oolerlte Subductlon

zone

palaeomagnetic data of Taylor & Shaw (1989) the Falkland Islands were located at that time adjacent to Dronning Maud Land and southern Africa. In adjacent Dronning Maud Land, Mesozoic basalts and dolerites have yielded a wide range of K/Ar dates. Dates older than 200 Ma have been obtained for lavas at Kraul Mountains (Vestfjella) (best estimate age of 205 Ma, F u r n e s & Mitchell 1978; see also next section). Some dykes at A h l m a n n Ridge have given slightly younger dates of 180-190 Ma (Wolmarans & Kent 1982), and there is some evidence for dolerite intrusion at about 190 Ma in the Theron Mountains (Brewer et al. 1991) and Whichaway Nunataks (Hofmann etal. 1980). Basalts in western Victoria, Australia, which include Ferrar province rocks have also yielded a date in this • range (190+10 Ma; McDougall & Wellman 1976). The silicic rocks in the Beardmore Glacier region lack reliable radiometric dates because of overprinting during the widespread zeolitization Fig. 6. Distribution of Lower Jurassic (about 190-195 Ma) igneous rocks in the South America-southern Africa-Antarctica-southeastern Australia sector of Gondwanaland. Reconstruction modified from that of Grunow et al. (1991) for 175 Ma, and of Lawyer et al. (1991, in press) for 160 Ma; the Lawyer et al. (1991) reconstruction eliminates the gaps in West Antarctica present in that of Grunow et al. (1991), but uses palaeomagnetic data (Grunow etal. 1987) to constrain the position of the Ellsworth-Whitmore mountains crustal block at 175 Ma. The Falkland Islands are placed according to the reconstruction of Taylor & Shaw (1989). The various crustal blocks of West Antarctica are assumed not to have changed positions significantlybetween the time under consideration (190-195 Ma) and 175 Ma. This may not be valid for the Ellsworth-Whitmore mountains crustal block which sometime in the Late Permian to Early Jurassic must be rotated c. 90° relative to its 175 Ma position to account for stratigraphic (Collinson et al. in press), palaeomagnetic (Grunow et al. 1987) and structural data (Dalziel et al. 1987). If deformation in the EIIsworth Mountains is related to deformation in the Cape Fold Belt (H~ilbich et al. 1983), the probable timing of the rotation was Late Triassic to Early Jurasic. The sparseness of the data for the inferred plate margin magmatic arc makes the continuityof the subduction zone uncertain; the subduction zone could be offset by transform faults, as suggested by Grunow et al. (1991) for the segment between Thurston Island and New Zealand. When all available data are taken into account, there is little doubt that a subductionrelated magmatic arc existed along the Pacific margin during the Jurassic. Published data for the three time intervals (c. 15 Ma each) under consideration, however, are very sparse. Hachures indicate the generalized distribution of rocks and the symbols the known or dated localities.

JURASSIC MAGMATISM & TECTONISM: ANTARCTICA associated with Ferrar magmatism. Their stratigraphic position indicates an age greater than c. 180 Ma (the age of the basalts) and younger than c. 208 Ma (the age of the Triassic/ Jurassic boundary and the inferred minimum age of the Upper Triassic plant-bearing beds of the Falla Formation). Changes in stratigraphic thickness of the plant-bearing lower part of the Falla Formation suggest the 15resence of a hiatus between the two parts of the formation (Elliot & Larsen in press). The volcanic beds span an uncertain interval of time; the presence of silicic shards in a few fine-grained beds in the Prebble Formation and in interbeds in the basalts suggests there might not be a significant time interval between the upper part of the Falla Formation and the Prebble Formation. A single granitic pluton in the Ellsworth-Whitmore mountains crustal block has yielded a slightly older errorchron date of c. 203 Ma, and differs from other granites in that region in having more calc-alkaline affinities (Pankhurst etal. 1991; see also next section). Early Jurassic magmatism is well documented in New Zealand by fore-arc volcaniclastic sequences of the Murihiku terrane (e.g. Bradshaw et al. 1981). Although the Murihiku terrane may have been accreted onto the plate margin by cloL sure of a marginal basin (Bradshaw 1989), it is not thought to have been a far-travelled allochthonous terrane (Grindley et al. 1981) and therefore it would reflect Gondwana plate margin processes. A Lower Jurassic granite (193 Ma, Wasserburg et al. 1963) has been reported from Bounty Island, and from Jones Mountains (197+4 Ma, Pankhurst 1990) in west Antarctica. The lack of exposure in the Antarctic sector (Campbell Plateau to the southern Antarctic Peninsula), due to subsidence below sea level or ice cover, may account for the apparent lack of subduction zone magmatism. A number of Lower Jurassic plutonic, metamorphic, and felsic extrusive rocks have been reported from restricted areas of Palmer Land (Meneilly et al. 1987; Hole et al. 1991; Piercy & Harrison 1991; Wever & Storey 1991), and provide support for an episode of both subduction-related, and bimodal back arc-related, magmatism. In South America, the batholith of Central Patagonia includes many high-silica peraluminous granitic plutons that are slightly older than 200 Ma (Pankhurst & Rapela 1991; Rapela et al. 1989) although younger dates (c. 195 Ma) have been reported (Rapela & Kay 1988). This batholith, or coeval intrusions, may extend southeast in the Deseado Massif region (Rapela & Kay 1988). Jurassic volcanic rocks of intermediate to silicic composition occur in this region of Patagonia

173

(Riccardi 1983; Llambias et al. 1984; Uliana & Biddle 1987), most commonly interbedded in marine sedimentary sequences deposited in extensional tectonic settings. Early Jurassic magmatism northward along the Andes is regarded as partly extension-related (Parada 1990).

Early-Mid-Jurassic (175-180 Ma, Fig. 7) Another peak in basaltic magmatism is postulated for southern Africa and was located in the Nuanetsi and north Lebombo regions (Fitch & Miller 1984); massive accumulation of rhyolitic rocks also occurred in the Lebombo belt. In addition, hypabyssal intrusions of this age are scattered through the Karoo Central Area. The

:k Marine volcanlclaatic rocks • Rhyolitic rocks

intermediate to eiNcic volcanic rocks ,'-, Folieted grenitolds "1" Granltoids

V ~,~ ~

Basalt

Dolarite Rift margins (inferred) 8ubduction zone ,e-- ..b Principal extension direction

-&-

S,T

Straumsvole. Tvora

Fig. 7. Distribution of lower Middle Jurassic (about 175-180 Ma) igneous rocks (for details, see caption to Fig. 6). The Falkland Islands are placed in their present position relative to South America; rotation and translation occurred sometime after about 190 Ma and before the opening of the South Atlantic in the Cretaceous. Only one rift margin is indicated for Antarctica because of uncertainties in the nature of the rift and the location of the other rift margin. Extension directions are taken from Wilson (in press a, b).

174

D.H. ELLIOT

adjacent part of Antarctica, Dronning Maud Land, has scattered outcrops of basalt of this general age, on the Kirwan Escarpment, Heimefront Range and Kraul Mountains (Vestfjella) (see Elliot et al. 1985 for a summary of pre-1983 radiometric dates). Recently Peters et al. (1991) have reported K/At plagioclase dates of 179+ 13 Ma and 189+ 10 Ma for lavas from the Kraul Mountains. The dykes and sills in these areas have given comparable dates (Peters et al. 1991; see also Elliot et al. 1985), as have basalt dykes in the Schirmacher Oasis area several hundred km to the east at 12°E (Kaiser & Wand 1985). Offshore a major dipping-reflector sequence, the Explora Wedge, has been identified geophysically (Hinz & Krause 1982); although its age is unknown, it is probably correlative with some part of the basalt and rhyolite sequence in the Lebombo Monocline. In addition to the basaltic rocks, syenite intrusions at Straumsvola and Tvora have given dates of 180+3 Ma (4°Ar/ 39Ar) and 170+4 Ma (Rb-Sr whole rock) (Grantham et al. 1988) and other syenites in this region may also be Jurassic. This Early-Mid-Jurassic time interval spans the principal episode of Ferrar magmatism. Best estimates of the age of magmatism (180+5 Ma; Elliot et al. 1985; Elliot & Foland 1986) are supported by 4°Ar/39Ar age determinations of 177+2 Ma on plagioclase separates from the capping lava in north Victoria Land (Fleming et al. in press a). 4°Ar/39Ar age determinations for dolerites from the Theron and Pensacola mountains (Brewer et al. 1991) support the inference of a major episode of magmatism. This is also the time of emplacement of the Dufek layered basic intrusion (Ford & Kistler 1980). The possibility of younger episodes of Ferrar magmatism, at 165 Ma as suggested by Kyle et al. (1981), cannot be excluded. Silicic activity continued at a low intensity throughout eruption of the flood basalts in the Beardmore Glacier region. The Tasmanian dolerites, basalts at Kangeroo Island and a tholeiitic dolerite in southeastern New South Wales also were emplaced during this time (respectively 175+& Ma, Schmidt & McDougall 1977; 170+10 Ma, McDougall & Wellman 1976). In geographical extent, this was the principal episode of tholeiitic magmatism associated with Gondwanaland rifting. Granitic rocks of this age (175-180 Ma) are well developed in the Ellsworth-Whitmore mountains crustal block (Millar & Pankhurst 1987; Pankhurst et al. 1991). These plUtons have S-type geochemical characteristics, and include one that is thought to possibly represent extreme differentiation of a Ferrar-like magma (Storey et al. 1988). In the northern part of Patagonia, late

members of the batholith have given a poorly constrained date of 172+15 Ma (Rapela et al. 1989). The Marifil Group of rhyolitic rocks (Llambias et al. 1984; Rapela & Kay 1988), erupted onto a mid-Palaeozoic igneous and metamorphic basement, are considered possibly coeval with these late granites (Rapela et al. 1989). To the south, silicic rocks assigned to the Chon Aike Group (Tobifera) form extensive outcrops but most are regarded as younger than this time interval (Gust et al. 1985; Riccardi 1983). As with the Early Jurassic interval, evidence for active plate margin processes in the Antarctic sector is scanty. The New Zealand record again lies in the fore-arc sequences of the Murihiku terrane (Bradshaw et al. 1981). There is no record from the Campbell Plateau and Marie Byrd Land, but a single silicic intrusion of this age (c. 180 Ma; R. J. Pankhurst in Grunow et al. 1991) crops out on Thurston Island. In northern Palmer Land, the Mt. Hill Formation which has an age of about 175 Ma (Meneilly et al. 1987; Thomson & Pankhurst 1983), is the earliest representative of extensive Jurassic and Cretaceous fore-arc and back-arc volcaniclastic sequences associated with the so-called Andean plate margin (Elliot 1983; Thomson et al. 1983). Other volcanic rocks and deformed granitoids have yielded ages in this time interval (Pankhurst 1983; Meneilly et al. 1987; Piercy & Harrison 1991), and on the east coast of the Antarctic Peninsula there are scattered plutonic and volcanic rocks and orthogneisses of similar age (Pankhurst 1982; Hole et al. 1991). Early Jurassic magmatic activity, with poorly constrained radiometric dates (Tanner & Rex 1979), occurred in South Georgia. In southern South America only toward the northern end of the Andean (Patagonian) Batholith, in the region of 40-42°S., have plutonic and volcanic rocks of this age been reported (Gonzfilez Diaz 1982; Hailer & Lapido 1982), although north of this area subduction-related magmatism is widespread (Riccardi 1983). L a t e M i d - J u r a s s i c ( 1 6 0 - 1 6 5 M a Fig. 8)

Tholeiitic magmatism of this age was confined to emplacement of hypabyssal intrusions in the Karoo Central Area (Fitch & Miller 1984). Extension-related silicic magmatism occurred only in Patagonia where it is represented by the voluminous Chon Aike rocks (Gust et al. 1985; Kay et al. 1989). The Chon Aike is found in outcrop on the Deseado Massif and to the north on other basement highs, but in much of Patagonia is present in subsurface, overlying the pre-Jurassic

JURASSIC MAGMATISM & TECTONISM: ANTARCTICA 160-165

Ma

'-':i:.".!!.:-~'::: :,::":

\

~ '

_7

~An_tarctic" "~"

~\ : ~

XThur~on,-,','~.

:

East

':~m~rT' I

Antarctica

"- -

)

? ?

..4

?

~-

Boardmore

I I 'larl~-?

?1

)g" .

r

l~'JZea~,.d

/

V

I

km tooo I ~~

!

I

Marine volcenlclastlc rocks

•

Rhyolltlc rocks slik31c volcaniclntermediaterockst°

V Basalt -il- Granitolds

e

"/-~

~.%~----..."

.'1 Tasmania

.// ~:~i.~,.:~ ,0" ~/.1,"I/ ~ EWB

175

1970, 1983), together with the intertonguing Mt Poster Formation (Laudon et al. 1983; Rowley et al. 1983) contain an extensive record of volcanism. The magmatic arc terrain includes a basaltic sill in northern Palmer Land (Rex 1972) and a number of plutons that form a belt along the east coast of the northern Antarctic Peninsula (Pankhurst 1982, 1990; Rex 1972). Coeval volcanic rocks from the northern Antarctic Peninsula are not known although probably present among the undated volcanic sequences associated with the arc. A magmatic arc along the Pacific margin of southernmost South America is recorded by widely separated andesites and dacites (Dalziel 1981; Ramos et al. 1982), by scattered plutons of the Andean (Patagonian) Batholith (Gonzt~lez Diaz 1982; Halpern 1973; Herv6 et al. 1981; Nelson et al. 1988; Weaver et al. 1990), and in volcaniclastic strata across Patagonia (Riccardi 1983). As in the Antarctic Peninsula, the principal activity occurred later in the Jurassic and Cretaceous (e.g. Pankhurst 1990; Parada 1990; Weaver et al. 1990).

"

.Australia .........

Oolerlte Rlft margins (inferred) Mid ocelm ridge system (Inferred) 8ubduction zone

F.Jlsworth-Whitmore block

Fig. 8. Distribution of upper Middle Jurassic (about 160-165 Ma) igneous rocks (for details, see caption to Fig. 6). The timing of the earliest stages of seafloor spreading between Africa and Antarctica is not well constrained and may be younger than the time interval considered here. The symbol for the mid-ocean ridge system is dashed where inferred.

basement. Radiometric dates for the Jurassic silicic rocks in Patagonia cluster around 155-165 Ma, but temporal relations to tectonism are difficult to evaluate; much of the silicic volcanism may be younger than the time interval considered here. Subduction-related magmatism, however, was much more widespread. The record in New Zealand again lies in the fore-arc sequences of the South Island, and between there and Thurston Island again there is a lack of data, mainly because of lack of outcrop. Volcanic rocks of this age (c. 165 Ma; R. J. Pankhurst in Grunow etal. 1991) crop out on Thurston Island. To the east, at the base of the Antarctic Peninsula, the oldest rocks in the Latady Formation, dated palaeontologically as Bajocian (Quilty

Discussion Field relations of the Jurassic volcanic rocks in the central Transantarctic Mountains argue for basement block faulting to create monoclinal flexures, provide a proximal source for coarse arkosic grits, and generate topography to contain thick ponded lavas. These features, together with structural data on dykes and faults (Wilson in press a, b), suggest eruption and deposition in a volcano-tectonic rift system. Although evidence exists for as much as 500 m of relief locally before onset of basaltic magmatism, absolute movement cannot be determined with assurance. Normal faulting and extension before flood basalt magmatism suggests passive lithospheric stretching (White & McKenzie 1989). The stretching factor is difficult to assess because younger episodes of rifting have affected West Antarctica. The maximum was 2, based on the crustal thickness of West Antarctica and assuming an initial thickness of 40 kin. It may well have been less in that, at present, the amount of extension associated with each of the various episodes of Late Mesozoic and Cenozoic tectonism is unknown. If major extension occurred at break-up (175-180 Ma), it was offset toward the plate margin and the Transantarctic Mountains region at that time would have formed one flank of any major rift system. The lack of data from West Antarctica precludes assessing whether any, or how much, subsidence or uplift occurred in association with the possible princi-

176

D.H. ELLIOT

pal riffs. Middle Jurassic marine sequences at the base of the seismically-identified grabens in the Ross Sea region, if found in the future, would argue for normal temperature mantle at the time of rifting. The linearity of the Ferrar province led Cox (1988) to propose a 'hot-line' rather than a hot spot to account for the Gondwana Jurassic tholeiitic magmatism. Jurassic structures along the. Transantarctic Mountains (Spaeth & Schull 1987; Wilson in press a, b) indicate extension directions perpendicular to the range, however the orientation of the range varies from Dronning Maud Land to north Victoria Land. The pattern suggests rifts that intersected possibly with a node in the Dufek Massif region where a layered basic intrusion was emplaced (Ford 1976; Ford & Kistler 1980; Fig. 9). A third arm at that node

could have intersected the Gondwana plate margin or passed between the Ellsworth-Whitmore mountains crustal block and the Antarctic Peninsula. Cox's (1988) hot-line may be a series of two or more linked centres of magmatic activity, only one of which, in southeastern Africa, was plume-related. Other centres, in addition to the Dufek Massif area, might be represented by the Beardmore Glacier region and south Victoria Land where tholeiitic rocks have stratigraphic thicknesses approaching 2 km. Zones of weakened lithosphere may have propagated away from the nodes and provided pathways for lateral migration of magmas derived from lithospheric sources. The length of the Ferrar province (c. 4000 km) suggests lithospheric control on the magmatism and tectonism which affected EW

Explora Wedge EIIsworthEWB Whitmore Block L Lebombo MBL Marie Byrd Land NZ New Zealand TAM Transantarctic Mtn8 Sri MinimumSr isotope initial ratio

F Paral~!

South America

~,

Africa Nuanetsl

Srt > 0.707 ~ -

~nc~,ai

4- .~ extenmon

~:

Q

direction

Dronm~

-I-F- Gondwanian

Maud Land

tectonic trends .:. "-:"• Zones of extension Subductlon zone Magmatic arc HTP High-Ti i~'ovinca'~ LTP Low-Ti province UJ Upper Juraa8tc sili¢ic rock1 MJ Middle Jurassic silicic rocks LJ Lower Jurassic blmodal rocks

Lower Antarctic Peninsula

Basaltic Silicic rocks

Antarctica

rock8

Tertiary ~===1 ~

Cretaceous

I]]]T~ ~

Jurassic

r ~

~fCampbell'~" ~ ~ f ' ~ '

~

(; / J Australia "Kangaroo I. W. Victorias.E. New ...-South Wales

Fig. 9. Reconstruction of Oondwanaland for the Early to Middle Jurassic to show the spatial relations of the lowTi...andsThigh'Ti basalt, provinces, the subprovinces within the iow-Tl"province" that are based on the minimum mmal Sr/S6Srrauos, the within-plate silicic provinces, and the magmatic arc along the plate margin., The orientation of the Gondwanian fold and thrust belt in the Ellsworth-Whitmore mountains crustal block indicates that 90° of rotation occurred between about 210 and 175 Ma. Low-Ti and high-Ti province boundaries from Cox (1988). See caption to Fig. 6 for comments on the inferred subduction zone. The rift basin off Dronning Maud Land identified by Kristoffersen & Hinz (1991) occurs within the southern part of the area indicated for the Explora Wedge. Extension directions are taken from Wilson (in press a, b).

JURASSIC MAGMATISM & TECTONISM: ANTARCTICA the retro-arc region of the inferred Gondwana active plate margin. The incompatible element geochemistry places the Ferrar tholeiites in the Gondwana low-Ti province. The Ferrar province, however, is distinguished by the highest initial 87Sr/a6Sr ratios (-> 0.7085), in marked contrast to the lowTi province rocks in Dronning Maud Land and southern Africa which have lower minimum ratios (Kirwan Escarpment, Dronning Maud Land: 0.7049; Karoo central area: 0.7047; and south Lebombo belt: 0.7038). It differs similarly from the Cretaceous Paran~i and Etendeka lowTi province rocks (minimum initial ratios - 0 . 7 0 7 ) . The source region itself may have been shallow-level lithospheric mantle, as proposed by Sweeney et al. (1991) for the voluminous low MgO and low-Ti flood basalts of the Karoo. There is no sign in the Ferrar province of the high-Ti rocks, as defined by Erlank et al. (1988), that Sweeney et al. (1991) postulated as precursors to Karoo low-Ti tholeiites. The lithosphere is considered (Sweeney et al. 1991) to have been refractory or residual and not to have suffered metasomatism or addition of fertile asthenosphere. However, the geochemistry of the low-Ti province indicates that the source region acquired crustal ratios of incompatible elements as well as a range of elevated Sr isotope ratios. The Ferrar has the lowest Sr abundances but the highest initial aTSr/86Srratios, reflecting a high isotope ratio component added to markedly Sr-depleted lithosphere. The possible belt of low-Ti magmatism north of the high-Ti province in the Karoo (Cox 1988) suggests that other processes, in addition to sediment subduction (Hergt et al. 1989b), may have had a role in determining the geochemical characteristics. The major part of the Ferrar Province was erupted through crust involved in Early Palaeozoic plate margin processes. Basement rocks forming the Shackleton Range and Littlewood Nunataks are Proterozoic and thus the Theron Mountains and Whichaway Nunataks crust is probably the same age. No Archean crust is known in this region, although Archean precursors have been inferred for some Proterozoic rocks in the Shackleton Range (Pankhurst et al. 1983). The boundary of the Ferrar province appears to coincide with the transition to Proterozoic crust. The spatial relations of the Ferrar province to the Gondwana plate margin differ from those of Dronning Maud Land and the Karoo; that sector of the low-Ti province was separated from the plate by an additional 1000 to 2000 km and by a zone of within-plate silicic magmatism.

177

Gondwanaland reconstruction for 175 Ma places the Ellsworth-Whitmore mountains crustal block, part of that within-plate silicic province, off the Theron Mountains. The Ellsworth-Whitmore mountains crustal block has been linked geochemically (Storey et al. 1988) to Ferrar-like sources, however the Pirrit Hills granite, the possible differentiate of a Ferrar-like magma in that region, has initial STSr/S6Sr ratios (0.7070__.16) lower than those of the majority of the Ferrar province although not incompatible with some the Theron Mountains dolerites. The Ellsworth-Whitmore mountains crustal block is more appropriately linked geochemically and spatially to the African sector of t h e low-Ti Gondwana province. Age assignments for the various parts of the flood basalt provinces, within-plate silieic provinces and plate margin subduction provinces provide only a broad overview of temporal relations. Apart from southern Africa, evidence for tholeiitic activity in the Early Jurassic (c. 190195 Ma) is poor; the geographically most extensive episode was Early-Mid-Jurassic (c. 175-180 Ma). Silicic magmatism predated tholeiitic activity in the Ferrar province but its former extent is unknown. The possibility cannot be excluded that this silicic activity reflects anatexis related to the early stages of basaltic magmatism associated with break-up. The principal extensionrelated silicic province was confined to the South Atlantic sector (Bruhn et al. 1978; Gust et al. 1985; Storey & Alabaster 1991) and occurred within what had been the widest belt of Late Palaeozoic-Early Mesozoic Gondwana plate margin magmatism and deformation. The age and duration of silicic volcanism in the Patagonian region is not well constrained which makes relations between tectonism and magmatism uncertain. Apart from the Ellsworth-Whitmore mountains crustal block, silicic magmatism postdated the principal episode of magmatism associated with Africa-Antarctica rifting but was related to rifting that preceded opening of the South Atlantic. The rifting between Africa and Antartctica (between East and West Gondwanaland) possibly propagated from the Tethys toward the Gondwanaland plate margin, the earliest marine incursion in East Africa being Early Jurassic (Dingle 1978; Cannon etal. 1981). Rifting may have followed or reactivated older fault-bounded basins filled by Gondwana sedimentary sequences (Reeves et al. 1987). Break-up reflects the integrated stresses on the Gondwana plate; intracontinental rifting, with which Karoo magmatism is associated, is one aspect and may have operated independently

178

D.H. ELLIOT

from, and in addition to, plate boundary processes that have been argued to be important (Cox 1978; Storey & Alabaster 1991).

Conclusions The Jurassic history of the Transantarctic Mountains indicates that basaltic magmatism was preceded and accompanied by extension and vertical movements. This supports the passive stretching and upwelling model, in the case of the Ferrar province, for continental flood basalt magmatism. The upper crustal rift zone associated with Ferrar magmatism may have been offset from the principal region of extension in West Antarctica. Silicic magmatism was an integral part of the break-up process but its relationship to tholeiitic activity varied in scale, space and time. Silicic volcanism preceded, and at low intensity accompanied, Ferrar magmatism, whereas in the Lebombo-Nuanetsi region silicic extrusive activity on a massive scale accompanied basaltic magmatism. Similarly in the Patagonian sector of Gondwanaland the scale was much greater but preceded the spatially far-removed basaltic magmatism associated with the opening of the South Atlantic. The Gondwana low-Ti province, identified on high field strength element abundances (Cox 1988) and ratios (Erlank et al. 1988), has a range of strontium and neodymium isotope characteristics. The Ferrar province shows the largest high field strength element depletions but the most isotopic enrichment in strontium relative to other parts of the low-Ti province. The Ferrar province reflects melting of more refractory and depleted lithospheric mantle than elsewhere in the low-Ti province, but mantle that had the strongest crustal imprint. Improved dating of magmatic episodes is required before details of the temporal and spatial relations between magmatism and tectonism can be established and models evaluated. Preparation of this paper has beefl supported by NSF Grant DPP 8917348. Discussions with T. H. Fleming on the Ferrar Group, its chemistry, tectonic setting and relations to other Gondwana tholeiites, and with T. J. Wilson on the structural evolution of the Transantarctic Mountains are particularly acknowledged. Constructive criticism by the reviewers, P. J. Barrett, L. A. Lawyer and R. J. Pankhurst, and the editor, B. C. Storey, has greatly improved the manuscript.

References BALLANCE,P. F. & WATTERS,W. A. 1971. The Mawson Diamictite and the Carapace Sandstone, for-

mations of the Ferrar Group at Allan Hills and Carapace Nunatak, Victoria Land, Antarctica. New Zealand Journal of Geology and Geophysics, 14, 512-527. BARRETT, P. J. 1965. Geology of the area between the Axel Heiberg and Shackleton Glaciers, Queen Maud Range, Antarctica; 2. Beacon Group. New Zealand Journal of Geology and Geophysics, 8, 344-470. & KOH~, B. P. 1975. Changing sediment transport directions from Devonian to Triassic in the Beacon Super-Group of South Victoria Land, Antarctica. In: CAMPBELL,K. S. W. (ed.) Gondwana Geology. Australian National University Press, Canberra, 15-35. & MCKELVEY,B. C. 1981. Cenozoic glacial and tectonic history of the Transantarctic Mountains in the McMurdo Sound region: recent progress from drilling and related studies. Polar Record, 20, 543-548. ~, ELLIOT, D. H. & LINDSAY, J. F. 1986. The Beacon Supergroup (Devonian-Triassic) and Ferrar Group (Jurassic) in the Beardmore Glacier area, Antarctica. In." TURNER, M. D. SPLETTSTOESSER,J. F. (eds). Geology of the Central Transantarctic Mountains. American Geophysical Union, Antarctic Research Series, 36, 339-428. BEHRENDT, J. C. & COOPER, A. K. 1991. Evidence of rapid Cenozoic uplift of the shoulder escarpment of the Cenozoic West Antarctic rift system and a speculation on possible climate forcing. Geology, 19, 315-319. , LEMASURIER,W. E., COOPER,A. K. TESSENSOHN, F., TREHU, A. & DAMASKE,D. 1991. The West Antarctic rift system: a review of geophysical investigations. In: ELLIOT, D. H. (ed.) Contributions to Antarctic Research II. American Geophysical Union, Antarctic Research Series, 53, 67-112. BELLIENI, G., BROTZU, P., COMIN-CHIARAMONTI,P., ERNESTO, M., MELFI, A., PACCA, I. G. & PICClRILLO, E. M. 1984. Flood basalt to rhyolite suites in the southern Parana Plateau (Brazil): paleomagnetism, petrogenesis and geodynamic implications. Journal of Petrology, 25, 579-618. BENTLEY,C. R. & CLOUGH,J. W. 1972. Antartic subglacial structure from seismic refraction measurements. In: ADIE, R. J. (ed.) Antarctic Geology and Geophysics. Universitetsforlaget, Oslo, 683691. BERG, J. H. 1991. Geology, petrology and tectonic implications of crustal xenoliths in Cenozoic volcanic rocks of southern Victoria Land. In: THOMSON, M. R. A., CRAME,J. A. & THOMSON,J. W. (eds). Geological Evolution of Antarctica. Cambridge University Press, Cambridge, 311-315. BORG, S. G. & DEPAOLO,D. J. 1991. A tectonic model of the Antarctic Gondwana margin with implications for southeastern Australia: isotopic and geochemical evidence. Tectonophysics, 196, 339-358. BORNS,H. W., Jr. & HALL,B. A. 1969. Mawson tillite in Antarctica: A preliminary report of a volcanic deposit of Jurassic age. Science, 166, 870-872.

JURASSIC MAGMATISM & TECTONISM: ANTARCTICA

BRADSHAW, J. D. 1989. Cretaceous geotectonic patterns in the New Zealand region. Tectonics, 8, 803-820. , ADAMS, C. J. & ANDREWS, P. B. 1981. Carboniferous to Cretaceous on the Pacific margin of Gondwana: the Rangitata phase of New Zealand. In: CRESSWELL,M. M. & VELLA, P. ( e d s ) Gondwana Five. Balkema, Rotterdam, 217-221. BRADSHAW, M. A. 1987. Additional field interpretation of the Jurassic sequence at Carapace Nunatak and Coombs Hills, south Victoria Land, Antarctica. New Zealand Journal of Geology and Geophysics, 30, 37-49. BREWER, T. S. 1989. Mesozoic dolerites from Whichaway Nunataks. Antarctic Science, 1,151-155. & BROOK, D. 1991. The geochemistry of Mesozoic tholeiites from Coats Land and Dronning Maud Land. In: THOMSON, M. R. A., CRAME, J. W. & THOMSON,J. W. (eds) Geological Evolution of Antarctica. Cambridge University Press, Cambridge, 569-572. & CLARKSOrq,P. D. 1991. Mesozoic magmatism in greater Antarctica: implications for Precambdan plate tectonics. In: THOMSON, M. R. A., CRAME, J. W. & THOMSON,J. W. (eds) Geological Evolution of Antarctica. Cambridge University Press, Cambridge, 117-121. , REX, D., GUISE, P. & STOREY, B. 1991. Ar40/ Ar39 age determinantions from the Theron and Pensacola Mountains, Antarctica: implications for the age of Mesozoic magmatism in Antarctica. In: Programme and Abstracts, Magmatism and the Causes of Continental Break-up. Geological Society of London, 6-7. BRUHN, R. L., STERN, C. R. & DE WIT, M. J. 1978. Field and geochemical data bearing on the development of a Mesozoic volcano-tectonic rift zone and back-arc basin in southernmost South America. Earth and Planetary Science Letters, 41, 32-46. CAMPBELL, I. H. & GRIFFITI-IS,R. W. 1990. Implications of mantle plume structure for the evolution of flood basalts. Earth and Planetary Science Letters, 99, 79-93. CANNON,R. T., SIMIYU-SIAMBI,W. M. N. & KARANJA, F. M. 1981. The proto-Indian Ocean and a probable Paleozoic/Mesozoic triradial rift system in East Africa. Earth and Planetary Science Letters, 52,419-426. CINGOLANI,C. A. & VARELA,R. 1976. lnvestigaciones geol6gicas y geochronol6gicas en el extremo sur de la Isla Gran Malvina, sector de Cabo Belgrano (Cabo Meredith), Islas Malvinas. In: Actas 6° Congreso Geol6gico Argentino, Buenos Aires, 1, 457-473. COLLINSON,J. W. 1991. The palaeo-Pacific margin as seen from East Antarctica. In: THOMSON, M. R. A., CRAME, J. W. & THOMSON, J. W. (eds). Geological Evolution of Antarctica. Cambridge University Press, Cambridge, 199-204. - - , PENNINGTON,D. C. & KEMP, N. R. 1986. Stratigraphy and petrology of Permian and Triassic fluvial deposits in northern Victoria Land, Antarctica. In: STUMP, E. (ed.) Geological Investiga-

179

tions in Northern Victoria Land. American Geophysical Union Antarctic Research Series, 46, 211-242. , VAVRA, C. L. & ZAWISKIE, J. M. in press. Sedimentology of the Polarstar Formation, Permian, Ellsworth Mountains, Antarctica. In: WEBERS, G. F., CRADDOCK,C. & SPLETI'STOESSER, J. F. (eds). Geology and Paleontology of the Ellsworth Mountains, Antarctica. Geological Society of America Memoir. COMPSTON, W., McDOUGALL,I. & HEIER, K. S. 1968. Geochemical comparison of the Mesozoic basaltic rocks of Antarctica, South Africa, South America and Tasmania. Geochimica et Cosmochimica Acta, 32, 129-149. COOPER,A. K., DAVEY,F. J. & BEHRENDT,J. C. 1987. Seismic stratigraphy and structure of the Victoria Land Basins, western Ross Sea, Antartica. In: COOPER, A. K. & DAVEY,F. J. (eds) The Antarctic Continental Margin: Geology and Geophysics of the Western Ross Sea. Circum-Pacific Council for Energy and Mineral Resources, Earth Science Series, 5B, 27-76. Cowm, J. W. & BASSEYr, M. G. 1989. International Union of Geological Sciences 1989 global stratigraphic chart. Episodes, 12, (Supplement). Cox, K. G. 1978. Flood basalts, subduction, and the break-up of Gondwanaland. Nature, 274, 47-49. 1988. The Karoo Province. In: MACDOUGALL,J. D. (ed.) Continental Flood Basalts. Kluwer, Boston, 239-271. , MACDONALD, R. & HORNUNG, G. 1967. Geochemical and Petrographic provinces in the Karoo basalts of southern Africa. American Mineralogist, 52, 1451-1474. DALZIEL, I. W. D. 1981. Back-arc extension in the southern Andes: a review and critical reappraisal. Philosphical Transactions of the Royal Society of London, A300, 319-335. , GARRETT, S. W., GRUNOW,A. M., PANKHURST, R. J., STOREY,B. C. & VENNUM,W. R. 1987. The Ellsworth-Whitmore Mountains crustal block: its role in the tectonic evolution West Antarctica. In: MCKENZIE, G. D. (6d.) Gondwana Six: Structure, Tectonics, and Geophysics. American Geophysical Union, Geophysical Monograph, 40, 173-.182. DINGLE, R. V. 1978. South Africa. In: MOULLADE,M. & NAIRN, A. E. M. (eds). The Phanerozoic Geology of the World H. The Mesozoic, A. Elsevier, Amsterdam, 401-434. DREWRY, D. J. 1972. Subglacial morphology between the Transantarctic Mountains and the South Pole. In: ADIE, R. J. (ed.) Antarctic Geology and Geophysics. Universitetsforlaget, Oslo, 693 -703. DUNCAN, R. A. & RICHARDS,M. A. 1991. Hotspots, mantle plumes, flood basalts, and true polar wander. Reviews of Geophysics, 29, 31-50. ELLIOT,O. H. 1974. The tectonic setting of the Jurassic Ferrar Group, Antarctica. In: GONZALEZ-FERRAN, O. (ed.) Proceedings of the Symposium on Andean and Antarctic Volcanology Problems. International Association of Volcanology and

180

D.H. ELLIOT

Chemistry of the Earth's Interior, Special Series, 357-372. 1983. The mid-Mesozoic to mid-Cenozoic active plate margin of the Antarctic Peninsula. In: OLIVER, R. L., JAMES,P. S. & JAGO, J. B. (eds). Atttarctic Earth Science. Australian Academy of Science, Canberra, 347-351. & FOLAm),K. A. 1986. K-Ar age determinations of the Kirkpatrick Basalt, Mesa Range. In: STUMP, E. (ed.) Geological Investigations in Northern Victoria Land. American Geophysical Union, Antarctic Research Series, 46, 279-288. .... & LARSON, D. in press. Mesozoic volcanism in the Transantarctic Mountains depositional environment and tectonic setting. In: Gondwana 8: Assembly, Evolution and Dispersal. A. A. Balkema, Rotterdam. , BIGHAM,J. & JONES, F. S. 1991. Interbeds and weathering profiles in the Jurassic basalt sequence, Beardmore Glacier region, Antarctica. In: ULEmCH, H. & ROCH^ CA~OS, A. C. (eds). Gondwana Seven Proceedings. Instituto de Geoci~ncias-USP, Sao Paulo, 289-301. , FLECK, R. J. & StrrmR, J. F. 1985. Potassiumargon age determinations of Ferrar Group rocks, central Transantarctic Mountains. In: TtmSER, M. D. & SPLETrSTOESSER,J. F. (eds). Geology of the Central Transantarctic Mountains. American Geophysical Union, Antarctic Research Series, 36, 197-224. , HAn~, M. A. & SIOERS,M. A. 1986a. The Exposure Hill Formation, Mesa Range. In: STUMP, E. (ed.) Geological Investigations in Northern Victoria Land. American Geophysical Union, Antarctic Research Series, 46, 267-278. , StOERS, M. A. & HAnAN, M. A. 1986b. Jurassic tholeiites in the region of the upper Rennick Glacier, north Victoria Land. In: STUMP,E. (ed.) Geological Investigations in Northern Victoria Land. American Geophysical Union, Antarctic Research Series, 46, 249-265. ERLANK, A. J., DUNCAN, A. R. MARSH, J. S., SWEENEY, R. J., HAWKESWORTH,C. J., MILNER, S. C., MILLER,R. McG.& ROOERS,N. W. 1988. A laterally extensive geochemical discontinuity in the subcontinental Gondwana lithosphere. In: Geochemical Evolution of the Continental Crust, Conference Abstracts, Brazil, 1-10. FAURE,G. & ELLIOT,D. H. 1971. Isotope composition of strontium in Mesozoic basalt and dolerite from Dronning Maud Land. British Antarctic Survey BuUetin,2$, 23-27. , BOWUA~,J. R. & ELLIOT,D. H. 1979. The initial STSr#Sr ratios of the Kirwan Volcanics of Dronning Maud Land: comparison with the Kirkpatrick Basalt, Transantarctic Mountains. Chemical Geology, 26, 77-90. , , & JONES, L. M. 1974. Strontium isotope composition and petrogenesis of the Kirkpatrick Basalt, Queen Alexandra Range, Antarctica. Contributions to Mineralogy and Petrology, 43, 153-169. l~rCH, F. J. & MILLER,J. A. 1984. Dating Karoo igneous rocks by the conventional K/At and 4°Ar/39Ar

spectrum methods. In: Era.Am 0.709). All of these geochemical features indicate a major contribution from the continental mantle lithosphere in the generation of these magmas. In contrast, the Dronning Maud Land magmatism has elevated trace element ratios and eNd values (Ti/Y 250-600; Zr/Y 3.0-9.0; eNd - 2 to +3) and lower initial STSr/86Sr ratios (< 0.707) relative to the Ferrar Magmatic Province. The trace element and isotopic correlations suggest that these magmas were derived by the mixing of an OIB like asthenospheric component with a continental lithosphere component. The transition between these two geochemical provinces is located in Coats Land. In Coats Land, the Mesozoic tholeiitic magmatism is represented by doleritic sills and minor dykes which intrude Permo-Triassic sedimentary rocks. The dolerites can be subdivided into two series based on their TiO2 contents. Series 1 dolerites (TiO2 < 1.5%) can be further subdivided into three groups, which give Ar/Ar ages of 171+6 Ma (Group 1) and 193+7 Ma (Groups 2 and 3). It is only Group 2 magmas which have trace element and isotopic signatures akin to the Ferrar Magmatic Province. Group 1 dolerites have geochemical signatures which are transitional between the Ferrar Magmatic Province and Dronning Maud Land magma types. The Ferrar Magmatic Province signature in Coats Land is confined to the early magmatic episode (193+7 Ma) and this appears to mark the initiation of rift related magmatism in this region. It is argued that extension was limited and that most of the melt was derived from the continental mantle lithosphere. In contrast, the younger rocks (176+5 Ma) have relatively lower initial STSr/S6Srand higher trace element ratios relative to the Ferrar Magmatic Province, and this appears to be asscoiated with the later stages of firing and relatively enhanced crustal extension which allowed for the eneorporation of a small asthenosphere component.

The role of mantle plumes in the generation of continental flood basalts (CFB), and in the break up of large supercontinents, remains a matter of speculation and debate. In certain flood basalt provinces there is also a residual debate concerning the origin of the distinctive geochemical characteristics of some of the tholeiites, although most authors now agree that these include a component derived from sub, continental lithospheric mantle. Remnants of Mesozoic magmatism associated with the break-up of Gondwana form large

tholeiitic flood basalt provinces in South America, Southern Africa, Antarctica and Australia (Fig. 1; Kyle et al. 1981; Bellieni et al. 1984; Erlank 1984; Mantovani et al. 1985; Hergt et al. 1989a, 1991). In Antarctica, Mesozoic thoeliitic magmatism extends from Dronning Maud Land (DML), along the Transantarctic Mountains (TAM) to Victoria Land (Fig. 1). Recent models have reaffirmed that the generation of continental flood basalts is closely linked with the presence of mantle plumes, although in some models magmatism is triggered by the

FromSTOle-t, B. C., ALABASTER,T. & PANKHURST,R. J. (eds), 1992, Magmatismand the Causes of ContinentalBreak-up, Geological Society Special Publication No. 68, pp. 185-208.

185

186

T.S. BREWER E T A L .

qJmlm _

"\,1

Paran6\l

130 M a X

.

\~

~fendeko

/ £

..3"

_b2

w ~ \ ~ /." - " ' - ~ ' ~ I ~ ,~-i ~ \/~ [4 ~ ~ • ~'~ I o~ o~ . / ~ ~ , ) ,~ , k - \ \ L\ e s o t h o , y z~a ! ,. . . . // ,~ ",,, "K . . . . . . . f f / -~,,.o--, ~ k e r r a r Megmehc Prownce ,%..., " - - ~ ~ \ . II/ k___._~~lnte~ffSWh, c h owo y c.lBO Mo t, __~'~.... N___.. ~'"Y-- -- >,,~unatoks /I "~. -^^'MO\/ '. "~.Du ~eK-- - - "-- ~ ' - j_ -i~u o " \'~-:. ~ ~ , , , ~ ----- -- --" / ¢ "

Fig. 1. Sketch map illustrating Mesozoic flood basalt provinces of Gondwana, modified from Hergt etal. (1991). The thin line in the Paranfi and Karoo provinces represents the southernmost extent of high-Ti continental flood basalts (after Erlank et al. 1988).

emplacement of the plume (e.g. Richards et al. 1989), whereas, in others magmatism is associated with rifting across an area of anomalously hot mantle (White & McKenzie 1989). The linear outcrop pattern along the TAM and into south Australia is difficult to reconcile with an active plume model, but it offers an excellent opportunity to evaluate suggestions that, at least in some areas, the generation of continental flood basalts may be related to extensional regional tectonics. This paper presents the results of a petrological and geochemical study from Coats Land, which is close to the transition between DML and the TAM, in order to evaluate the chemical differences in the two areas, and the wider implications for the causes of CFB generation. Within Coats Land, Mesozoic dolerites outcrop in the Theron Mountains, the Shackleton Range and the Whichaway Nunataks (Figs l&3). Background geology

The Mesozoic magmatic rocks of DML and the TAM consist of basaltic (and minor acidic) lava flows, dykes, sills and the singularly large Dufek gabbroic intrusion, the total volumes of which are in excess of 1 ×106Km 3 (Kyle et al. 1981). Previous geochemical work emphasised the bimodality in the initial 87Sr/~Sr ratios of the tholeiites, which in turn led to the subdivision of

the tholeiites into two provinces, one centred on DML, and the other now termed the Ferrar Magmatic Province (FMP), which includes the TAM (Faure et al. 1979; Ford & Kistler 1980; Hergt et al. 1991). In DML magmatism is characterized by low initial ~rSr/a6Sr ratios (< 0.707, Faure et al. 1972; Harris et al. 1990), whereas in the FMP it is distinguished by elevated initial 87Sr/.~Sr ratios (> 0.709) and relatively high SiO2 and LIL element abundances (Kyle et al. 1983]' In order to explain the unusual geochemistry of the FMP, a number of alternative models have been proposed, but these can be broadly subdivided into those which invoke varying degrees of crustal contamination and those which require distinctive, chemically enriched source regions in the sub-continental mantle (Compston et al. 1968; Faure et al. 1972, 1974; Pankhurst 1977; Kyle 1980; Kyle et al. 1983, 1987; Mensing et al. 1984; Hergt et al. 1989a, b, 1991). Ford & Kistler (1980) suggested that the boundary between the two provinces lies between the Theron Mountains in Coats Land and the Pensacola Mountains (Fig. 1). However, Brewer (1990) demonstrated the occurrence of Ferrar signatures within the Coats Land dolerites and proposed that the boundary be located there. Contemporaneous with magmatism in DML, are the thick sequences of volcanic rocks in the Explora-Andenes Escarpment on the eastern

ANTARCTIC FLOOD BASALTS edge of the WeddeU Sea (Hinz & Krause 1982; Kfistoffersen & Hinz 1990). The presence of magmatic activity in both the DML and the Weddell Sea regions indicates substantial crustal thinning linked to the failed Weddell Rift system (Kristoffersen & Hinz 1990). This episode of crustal thinning predated any rift separation of Antarctica from South Africa, since the first stage of true seafloor spreading between the two continents is indicated by magnetic anomaly M22 (155 Ma) from near the margins of the Mozambique basin (Martin & Hartnady 1986). This episode of crustal thinning is considerably younger than the Karoo-DML magmatism, and was related to the strike-slip regime created by the separation of east and west Gondwana (Lawver et al. 1985). The extension of Mesozoic magmatism from DML through Coats Land and along the TAM is now represented by a narrow discontinuous linear outcrop pattern. The minimum estimated volume of this Antarctic magmatism is c. 1 x 106

I

i

Jt~.

Dronning Maud Land Province [ ] Ferrar Magmatic Province C) Extent of DML Karoo plume

Fig. 2. Schematic reconstruction of Gondwana for the lower-mid-Jurassic, also indicated are the Jurassic stress regimes for Antarctica (after Storey 1990). The position of the Karoo-Dronning Maud Land plume trace from White & McKenzie (1989). Note the extensional regime in the Weddell Sea region and the predominance of strike-slip tectonics between the Transantarctic Mountains and the collage of microplates. It would therefore appear that the Weddell Sea embayment was a site of relatively large amounts of extension compared to the Transantarctic Mountains. AP, Antarctic Peninsula; EWM, Ellsworth Whitmore Mountains; MBL, Marie Byrd Land; TI, Thurston Island; WSE, Weddell Sea embayment.

187

km 3, although no estimate of the amounts of crustal extension can be made because of the limited outcrop pattern and lack of geophysical data. However, in all Jurassic reconstructions, DML and the TAM are separated from the proto-Pacific subducting margin by c. 2000 km (Fig. 2), and within this zone there is a complex collage of microplates (Lawver etal. 1985; Elliot 1990; Grunow et al. 1991). The substantial distance between the proto-Pacific margin and the DML-FMP magmatism would appear to preclude the introduction of a contemporaneous Jurassic subduction component to the source of these magmas although Cox (1978) and Elliot (1990) have suggested that the TAM magmas were generated in a back-arc setting to the proto-Pacific margin. If such a model is correct, then early to mid-Jurassic extension and magmatism may have been associated with back-arc extension processes (Cox 1978). Recently, Storey et al. (1988) and Storey & Alabaster (1991) have identified Ferrar type signatures in igneous .rocks to the west of the Transantarctic Mountains. The westward extension of the Ferrar province places such magmatic signatures closer to the subducting proto-Pacific margin, although there is as yet no geochemical evidence to suggest the involvement of a significant contribution from contemporaneous Jurassic subduction in these areas. The central Transantarctic Mountains were the focus of active subduction processes and micro-continent collision between c. 760 Ma and 500 Ma (Borg et al. 1990). However, in the H.U. Sverdrupfjella of DML, the basement complex is composed of highly deformed calc-alkaline gneisses interpreted as 1.2-1.1 Ga volcanic sequences (Groenewald et al. 1991). Similar rocks are probably present in the basement of Kirwanveggen and Heimefrontfjella, and these 1.2-1.1 Ga volcanic sequences may also have formed in arc environments. Thus, in the geological evolution of this segment of Antarctica there have been a number of subductionrelated episodes which may have modified the subcontinental mantle. Specifically, subductionrelated geochemical signatures may have been introduced to otherwise depleted material in the continental mantle lithosphere at different times in different areas; viz the different ages of probable subduction-related rocks in the basement of DML and the TAM. Finally, in this section, we note that the Mesozoic magmatism of DML and the adjacent Karoo province has been related to the Crozet hot spot (Morgan 1981), to a hot line (Cox 1988) and to a slightly more southerly centred plume (White & McKenzie 1989). In the White and

188

T.S. BREWER E T A L .

McKenzie model, the plume supplies melt ovOr a c. 1000 km radius zone, but it is not the driving force for separation of Africa and Antarctica. White and McKenzie (1989) conclude that the FMP magmatism is unrelated to the plume which generated the DML magmatism, consistent with the linear nature of the FMP (Fig. 1) and the relatively limited extensional regimes operating during Early-Mid-Jurassic times in the TAM. However, considerable uncertainty persists over the relationship between CFB magmatism in DML and the TAM, and over any association between the mantle plume invoked beneath DML and the distinctive geochemical signature of the FMP magmatism. The key region to investigate such inter-relationships is Coats Land, since this region is marginal to the inferred mantle plume (White & McKenzie 1989), and is the most northerly extension of the FMP signatures. Geographical location Coats Land is situated to the east of the Filchner Ice Shelf at the head of the Weddell Sea, and lies on the edge of the East Antarctic craton between the TAM and DML (Fig. 3). In Coats Land there are three areas in which Mesozoic dolerite sills and/or dykes are exposed: the Theron Mountains (Brook 1972; Brewer & Brook 1990), the Shackleton Range (Clarkson 1981) and the Whichaway Nunataks (Stephenson 1966; Brewer 1990). The Theron Mountains form an approximately 120 km long NE-trending escarpment on the southern edge of the Bailey Ice Stream (Fig. 3). The main part of the escarpment has a maximum exposed relief of 900 m, which is composed of near horizontally bedded Lower Permian sedimentary rocks intruded by doleritic sills and dykes (Brook 1972; Brewer & Brook 1990). The sub-horizontal sills range in thickness from < 1 to > 200 m, averaging 30-40 m; in contrast, the dykes are impersistent and range from 1-6 m in width (Brook 1972). In the larger sills mineralogical layering has been identified, and where present, xenoliths are only of locally derived sediments (Brook 1972). The Shackleton Range lies between the Slessor and Recovery glaciers (Fig. 3) and is composed of a Precambrian basement complex unconformably overlain by Cambro-Ordovician sedimentary rocks (Clarkson 1972, 1981). Basic dykes, usually < 3 m thick, intrude both the metamorphic and sedimentary rocks, and on geochemical criteria Clarkson (1981) identified one Mesozoic dolerite. The remaining dykes have whole-rock K-At ages in excess of 300 Ma (Rex 1972; Clarkson 1981).

To the south of the Recovery Glacier lie the Whichaway Nunataks (Fig. 3). These nunataks form a series of small conical shaped hills composed of flat-lying Lower Permian sedimentary rocks (Whichaway Formation) intruded by Mesozoic doleritic sills and dykes (Omega dolerites, Stephenson 1966; Hofmann et al. 1980; Brewer 1990).

Timing of Coats Land magmatism Lower Permian sedimentary rocks exposed in the Theron Mountains (Theron Formation, Stephenson 1966; Victoria Group, Brook 1972) and Whichaway Nunataks (Whichaway Formation, Stephenson 1966) are composed of sandstones, siltstones, shales and coals. Plant remains within the shales and coals indicate a Lower Gondwana age (Plumstead 1962) and led to the correlation of these rocks with those of the Permo-Triassic Beacon Supergroup of the TAM (Gunn & Warren 1962; Plumstead 1962; Stephenson 1966; Brook 1972; Barrett et al. 1986). In the Shackleton Range, Mesozoic dolerites have been identified only on the basis of geochemical similarities to dolerites which have yielded Mesozoic radiometric ages in other areas (Clarkson 1981). However, conventional wholerock K-At studies on dolerites from the Theron Mountains indicate that they were emplaced between 158 +6 Ma and 173 +6 Ma (Rex 1972), and the Omega dolerites have yielded K-Ar ages between 163+13 Ma and 171+14 Ma (Hofman et al. 1980). The age range implied by these K-Ar results may be too large to be reliable, whereas, recent Ar-Ar palagioclase ages from the Theron Mountains more tightly constrain the magmatic episode to 176+5 Ma, with the possibility of an early event at 193+7 Ma (Brewer et al. 1991). Sample distribution and analytical procedures All of the samples analysed in this study are from sillsor dykes exposed in the Theron Mountains (105 samples (Brook 1972), Shackleton Range (2 samples, Clarkson 1981) or the Whichaway Nunataks (6 samples, Stephenson 1966). The Theron Mountains contain the most extensive outcrops of Mesozoic dolerites in Coats Land. However, owing to the nature of the terrain, no complete sections through individual sills were obtained. All samples were crashed and then powdered in agate; pressed powder pellets were used for trace clement analysis, major elements were determined on glass fusion beads, and major and trace elements were determined by X-ray flourcscencc spectrometry at Nottingham University (Table 1) following the procedure

ANTARCTIC FLOOD BASALTS

30o

/7

189

200

STFJ ELLA

200 km

'

'

o -'75 S

'

.,,,"""

DRONNING

:"

~

~,~~/soo

,y, /~/ , / . / ' / " ~r.~. / /'•

~'O~I'2 / "

(

..J~' / /(~)/" " / /-~i' / COATS

, ..............

•r

/)

jr

~

1"/''-"/

"-..

/"

t. I -

•,

~

-

. . . . .

LANU

FILCHNER ~ ICE S H E L F / / / / I

~.~

/'-

I I

I

/ 40 °

"\

",L': ,I r..-,

, "',

.

)

t

%

/~ /'~ / , , ; " ~ l C HAWAY1~/ . ~" ~ NUNATAKS

~1

20 ° l

~

i~

' -/_~

,~,~°1

~.............- e - ~ c ~ _ ~ , , " ~)~., ~> ~ 0 y w .

~

I

J

I

I

¢@~ /W . AENSTTAR?T~GA"~}" EAST I "~

ANTARCTICA i

~"=t"'o~ ~ + -'~"~--"('q/ .,~. ~ ~

n

,o.~1

~'.z~.

----v..~ .......

L,

~:::---i

I

Yl

I

...."....-~,~i

I

I

i

-oo,< s~A

;~.

~'~HAO~'ET0~" - ,, ,1~' ,~,I~RANGE / ~ll~..._.etJ~• . / "\ ~£~ /

/"

/ ./ Areowithexposed rock

coo,~ "\ surfocecontours(m) I ,," ~

I . I

30 °

/'L A/N D

~ . ,~,-Ice

R~'ov~.,~ o~-~'~

I

.~,~--

"'HEIMEFRONTFJELL~ ~ /" / !

......",",,, Ice front~

~

x . . ~ ~ o ~ ~c,~, / /

/ .1C~

1~

~

t x. _ . . I,~THERON ~.Mt..Fo,,oy_,. V . . . . . . _.., ¢l MOUNTAINS "'~ 40 W ~PorryPoint --",, .

.80°S

•

....i

.~,ooo - - -

,.-----

M(AUD

{,. BRUNT f./ ) ICE SHELF ( ~) ............ ... J ." {.." ~,,,~~/. ~-"-T.J

WEDDELL SEA

I,

I" /

I

• I ~~*"° I ~!iivIOTORIAI

o ,ooo ~-~l'~'~lm 18m

-'"

1-_

~ J

"l

10°W / I

Fig. 3. Geographical location of Coats Land; ice contours from Marsh (1986). Inset map shows the outcrop pattern of the Antarctic Mesozoic Province; black, Mesozoic magmatic rocks; stipple, other outcrop areas in DML and TAM. CL, Coats Land; DI, Dufek intrusion; DML, Dronning Maud Land; TAM, Transantarctic Mountains; TM, Theron Mountains; WN, Whichaway Nunataks.

1

14.5 33.7 4.4 17.8 5.1 1.66 5.03 5.45 1.17 3.31 3.49 0.54

1

12.2 26.7 3.4 15.7 4.4 1.37 4.42 4.94 1.02 2.89 3.40 0.44

249 43 147 109 18 55 7 bd 23 190 bd bd 215 29 86 115

99.59

50.88 0.92 14.81 11.53 0.19 6.35 10.79 2.40 0.90 0.13 0.69

Z487.2

1

13.6 31.5 4.2 17.6 4.9 1.47 5.02 5.83 1.27 3.61 3.92 0.62

388 55 61 115 19 70 12 3 27 161 3 bd 285 43 124 150

100.11

52.78 1.24 13.03 14.48 0.21 4.98 8.78 2.70 1.09 0.21 0.61

Z509.1

1

na na na na na na na na na na na na

243 45 669 19 17 25 8 2 17 182 2 bd 140 24 92 89

99.95

50.90 1.01 15.52 11.24 0.19 7.64 10.00 2.07 0.76 0.15 0.47

Z453.2

1

na na na na na na na na na na na na

413 47 55 116 20 36 13 4 29 155 bd 3 274 45 112 174

99.58

53.17 1.33 12.99 14.54 0.20 4.38 8.37 2.44 1.35 0.23 0.58

Z463.2W

1

na na na na na na na na na na na na

257 43 155 108 17 49 7 5 31 175 3 2 230 34 98 136

99.89

51.25 1.13 14.25 12.54 0.19 6.01 10.29 2.47 1.03 0.15 0.58

Z487.3

1

na na na na na na na na na na na na

221 39 527 18 17 20 6 bd 17 176 2 2 154 24 96 88

99.77

50.23 0.93 15.76 11.07 0.17 8.52 9.94 1.98 0.66 0.13 0.38

Z488.1

1

na na na na na na na na na na na na

197 41 541 20 16 19 6 2 19 170 bd bd 156 24 88 78

99.78

49.85 0.88 15.67 10.89 0.23 8.66 10.25 1.78 0.58 0.13 0.86

Z489.1

na na na na na na na na na na na na 1

na na na na na na na na na na na na 1

99.84

100.14

138 43 818 22 16 18 4 2 28 166 bd bd 193 21 74 61

49.66 0.78 15.82 10.24 0.19 9.41 9.94 1.86 0.42 0.09 1.43

49.93 0.98 15.76 11.10 0.16 9.68 9.44 1.79 0.72 0.15 0.43

246 42 459 18 19 11 7 bd 18 179 bd bd 159 27 105 94

Z498.6

Z453.1

2

13.1 29.3 3.8 17.9 4.3 1.16 4.56 5.22 1.12 3.25 3.26 0.49

350 38 732 47 16 49 9 2 30 14t bd 2 170 33 87 137

99.60

52.40 0.94 14.63 9.91 0.17 8.15 9.62 1.86 1.07 0.21 0.64

Z483.7

Sample localities: Theron Mountains, B r o o k (1972), Shackleton Range, Clarkson (1981). G r o u p number refers to geochemical groups as discussed in the text. bd, below detection; na, not analysed.

Group

La Ce Pr Nd Sm Eu Gd Dy Ho Er Yb Lu

Ba Co Cr Cu Ga Ni Nb Pb Rb Sr Th U V Y Zn Zr

330 30 250 18 19 7 11 7 27 173 bd bd 206 38 126 139

99.74

Total

P205 LOI

K20

53.16 1.47 13.56 13.03 0.20 5.21 9.00 2.23 1.15 0.25 0.48

SiO2 TiOz AI20~. FeEOJ MnO MgO CaO Na20

Z461.1

Table 1. Representative chemical analysis of dolerites from the Theron Mountains and Shackleton Range (Z726.1,. 4 and. 4 W)

2

19.3 39.2 5.3 21.6 4.7 1.51 4.97 5.49 1.05 3.20 3.39 0.47

359 34 604 19 18 68 9 2 32 170 3 2 159 34 100 149

99.69

53.78 1.04 15.05 9.61 0.15 7.40 8.56 2.00 1.27 0.17 0.66

Z475.2

t'.,

t~ ~q

m

m

Group

Ce Pr Nd Sm Eu Gd Dy Ho Er Yb Lu

La

Co Cr Cu Ga Ni Nb Pb Rb Sr Th U V Y Zn Zr

99.59

17.8 39.6 5.0 22.8 5.2 1.56 5.05 5.71 1.26 3.40 3.54 0.49

2

2

428 28 430 19 18 26 10 5 40 179 2 2 133 32 87 160

14.0 31.0 3.9 18.4 4.4 1.21 4.51 5.21 1.08 3.08 3.15 0.45

100.03

270 38 806 46 15 63 7 6 26 142 2 bd 162 33 97 131

54.38 1.12 15.19 9.40 0.15 6.37 8.36 2.05 1.61 0.19 0.77

52.25 0.93 14.55 9.88 0.16 8.54 9.68 1.57 0.75 0.22 1.50

la

Z479.2

ZA77.7

Total

LOI

P205

CaO Na20 K20

MgO

Al20/. Fe20~ MnO

TiO~

SiO2

Table 1. Cont.

2

na na na na na na na na na na na na

349 34 503 17 17 42 9 7 33 180 2 2 148 34 106 155

99.65

53.46 1.08 15.60 9.35 0.16 6.78 8.93 1.71 1.29 0.17 1.12

Z453.3

2

na na na na na na na na na na na na

362 33 602 19 18 54 8 5 35 190 1 bd 128 30 89 139

99.69

53.78 0.98 16.03 8.65 0.15 7.19 8.82 1.82 1.34 0.17 0.76

ZA79.3

2

na na na na na na na na na na na na

324 39 712 39 14 35 7 3 24 137 2 2 166 32 84 126

99.66

52.60 0.90 14.56 9.99 0.17 8.32 9.77 1.66 0.98 0.20 0.51

ZA81.4

3

7.5 16.9 2.2 10.1 2.6 0.82 2.89 3.76 0.75 2.36 2.62 0.37

120 45 959 38 14 28 5 4 24 120 1 2 194 23 71 75

99.76

50.80 0.63 14.98 9.99 0.17 9.47 10.63 1.83 0.31 0.09 0.86

Z471.13a

3

na na na na na na na na na na na na

128 34 478 17 16 7 5 bd 12 211 bd bd 166 16 71 61

99.70

49.98 0.67 17.44 8.45 0.15 8.75 10.65 1.89 0.40 0.09 1.23

ZA66.3

3

na na na na na na na na na na na na

129 37 547 11 15 6 4 bd 9 147 bd 2 197 20 72 70

99.92

50.53 0.66 16.50 8.87 0.12 8.58 11.29 1.65 0.37 0.09 1.26

ZA71.14

3

na na na na na na na na na na na na

113 35 483 14 15 5 3 7 13 176 4 bd 161 19 66 64

99.80

49.34 0.64 16.55 9.13 0.11 9.12 11.49 1.28 0.25 0.10 1.79

ZA98.9B

3

na na na na na na na na na na na na

173 42 636 32 14 11 5 5 21 115 1 2 199 25 76 76

99.76

51.24 0.68 15.06 9.95 0.17 8.43 10.79 1.64 0.50 0.11 1.19

ZA71.13C

A

17.6 42.5 5.7 25.2 7.0 1.99 7.23 8.17 1.50 4.15 4.82 0.67

300 56 81 131 23 33 11 bd 41 187 4 2 262 52 157 210

99.63

48.65 2.25 14.13 16.77 0.23 4.55 8.82 2.48 1.23 0.29 0.23

ZA77.4

A

18.4 45.5 6.1 26.1 7.4 2.17 7.70 8.46 1.54 4.26 4.40 0.54

292 53 96 168 23 28 12 6 46 172 4 2 312 54 179 226

99.85

48.74 2.43 13.14 17.77 0.23 4.75 8.53 2.49 1.45 0.32 0.00

Z480.1b

Group

Ho Er Yb Lu

Dy

La Ce Pr Nd Sm Eu Gd

Co Cr Cu Ga Hi Nb Pb Rb Sr Th U V Y Zn Zr

ea

Total

P205 LOI

K20

CaO Na20

MgO

AI20~. Fe2O~ MnO

TiO2

SiO2

Table l. Cont

22.4 51.6 6.5 30.8 7.7 2.00 8.82 8.96 1.94 5.31 5.21 0.78

A

A

276 59 84 96 23 30 9 3 38 192 lad 2 241 49 148 197

21.1 43.5 5.5 25.7 7.0 1.82 8.76 7.87 1.59 4.10 3.52 0.54

286 49 93 148 24 32 11 bd 45 167 5 2 284 52 164 211

99.59

48.22 2.08 14.41 16.61 0.20 4.91 8.95 2.55 1.17 0.28 0.21

48.50 2.37 13.47 17.28 0.24 4.72 8.65 2.38 1.35 0.31 0.33

99.60

Z483.6

Z480.2

A

18.3 43.9 5.8 26.4 6.3 1.68 7.53 7.46 1.64 4.67 4.33 0.61

255 54 82 113 25 31 10 bd 40 195 5 2 240 48 150 190

99.62

48.00 2.06 14.53 16.68 0.21 4.89 8.85 2.33 1.16 0.27 0.64

Z481.11

A

18.2 41.3 5.3 24.2 6.7 1.73 8.09 7.75 1.60 4.57 4.44 0.67

246 54 87 144 22 30 9 bd 45 162 1 2 282 51 150 216

99.95

48.18 2.20 14.58 15.77 0.19 4.95 8.54 2.74 1.34 0.29 1.17

Z478.11

A

20.5 46.4 6.1 29.8 7.1 1.87 8.27 8.42 1.78 5.18 4.71 0.73

313 52 84 134 24 29 11 4 42 184 6 3 260 49 152 200

99.71

48.57 2.22 14.02 17.10 0.22 4.82 8.70 2.47 1.29 0.30 0.00

Z500.1

A

21.0 47.3 6.3 30.0 7.2 1.91 8.54 8.64 1.83 5.06 4.78 0.74

263 47 87 148 24 31 11 6 39 205 4 bd 276 52 177 209

99.88

46.50 2.32 14.13 17.19 0.22 5.00 9.30 2.14 1.09 0.30 1.69

ZA97.6

A

na na na na na na na na na na na na

367 48 90 133 24 35 11 bd 36 228 3 3 282 48 154 194

99.62

47.60 2.19 14.33 16.39 0.22 4.84 8.29 2.48 1.03 0.29 1.96

ZA98.7

B

33.8 76.6 9.5 41.1 11.5 3.05 11.10 12.18 2.19 5.79 5.70 0.92

696 34 56 260 24 37 19 10 46 244 bd 3 388 72 209 323

99.60

51.80 2.90 12.49 14.69 0.23 3.64 8.09 2.55 1.76 0.50 0.95

ZA71.11

B

27.9 68.3 9.1 41.3 11.1 2.83 10.61 12.27 2.17 6.14 5.68 0.85

601 37 62 202 23 35 18 5 45 232 5 1 390 70 147 297

99.47

51.47 2.88 12.72 14.91 0.19 3.70 7.95 2.74 2.01 0.48 0.42

ZA72.4

B

30.3 65.4 8.4 38.6 9.9 2.34 11.65 10.99 2.31 6.53 6.19 0.89

646 42 60 201 24 33 19 4 44 245 3 2 390 70 138 310

100.13

51.60 2.96 12.79 14.99 0.21 3.59 8.03 2.74 1.91 0.50 0.81

Z471.16

B

26.5 64.5 8.5 39.5 9.1 2.28 11.40 10.41 2.22 5.96 4.98 0.74

568 39 60 198 24 39 17 9 45 236 1 2 387 67 146 297

99.53

51.16 2.86 12.47 15.05 0.22 3.72 8.05 2.70 1.90 0.47 0.93

Z483.9

C

20.3 38.7 4.6 20.2 5.5 1.40 6.84 6.74 1.42 3.91 3.77 0.47

381 48 36 154 21 20 10 7 56 142 8 1 428 43 125 163

99.66

53.64 2.02 12.05 16.25 0.19 3.21 7.57 2.31 1.62 0.23 0.57

Z508.1

t~

.]

ANTARCTIC FLOOD BASALTS

193

.~.~.~.~.~.~.~.~

.

.

.

.

.

.

.

.

.

.

•

. . . . . . . . . . .

~

~

•

•

o

,

•

•

°

.

.

~

.

.

.

o

.

o

.

.

~

.

.

.

.

.

.

.

.

°

~

.

o

•

~

°

~

'

,

.

•

•

°

•

N

v ~ . ~

~

I

'

~

l

'

~

l

'

~

.

.

'

~

'

~

O

~

'

f

~

r

~

N

•

.

•

.

•

.

•

.

.

•

•

•

•

•

.

m.~.~.~.~.o.

c~o~o~o o o Oo c~ ~

.

.

.

°

.

.

.

.

.

.

.

.

.

.

.

.

•

.

~

.

.

•

.

•

18.4 4.4 0.512300

18.166 15.599 38.047

2°6Pb/2°4Pb 2°7pb/2°4pb 2°apb/2°4Pb

26.1 141.7 0.71109

Nd Sm 143Nd/l~Nd

Rb Sr aVSr/~Sr

18.152 15.560 38.044

2°6Pb/2°4Pb 2°7pb/2°4pb ~pb/2°4pb

Z477.7

17.8 5.1 0.512320

27.3 172.8 0.70986

Z461.1

18.214 15.607 38.157

22.8 5.2 0.512260

40.3 178.6 0.71463

Z479.2

20.122 15.724 39.205

15.7 4.4 0.512400

23.2 189.5 0.70888

Z487.2

na na na

na na na

33.0 180.4 0.71382

Z453.3

18.377 15.554 38.042

17.9 4.9 0.512510

26.7 161.4 0.70765

Z509.1

7_.479.3

na na na

na na na

35.1 190.3 0.71418

na na na

na na na

16.9 181.5 0.70897

ZA53.2

Z481.4

na na na

na na na

24.3 136.5 0.71117

na na na

na na na

29.3 154.6 0.70797

Z463.2W

na na na

Z466.3 11.5 211.1 0.70763

na na na

na na na

17.1 176.3 0.70886

Z488.1

na na na

i0.1 na 2.6 na 0.512380 n a

23.5 119.8 0.71028

Z471.13a

na na na

na na na

30.7 175.4 0.70943

7_,487.3

na na na

na na na

9.3 146.7 0.70893

Z471.14

na na na

na na na

19.4 170.1 0.70883

Z489.1

na na na

na na na

13.0 176.0 0.70937

Z498.9B

na na na

na na na

17.9 178.6 0.70921

ZA53.1

na na na

na na na

21.2 115.2 0.70990

Z471.13G

na na na

na na na

28.1 166.4 0.70866

7_,498.6

7_,477.4

18.350 15.582 38.550

25.2 7.0 0.512480

41.3 187.3 0.70818

na na na

17.9 4.3 0.512320

30.2 141.3 0.71132

ZA83.7

Representative St, Nd and Pb isotopic compositions for dolerites from the Theron Mountains and the Whichaway Nunataks (samples prefixed TAE)

Nd Sm 143Nd/l~Nd

Rb Sr aTSr/a6Sr

T a b l e 2.

na na na

26.1 7.4 0.512510

46.1 171.7 0.70816

Z480. l b

18.236 15.606 38.152

21.6 4.7 0.512330

31.6 169.5 0.71390

Z475.2

"~

38.4 191.9 0.70800

44.7 166.8 0.70817

25.7 7.0 0.512525

18.306 15.557 38.398

Rb Sr 87Sr/86Sr

Nd Sm 143Nd/144Nd

2°6pb/2°4pb 2°7pb/2°4pb 2°8pb/2°4pb

68.8 149.3 0.71334

28.2 6.8 0.512360

56.0 140.1 0.71294

24.2 5.5 0.512349

18.871 15.627 38.688

Rb Sr 87Sr/86Sr

Nd Sm 143Nd/144Nd

2o6pb/2O4pb ZOTpb/2O4pb 2o~pb/2O4pb 18.840 15.580 38.570

32.4 7.4 0.512405

72.0 135.9 0.71389

Z484.1

na na na

Z500.1

18.910 15.666 38.822

18.2 4.3 0.512350

46.9 149.6 0.71248

Z476.1

18.291 15.539 38.343

29.8 7.1 0.512513

42.4 184.4 0.70814

30.8 7.1 0.512360

78.1 140.5 0.71420

Z490.1

18.455 15.527 38.529

24.2 6.7 0.512497

44.7 162.1 0.70862

Z,478.11

Z498.7

Z486.1 60.1 134.4 0.71318

na na na

na na na

na na na

19.6 22.8 5.6 5.8 0.512370 0.512357

43.8 142.5 0.71227

Z487.1

18.684 15.585 38.590

30.0 na 7.2 na 0.512380 0.512547

39.3 35.6 205.4 227.5 0.70869 0.70860

Z497.6

na na na

41.3 11.1 0.512460

44.8 231.5 0.70654

Z472.4

18.243 15.540 37.950

38.6 9.9 0.512494

43.6 244.7 0.70690

Z471.16

18.089 16.398 37.810

39.5 9.1 0.512505

44.7 236.4 0.70646

Z483.9

18.947 15.721 38.987

20.2 5.5 0.512381

56.2 141.9 0.71283

Z508.1

na na na

24.9 6.0 0.512402

56.1 162.3 0.71291

52.1 159.7 0.71267

18.165 15.523 37.816

18.520 15.554 38.176

17.8 22.3 3.9 5.2 0.512400 na

42.4 146.7 0.71396

18.335 15.554 38.013

30.9 7.1 0.512465

68.6 142.1 0.71384

18.390 15.563 38.099

30.4 6.9 0.512477

69.7 155.4 0.71365

TAE301/1 TAE302/5 TAE304/2 TAE304/6 TAE304/7

18.558 15.688 38.701

41.1 11.5 0.512490

46.2 243.7 0.70734

Z471.11

Sr, Nd and Pb isotopes values represent the measured values. Sample localities for the Whichaway Nunataks from Stephenson (1966).

18.895 15.611 38.641

Z478.1

18.286 15.563 38.426

26.4 6.3 0.512473

40.0 194.9 0.70810

Z481.11

Z485.1

18.290 15.588 38.464

30.8 7.7 0.512466

Z483.6

7_,480.2

Table 2. Cont.

0 0

196

T.S. BREWER E T A L .

of Harvey & Atkin (1980). Rare-earth elements were determined on a subset of samples from the Theron Mountains and all samples from the Shackleton Range and Whichaway Nunataks (Table 1) using inductively coupled plasma optical emission spectrometry, following the procedures of Walsh et al. (1980). Nd, Pb and Sr isotopes were determined on all of the Omega dolerites and a sub-set of Theron Mountain samples (Table 2). The Pb, Sr and Nd fractions analysed for their isotopic compositions were prepared using standard dissolution and ion exchange techniques. Blanks were negligible at < I ng for Pb and Nd, and < 5 ng for Sr, and no blank corrections were applied. The Pb, Sr and Nd isotope ratios were analysed on the multiple collector Finnigan MAT261 instrument at the Open University. Nd was run in dynamic mode, and Pb and Sr were run in static mode. For STSr/at'Sr,NBS 987 gave a mean value of 0.71023 +4 (2 standard deviations) over the period of this study, and the in-house Nd standard results were reproducible to +0.00002 based on 18 runs with 143Nd/144Ndfor BCR-1 = 0.51264. The Pb data were collected for a 1% a.m.u. Corrected fractionation relative to values of 2°6pb/2°4Pb = 16.937, 2°TPb/ 2°4pb = 15.491 and ~Pb/Z°4Pb = 36.700 for NBS 981.

Petrography In the TAM, Ferrar dolerites have been subdivided into olivine tholeiites, hypersthene tholeiites and pigeonite tholeiites (Gunn 1966). All three petrological types occur in Coats Land, but the olivine tholeiites are extremely rare and have been reported only from the Theron Mountains. Olivine tholeiites in the Theron Mountains are represented by two textural varieties. In the first, olivine forms small (0.1-1 mm) anhedral grains in a medium-grained holocrystalline mosaic of clinopyroxene, plagioclase and opaque phases. The second variety is far more restricted, and olivine forms larger (2-5 mm) subhedral-anhedral grains which define a mineralogical layering (Brook 1972). The bulk compositions of such samples have relatively high MgO values (> 12 wt%). Hypersthene and pigeonite tholeiites range texturally from holocrystalline to hypocrystalline varieties. Hypocrystalline dolerites contain variable proportions of glass (3-90%) and the phenocrysts are composed of plagioclase and/or clinopyroxene. The holocrystalline dolerites are composed of mosaics of clinopyroxene, plagioclase and opaques, with variable amounts (0-10%) of either orthopyroxene or granophyre. The granophyre is interstital, and composed of quartz and K-feldspar. Within individual intrusions there is evidence of localized low temperature alteration (devitrification, carbonation and chloritization), but such samples have been excluded from the geochemical data base.

Geochemistry Coats Land dolerites are sub-alkaline tholeiites, which are either quartz or olivine normative, with M g # ranging from 71.0 to 21.4 ( M g # - - 1 0 0 [Mg/Mg+Fe 2+] molecular, with FeY+/Fe 2+ = 0.15). SiO2 ranges from 46.5 to 56.3 wt%, MgO from 2.0 to 9.9 wt%, KzO from 0.25 to 2.26 wt%, TiO2 from 0.60 to 3.09 wt%, P205 from 0.08 to 0.54 wt%, Ba from 90 to 743 ppm, Cr from 10 to 1383 ppm, Nb from 3 to 22 ppm, Rb from 9 to 81 ppm, Sr from 105 to 256 ppm, Y from i6 to-74 ppm and Zr from 6i to 347 ppm. Within the Karoo and Parand Mesozoic CFB provinces, tholeiites have been usefully subdivided into 'high-Ti' and 'low-Ti' magma types. This was initially done on the basis of TiOz contents. However, a better subdivision is achieved by use of element ratios which are less sensitive to variations in the degree of partial melting and fractional crystallization (Erlank et al. 1988; Peate et al. 1991). Some selected trace element ratios are summarized in Table 3, and adopting this approach it is clear that all the Coats Land

Table 3. Elemental characteristics o f Mesozoic continental flood basalts associated with the break-up o f Gondwana. The subdivision o f high and 1ow- Ti magma types is based on the criteria applied in the Karoo Province by Erlank et al. (1988)

TiO2

Ti/Y

ZrfY

1.1-1.8 1.1-2.6 1.0-4.8

230-350 250-460 200-650

3.1-4.0 2.7-4.9 3.4-8.1

0.6-1.5 1.9-3.1

160-260 170-400

2.9-5.1 3.6-4.8

0.4-2.0

140-220

2.7-4.3

> 2.5

> 420

> 6.0

> 2.9 1.5-3.2 0.7-1.9

> 350 > 300 < 330

> 5.5 3.5-7.0 2.0-6.5

DML

Kirwan Vestfjella Heimefrontfjella Coats Land

Series 1 Series 2 FMP Karoo

High-Ti Parana

High-Ti Intermediate Low-Ti

Data sources: Dronning Maud Land; Heimefrontfjella Juckes (1968); Brewer & Brook (1990); Brewer (unpublished data); Kirwan, Harris et al. (1990); Vestfjella, Furnes et al. (1987); Ferrar Magmatic Province; Siders & Elliot (1987); Hergt et al. (1989 a & b); Kyle & Pankhurst (unpublished data); Karoo; Erlank et al. (1988). Paranfi; Peate et al. (1991).

ANTARCTIC FLOOD BASALTS rocks and the FMP classify as low-Ti, whereas, in DML both high-Ti and low-Ti magma types occur (Harris et al. 1991). As indicated earlier, the Antarctic Province had previously been subdivided into the DML and Ferrar Provinces (Faure et al. 1979). This classification was based upon initial 875r/a6Sr ratios, such that the Ferrar was characterized by ratios < 0.709 and the DML compositions had ratios between 0.703 and 0.707. In a recent synthesis of low-Ti Gondwana CFBs, Hergt et al. (1991) further identified the FMP as an extremely depleted form of the low-Ti magma type (Table 4). Geochemically, the FMP is characterized by low TiO2, P205, Na20, Fe203, Ti/Zr, and eNd, and high SiO2, Rb/Ba, Rb/Sr, S7Sr/a6Sr and 2tnpb/2°4pb relative to oceanic basalts (Hergt et al. 1991), and appears to be remarkably uniform in composition along the available outcrop (c. 3000 km). In detail the Coats Land dolerites can be subdivided into two Series in the basis of TiO2 content (Fig. 4), and these can then be further subdivided into a number of groups based upon major and trace element correlations (Brewer & Brook 1990). The recognition of at least two groups within Series 1 dolerites is substantiated by Ar-Ar plagioclase ages of 173+6 Ma for Group 1 and 193+7 Ma for Groups 2 and 3 dolerites respectfully (Brewer etal. 1991). All of the Series 2 compositions have Ar-Ar ages in the range 176+5 Ma. If the specific definition for the FMP (Hergt et al. 1991) is applied, then only

197

Group 2 dolerites may be classified as having FMP signatures sensu stricto. However, many of the rocks studied have broadly similar minor and trace element characteristics, and they all represent low-Ti CFB (see Table 4). A detailed account of the petrogenesis of the Coats Land dolerites will be presented elsewhere. A plot of Ti;Y versus initial S7Sr/S6Sr may be used to illustrate the regional geochemical variation in Antarctica (Fig. 5). Ti/Y has been shown to be relatively insensitive to crustal contamination (Hergt et al. 1991) and s o should reflect 3.3

3.0 TtO~ (m. *i)

1-5

l..q

~ " ~

o.s

~

s,,u,l

~

Gt.,o~ 1

.... .~....j pbbro ~rsetJomgion

0.o

,

46

i

48

,

i

,

50

i ~2

,

i ~,

,

i .$6

s;o2(wc~)

Fig. 4. Subdivision of Coats Land dolerites on the basis of TiO2 content. Vector indicates the effect of gabbro (plagioclase+clinopyroxene) fractionation. FMP, Ferrar Magmatic province, data from Siders & Elliot (1985); Hergt et al. (1989a, b). Symbols: Series 1: open squares, Group 1; solid squares, Group 2; open circles, Group 3; Series 2: open triangles, group A; solid triangles; Group B; solid circles, Group C; crosses, Group D.

Table 4. Geochemical characteristics of FMP, D M L P and Coats Land Series 1 magma types, in which MgO >

5%. All of the Coats Land Series H magma have been excluded because MgO is < 5.0%

Series 1 Coats Land FMP

DMLP

1

Groups 2

3

TiO2

< 1.0

1.0-2.6

0.8-1.5

0.8-1.2

P205

< 0.15

0.15-0.5

0.1-0.25

0.15-0.25

0.6-0.7

Na20 Fe203 SiO2

1.0-2.8 7.0-13.7 49.5-55.0

1.5-4.1 10.4-16.5 46.0-54.0

1.7-2.7 10.5-15.0 49.0-53.5

1.5-2.1 8.5-11.5 51.5-54.5

1.2-1.9 8.5-10.0 49.3-51.5

Ti;Y Ti/Zr Rb/Ba Rb/Sr ENd 87Sr/a6Sr 2°Tpb/2°4pb

< 230 < 85 0.03-0.25 0.02-0.80 >-3.50 > 0.709 > 15.60

230-650 48-139 0.01-0.25 0.01-0.16 5.2--2.1 < 0.7070 < 15.56

170-280 46-103 0.07-0.20 0.06-0.21 -1.19--5.17 0.7064-0.7087 15.55-15.72

160-250 39-65 0.07-0.15 0.11-0.25 > -4.5 > 0.7097 15.59-15.61

160-250 50-67 0.06-0.25 0.05-0.25 > -4.0 0.7070-0.7089 na

< 0.12

na, not analysed. Data sources: FMP; Siders & Elliot (1985), Hergt et al. (1989a & b), Kyle et al. (1987), Kyle & Pankhurst (unpublished data). DMLP; Fumes et al. (1987), Harris et al. (1990), Brewer & Brook (1990), Brewer (unpublished data).

198

T.S. BREWER E T A L . 0.714

~

0.710

0.706 0.704 0.702 1111

,

i 200

F ~ 300

i 400

,

i ~0

,

! 600

Fig. 5. Geochemical spatial variation within the Antarctic Mesozoic Province, for clarity only fields are shown for Dronning Maud Land (Furnes ~,~al. 1987; Harris et al. 1990; Brewer, unpublished data) and Ferrar Magmatic Provinces (Hergt etal. 1990b; Kyle & Pankhurst, unpublished data), Coats Land symbols as for Fig. 4. source compositions and/or partial melting processes. Although STSr/~Sr is sensitive to crustal contamination it has been widely used to distinguish the FMP and DML magmas (Faure et al. 1979). In Antarctica, both Ti/Y and the intitial a7Sr/S6Sr ratios vary systematically from the DML to the FMP (Fig. 5). DML magmas have Ti/Y ratios > 230 and initial ~Sr/~'Sr < 0.7070 whereas the FMP magmas have Ti/Y < 200 and initial STSr/86Sr > 0.7076. Coats Land dolerites have Ti/Y (163-395) and initial 87Sr/S6Sr ratios (0.7052-0.7141) which are transitional and overlapping with the values typical of the FMP and DML magmas (Fig. 5).

lations between major elements in Group 1 and Group 2 magma types are similar to those that result from fractional crystallisation of a gabbroic assemblage (clinopyroxene+plagioclase) from a basaltic magma (Fig. 6). Within both Groups 1 and 2 some of the divergence from the gabbro vector (Fig. 6) can be partly explained by variation in the mineral proportions in the fractionating assemblage. The Group 1 tholeiites have near constant Sr values (150-210 ppm) over a range of MgO (4.3-9.7 wt%), suggesting that significant amounts of plagioclase were present in the fractionating assemblage (of. Cox & Hawkesworth 1985). In contrast, the Groups 2 and 3 dolerites exhibit a range of Sr ~,alues over a restricted range of MgO (Group 2: Mgo 6.4-9.9 wt%, Sr 130-240 ppm, Group 3" MgO 8.4-9.5 wt%, Sr 105-220 ppm), suggesting less fractionation, and perhaps less plagioclase in the fractionating assemblage. Fractionation of a gabbroic assemblage cannot explain the lower Na20, TiO2, Fe203 and

52 J0 $c

Major elements

18 F,20~ (wt.~)

16

Series 1 14

Series I tholeiites are primarily distinguished by TiO2 < 1.5 wt%. From major element correlations, Series 1 can be further subdivided into 3 groups (Figs 4 and 6), of which Group 1 shows the largest compositional range (SiO2 49-55 wt%) and strong correlations between major elements. In contrast, Group 2 has a more re= stricted range of SiO2 (51-55%), and elements such as AI, Fe, Na, Mg, Mn, and P do not correlate with SiO2. The third group has a very restricted compositional range, AI, Fe, Mg, Na, Mn, P and MgO are not correlated with SiO2. This group appears to be related to the Group 2 compositions by fractonation of olivine+plagioclase+clinopyroxene. In order to evaluate the effects of fractional crystallization, an index.is required that is also relatively insensitive to crustal contamination processes. In basic rocks such an index is the MgO value (Cox & Hawkesworth 1984). Corre-

• 12

1o

CaO

11 C .

(mt.~) IO 9 8 "7 6

t..5

2.5

3.5

4~

$,5

6.5

7.5

8..$

9~q

MZO ( w L ~ )

Fig. 6. Major element variation diagrams for Coats Land magma types. On each diagram the vector for gabbro (plagioclase+clinopyroxene) fractionation is illustrated. Symbols as for Fig. 4.

ANTARCTIC FLOOD BASALTS higher SiO2 contents of Group 2 dolerites relative to Group 1. Relatively low Na20 and Fe203 at elevated SiO2 values is a feature of the FMP, and has been attributed to partial melting of a less fertile source under hydrous conditions (Hergt et al. 1991). The identification of Group 2 magma type as an extension of the FMP suggests a similar source and melting regime, whereas, the higher TiO2 and Na20 contents of Group 1 magmas may be due to smaller degrees of partial melting or slightly different source compositions.

I000

199

ia.

Ba Th

g

~

La C.e Pb

Sr bid

'=[ l,.

P

Sm Zr Ti

Y

f S,ri,, 2

[

I00 /

/--'-

472.4

/ ~

47"/.4

Series 2

Series 2 tholeiites are characterised by TiO2 > 1.8 wt%, and high Fe203, P205 and low A1203 and MgO relative to Series I samples (Figs 4 and 6). On the basis of major and trace element correlations 4 groups (A-D) can be identified, although the range of compositions within each group is small (Figs 4 and 6). The previously described high TiO2 tholeiites from the TAM (Siders & Elliot 1989) all had high SiO2 (> 56 wt%) and very low MgO (< 3%) values, in contrast the Coats Land compositions have a range of SiO2 values (46-57 wt%). It is also evident from Figure 6 that Series 1 and 2 rocks cannot be related by low pressure fractionation of a gabbroic assemblage. Trace d e m e n t s

All Coats Land tholeiites have similar Primitive Mantle Normalized (PMN) profiles (Fig. 7), with relatively high LIL element contents, high Pb relative to Ce and Sr, and low Nb, Ti, P and Sr and HFSE abundances. In general Series 1 rocks have lower overall HFSE concentrations than those in Series 2 (Fig. 7). The PMN profiles are similar to those described from the FMP by Hergt et al. (1991), and exhibit many of the characteristics of calc-alkaline magmas and upper crustal compositions rather than those commonly observed in partial melts from asthenospheric upper mantle. Thus, Coats Land dolerites have low Nb/La, Zr/Y and Ti/Y and high Rb/Sr, Rb/Ba, and Ba/La ratios relative to oceanic basalts (Fig. 7 and Table 4). The Rb/Sr and Rb/Ba ratios are variable, but the high values are more typical of upper crustal rocks than those derived from the upper mantle. Overall, Coats Land rocks have the distinctive minor and trace element signatures of many low-Ti CFB, with many Series 1 rocks having the same element ratios as the FMP (Table 4, Hergt et al. 1991). None of the above trace element ratios correlate with an index of fractionation such as

t

Rb B I

"I'll

K

~

~

C.~ Pt) Sr lqd

P Sal Zr

1'I

Y

Fig. 7. Representative Primitive Mantle Normalized diagrams for Coats Land magma types. In both (a) and (b) the average FMP magma (Hergt et al. 1989a) is represented by a solid line with no symbols. Normalizing values from Sun & McDonough (1987). Symbols for Coats Land as in Fig. 4.

MgO, which further suggests that the trace element signatures were a feature of the magmas prior to emplacement in the continental crust. Incompatible element ratios can be used to infer whether continental basalts derived their minor and trace element signatures largely from the asthenosphere or from the sub-continental lithospheric mantle. N- and E- type MORB have Ba/La ratios similar to or less than the Primordial Mantle (PM) value (10.2, Sun & McDonough 1989), but Nb/La ratios ranging from 0.5-1.0 for N-MORB and 1.0-1.5 for EMORB (Gill 1981; Sun & McDonough 1987). In contrast, Ocean Island Basalts from the South Atlantic islands of Gough and Tristan da Cunha, together with basalts from the Walvis Ridge, are characterized by somewhat higher Ba/La (8-20, Weaver et al. 1986), over a similar range of Nb/ La, and are displaced from the PM towards sediment values and the field of orogenic andesites (see Fig. 8). The relatively low Nb/La (< 0.9), together with high Ba/La (> 15) and Ba/Nb ratios in these oceanic basalts have been attributed to the incorporation of small amounts of ancient (Proterozoic) pelagic sediment within their source regions (Weaver et al. 1986; Sun & McDonough 1989). In Fig. 8, Coats Land magmas have high Ba/ La (> 10) and low Nb/La (> 1.0) ratios, similar

200

T.S. BREWER E T A L .

m'.voI V ~ : ~ 9

I

0.5

I ..=.,~,..: .~"

:,l,

~

_

~.- a , . V l ; , ~

,

•

•

i 20

10

.

-

,

i 30

Fig. 8. Variation of Ba/La versus Nb/La for the Coats Land magma types. N-MORB and E-MORB values from Gill (1981) and Sun & McDonough (1987), orogenic andesites from Gill (1981). Open squares are the average sedimentary values from Taylor & McLennan (1985). Large circle indicated Tas. represents field for Tasmanian dolerites, data from Hergt et al. (1989a). Small circle marked PM is the primordial mantle from Sun & McDonough (1989). to those in the average sediment values of Taylor & McLennan (1985) and displaced towards the field for orogenic andesites (Gill 1981). The degree of displacement relative to MORB values is more extreme than Gough-Tristan-Walvis data, which may reflect a larger amount of sedimentary material, and/or a more depleted

[/ -"-

471.1~ /

L/

,-5~

:

,IL

mantle component. One possibility for such sediment-mantle mixing is in the earlier (Precambdan) subduction episodes, when sedimentary material is likely to have been dragged down and incorporated into the sub-continental mantle lithosphere. All the rocks analysed are light rare-earth element (LREE) enriched with flat heavy rare earth element (HREE) profiles (Fig. 9). Some of the samples also have negative Eu anomalies (Fig. 9). Series 2 compositions show greater overall abundances of the REE, often with significant negative Eu anomalies. Sr and Nd isotopes Coats Land dolerites have a relatively large range in initial 87Sr/a6Sr (0.7052-0.7138; Table 2), which overlaps that of the DML and FMP magma types (Fig. 10). Dolerites from the Theton Mountains exhibit variable Sr isotope ratios, similar to those of both the DML and FMP. Dolerites from the Whichaway Nunataks have FMP geochemical signatures, and slightly less radiogenic Nd values relative to the Theron Mountains dolerites (Fig. 10), although this may in part reflect the small number of samples available. In detail, the different Coats Land magma types have relatively restricted initial 87Sr/a6Sr ratios (Fig. 10). Series 2 groups are characterised by either relatively low initial 87Sr/a6Sr (< 0.707, Groups A and B) or relatively high values (> 0.709, Groups C and D). Series 1 compositions have a range (Fig. 10), with the FMP magma types having values > 0.708. The Group 3 magma type has relatively high c MgO and a restricted range of initial 87 Sd ~ o Sr (0.7070-0.7089) compared with the Group 2

/

' ,j . . . . . . . .

!~ Cz Pr Nd Pm Sm Eu Od To 1~ lio Er Tm Yb Lu 100

0.5127 0J126

~

0~L24 IUrmm

i

•

•

:

¢r~A

: :

~rL4 ~OLI i

i

i

•

•

0.5123 0.5122 0.5121 , ,

.

.

.

i

|

i

,

,.

.

0.5120 0.703

~ 0.70~

....

a

0.707 ~Sr/~r

Fig. 9. Representative chondrite normalized rare earth element plots for Coats Land magma types. Normalized to the C1 chondrite values of Sun & McDonough (1987). Symbols: Series 1: open squares, Group 1; solid squares, Group 2; open circle, Group 3; Series 2: open triangles, Group A; solid triangles, Group B; solid circles, Group C; crosses, Group D.

0 _ 0.709

0.711

0.713

0.715

(at lS0 M L )

Fig. 10. Variation of initial STSr/S6Srversus t43Nd/144Nd calculated at 180 Ma. Data sources for fields: Ferrar; Kyle et al. (1987), Hergt et al. (1989a,/~), Dronning Maud Land (Brewer unpublished data), Kirwan basalts (Harris et al. 1990), Whichaway Nunataks (Brewer 1990). Symbols as for Fig. 9.

ANTARCTIC FLOOD BASALTS

201

0.714 39~ i

0.712

t.

l

l

39.O 38.5

0.710

38.0 ~.5 0.706

37.0 _fr~____t~____aaa !

0.704 1.5

3.5

,

I 5.5

! 7.5

,

.

l

,

!

A

,

I

,

I

|

1

•

, •

9.5

MsO(wt.~) 15.(x5

Fig. 11. Variation of MgO versus initial STSr/S6Srratio, calculated at 180 Ma. Also shown is a vector for gabbro fractionation. Symbols as for Fig. 9. magmas (Fig. 11), and has a similar Sr isotopic composition to that of the Dufek intrusion of the FMP (initial 87Sr/86Sr > 0.707, Ford etal. 1986). It has been argued that the Dufek intrusion gabbros represent parental FMP magmas, unaffected by crustal contamination (Ford et aL 1986; Kyle et al. 1987), and thus the isotopic similarity of Group 3 rocks to the Dufek gabbros, suggests that this group represent the least contaminated FMP magma type within Coats Land. The range in eNd for Coats Land dolerites is more restricted (Fig. 10), but again is transitional between FMP and DML magma types. The correlation between Nd and Sr isotopes is such that the Thero~ Mountains dolerites form an array overlapping and between the FMP and DML compositions (Fig. 10), whereas, dolerites from Whichaway Nunataks have similar Sr but elevated Nd isotope compositions relative to FMP (Fig. 10). All of the analysed samples have Nd TI)M model ages falling in the range 12001000 Ma, which is similar to that previously reported for the Ferrar dolerites from the TAM (Menzies et al. 1985; McGibbon et al. 1987; Hergt et al. 1989a). Sr and Nd isotope ratios for the Coats Land dolerites do not show any significant correlation with the degree of fractionation (Fig. i l l This suggests that crustal A F C processes (DePaolo 1981) did not have an important role in the Petrogenesis of the tholeiites, although one exception might be the more evolved rocks within Group 2.

1J.sJ

l& 4 ~

15.3J 17.5

i 18.0

i 18.5

| 19.0

i

i 19.5

,

i 20.0

i

Fig. 12. Variation of 2°6pb/2°4pbversus 2°Tpb/2°4pband Z°sPb/Z°4pbfor Coats Land magma types. Data sources: Ferrar, Kyle etal. (1987); Hergt et al. (1989a, b); Kyle & Pankhurst unpublished data; Dufek (Brewer, unpublished data); Dronning Maud Land (Brewer, unpublished data). NHRL: Northern Hemisphere Reference Line after Hart (1984). Symbols as for Fig. 9.

suggest involvement of an old radiogenic component (> 1 Ga). Moreover, this component was probably present in the magma source regions as it can be identified in all compositions irrespective of the degree of fractionation. The initial 87Sr/S6Srratios show a relatively large range ~0.705-0.714) for a very limited range of 2°Tpbi '~"~Pb values (18.0-19.0). All the rocks analysed have radiogenic isotope ratios displaced from typical mantle values. The lack of correlation between the Sr and Pb isotopes may indicate that contamination by upper crustal (positive Sr-Pb isotope correlation, Hawkesworth etal. 1986) or lower crustal (negative Sr-Pb isotope correlation, Church 1985) material, was not an important late-stage process.

Discussion

Pb isotopes

Origin o f the g e o c h e m i c a l signatures

All the Coats Land tholeiites plot above the Northern Hemisphere Reference Line (NHRL) and fall between the fields for DML and FMP rocks (Fig. 12). Series 1 and 2 magmas overlap, and the relatively elevated 2°Tpb/2°4pb ratios

Spatial geochemical variations observed within the igneous rocks of the Antarctic Mesozoic Province have led to the development of different petrogenetic models for DML and FMP. The unusual geochemistry of the FMP has been

202

T.S. BREWER E T A L .

explained by models which can broadly be divided into those requiring some form of crustal contamination, and those invoking modification of their upper mantle source regions (Compston et al. 1968; Faure et al. 1972, 1974; Pankhurst 1977; Kyle 1980; Kyle etal. 1983, 1987; Mensing et al. 1984; Harris et al. 1990; Hergt et al. 1989b, 1991). The following discussion summarizes the models for crustal contamination and mantle source modification and outlines a model which involves a contribution from the lithospheric mantle in the generation of these Antarctic CFB. Crustal contamination

The FMP extends for some 3000 km, within which the compositions are remarkably homogenous (Kyle 1980; Hergt et al. 1991). If crustal assimilation processes were responsible for the distinctive geochemistry of the FMP, the effects were strikingly similar along the length of the FMP. Hergt et al. (1991) further argued that in order for FMP signatures to have been generated by crustal contamination processes it was necessary that:

veloping the geochemical signatures of the Kirwan basalts, and contamination, if it occurred, was limited to < 7%. In Coats Land, a lack of correlation between isotope ratios and major and trace elements would suggest that AFC processes were not important in the generation of the bulk of the magmas, although, as noted above, one exception may be the evolved compositions of the Group 2 dolerites. If contamination processes within the continental crust were not responsible for the distinctive FMP signature, the most obvious alternative is that they were derived from the lithospheric mantle. However, the production of significant volumes of basalt from the lithospheric mantle must be reconciled with the geophysical data which suggest that this is a cold infertile domain in which large melt volumes are not readily generated or extracted.

Role of the lithospheric mantle

Karoo picrites from the Nuanetsi area in southern Africa are 'high-Ti' CFB, with enriched radiogenic isotope compositions requiring at least a contribution from old source regions in (i) the parental magmas were extremely de- the upper mantle (e.g. Ellam & Cox 1991). Since such source regions are probably sited in the pleted in incompatible trace elements; (ii) the parental magmas had unusual major continental mantle lithosphere and melting calelement compositions, being different from culations at the dry peridotite solidus indicate N-MORB, but also having a number of fea- that such mantle is unlikely to melt significantly under reasonable tectonic conditions (McKentures of both picritic and boninitic melts; (iii) the crustal assimilant had high Rb/Ba, Rb/ zie & Bickle 1988), Ellam & Cox (1991) propSr, SiO2 and S7Sr/S6Sr, consistent with deri- osed that the Karoo picrite basalts resulted from the interaction of asthenospheric magmas and a vation from the middle to upper crust; (iv) the calculated primary magmas had Y, Ti component derived from the lithospheric manand Zr contents similar to those of depleted tle. The lithospheric mantle component was mantle peridotites, and the amounts of as- modelled as a small degree, lamproitic melt, and similation were large (25-30%), in order to the asthenosphere endmember was similar to explain the observed depletion of Ti rela- depleted MORB. To evaluate such a model for the Mesozoic tive to Zr and Y. CFB from Antarctica, Ti/Y ratios have been In the FMP, tholeiites with initial S7Sr/S6Sr plotted against end in Fig. 13. In this diagram, ratios of 0.710, have 81So of +6%o (Mensing et most asthenosphere-derived magmas fall in a al. 1984; Hergt et al. 1989a); such oxygen values broad array between average MORB and OIB would be difficult to maintain after 25--30% mid- values. The sediment value in Fig. 13 is taken to dle to upper crust assimilation. To compound be representative of continental material substill further the problems with the crustal con- ducted back into the upper mantle along detamination model there are questions of both the structive plate margins, and it may therefore be areal extent of the province and the production regarded as an estimate of sediment-contaminof sufficient volumes of high degree mantle melts ated lithospheric mantle. If such lithospheric (i.e. the parental magmas) over such a large mantle was subsequently melted, the resultant area. It is therefore regarded as unlikely that basalts would have low Ti/Y ratios and negative crustal contamination processes were responsieNd values. ble for the development of the key features of The DML rocks exhibit a positive correlation the distinctive FMP signature. between Ti/Y and eNd (Fig. 13). This is opposite In DML, Harris et al. (1990) demonstrated of what would be expected if these rocks had been derived from old source regions which had that AFC processes were not significant in de-

ANTARCTIC FLOOD BASALTS

o'l

7

a

FL~

=° ' "

"

+ +

%~

~ ~tament 10o

2o0

3oo

l~.mral~ 4oo Ti/Y

50o

60o

7oo

Fig. 13. Variation of Ti/Y versus end for the Antarctic Mesozoic Province. Lamproite values from Bergman (1987), Ellam & Cox (1991). Sediment from Taylor & McClennan (1985); BE, Bulk Earth value from Sun & McDonough (1987). Ferrar field from Hergt et al. (1989a, b), Pankhurst unpublished data. Dronning Maud Land field from Harris et al. (1990), Brewer unpublished data. Arrow indicates the effect of mixing small amounts of lamproite with asthenosphere (Bulk Earth). Symbols as for Fig. 9.

variable Ti/Y and Sm/Nd, since with time that would result in a negative correlation between Ti/Y and Nd isotopes. Instead, the observed trend for DML rocks is consistent with the mixing of a high Ti/Y, high end component (similar to many OIBs), and a low Ti/Y, low eNd component presumably derived from the continental lithosphere. The FMP rocks all have low Ti/Y and negative end values, similar to those suggested for the likely sediment component (Fig. 13), and indicating a major contribution from lithospheric mantle in the generation of these magmas. Finally, the Coats Land dolerites have Ti/Y and Nd isotope ratios which overlap and are transitional between those of the DML and FMP fields. This may suggest that in Coats Land both the asthenosphere and lithospheric mantle have contributed in varying proportions to the petrogenesis of the different magma types. If the lithospheric mantle is regarded as a potential source for large volumes of basalt, two fundamental questions must be addressed. First, how did partial melting take place in the lithospheric mantle, and secondly, what controls the degree of asthenospheric involvement?

intrusion, Himmelberg & Ford 1976), these ratios may be used in the following discussion of petrogenetic models. On Ti/Y versus Zr/Y plots, a broad mantle array can be identified from the values of MORB and OIB (Fig. 14). Variations along the mantle array reflects variations in the degree of partial melting and/or different Ti/Y and Zr/Y values in different source regions. It is inferred that partial melting of old segments of mantle lithosphere in which the minor and trace element ratios were themselves controlled by the extraction or introduction of small degree melts, will result in magmas with broadly similar Ti/Zr. However, if the minor and trace element inventory of segments of the mantle lithosphere are dominated by the introduction of a sedimentary component, their Ti/Y and Zr/Y ratios will be displaced towards that of the Post Archean Shale Composite (Taylor & McLennan 1985; Fig. 14). If the generation of basalts involved mixing of material from the asthenosphere and from mantle lithosphere containing a contribution from subducted sediment, the resultant trends should be displaced from the MORBOIB array towards the composition for subducted sediment. Moreover, the intersection of such mixing trends with the MORB-OIB array should constrain the nature of the asthenospheric component. The majority of the DML magmas fall on or near the mantle array in Fig. 14, and they have relatively low initial aTSr/a6Sr ratios (Fig. 5). However, they have some distinctive trace element signatures and the positive correlation between Nd isotopes and Ti/Y requires a signi700

580 --

In basaltic melts Ti/Y and Zr/Y ratios are both little effected by low pressure fractionation processes, in the absence of significant Ti-magnetite fractionation. As Ti-magneti'te is not a major fractionating phase in Coats Land, DML and the majority of the FMP magmas (except the Dufek

O Bulk Eorth X N-MORB A E-MORB

ri / Y

I .t .I / f//

00IB + PAS

""~) / ./f

DML n

. , . / ~

I " ' ~

,.. /

-

l

e'~,>'~

3&0-~ 220 100

Lithosphere melting: an Antarctic perspective

203

0

_

^ ,c

) i> /

5e ies

-1- ~

uroupl --~L - - ~ : = _ _ ' - - - - , Group2 - - - - ~ ; ~ - - " " " I Ferror I 2.0 /,.0

II

_ "II 6"0

I 8.0

10.0

Zr/Y

Fig. 14. Variation of Zr/Y versus Ti/Y ratio within the Antarctica Mesozoic Province, for clarity only fields shown. Data sources, FMP; Siders & Elliot (1985), Hergt et al. (1989a, b), Pankhurst unpublished data, DMLP; Fumes et aL (1987), Brewer & Brook (]990), Harris et aL (1990), Brewer unpublished data. PAS represents the Post Archean Shale Composite from Taylor & McLennan (1985).

204

T.S. BREWER E T A L .

ficant contribution from old, low Sm/Nd source regions, presumably in the continental lithosphere. Discussion of the origin of the FMP is hampered by the shortage of complete published geochemical data sets, but the magmatism represents an extreme form (depleted) of the low-Ti Gondwana CFB (Hergt et al. 1991). On Fig. 14 Group 2 rocks, for example, exhibit a flat-lying trend that could be interpreted in terms of mixing between a sedimentary component, inferred to be in the continental lithosphere, and an asthenospheric component from an extension of the MORB-OIB array. However, that requires the asthenospheric component to have unrealistically low Ti/Y and Zr/Y ratios relative to those observed in most oceanic basalts. An alternative interpretation is that the magmas, such as those in Group 2, formed by melting of the lithospheric mantle without a significant asthenospheric component. In this model, the distinctive minor and trace element signature was stabilised in the lithospheric mantle during earlier subduction episodes, and was remelted in the Mesozoic CFB event. Gallagher & Hawkesworth (1992) and Hawkesworth et al. (this volume) discuss models in which CFBs in some areas may be generated from within the continental mantle lithosphere by partial melting of hydrous peridotite. The important parameters in such a melting model are mantle potential temperature, peridotite solidus temperature, thickness of the mechanical boundary later (MBL), taken here to be equivalent to the lithospheric mantle, and amount of crustal extension. For < 20% crustal extension in the presence of a mantle plume, significant volume of melt may be produced by dehydration melting within the mantle lithosphere, and without a significant contribution from the underlying asthenosphere. At larger amounts of extension partial melts from the upwelling asthenosphere rapidly dominate the resultant magmas. The composition of hybrid magmas containing material from both the lithosphere and the underlying asthenosphere is primarily a function of the amount of crustal extension. Most calculated geotherms indicate that partial melting may take place in hydrous peridotite even without the introduction of a mantle plume (e.g. Gallagher & Hawkesworth 1992). Thus melting within the MBL may take place in response to extension and relatively small increases in temperature that occur in, for example, behind-the-arc tectonic settings. Within the FMP, it is very diffficult to obtain quantitative estimates for any associated crustal extension, but Jurassic plate tectonic reconstructions of Antarctica suggest that rift basins developed be-

tween the TAM and the adjacent college of microplates (Elliot 1990; Storey 1990; Storey & Alabaster 1991; Grunow et al. 1991). These rift basins, some of which were synchronous with magmatism, were produced by limited amounts of crustal extension that some authors have linked to subduction along the Proto-Pacific margin (Elliot 1975, 1990; Cox 1978). It would therefore appear that the FMP magmatism developed in a region undergoing limited crustal extension, consistent with the geochemical arguments that significant volumes of melt were generated within the continental mantle lithosphere. In Coats Land, the situation is complicated by the presence of two magmatic episodes. The older Group 2 dolerites (193+7 Ma) have FMP signatures, and defined a linear trend towards the sediment value, which overlaps the FMP field in Fig. 14. The extrapolation of this trend onto the mantle array gives a similar value to the FMP, i.e. an unrealistic (extremely depleted) asthenospheric component. This early magmatic episode marks the initiation of rift related magmatism in Coats Land, during which crustal extension was limited and most of the melt was derived from the lithosphere by dehydration melting, in a similar manner to that inferred for FMP magmatism. In constrast, all of the 176+5 Ma dolerites have lower Sr isotope and higher trace element ratios relative to FMP (Fig. 5, 10 & 14). Although the 176+5 Ma dolerites have elevated trace element ratios relative to FMP, extrapolation of the trends in Figure 14 onto the mantle array still produces extremely depleted values. This younger magmatism correlates with the latter stages of rifting and relatively enhanced crustal extension, which may have allowed for a greater asthenospheric component in these magmas. The actual proportions of lithospheric mantle relative to asthenospheric component cannot as yet be quantified. It would, however, appear from Figure 14 that the majority of compositions are signficantly displaced from t h e mantle, suggesting a relatively small asthenospheric component.

Conclusions New geochemical and radiogenic isotope data indicate that Mesozoic tholeiitic magmatism of continental Antarctica can be usefully divided into the Ferrar and Dronning Maud Land Magmatic Provinces, and these appear to have been generated in different tectonic regimes. Tholeiites of the Ferrar are distinguished by low Ti/Y, Zr/Y and Nd isotope ratios and relatively high

ANTARCTIC FLOOD BASALTS LILE, SiO2 and STSr/86Sr. They are not readily attributed to crustal contamination of asthenosphere derived magmas, but rather they appear to have been derived by partial melting of continental mantle lithosphere which contained a significant component of subducted sediment (see also Hergt et al. 1991). The FMP is currently exposed in a long ( > 3000 km) linear belt, and in the preferred model magmatism was triggered by intracontinental extension associated with subduction along the proto-Pacific, rather than to the emplacement of a mantle plume. In contrast, DML tholeiites have high Ti/Y, Zr/Y and eNd values and lower 87Sr/~'Sr ratios than those from FMP. These tholeiites were generated in an area close to the inferred mantle plume (White & McKenzie 1989) and they contain a greater contribution from the asthenospheric upper mantle (Harris et al. 1990). The Coats Land region is situated at the margin of the Ferrar and DML magmatic Provinces. It is argued that the earliest magmatic episode (193+7 Ma) was related to the initial stages of rifting when crustal extension was limited and most melt was derived from the lithosphere by dehydration melting. The resultant magmas are thus indistinguishable from the FMP magmas which were formed by a similar process. Younger magmatism (176+_5 Ma) formed during the more advanced stages of rifting when the crust was more attenuated and allowed for the incorporation of a small asthenospheric component into melts. The amount of asthenosphere derived magma appears to have been less than that in DML and so these basalts are transitional between the D M L and FMP rocks. A somewhat similar model has been proposed by Storey & Alabaster (1991) for the Ferrar signatures identified in South Georgia and the northern Antarctica Peninsula. In South Georgia the initial rift magmas (low extension) have Ferrar signatures but with time the early drift and main drift magmas have asthenospheric characteristics, which again demonstrates that the controlling factor for the geochemistry is related to the degree of crustal extension.

Rare earth element and isotopic analytical work was carried out at RHBNC London and the Open University. This paper has benefited from discussions with P. D. Clarkson (SCAR), R. J. Pankhurst (BAS) and P. E. Baker (University of Leeds). P. R. Kyle and R. J. Pankhurst are thanked for use of unpublished trace element and isotope data from the Transantarctie Mountains. The paper has also benefited from reviews by C. Harris and P. R. Kyle.

205

References BARREI'r, P. J., ELLIOT,D. H. & LINDSAY,J. F. 1986. The Beacon Supergroup (Devonian-Triassic) and Ferrar Group (Jurassic) in the Beardmore Glacier area, Antarctica. In: TURNER, M. D. & SPLETTSTOESSER,J. F. (eds) Geology of the Central Transantarctic Mountains. Antarctic Research series, American Geophysical Union, Washington DC, 36, 339-428. BELLIENI, G., BROTZU, P., COMIN-CHIARAMONTI,P., ERNESTO, M., MELFI,A. J., PACCA,1. G. & PICCmILLO, E. M. 1984. Flood basalt to rhyolite suites in the southern Parana Plateau (Brazil): paleomagnetism, petrogenesis and geodynamic implications. Journal of Petrology, 25, 579-618. BERGMAN, S. C. 1987. Lamporites and other potassium-rich igneous rocks: a review of their occurrence, mineralogy and geochemistry. In: Fn-roN, J. G. & UFroN, B. G. J. (eds) Alkaline Igneous Rocks. Geological Society, London, Special Publication, 30, 103-~190. BORG, S. G., DEPAOLO,D. J. & SMITH,B. M. 1990. Isotopic structure and tectonics of the central Transantarctic Mountains. Journal of Geophysical Research, 95, 6647-6667. BROO):, D. 1972. Geology of the Theron Mountains, Antarctica. PhD thesis, University of Birmingham. BREWER,T. S. 1990. Mesozoic dolerites from Whichaway Nunataks. Antarctic Science, 1, 151-155. & BRoog, D. 1990. The geochemistry of Mesozoic tholeiites from Coats Land and Dronning Maud Land. In: THOMSOr~, M. R. A., CRAME, J. A. & THOMSON,J. W. (eds) Geological Evolution of Antartica. Cambridge University Press, Cambridge, 569-572. , REX, D. C. & STOREY,B. C. 1991. Ar-Ar dating of Mesozoic tholeiitic magmatism from Coats Land and the Pensacola Mountains, Antarctica. Terra Abstracts, 2, 1. CnoRcn, S. G. 1985. Genetic interpretation of leadisotope data from the Columbia River Basalt Group, Oregon, Washington and Idaho. Geological Society of America Bulletin, 96, 676-690. CLARgSON, P. D. 1972. Geology of the Shackleton Range: a preliminary report. Bulletin of the British Antarctic Survey, 51,257-283. 1981. Geology of the Shackleton Range: IV. The dolerite dykes. Bulletin of the British Antarctic Survey, 53, 201-212. COMPSTON,W., McDOUGALL,I. & HEIER, K. S. 1968. Geochemical comparison of the Mesozoic basaltic rocks of Antarctica, South Africa, South America and Tasmania. Geochimica et Cosmochimica Acta, 32, 129-149. Cox, K. G. 1978. Flood basalts, subduction and the break-up of Gondwanaland. Nature, 274, 47-49. 1988. The Karoo Province. In: MACDOUrALL,J. D. (ed.) Continental Flood Basalts. Kluwer, Dordrecht, 239-271. & HAW~:ESWORTH,C. J. 1984. Relative contribution of crust and mantle to flood basalt magmatism, Mahableshwar area, Deccan Traps. Philosophical Transactions of the Royal Society

T.S. BREWER E T A L .

206

of London, A310, 627-641. & 1985. Geochemical stratigraphy of the Deccan Traps at Mahabaleshwar, Western Ghats, India, with implications for open system magmatic processes. Journal of Petrology, 26, 355-377. DEPALO, D. J. 1981. Trace element and isotopic effects of combined wall rock assimilation and fractional crystallisation. Earth and Planetary Science Letters, 53, 189-202. ELI_AM,R. M. & Cox, K. G. 1991. An interpretation of Karoo picrite basaits in terms of interaction between asthenospheric and mantle lithosphere. Earth and Planetary Science Letters, 105, 330-342. ELLIOT, n . H. 1975. Tectonics of Antarctica: a review. American Journal of Science, 275-A, 45-106. 1990. Triassic-Early Cretaceous evolution of Antarctics. In: THOMSON,M. R. A., CRAME,J. A. & THOMSON, J. W. (eds) Geological Evolution of Antarctica. Cambridge University Press, Cambridge, 541-548. ERLANK, A. J. (ed.) 1984. Petrogenesis of the volcanic rocks of the Karoo Province. Special Publication of the Geological Society of South Africa, 13. , DUNCAN, A. R. MARSHJ. S., SWEENEYR. J., HAWKESWORTH,C. J., MILNER,S. C., MILLER,R. McG. & ROGERS,N. W. 1988. A laterally extensive geochemical discontinuity in the subcontinental Gondwana Lithosphere. In: Geochemical -

evolution of the Continental Crust, Conference Abstracts, Pocos de Caldas, Brazil, 1-10. FAURE,G., HILL, R. L., JONES,L. M. & ELLIOT,D. H. 1972. Isotope composition of strontium and silica contents of Mesozoic basalt and Dolerite from Antartica. In: ADIE, R. J. (ed.) Antarctic Geology and Geophysics. Universtetforlaget, Oslo, 617-624. , BOWMAN,J. R. & ELLIOT,D. H. 1979. The initial 87Sr/S6Sr ratios of the Kirwan volcanics of Dronning Maud Land: comparison with Kirkpatrick Basalts, Transantarctic Mountains. Chemical Geology, 26, 77-90. ~Z. JONES,L. M. 1974. Strontium isotope composition and petrogenesis of the Kirkpatrick Basalt, Queen Alexandra Range, Antarctica. Contributions to Mineralogy and Petrology, 48, 153-169. FORD, A. B. & KISTLER,R. W. 1980. K-Ar age, composition and origin of Mesozoic mafic rocks related to the Ferrar Group, Pensacola Mountains, Antarctica. New Zealand Journal of Geology and Geophysics, 23, 371-390. FURNES,H., MAD,E., AUSTRHEIM,H., MITCHLL,J. G. GARMAN, L. B. 1987. Geochemistry of basalt lavas from Vestfjella and adjacent areas, Dronning Maud Land, Antarctica. Lithos, 20, 337-356. GALLAGHER, K. & HAWKESWORTH,C. J. 1992. Dehydration melting and the generation of continental flood basalts. Nature, 358, 57-59. GILL, J. B. 1981. Orogenic andesites and plate tectonics. Springer-Verlag, New York. GROENEWALD,P. B., GRANTHAM,G. H. & WATKEYS, ,

,

-

-

M. K. 1991. Geologmai evidence for a Proterozoic to Mesozoic link between southeastern Africa and Dronning Maud Land, Antarctica. Journal of the Geological Society, London, 148, 1115-1124. GRUNOW, A. M., KENT, D. V. & DALZIEL, I. W. D. 1991. New paleomagnetic data from Thurston Island and their implication for the tectonics of West Antarctica. Journal of Geophysical Research, 96, DII, 17935-17954. Gumq, B. M. 1966. Modal and element variation in Antarctic tholeiites. Geochimica et Cosmochimica Acta, 30, 881-920. • WARREN, G. 1962. Geology of Victoria Land between the Mawson and Mulock Glaciers, Antarctica. Bulletin of the New Zealand Geological Survey, 71, 1-157. HARRIS, C., MARSH,J. S., DUNCAN,A. R. & ERLANK, A. J. 1990. The petrogenesis of the Kirwan Basalts of Dronning Maud Land, Antarctica. Journal of Petrology, 31,341-369. , WATrERS, B. R. & GROENEWALD,P. B. 1991. Geochemistry of the Mesozoic regional basic dykes of western Dronning Maud Land, Antarctica. Contributions of Mineralogy and Petrology, 107, 100-111. HART, S. R. 1984. The DUPAL anomaly: a large-scale isotopic anomaly in the southern hemisphere. Nature, 309, 753-756. HARVEY,P. K. & ATgIN, B. P. 1982. Automated X-ray Fluorescence analysis. In: Sampling and Analysis for the Minerals Industry. Special Publication of the Institute of Mining and Metallurgy, 17-26. HAWKESWORTH, C. J., MANTOVANI, M. S. M., TAYLOR, P. N. & TAYLOR,P. N. 1986. Evidence from the Parana of south Brazil for a continental contribution to Dupal basalts. Nature, 322, 356-359. - - , GALLAGHER,K., KELLEY,S., MANTOVANI,M., PEATE, D. W., REGELOUS,M. & ROGERS,N. W. 1992. Paran~i magmatism and the opening of the South Atlantic. This Volume. HERGT, J. M., CHAPPELL,G., FAURE, G. & MENSING, T. M. 1989b. The geochemistry of Jurassic dolerites from Portal Peak Antarctica. Contributions to Mineralogy and Petrology, 102,298-305. - - , MCCULLOCH,M. T., McDOUGALL, I. & CHIVAS,A. R. 1989a. Geochemical and isotopic constraints on the origin of the Jurassic dolerites of Tasmania. Journal of Petrology, 30, 841-883. , PEATE,D. W. & HAWKESWORTH,C. J. 1991. The petrogenesis of Mesozoic Gondwana low-Ti flood basalts. Earth and Planetary Science Letters, 105, 134-148. HINZ, K. & KRAUSE,W. 1982. The continental margin of Queen Maud Land/Antarctica: seismic sequences, structural elements and geological development. Geologisches Jarbuch, E23, 17-41. HIMMELBERG,G. R. & FORD,A. B. 1976. Pyroxenes of the Dufek Intrusion, Antarctica. Journal of Petrology, 17, 219-243. HOFMANN,J., KASIER,G., KLEMM,W. & PAECH, H-J. 1980. K/Ar-Alter von Doleriten und Metamor-

ANTARCTIC FLOOD BASALTS phiten der Shackleton Range und der Whichaway Nunataks, Ost-und Sudostumrandung des Filchner-Eisschelfs (Antarkis). Zeitschrift fur Geologische Wissenschafen Berlin, 8,1227-1232. JucgES, L. M. 1968. The geology of Mannefalknausane and part of Vestfjella, Dronning Maud Land. British Antarctic Survey Bulletin, 18, 65-78. KmSTO~RSEN, Y. & HINZ, K. 1990. Evolution of the Gondwana plate boundary in the Weddell Sea area. In: THOMSON,M. R. A., CRA~, J. A. & THOMSON, J. W. (eds) Geological Evolution of Antarctica. Cambridge University Press, Cambridge, 225-230. KVLE, P. R. 1980. Development of heterogeneities in the sub-continental mantle: evidence from the Ferrar Group, Antarctica. Contributions to Mineralogy and Petrology, 73, 89-104. ~, ELLIOT, D. H. & SUYrER, J. F. 1981. Jurassic Ferrar Supergroup tholeiites from the Transantarctic Mountains, Antarctica, and their relation to the initial fragmentation of Gondwana. In: CRESSWELL,M. M. & VELLA,P. (eds) Gondwana

Five, Proceedings of the Fifth International Gondwana Symposium, Wellington New Zealand. A. A. Balkema, Rotterdam, 283-287. PANKHURST, R. J. & BOWMAN, J. R. 1983. Isotopic and chemical variations in Kirkpatrick Basalt Group rocks from southern Victoria Land. In: OLIWR, R. L., JAMES,P. R. & JAGO,J. B. (eds) Antarctic Earth Science. Australian Academy of Science, Canberra and Cambridge University Press, Cambridge, 234-237. , PANKHURST,R. J., MOORBATH,S. & BOWMAN,J. 1987. Nature and development of an enriched mantle source for Jurassic Ferrar tholeiites, Transantarctic Mountains, Antarctica. Terra Cognita, 7, 614. LAWVER, L. A., SCLATER,J. G. & MEINKE, L. 1985. Mesozoic and Cenozoic reconstructions of the south Atlantic. Tectonophysics, 114, 233-254. MARSH, P. D. 1986. Ice surface and bedrock topography in Coats Land and part of Dronning Maud Laud, Antarctica, from satellite imagery. Bulletin of the British Antarctic Survey, 68, 19-36. MARTIN, A. K. & HARTNADY,C. J. H. 1986. Plate tectonic development of the South Indian Ocean: A revised reconstruction of East Antarctica and Africa. Journal of Geophysical Research, 91, 4767-4786. MATOVANI,M. S. M., MARQUES,L. S., SOUSA,M. A. de., CIVETTA,L. ATALLA,L. T. & INNOCENTI,F. 1985. Trace element and strontium isotope constraints on the origin and evolution of Parana continental flood basalts of Santa Catarina State (Southern Brazil). Journal of Petrology, 26, 187-209. MCDONOUGH. W. F. 1990. Constraints on the composition of the continental lithospheric mantle. Earth and Planetary Science Letters, 101, 1-18. MCGIBBON, F. M., HAWKESWORTH,C. J., MENZIES, M. A. & KYLE, P. R. 1987. Geochemistry of ultramafic xenoliths from McMurdo Sound, Antarctica: implication of Jurassic Ferrar magmatism. Terra Cognita, 7, 396.

~,

207

MCKENZIE, D. P. & BICKLE,M. S. 1988. The volume and composition of melt generated by extension of the lithosphere. Journal of Petrology, 29, 625-679. MENSINGT. M., FAURE,G., JONES,L. M., BOWMAN,J. R. & HOEFS,J. 1984. Petrogenesis of the Kirkpatrick Basalt, Solo Nunatak northern Victoria Land, Antarctica, based on isotopic compositions of strontium, oxygen and sulfur. Contributions to Mineralogy and Petrology, 87, 101-108. MENZIES, M. A., KYLE, P. R. & GAMBLE, J. 1985. Fragments of aged enriched sub-continental lithosphere entrained in asthenospheric alkaline magmas erupted at Foster Crater, Antarctica.

EOS Transactions of the American Geophysical Union, 66, 409. MORGAN, W. J. 1981. Hotspot tracks and the opening of the Atlantic and Indian Oceans. In: EMILIANI, C. (ed.) The Oceanic Lithosphere. John Wiley, New York, 443-487. PANKHURST, R. J. 1977. Strontium isotope evidence for mantle events in the continental lithosphere. Journal of the Geological Society, London, 134, 255-268. PEATE, D. W., HAWKESWORTH,C. J., MANTOVANI,M. S. M. & SHUKOWSK£,W. 1991. Mantle plumes and flood basalt stratigraphy in the Parana, South America. Geology, 18, 1223-1226. PLUMS'rEAD, E. P. 1962. Geology: 2. Fossil floras of Antarctica (with an Appendix on Antarctic fossil wood by R. Krausel). Trans-Antarctic Expedition Scientific Reports, 9. REX, D. 1972. K-Ar age determinations on volcanic and associated rocks from the Antarctic Peninsula and Dronning Maud Land. In: ADm, R. J. (ed.) Antarctic Geology Geophysics. Universtetforlaget, Oslo, 133-136. RICHARDS,M. A., DUNCAN,R. A. & COURTILL,V. E. 1989. Flood basalts and hot-spot tracks: plume heads and tails. Science, 246, 103-107. SIDERS, M. A. & ELLIOT,D. H. 1985. Major and trace element geochemistry of the Kirkpatrick Basalt, Mesa Range, Antarctica. Earth and Planetary Science Letters, 72, 54-64. STEPHENSON,P. J. 1966. Geology: 1. Theron Moun-

tains, Shackleton Range and Whichaway Nunataks (with a section on palaeomagnetism of the dolerite intrusions by D. J. Blundell). TramAntarctic Expedition Scientific Reports, 8. STOREY,B. C. 1990. The crustal blocks of West Antarctica within Gondwana: reconstruction and break-up model. In: THOMPSON, M. R. A., CRAME,J. A. & THOMSON,J. W. (eds) Geological Evolution of Antarctica. Cambridge University Press, Cambridge, 587-592. & ALABASTER,T. 1991. Tectonomagmatic controls on Gondwana break-up models: evidence from the Proto-Pacific margin of Antarctica. Tectonics, 10, 1274-1288. ~ - , HOLE, M. J., PANKHURST,R. J., MILLAR,I. L. & VENNUM, W. 1988. Middle Jurassic within-plate granites in west Antarctica and their bearing on the break-up of Gondwanaland. Journal of the Geological Society, London, 145,999-1007.

208

T. S. BREWER E T A L .

SUN, S. S. & MCDONOUOH,W. F. 1989. Chemical and isotopic systematics of oceanic basalts: implications for mantle composition and processes. In: SAUNDERS, A. D. & NORRV M. J. (eds). Magmatism in Ocean Basins. Geological Society, London, Special Publication, 42, 313-345. TAYLOR, S. R. & McLENNAN, S. M. 1985. The continental crust: its composition and evolution. Blackwell Scientific Publications. WALSH, J. N., BUCKLEY,F. & BARKER,J. 1981. The simultaneous determination of the Rare-earth elements in rocks using inductively coupled

plasma spectrometry. Chemical Geology, 33, 141-153. WEAVER, B. L., WOOD, D. A., TARNEY,J. & JORON, J. L. 1986. Role of subducted sediment in the genesis of ocean-island basalts: Geochemical evidence from South Atlantic Ocean Islands. Geology, 14, 275-278. WHITE, R. S. & MCKENZIE, D. P. 1989. Magmatism at rift zones: The generation of volcanic continental margins and flood basalts. Journal of Geophysical Research, 94, 7685-7729.

The granites of northern Patagonia and the Gastre Fault System in relation to the break-up of Gondwana C. W. R A P E L A 1 & R. J. P A N K H U R S T 2

1Centro de Investigaciones Geol6gicas, L a Plata University, Calle 1 N o 644, 1900 La Plata, Argentina 2British Antarctic Survey, c/o N E R C Isotope Geosciences Laboratory, Keyworth Nottingham NG12 5GG, U K

Abstract: The transcurrent Gastre Fault System in central Patagonia, which is closely associated with subvolcanic granite emplacement, is recognized as a major dextral shear-zone and geological boundary. We propose its equivalence to a Late Triassic-Jurassic precursor of the Aghulas Fracture Zone, allowing dextral displacement of the Southern Patagonian Block relative to the rest of South America during the earliest rifting phase of Gondwana break-up. This model could explain some of the inferred movement of the Falkland/ Malvinas Islands and alleviate geometrical problems inherent in reconstructions of the South Atlantic region. It can also explain unique geological features of southern Patagonia, such as the Upper Triassic to Lower Jurassic calc-alkaline granitoids of the North Patagonian Massif and the extensive silicic volcanism of Mid-Late Jurassic times. The magmatism is seen as a consequence of the mechanism of Gondwana disintegration and it is not necessary to invoke a relationship to deep mantle structure or plume activity.

The igneous rocks of Patagonia extend from Palaeozoic to Recent in age and are vital to an understanding of the geological evolution of southern South America. From early Cretaceous times, the focus of activity has mostly been within the Patagonian Batholith of the western seaboard, interpreted as an expression of Pacific ocean floor subduction. It is the origin and significance of the earlier rocks, notably the granitoids of the North Patagonian Massif and the predominantly silicic volcanic rocks that occur throughout Patagonia that concern us here, particularly in relation to the history of Gondwana break-up. Since their tectonic context is important, we begin with some critical observations on the presumed regional situation in Jurassic times.

The South Atlantic space problem Although the general form of Gondwana is now generally accepted, the precise fit of the South Atlantic region and West Antarctica continues to be problematical. Figure 1 shows part of the Jurassic reconstruction of L a w v e r & Scotese (1987), in which the interior of the supercontinent contains gaps (south and east of the Malvinas/Falkland Plateau) unfilled by the available continental fragments. Other authors (Barron & Harrison 1978; Miller 1983; Grunow et al. 1987, 1991; de Wit et al. 1988; Lawver et al. 1991;

Storey 1991), produced significantly differing concepts for the South Atlantic fit, according to the weight ascribed to the available palaeomagnetic evidence and/or geological reasoning. Of these, some treated Antarctica as a single rigid plate, and only those of Grunow et al. (1991) and Lawver et al. (i991) combine a databased evaluation of micro-plates together with computer-fitting. Nevertheless, in all of these treatments some form of 'excess space' problem persists: it is usually explained as resulting from tectonic processes during or following break-up that may have distorted the original shape of the continental masses. In fact, the operation of such processes appears to be inescapable if major movement and rotation of the Falkland/ Malvinas Islands relative to Africa and South America is accepted (Mitchell et al. 1986; Taylor & Shaw 1989), since this requires that a significant portion of the Malvinas/Falkland Plateau shown in Fig. I did not exist as rigid continental crust in its present form within Gondwana. Some authors have attempted to alleviate this problem by suggesting dextral strike-slip movement within the South American plate (Urien et al. 1976; Unternher et al. 1988) but, until now, field evidence from South America has been too scarce to support the large-magnitude shear across the continent required by these hypotheses. We suggest, for the first time, that t h e most suitable candidate of this type is the Gastre

From STOREY,B. C., ALABASTER,T. & PANKHURST,R. J. (eds), 1992, Magmatism and the Causes of Continental Break-up, Geological Society Special Publication No. 68, pp. 209-220.

209

210

C.W. RAPELA & R. J. PANKHURST

Africa

East Antarctica

MFI

s,\ •

/

li

i'll. 1. Detail of previous Jurassic reconstruction of Gondwana (Lawver& Scotese 1987) indicating the marked coincidence of the traces of the Aghulas Fracture Zone (AFZ) and the Gastre Fault System (GFS). MFI, Malvinas/Falkland Islands; AP, Antarctic Peninsula; EWM, Ellsworth-Whitmore mountains crustal block; HN, Haag Nunataks crustal block. The shaded area is unfilled ocean floor of the South Atlantic Ocean and the Weddell Sea. Fault System (GFS) of north-central Patagonia (Fig. 1). As demonstrated here, this may be regarded as the precursor of the Aghulas Fracture Zone which is known to have been a controlling structural feature during the break-up of Gondwana (e.g. Rabinowitz & LaBrecque 1979). The GFS divides cratonic-based crust to the north from the relatively thin continental crust of southern Patagonia, the Falkland/Malvinas Plateau and West Antarctica to the south. It controlled the emplacement and distribution of Triassic-Late Jurassic magmatism.

The Gastre Fault System as a major geological boundary The Gastre Fault System is a prominent N W - S E shear-zone (Coira et al. 1975), with individual faults inferred to occupy a zone at least 30 km wide, in northern Patagonia. It passes south of the North Patagonian Massif (a large-scale inlier of crystalline basement within later Mesozoic

and Cenozoic cover rocks), through an area of granitoid rocks previously thought to be Palaeozoic in age, but now recognized as the Late Triassic to Early Jurassic Batholith of Central Patagonia (Rapela & Kay 1988; R a p e l a et al. 1991, in press). The GFS is a remarkably significant geological boundary (Fig. 2), across which there are numerous major changes in geology. Its northwesterly projection runs towards the Chilean Lake Region at about 40°S, where NW-SE faults are numerous (Munizaga et al. 1988). Here it marks the northern limit of the MesozoicCenozoic Patagonian Batholith (between 40°S and 38°S the Andean Cordillera contain only scattered outcrops of Mesozoic intrusive rocks). It is possible that the western termination of the GFS may be offset some 120 km to the north by the Liquifie-Ofqui strike-slip fault zone, to intersect the Pacific coast at 38°S. This is just south of the Nahuelbuta Mountains, the southernmost exposure of late Palaeozoic rocks in the Southern Coastal Batholith (Herv6 et al. 1987). In this

PATAGONIAN GRANITES

211

et a l . 1987), and include deep-sea cherts and marie and ultramafic bodies (Frutos & Alfaro 1987). In the Coastal Cordillera of Chile north of the GFS, a Jurassic plutonic belt occurs to the west of the Andes (this is the basis for the wellknown eastwardmigration of McNutt e t a l . 1975; Pankhurst et al. 1988): to the south, in contrast, a Triassic-Jurassic plutonic belt is situated on

area there is also a significant change in the nature of the Palaeozoic basement rocks of the Coastal Cordillera (Herv6 et al. 1981; Herv6 1988): those to the north consist mainly of continent-derived turbidites overlain by unmetamorphosed Late Triassic continental sedimentary rocks, those to the south are low-grade, with evidence of Jurassic deformation (Davidson

,m, C

LEGEND ~

(g

Late Jurassic-Tertiary bathoUths 1~ Patagonian bathoUth Jurassic volcanic rocks

MF: Marifil Group LT : Lonco Trapial Group CA: Chon -Aike (Tobifera) Group r ~ - ~ Bajo Pobre basalts

t~

C~-

~

Exotic~ terranes

Upper Palaeozoic-Middle Jurassic batholiths

(~) Southern Coastal

~

(~ Somuncura (~) Central Patagonia Upper Palaeozoic to Lower Mesozoic coastal metamorphic basement Eastern Series (north of 38" S)

~

Western Series (south of 38" S)

" t"

/i

tt

Fig. 2. Geological sketch map of southern South America showing the relationship between the Gastre Fault System and the main features of the pre-Cretaceous geology of Patagonia referred to in the text. Insets show details of the coastal region at 38°S and of the region around Gastre. N.B. The ages of the pre-Jurassic batholiths are undifferentiated.

212

C.W. RAPELA & R. J. PANKHURST

the eastern side of the Patagonian batholith (Cingolani et al. 1991), as is also the case in the Antarctic Peninsula (Pankhurst et al. 1988). Southeastwards from Gastre, the GFS disappeals beneath Upper Jurassic volcanic and younger sedimentary rocks, re-appeared on the Atlantic Coast at about 45°S where it is associated with Lower Jurassic volcanic rocks of the Marifil Group (see below).

The Gastre Fault System as a major shear zone

The structural characteristics of the GFS are best exposed along its central and eastern sectors in northern Patagonia, since farther to the west Andean magmatism and brittle tectonics have strongly overprinted any pre-Cretaceous deformation. The GFS was first recognized as a major fracture system by Coira et al. (1975), and has recently been studied in more detail in the Gastre area by Rapela et aL (1991). It forms a subparallel anastomosing system of faults about 40 km wide. Ductile shear zones and heterogeneously distributed outcrops of mylonitic rocks of up to 3 km wide and 7 km long are associated with the main faults, and sometimes contain porphyroclastic mylonites and foliated protomylonites. The mylonitic foliation is defined by alternating layers of chlorite, biotite, amphibole and opaque minerals, layers of recrystaUized quartz, and porphyroclasts of microcline and plagioclase. The mylonitic foliation is vertical with an average strike of N 130°, consistently parallel to the main faults. Cataclasites that appear in the outer parts of the mylonitic bodies grade into granitoid rocks displaying shear fractures with a similar orientation (N125-900). Asymmetric structures uniformly indicative of dextral shear were recognized in orientated thin sections. These include S--C fabrics, asymmetric plagioclase and K-feldspar porphyproclasts, and oblique foliation in quartz layers as well as asymmetric pressure shadows on feldspar (Rapela et al. 1991). Development of the mylonitic fabrics is closely associated geographically with the Triassic-Early Jurassic plutonic igneous rocks of the Batholith of Central Patagonia (BCP: Rapela et al. ~1992). The older intrusive rocks of the BCP (Gastre suite, 220+3 Ma) shows a moderate magmatic-flow foliation (Rapela et al. 1991). This rock is the protolith of mylonitic gneisses outcropping in the vicinity of the main faults, which exhibit a strong foliation defined by recrystallized elongate aggregates of chlorite, biotite, hornblende and quartz. The mylonitic areas

are also spatially associated with subvolcanic dykes of acid to intermediate composition, felsites, and small granitoid bodies of the younger suites of the BCP (mainly the Lipetr6n superunit). Felsites and aplite dykes appear throughout the batholith but are concentrated near the mylonites: some cut the foliation while others appear along the strike of the faults associated with the mylonites and show microscopic shear bands. These relations indicate that development of the solid-state mylonitic foliation overlapped in time with emplacement of the latest phases of the younger granitoid suites represented by the aplitic dyes, porphyries and felsites. Five samples of these subvolcanic rocks plot on the same well defined Rb-Sr WR isochron for the granitic units of the Lipetr6n suite (208+2 Ma). It seems that the 220 Ma foliated granitoids are related to the early stages of ductile deformation, whereas the 208 Ma subvolcanic rocks are affected by later-stage deformation. The GFS appears on the Atlantic coast at about 45°S as a myriad of faults associated with the Lower to Middle Jurassic volcanic rocks of the Marifil Group. Between 44°20'S and 45°15'S this group consists of rhyolites and associated dykes, ignimbrites and ash-flow tufts, affected by hundreds of vertical faults with a general strike N125-130 °. These faults commonly form furrows and ravines along the coast: they are all vertical, developing brecciation and cataclasites, and the major ones show interlayering between cataclastic material and acid volcanics. Shear fractures with the same orientation as the faults are developed in some places:extension fractures and dykes consistently strike N165 °. The dykes have compositions very similar to the rocks they intrude, with SiO2 ranging from 68 to 72%. We have obtained a Rb-Sr WR age of 178+1 Ma (N = 10, MSWD = 1.5, (STSr/ 86Sr)0 = 0.7067) for the rhyolitic dykes and host rhyolites at 45°03'S (Pankhurst, Rapela & Hailer, unpublished data), well within the range of K-Ar ages summarized by Cortrs (1981). A recent structural analysis of fluorite mineralization associated with the Marifil Group considered that it was controlled by a regional NW (N130-155 °) dextral shear system that generated N-S and NNE-SSE tensional fractures and hemigraben (Demichelis et al. 1991). Thus the dextral transcurrent system that controlled and affected the Marifil Group on the Atlantic coast is considered to be the easternmost part of the GFS. The age of the dykes and associated igneous rocks related to the GFS in the Gastre area and the Atlantic coast indicate a history dominated by dextral strike-slip in the

PATAGONIAN GRANITES interval 208-172 Ma (Early to lower-Middle Jurassic). This interval of deformation is coincident with that for the displacement of the Malvinas/Falkland Plateau (see below). Throughout northern Patagonia, Jurassic acidic volcanism is associated with NW-trending graben and half-graben that may have first formed in Triassic times (Uliana & Biddle 1987), perhaps overlapping with the plutonism of the BCP and the first movements on the GFS. The continued importance of the GFS in the geological evolution of central Patagonia is evident from the deposition of Upper CretaceousLower Tertiary continental sediments south of Gastre in intraplate NW-SE pull-apart basins (Spalletti et al. 1989), suggesting that strike-slip faulting may extend over a total width of > 50 km. There is no evidence of recent seismicity in the GFS.

Relationship to the Aghulas Fracture Zone The trace of the Aghulas Fracture Zone (AFZ) and its extension as the Falklands/Malvinas Fracture Zone, separates oceanic from continental crust (Rabinowitz & LaBrecque 1979). Computer analysis (kindly carried out by L. Gahagan of The University of Texas Institute of Geophysics) shows that a small circle based on the pole of first opening of the southern Atlantic Ocean is essentially continuous through the AFZ and the GFS, with a misfit of no more than about 5°C. This coincidence suggests that the transform fault (i.e. AFZ) originated along the trace of a previously active continental transcurrent fault (i.e. GFS), comprising a single system similar to those described for the early opening of the North Atlantic (Thomas 1988). We interpret the GFS as an intraplate boundary between a Southern Patagonian continental block (SPB) and the rest of South America including the North Patagonian Massif, along which dextral strike-slip movement occurred during the early rifting of Gondwana, in Late Triassic-Early Jurassic times (Fig. 3a). Although direct evidence for movement on the AFZ pertains to Cretaceous opening of the South Atlantic Ocean, there is a strong case for believing that a co-incident feature was present in earlier times. Following Le Pichon & Hayes (1971), Lorenzo & Mutter (1988) identify this as a marginal fracture ridge along the north side of the Malvinas/Falkland Plateau and the Maurice Ewing Bank, and conclude that it had been active as a strike-slip fault since 'at least the Middle Jurassic'. For convenience we refer to this as the Aghulas Fracture Ridge (AFR, Fig. 3a).

213

Magnitude of dextral displacement Within the South American continent there are no geological constraints on the magnitude of dextral movement on the GFS. Figure 3a illustrates our model for approximately 500 km dextral displacement of the SPB in Lower-MidJurassic times, as a working hypothesis. It is consistent with the importance attributed to the AFR by Lorenzo & Mutter (1988), along which they postulate at least 400 km of crustal extension within the Falkland/Malvinas Plateau. This hypothesis is highly relevant to the tectonics of continental rifting and drifting in the South Atlantic region. According to Lawver et al. (1991), separation of Madagascar and East Antarctica from Africa may have started as early as 170 Ma ago, although sea-floor was not formed between Africa and the MFP until M10 (c. 130 Ma ago; DNAG time-scale of Palmer 1983). Martin & Hartnady (1986) disputed a previous identification of Jurassic anomalies (M29) in the northeast Weddell Sea (LaBrecque & Barker 1981), which would be consistent with the hypothesis suggested here (see Fig. 3b). However, EarlyMid-Jurassic movement could have been accommodated by intra-crustal processes such as strike-slip deformation and crustal thinning prior to the production of oceanic floor (Ludwig 1983), as has been considered necessary to explain the proposed translation of the Malvinas/ Falkland Islands (Mitchell et al. 1986; Taylor & Shaw 1989). Taylor & Shaw (1989) argue that movement of the MFI must have postdated the 192+10 Ma dykes on which they obtained palaeomagnetic data and Lorenzo & Mutter (1988) claim that active faulting on the AFR affected only the lower part of the Jurassic sedimentary sequence: taking into account the association with the BCP and the Marifil Group noted above, such movements are apparently constrained to Early-Mid-Jurassic times.

Implications for the pre-Jurassic geology of Patagonia The suggested large dextral displacement strongly affects the palaeogeography of Patagonia prior to Middle-Upper Jurassic times (Fig. 3a). It requires a large-scale bend in the continental margin (similar to the Arica deflection of southern Peru). The northern flank of the GFS, from Gastre northwestwards, would have represented a Pacific coastal region in pre-Jurassic times, rather than the present intracontinental environment. We suggest that this hypothesis can explain two otherwise curious aspects of the pre-Jurassic geology of Patagonia: the marked

214

C.W. RAPELA & R. J. PANKHURST

a I

LATE TRIASSIC-EARLY JURASSIC

Africa

/ l~ \ "-,.,

--

l!

b i

MIDDLE JURASSIC - EARLY CRETACEOUS

,,*a5'--

~/\~" ~%~J

•,.~," r

/

/

',',~---~_~" oo

/

%q,.vf.~--~

arc

/ ~

1 Pacific o c e a n ~ crust

volcanics

A"~,~/! ! 1"~ EA -'"

~.,(

DN. /¢~.,of&~r.

L a t e Palaeozoic accretionary prim

j~~

Africa

\

Main arc

Trlmstensionel volcanism: M a r i f i l Group

"

*

1~

Termal focus under e x t e n d e d continent: . Chon- AikeGroup .

MFI Oolerite d y k e s

I uelorme(I ~ . Palaeozoic . ~ t t . accretionary prism ~ with slices of J u r a s s i c o c e a n crust

"

-

~ I

t ,

MFI

-j [

~

Pacific o c e a n c r u s t

Fig. 3. Sketches of western Gondwana, illustrating the model suggested in this paper, together with schematic cross-sections through the northern part of the SPB. AP, Antarctic Peninsula; EA, Eastern Antarctica; AFZ, Aghulas Fracture Zone; GFS, Gastre Fault System; J, position of disputed Jurassic sea-floor anomalies (LaBrecque & Barker 1981; Martin & Hartnady 1986); M, Moho; MFI, Falkland/Malvinas Islands; AFR, Aghulas Fracture Ridge (after Lorenzo & Mutter 1988); MFP, Malvinas/Falkland Plateau; SPB, Southern Patagonian Block; TI, Thurston Island; other abbreviations as in Fig. 1. The West Antarctic crustal blocks are shown essentially in the context of the 'mosaic Weddellia' model (Grunow et al. 1987), except that the pre-Jurassic rotation of the EWM is 70° rather than 90°: this is well within the limits allowed by the palaeomagnetic data. (a) Pre-break-up palaeogeography prior to dextral displacement. The proposed Late Triassic-Early Jurassic magmatic arc is recognized in the Batholith of Central Patagonia (BCP) and outcrops in the eastern Andean Cordillera as far south as 44°S in Patagonia (Franchi & Page 1980), and in the eastern part of the Antarctic Peninsula. The Late Triassic position of the MFI is considerably closer to that suggested by the palaeomagnetie data (Taylor & Shaw 1989). The distribution and ages of Karoo volcanic rocks are from Duncan (1987) and Marsh (1987). The dashed line in Africa is the inferred boundary between the northern ('within-plate') and the southern ('calc-alkaline and low-K tholeiite') provinces of Duncan (1987). The patterned field is the area of thinned continental lithosphere produced during the early Karoo event (.Marsh 1987). (b) Configuration after displacement of SPB, when it has over-ridden the site of the earlier subduction zone. The AP is left in its pre-drift position for reference and to allow direct alignment with the Andean Cordillera by the end of Jurassic times (see text). The patterned field shows the distribution of volcanic rocks of Chon-Aike age (presumed to be 175-150 Ma). Sea-floor spreading associated with the drifting phase of continental separation was simultaneous with the 130-120 Ma basaltic volcanism in southwest Africa (Etendeka) and the marginal arc magmatism of Patagonia and the Antarctic Peninsula (see Fig. 2).

PATAGONIAN GRANITES change in character of the Palaeozoic accretionary prism of southern Chile, and the chemical variations in the late Palaeozoic and early Mesozoic calc-alkaline plutonism of northern Patagonia. We have already pointed out that the Pacific emergence of the GFS at 38°S marks a significant change in the Palaeozoic metasedimentary basement. It is suggested that the Mesozoic reactivation of the accretionary complex south of this point (see also Herv6 et al. 1988) was associated with westward displacement of the SPB. It is even possible that some of the mafic-ultramafic bodies that occur here, which are MORB-Iike (Godoy et al. 1984) and have yielded possible Jurassic Rb-Sr ages (Herv6 et al. 1990), could represent slices of ocean floor tectonically emplaced during this displacement. The general absence of Mesozoic components in the prism itself could be explained by delamination and tectonic erosion beneath the older Palaeozoic (? and already metamorphosed) part as it overrode the Jurassic trench position (Fig. 3b).

Implications for magmatism associated with displacement The Early Mesozoic I-type granitoids of the Batholith of Central Patagonia would have formed close to the Pacific margin rather than in their present, apparently anomalous, intraplate position. These are granitoids and rhyolites (63-77% SiO2) of calc-alkaline I-type: their geochemistry is distinct from the alkaline A-type tendency of the Marifil Group. Together with intermediate isotopic signatures (initial 87Sr/a6Sr ratios of 0.7055-0.7060, eNdt values of - 2 to -4) this suggests modified 1-type magma generation in a continental magmatic arc setting (e.g. Pankhurst 1990). The Somuncura batholith to the northeast (Fig. 2), which consists of calcalkaline granitoids with more of a crustal component than the BCP (authors' unpublished Nd isotope data), has been interpreted by Ramos (1986) as having formed within an allochthonous Patagonia-Antarctic Peninsula terrane, by subduction of ocean floor from the northeast prior in collision and accretion onto the rest of continental South America. However, the palaeogeography shown in Fig. 3a would allow magma generation by subduction of proto-Pacific ocean floor from the southwest, without the need for internal oceanic subduction and collision events within Gondwana at this time. Similarly, the plutonic rocks of the BCP could be regarded as a southern continuation of the Late-TriassicEarly Jurassic plutonic suites that crop out in association with the Late Palaeozoic granitoids of

215

the Southern Coastal Batholith of Chile (Parada 1990). It is implicit in this model that the crust of the SPB may have undergone much more crustal extension and thinning than the craton-cored continental regions before and/or during the initial break-up of Gondwana. An Early Jurassic extensional event in the Falkland/Malvinas Islands is represented by a mafic dyke suite with a single K-Ar age of 192+10 Ma (Cingolani & Varela 1976). In southern Africa, an early episode of Karoo volcanism (Duncan 1987; Marsh 1987) is dated at 204-179 Ma and was thus synchronous with the transtensional magmatism of the BCP and Marifil Group (see Fig. 3a). The predominantly Early-Middle-Jurassic Ferrar, Karoo and Tasman mafic igneous suites, together with roughly coeval acidic counterparts, represent a massive deep-seated thermal disturbance, generally associatedwith supercontinent disintegration (Dalziel et aL 1987; Storey et al. 1988; White & McKenzie 1989). Various tectonic ideas have been advanced to explain this association, including mantle plumes (Richards et al. 1989; White & McKenzie 1989), back-arc extension related to sudden steepening of 'fiat-stab' subduction (Cox 1978; Gust et al. 1985; Storey & Alabaster 1991), and cessation of subduction following a long period of accretion (Kay et al. 1989). Two significant groups of early-mid Jurassic volcanic rocks occur northeast of the GFS, whose timing and geochemistry may also reflect their relationship to the translation of the SPB. The Marifil Group, farthest east (Fig. 2) is dominated by high-SiO2 rhyolites and shows alkaline affinities (Hailer et al. 1990). It has been dated at close to 180 Ma (our unpublished data) and may be seen as the tail-end of magmatism associated with movement on the GFS. The Lonco Trapial Group occurs immediately northeast of Gastre and even partly within the GFS: it is a more typical calc-alkaline assemblage of basaltic andesite to rhyolite (Page & Page 1987; Hailer et aL 1990) that may be related to subduction from the southwest, but its age is poorly defined.

Implications for magmatism subsequent to displacement The predominantly silicic Chon-Aike Formation (Tobifera) south of the GFS is dated by Late Jurassic fossils in interstratified marine mudstones and 170-150 Ma K-Ar ages (see Wilson 1991 for review). This timing is supported by a recently obtained Rb-Sr isochron age for acid volcanics of the Deseado Massif in southern Patagonia (161+5 Ma; initial 87Sr/S6Sr = 0.7058;

216

C.w. RAPELA & R. J. PANKHURST

De Barrio 1989). The Chon-Aike Formation exhibits some contrasting geochemical characteristics; whereas it is usually considered to represent crustal anatexis, the presence of mafic and intermediate compositions in the more westedy outcrops has been taken as indicating a subduction-related component (Gust et al. 1985) as has also been argued on the basis of a new geochemical data from Cordillera Darwin, Tierra del Fuego (Storey & Alabaster 1991). Thus, we distinguish the extensive Middle Jurassic volcanic record represented by the Chon-Aike and equivalent formations of silicic and intermediate extrusive and pyroclastic rock (see Rapela & Kay 1988) as subsequent to the 200-180 Ma igneous episode that is apparently restricted to the northeast side of the GFS. According to the evolutionary scenario shown in Fig. 3b, Chon-Aike volcanism largely postdated the major transtensional displacement which resulted in the continental plate overriding an active subduction system to the west. The latter continued to produce calc-alkaline magmas, but this time at least 200 km farther east. The relatively rapid shift with respect to the thermal and magmatic anomalies associated with subduction resulted in over-heating of the continental crust (and the upper lithospheric mantle) much farther to the east, where crustal anatexis occurred on a large scale. Parts of the SPB between the GFS and the Deseado Massif, such as the San Jorge basin and the Malvinas/ Falkland Plateau, may represent subsided and stretched continental crust, on which the extension-related acid volcanic rocks form the 'basement' of later sedimentary basins (Urien 1981).

Implications for Patagonian-West Antarctic relations The hypothesis represented in Fig. 3 also has significance for the relationship between Patagonia and West Antarctica within Gondwana. The most recent analysis of palaeomagnetic data (Grunow et al. 1987) indicates a post-MidJurassic 30° clockwise rotation of the Antarctic Peninsula and the Ellsworth-Whitmore mountains (Fig. 1) relative to East Antarctica, but do not allow statistical distinction between maintenance of their present positions relative to each other ('rigid Weddellia' model, Grunow et al. 1987) and independent movement ('mosaic Weddellia'). Nor is the E-W position with respect to South America constrained. Nevertheless, in the position shown in Fig. 3a, the other known micro-continental fragments, such as the Ellsworth-Whitmore mountains and Haag Nunataks crustal blocks (Dalziel et al.

1987), are available to fill a South Atlantic gap which is significantly smaller than in conventional reconstructions. The model illustrated in Fig. 3 requires 'side by side' overlap of the Antarctic Peninsula and Patagonia prior to break-up, with the northern tip of the peninsula in a 'Pacific' position relative to Patagonia, as in the microplate models of Grunow et al. (1987, 1991) and Lawyer et al. (1991). In Fig. 3a, the tip of the peninsula reaches a latitude of about 50°S compared to present-day South America. The geology of the Antarctic Peninsula is dominated by its Cretaceous-Tertiary granitoid batholith, a continuation of the Patagonian Batholith, so that the SPB and Antarctic Peninsula must have attained an 'in-line' relationship by the end of Jurassic times (Dalziel & Elliot 1982; Pankhurst 1990). The Mesozoic magmatic history of the Antarctic Peninsula began with a minor Triassic?Early Jurassic episode of predominantly granitic intrusion (Pankhurst 1990) and continued with more extensive Mid-Late Jurassic calc-alkaline plutonism and volcanism. The latter event has been traditionally regarded as subduction-related, but the discovery of high-Mg andesites suggests a back-arc or rifted extensional setting (Alabaster & Storey 1990). The Mid-Jurassic volcanism of the Antarctic Peninsula Volcanic Group is predominantly andesitic (but more rhyolitic toward the east; Weaver et al. 1982) and is probably equivalent to those Chon-Aike volcanic rocks of Patagonia at the calc-alkaline end of the compositional spectrum (see also Storey & Alabaster 1991), having formed in a similar position relatively close to the site of active subduction during break-up. The presence of back-arc sedimentary basins to the east, together with basic dyke emplacement, has also been interpreted as indicating extension and rifting along the eastern margin of the peninsula in Mid-Late Jurassic times (Sufirez 1976; Meneilly et al. 1987; Wever & Storey 1992). These interpretations are all consistent with the scenario depicted in Fig. 3, where latest Triassic-earliest Jurassic opening occurs between the Patagonia-Antarctic Peninsula 'terrane' and East Antarctica. It seems probable that this opening was accommodated by (a) crustal dilation in the Falkland/Malvinas Plateau and (b) back-arc extension behind the Antarctic •Peninsula.

Conclusions Continental rifting and extension of Gondwana began in Late Triassic to Early Jurassic times (c. 210-180 Ma ago). During this phase,

PATAGONIAN GRANITES the southern part of South America did not behave as a single rigid plate: it was at least divided into two blocks by the dextral Gastre Fault System, a transcontinental shear zone that was the precursor to the Aghulas Fracture Zone. In Early-Mid-Jurassic times, the continental mass south of this line (the Southern Patagonian Block and the Malvinas/Falkand Plateau) moved from a position some 500 km closer to Africa and East Antarctica. This plate motion was probably associated with the rotation and translation of the Falkland/Malvinas Islands inferred from palaeomagnetic analysis (Taylor & Shaw 1989). We propose a pre-existing offset in the coastline, with active subduction of proto-Pacific ocean floor. Magmatism associated with this phase of the GFS includes the late calc-alkaline granitoids of the B C P (208 Ma), followed by more alkaline rhyolites of the Marifil Group in eastern Patagonia (180 Ma), together with early Karoo mafic-acid volcanism. Rapid displacement of the SPB/MFP plate was in part accommodated by crustal stretching and resulted in thinned crust over-riding the subduction zone. This was followed by the extensive eruption of silicic volcanic rocks of the Chon-Aike Formation of M i d - L a t e Jurassic times (c. 165 Ma), with subduction-related characteristics to the west and anatectic characteristics to the east. Relationships between southernmost Patagonia and the adjacent parts of West Antarctica are emphasized in that the two areas were in close (?side-by-side) contact in early Jurassic times. The Antarctic Peninsula Volcanic Group andesites and rhyolites are at least in part contemporaneous and cogenetic with Chon-Aike/ Tobifera volcanism (Storey & Alabaster 1991). Thus, many of the most important geological features that distinguish Patagonia and the Antarctic Peninsula from adjacent continental areas to the north, including their characteristic magmatic evolution, may be seen as consequences of the mechanism of Gondwana disintegration. In particular, their Jurassic magmatism resulted from events during and after initial rifting that were controlled by interaction between a preexisting subduction regime and the dynamics of break-up. The early (BCP) subduction-related magmas merely exploited the developing shear zone during their emplacement, whereas the Late Jurassic magmatism was amplified by anatexis of the thin and extended crust as it overrode the subduction zone. The mechanism for break-up is not specifically identified, but it is not necessary to invoke a relationship to deep mantle structure or plume activity in order to explain the observed magmatism.

217

Field and laboratory work in Patagonia were initially supported by CONICET research grant 3-006 000/88 to C. W. Rapela and as the start of a joint CONICETRoyal Society research programme on the magmatism of the North Patagonian Massif. We are grateful to the Royal Society for continued support. We also thank many colleagues for stimulating discussions, both in the field and during the formulation of this paper, especially C. A. Cingolani, L. Dalla Salda, A. M. Grunow, F. Hervr, E. S. Oviedo, M. A. Parada and L. Spalletti. I. W. D. Dalziel and L. Gahagan are thanked for their use of the PLATES program at UTIG to check the geometry of the GFS. M. A. Uliana, P. F. Barker and B. C. Storey, I. W. D. Dalziel and W. E. LeMasurier kindly provided critical reviews of earlier versions. This paper is a contribution to IGCP Project Nos. 249 (Andean Magmatism and its Tectonic Setting) and 279 (Terranes in Latin America).

References ALABASTER,T. & STOREY,B. C. 1990. Modified Gulf of California model for South Georgia, North Scotia Ridge, and implications for the Rocas Verdes back-arc basin, southern Andes. Geology, 18, 497-500. BARRON, E. J., HARRISON,C. G. A. & HAY, W. W. 1978. A revised reconstruction of the southern continents. EOS Transactions, American Geophysical Union, 59, 436-439. OSGOLANI, C. A. & VARELA,R. 1976. Investigaciones geolpgicas y geocronolpgicas en el extremo sur de la Isla Gran Malvina, sector de Cabo Belgrano (Cape Meredith), Islas Malvinas. Actas VI Congreso Geolpgico Argentino, Buenos Aires, 1, 457-474. CINGOLANI, C., DALLA SALDA, L., HERV~, F., MUNIZAGA,F., PANKHURST,R. J., PARADA,M. A. & RAPELA, C. W. 1991. Evolution of the North Patagonian Andes and the adjacent continental massif: new impressions of Andean and pre-Andean tectonics. In: HARMON, R. S. & RAPELA,C. W. (eds) Andean Magmatism and its Tectonic Setting. Geological Society of America Special Paper, 265, 29-44. COIRA,B. L., NULLO,F., PROSPERPIO,C. • RAMOS,V. A. 1975. Tect6nica de basamento de la regi6n occidental del Macizo Norpatagrnico. Revista Asociacirn Geolrgica Argentina, 30, 361-383. CORTES, J. M. 1981. El substrato precretficico del extremo noreste de la provincia del Chubut. Revista Asociaci6n Geolrgica Argentina, 36, 217-235. Cox, K. G. 1978. Flood basalts, subduction, and break-up of Gondwanaland. Nature, 274, 47-49. DALZmL, I. W. D. & ELLIOT; D. H. 1982. West Antarctica: problem child of Gondwanaland. Tectonics, 1, 3-19.

218

C.W. RAPELA & R. J. PANKHURST

STOREY, B. C., GARRET, S. W., GRUNOW, A. M., HERROD,L. D. B. & PANKHURST,R. J. 1987. Extensional tectonics and fragmentation of Gondwanaland. In: COWARD,M. P., DEWEY,J. F. & HANCOCK, P. L. (eds) Continental Extensional Tectonics. Geological Society, London, Special Publication, 28, 433-441. DAVIDSON,J., MPODOZIS,C., GODOY, E., HERVI~,F., PANKHURST, R. J. & BROOK, M. 1987. Late Paleozoic accretionary complexes on the Gondwana margin of southern Chile: Evidence from the Chonos Archipi61ago. In: MCKENZIE,G. D. (ed.) Gondwana Six: Structure, Tectonics & Geophysics. American Geophysical Union Geophysical Monographs, 40, 221-227. DE BARRIO, R. E. 1989. Aspectos geol6gicos y

w~tnico. Actas XI Congreso Geol6gico Argentino San Juan, I, 56-59. HERV~, F. 1988. Late Paleozoic subduction and accretion in southern Chile. Episodes, 11,183-188. , DAVIDSON, J., GODOY, E., MPODOZIS, C. & COVACEVlCH, V. 1981. The Late Paleozoic in Chile: stratigraphy, structure and possible tectonic framework. Anais da Academia Brasileira de Ciencias, 53, 361-373. , GODOY, E., PARADA, M. A., RAMOS, V., RAPELA,C., MPODOZIS,C. & DAVIDSON,J. 1987, A general view of the Chilean-Argentinian Andes with emphasis on their early history. In: MONGER, J. W. H. & FRANCHETEAU,J. (eds)

geoquimicos de la Formaci6n Chon-Aike (Grupo Bahia Laura) en el noroeste de la provincia de Santa Cruz. Ph.D. Thesis Universidad Nacional

Union Geodynamic Series, 97-113. KAY, S. M., RAMOS,V., MPODOZIS,C. & SRUOGA,P. 1989. Late Paleozoic to Jurassic silicic magmatism at the Gondwana margin: Analogy to the Middle Proterozoic in North America? Geology, 17, 324-238. LABRECQUE, J. L. & BARKER,P. F. 1981. The age of the Weddell Basin. Nature, 290, 489-492. LAWVER, L. A. & SCOTESE,C. R. 1987. A revised reconstruction of Gondwanaland. In: MCKENZm, G. D. (ed.) Gondwana Six: Structure, Tectonics & Geophysics. American Geophysical Union Geophysical Monographs, 40, 17-23. , ROYER, J.-Y., SANDWELL,D. T. & SCO~SE, C. R. 1991. Evolution of the Antarctic continental margin. In: THOMSON, M. R. A., CRAME,J. A. & THOMSON, J. W. (eds) Geological Evolution of Antarctica. Cambridge University Press, 533-539. LE PICHON, X. & HAYERS, D. E. 1971. Marginal offsets, fracture zones, and the early opening of the South Atlantic. Journal of Geophysical Research, 76, 6283-6293. LORENZO, J. M. & Mu'rrER, J. C. 1988. Seismic stratigraphy and tectonic evolution of the Malvinas/ Falkland plateau. Revista Brasileira de Geociendas, 18, 191-200. LUDWIG, W. J. & KRASHENINNIKOV, et al. 1983. Geological framework of the Falkland Plateau.

~,

de La Plata. DE WIT, M., JEFFERY,M., BERGH, H. & NICOLAYSEN, L. 1988. Geological Map of Sectors of Gondwana. American Association of Petroleum Geologists. DEMICHELIS, A. H., HALLER, M. J. & MEISTER, J. 1991. Tect6nica de basamento y mineralizaci6n de fluorita en el subdistrito E! Moro, Macizo Norpatag6nico, Argentina. 6 Congreso Geo16gico Chileno, Vbia del Mar, Actas, 1,106-110. DUNCAN,A. R. 1987. The Karoo igneous province- A problem area for inferring tectonic setting from basalt geochemistry. Journal of Volcanology and Geothermal Research, 32, 13-34. FRANCHI, M. R. & PAGE, R. F. N. 1980. Los basaltos cretficicos y la evoluci6n magm~ttica del Chubut occidental. Revista Asociaci6n Geol6gica Argentina, 35, 208-229. FRUTOS, J. & ALFARO,G. 1987. Metallogenic and tectonic characteristics of the Paleozoic ophiolite belt of the southern Chile Coastal Cordillera. Geologische Rundschau, 76, 343-356. GODOY, E., DAVIDSON,J., HERVI~,F., MPODOZIS, C. & KAWASHITA,K. 1984. Deformaci6n sobreimpuesta y metamorfismo progresivo en un prisma de acreci6n paleozoico: Archipi61ago de Chonos, Ays6n, Chile. Actas IX Congreso Geol6gico Argentino, Bariloche, 4, 211-232. GRUNOW, A. M., KENT, D. V. & DALZIEL,I. W. D. 1987. Mesozoic evolution of the Weddell Sea Basin: new paleomagnetic constraints. Earth and Planetary Science Letters, 86, 12-26. , & ~ 1991. New palaeomagnetic data from Thurston Island: Implications for the tectonics of West Antarctica and Weddell Sea opening. Journal of Geophysical Research, 96, 17935-17954. GUST, D. A., BIDDLE, K. T., PHELPS, D. W. & ULIANA, M. A. 1985. Associated Middle to Late Jurassic volcanism and extension in southern South America. Tectonophysics, 116, 223-253. HALLER, M. J., DEMICHELIS,A. H., CABREROS,J. B. & PARDO, M. I. 1990. Petrologia y geoqufmica del vulcanismo Tri,qsico-Jurfisico en la Patagonia Extrandina y su relaci6n con el firing gond-

Circum-Pacific Orogenic Belts and Evolution of the Pacific Ocean Basin. American Geophysical

In: Initial Reports of Deep Sea Drilling Project, Washington (US Government Printing Office), 71,281-293. MARSH, J. S. 1987. Basalt geochemistry and tectonic discrimination within continental flood basalt provinces. Journal of Volcanology and Geothermal Research, 32, 35-49. MARTIN, A. K. & HARTNADY,C. J. H. 1986. Plate tectonic development of the South Indian Ocean: A revised reconstruction of East Antarctica and Africa. Journal Geophysical Research, 91, 4767-4786. McNuyr, R. H., CROCKET, J. H., CLARK, A. H., CAELLES, J. C., FARRAR, E., HAYNES, S. J. & ZEN~nLLI, M. 1975. Initial S7Sr/S6Sr of plutonic and volcanic rocks of the central Andes between latitudes 26 and 29 South. Earth and Planetary Science Letters, 27, 305-313. MENEILLY,A. W., HARRISON,S. M., PIERCY,B. A. &

PATAGONIAN GRANITES

219

STOREY, B. C. 1987. Structural evolution of the Revista Geolrgica de Chile, 18(?), 121-138. magmatic arc in northern Palmer Land, Antarc- RICHARDS, M. A., DUNCAN,R. A. & COURTILLOT,V. tic Peninsula. In: MCKENZIE, G. D. (ed.) Gond1989. Flood basalts and hot spot tracks: plume wana Six: Structure, Tectonics & Geophysics. heads and tails. Science, 246, 103-107. American Geophysical Union Geophysical SPALLETI'I, L. A., DEE VALLE,A., MANASSERO,M. J. & MATHEOS,S. D. 1989. Procedencia y ambiente Monographs, 40, 209-219. tect6nico delas areniscas Cretficico-Terciarias en MILLER, H. 1983. The position of Antarctica within el sector notre de la Patagonia. In: SPALLETn,L. Gondwana in the light of Palaeozoic orogenic deA. (ed.) Simposio del CreMcico de America velopment. In: OLIVER, R. L., JAMES, P. R. & Latina Part A: Eventos y Registro Sedimentario. JAGO, J. B. (eds) Antarctic Earth Science. SydGlobal Sedimentary Geology Programme ney, Australian Academy of Science, 579-581. (IUGS), Buenos Aries 149-163. MITCHELL,C., TAYLOR,G. K., Cox, K. G. & SHAW,J. STOREY, B. C. 1991. The crustal blocks of West Ant1986, Are the Falkland Islands a rotated microarctica within Gondwana: reconstruction and plate? Nature, 319, 131-134. MUNIZAGA, F., HERVI~, F., DRAKE, R., PANKHURST, break-up model. In: THOMPSON, M. R. A., CRAME,J. A. • THOMSON,J. W. (eds) Geological R. J., BROOK, M. & SNELLING, N. 1988. Geochronology of the Lake Region of South-Central Evolution of Antarctica. Cambridge University Press, 429-435. Chile (39°-42°S): Preliminary results. Journal of & ALABASTER,T. 1991. Tectonomagmatic conSouth American Earth Sciences, 1,309-316. PAGE, S. M. N. & PAGE,R. F. N. 1987. El Jur~isicovoltrois on Gondwana break-up models: evidence from the proto-Pacific margin of Antarctica. Tec6anico de la regi6n de Gastre-Pire Mahuida, protonics, 10, 1274-1288. vincia del Chubut. Actas X Congreso Geolrgico ~, HOLE, M. J., PANKHURST,R. J., MILLAR,I. L. & Argentino Tucumsn, 4, 174-176. VENNUM, W. R. 1988. Middle Jurassic withinPALMER, A. R. 1983. The decade of North American geology 1983 geologic time scale. Geology, 11, plate granites in West Antarctica and their bear503-504. ing on the break-up of Gondwanaland.Journal of Geological Society, London, 145,999-1007. PANKHURSr, R. J. 1990. The Paleozoic and Andean magmatie arcs of West Antarctica and southern SUAREZ, M. 1976. Plate tectonic model for southern Antarctic Peninsula and its relation to southern South America. In: KAy, S. M. & RAPELA,C. W. Andes. Geology, 4, 211-214. (eds) Plutonism from Antarctica to Alaska. Geological Society of America Special Paper, TAYLOR, G. K. & SHAW, J. 1989. The Falkland Is241, 1-7. lands: new paleomagnetic data and their origin as , HOLE, M. J. & BROOK, M. 1988. Isotope evida displaced terrane from southern Africa. In: ence for the origin of Andean granites. TransHILLHOUSE, J. W. (ed.) Deep structure and past actions of the Royal Society of Edinburgh, 79. kinematics of accreted terranes. American Geophysical Union Geophysical Monographs, 50, 123-133. 59-72. PARADA, M. A. 1990. Granitoid plutonism in central Chile and its geodynamic implications; a review. THOMAS, W. A. 1988. Early Mesozoic faults of the northern Gulf coastal plain in the context of In: KAY,S. M. & RAPELA,C. W. (eds) Plutonism from Antarctica to Alaska. Geological Society of opening of the Atlantic Ocean. In: MANSPEIZER, America Special Paper, 241, 51-66. W. (ed.) Triassic-Jurassic Rifting. Elsevier, DeRABINOWITZ, P. D. & LABRECQUE, J. 1979. The velopment in Geotectonics, 22, Part A, 463-476. Mesozoic South Atlantic Ocean and evolution of ULIANA,M. A. & BIDDLE,K. T. 1987. Permian to Late Cenozoic evolution of northern Patagonia, main its continental margin. Journal of Geophysical Research, 84, 5973-6002. tectonic events, magmatic activity, and deposiRAMOS, V. A. 1986. Tectonostratigraphy, as applied tional trends. In: MCKENZIE, G. D. (ed.) Gondwana Six: Structure Tectonics & Geophysics. to analysis of South African Phanerozoic Basins American Geophysical Union Geophysical by H. de la R. Winter, discussion and reply. Monographs, 40, 271-286. Transactions of the Geological Society of South UNTERNHER, P., CURIE, D., OLIVET,J. L., GOSLIN,J. Africa, 119,427-429. RAPELA, C. W. & KAY, S. M. 1988. The Late & BEUZART, P. 1988. South Atlantic fits and intraplate boundaries in Africa and South Paleozoic to Recent magmatic evolution of America. Tectonphysics, 155, 169-179. northern Patagonia. Episodes, 11,175-182. ~, PANKHURST, R. J. & HARRISON, S. M. 1992. URIEN, C. M. 1981. The basins of southern South America (southern Brazil, Uruguay and eastern Triassic 'Gondwana' granites of Gastre District, Argentina) including the Malvinas Plateau and North Patagonian Massif. In: BROWN, P. E. & South Atlantic paleogeographic evolution. In: CHAPPEL, B. W. (eds) Second Hutton symposium VOLKHEIMER, W. & MUSACCHIO, E. A. (eds) on granites and related rocks. Transactions of Cuencas Sedimentarias del Jurrsico y Cretticico de Edinburgh Royal Society: Earth Sciences 83, America del Sur. II Congreso Latinoamericano 291-304. de Paleontologia Porto Alegre, 45-126. , DIAS, G. F., FRANZESE, J. R., ALONSO, G. & MARTINS,L. R. & ZAMBRANO,J. J. 1976. GeoBENVENtYro, A. R. 1991. El Batolito de la ~ , logy and tectonic framework of southern Brazil, Patagonia central: evidencias de un magmatismo Uruguay and northern Argentina continental tri,'isico-jurgsico asociado a fallas transcurrentes.

220

C. W. R A P E L A & R. J. PANKHURST

margin: their behaviour during the South Atlantic opening. In: ALMEIDA, F. F. M. (ed.) Continental Margins of Atlantic Type. Annais Academia Brasileira Ciencias, 48, Supl., 365-376. WEAVER, S. D., SAUNDERS,A. D. & TARNEY,J. 1982. Mesozoic-Cenozoic volcanism in the South Shetland Islands and the Antarctic Peninsula: geochemical nature and plate-tectonic significance. In: CRADDOCK, C. (ed.) Antarctic Geoscience. University of Wisconsin Press, 263-273. WEVER, H. E. & STOREY, B. C. 1992. Bimodal magmatism in northeast Palmer Land, Antarctic

Peninsula: Geochemical evidence for a Jurassic ensialic back-arc basin. Tectonophysics, 205, 239-260. WroTE, R. & MCKENZIE, D. 1989. Magmatism at flit zones: the generation of volcanic continental margins and flood basalts. Journal of Geophysical Research, 94, 7685-7729. WILSON, T. J. 1991. Transition from back-arc to foreland basin development in the southernmost Andes: Stratigraphic record from the Ultima Esperanza District, Chile. Geological Society of American Bulletin, 103, 98-111.

Paran~i magmatism and the opening of the South Atlantic C. J. H A W K E S W O R T H , K. G A L L A G H E R , S. K E L L E Y , M. M A N T O V A N I 1, D . W . P E A T E 2, M. R E G E L O U S & N. W . R O G E R S

Department of Earth Sciences, The Open University, Walton Hall, Milton Keynes MK7 6AA, UK 1Departamento de Geofisico, IA G-USP, Caixa Postal 9638, Sao Paulo, 01065 SP, Brazil 2Present address: Department of Geological Sciences, University of Durham, South Road, Durham DH1 3HE, UK Abstract: New chemical and isotope results are presented on dyke rocks associated with the Paran~ CFB, together with preliminary laser 'l°Ar/39Aranalyses on selected Paran,'i basalts. Dyke rocks from the Ponta Grossa Arch are similar to the Pitanga and Paranapanema magma types in the Paran~i lavas, but dykes from the Santos--Rio de Janeiro section include samples with compositions not observed in the overlying lavas. Rather their minor and trace elements are strikingly similar to basalts recently erupted on Tristan de Cunha, and thus these late stage dykes may represent the first direct evidence for the involvement of typical plume-related OIB in the Paran~i province. Laser 4°Ar/39Aranalyses of two Gramado low Ti basalts have yielded preferred isoehron ages of 132.4+ 1.4 and 132.9+2.8 Ma. These indicate a short eruption time for at least the Gramado magma type, and that magmatism took place several million years after the species extinction in the Tithonian (c. 141 Ma). The majority of basalts and basaltic andesites in the Paran~i CFB have distinctive trace elements ratios (low Nb/La and Nb/Ba), and relatively enriched Sr, Nd, and Pb isotope compositions. Since such features are not commonly observed in oceanic basalts, and they occur in CFBs which have been screened for the effects of crustal contamination, they are typically attributed to old, incompatible element enriched source regions in the continental mantle lithosphere. In some models the minor and trace element 'mantle lithosphere' component was introduced in small degree melts (lamproites)added to asthenosphere derived magmas. However, such models appear to be inconsistent with the data from low Ti CFB, and they also require that the asthenosphere derived magmas have very low incompatible element contents, in marked contrast to the high Nb/La late stage dykes in the Paran~i. Alternatively some CFBs may have been generated within the mantle lithosphere in the presence of small amounts of water. The results of preliminary calculations indicate that in the presence of a mantle plume up to 5 km of melt may be generated entirely from within the mechanical boundary layer, for/3 values of less than 1.2.

Continental Flood Basalts (CFB) represent major magmafic events, and they may be significant contributions in the generation of new continental crust (White & McKenzie 1989). They have been attributed to meteorite impact (Alt et al. 1988), and regarded as the cause of mass extinction (Officer & Drake 1985; Rampino & Stothers 1988), but there remains considerable uncertainty over the causes of magmatism, and the extent to which the continental mantle lithosphere may be remobilized during these major magmatic events. The Deccan CFB, for example, were generated at the time of mass extinction, and a major meteorite impact of the K - T boundary (Courtillot & Cisowski 1987; Duncan & Pyle 1988). Some of the magma types are similar to those erupted recently on the island of Reunion, and so there are close geochemical

and spatial links between the Deccan CFB and the Reunion hot spot (Mahoney 1988; Lighffoot & Hawkesworth 1988). In contrast to the Deccan, the eruption of the Paran~i-Etendeka CFB did not coincide with any well documented meteorite impact, nor mass extinction. Rather they were associated with the opening of the South Atlantic, and specifically with extension across the mantle plume which is presently beneath Tristan da Cunha. However, unlike the Deccan, no Paran~i lavas have been identified with similar compositions to the recent hot spot related lavas on Tristan da Cunha, and most Paran~i lavas have enriched radiogenic isotope ratios indicative of significant contributions from the continental lithosphere (Hawkesworth et al. 1986, 1988; Petrini et al. 1987; Cordani et al. 1988). Stratigraphical studies have an important role

From STOI~Y, B. C., ALABASTER,T. & PANra-IURST,R. J. (eds), 1992, Magmatismand the Causes of ContinentalBreak-up, Geological Society Special Publication No. 68, pp. 221-240.

221

222

C.J. HAWKESWORTH E T A L .

edge of Paran~ sedimentarybasin ~ ~

youngersedimentaryrocks overlyinglavas rhyolites

~

flood basalts

~

dykes

I /

I

/

AN

° P ZO

SOUTH AMERICA sills

///~"~ \

//

'~ \

\

~n6 ~T \

\

% \ \

\

AFRICA

~'/y Etendeka eWlndhoek

\

~v~ .~.

Buenose

Aires

0

500km

I

I

Fig. 1. Pre-drift reconstruction showing the extent of Paranfi-Etendeka magmatism in relation to the Paranfi sedimentary basin and the marked asymmetry of flood basalts relative to the proto-Atlantic rift (after Peate et al. 1990). Late-stage rhyolites are restricted to the continental margins. Dykes are concentrated in four main areas: Namibia, eastern Paraguay (east of Asuncion) (PA), the Ponta Grossa Arch (PG), and along the coast between Sao Paulo and Rio de Janeiro, here termed the Santos-Rio de Janeiro (SRJ) section. The inferred location of the Tristan plume and hotspot track are from Duncan (1984). A-A' is the location of the cross section in Fig. 5; + shows present-day latitude and lon~titude at 5 ° intervals.

PARANA MAGMATISM AND SOUTH ATLANTIC OPENING in the investigation of CFB provinces in revealing the internal structure and sequential development of the lava pile (e.g. Columbia River Basalts: Swanson et al. 1979; Hooper 1982; Mangan et al. 1986, and Deccan Trap: Cox & Hawkesworth 1985; Beane et al. 1986; Devey & Lightfoot 1986). Recent investigations in the Parami identified eight major magma types, six in basalts and two in rhyolites, and documented a stacking of units of different basaltic magma types overlapping towards the north (Peate 1990; Peate et al. 1990). This suggests that the main locus of magmatism moved north with time in the ParanCi basin, and this is currently the single most important observation linking the generation of the Paran~i CFB with the northwards propagation of rifting in the South Atlantic. However, in addition to the extrusive rocks of the Paranti, there are a number of well-developed dyke swarms orientated both at a high angle to the present coastline and sub-parallel to it (Piccirillo et al. 1990). This contribution reports the initial results of a detailed study of the Paranfi dykes, to establish what magma types are present and the extent to which they may be linked with the magma types recognized in the overlying lava pile. The results of the preliminary laser Ar/Ar study on selected lavas from the southern Paran~i are presented, and then two models for CFB generation involving significant contributions from the continental mantle lithosphere are discussed.

The Paranfi--Etendeka CFB province The extensive Parami lava field in central South America and the minor Etendeka remnants in Namibia originally formed a single magmatic province (Erlank et al. 1984; Bellieni et al. 1984) which was closely associated with the opening of the South Atlantic ocean during the Early Cretaceous. The majority of the lavas yield 115-135 Ma K-Ar ages (see summaries by Erlank et al. 1984 and Rocha-Campos etal. 1988), but 130 Ma is generally taken as the minimum age of eruption. More recently Baksi et al. (1991) reported preliminary stepwise 4°Ar/39Ar plateau ages of

223

130-135 Ma on lavas from the southern Parami. The external shape of the Paran~i-Etendeka province is well constrained by surface mapping and, for the Parami, by data from over 70 oil exploration boreholes. The distribution of the lavas about the South Atlantic ocean is highly asymmetrical, with the Parami lavas covering an area in excess of 1.2x 106 km 2, over 15 times greater than the present-day extent of the Etendeka lavas (c. 0.08x 106 km 2) stranded on the African plate (Fig. 1). The thickness of the lava pile in the Paranti province mirrors the overall structure of the underlying sedimentary basin (Zalan et al. 1991). The thickest preserved accumulation of lavas is in the northern half of the province, coincident with the deepest part of the sedimentary basin, and the main area of lava thickness (> 1 km) runs down through the central area, roughly parallel to the northeast-southwest elongation of the basin. The products of the Paranti magmatic event are dominated (> 90 %) by tholeiitic basalts and basaltic andesites which are accompanied by significant quantities of acidic rocks (rhyolites and rhyodacites), notably along the Brazilian continental margin, and also in the Etendeka. There is a virtual absence of any samples with SiO2 contents between 60 and 64 wt%, and this makes a natural division to use in the classification of the Paranfi volcanics. Thus rocks with -> 64% SiO2 are loosely termed 'Rhyolites', and those with -< 60% SiO2 are termed 'Basalts'. Initial classifications divided the basalts into two groups based on their Ti contents (Bellieni et al. 1984; Mantovani et al. 1985), but more recently Peate (1990) and Peate et al. (1990, in press) proposed a more comprehensive classification scheme based on the 2000 major and traced element analyses which are available at present. The main geochemical features of each magma type are summarized in Table 1, and the average minor and trace element contents of the basaltic magma types are presented in Table 2 and Fig. 2. The newly defined magma types broadly correspond to the previous division of the Paranfi basalts into 'low-Ti' or 'high-Ti' groups, in that the Gramado, Esmeralda and Ribeira magma

Table 1. Main geochemical features of the different Parant~ magma types

Urubici Pitanga Paranapanema Ribeira Gramado Esmeralda

Type

TiO2

High-Ti, south (Khumib) High-Ti, north Intermediate-Ti, north Intermediate-Ti, south Low-Ti, south (Tafelberg) Low-Ti, south

> 3.3 > 2.9 1.7-3.2 1.5-2.3 0.75-1.9 1.1-2.3

Ti/Y > > > > <
6.5 > 5.5 4-7 3.5-7 3.5-6.5 2-5

> 550 > 350 200-450 200-375 140-400 120-250

224

C.J. HAWKESWORTH E T A L .

Table 2. Average compositions of basaltic magnm types, and selected dyke units Lavas

Gramado (n = 121)

SiO2 TiO2 A1203 Fe203(t) MnO MgO CaO Na20 K20

P202 Ni Rb Sr Y Zr Nb Ba

Esmeralda (n = 70)

Ribeira (n = 26)

Paranapanema (n = 26)

Pitanga (n = 82)

Urubici (n = 65)

av.

s.d.

av.

s.d.

av.

s.d.

av.

s.d.

av.

s.d.

av.

53.68 1.43 14.26 12.64 0.19 4.90 8.64 2.68 1.33 0.21

1.71 0.24 0.83 1.26 0.02 1.03 1.10 0.34 0.51 0.04

51.33 1.55 13.82 14.25 0.20 5.39 9.80 2.63 0.83 0.20

1.26 0.26 0.73 1.05 0.02 0.90 1.14 0.26 0.37 0.05

50.48 1.80 14.56 13.74 0.20 5.62 10.06 2.49 0.79 0.25

0.44 0.16 0.87 0.67 0.02 0.66 0.39 0.26 0.16 0.06

50.12 2.31 13.20 14.92 0.22 5.36 9.69 2.79 0.99 0.31

0.62 0.34 0.50 0.89 0.02 0.74 0.81 0.45 0.35 0.10

50.52 3.53 12.90 15.19 0.22 4.36 8.34 2.92 1.48 0.56

1.22 0.26 0.37 0.80 0.02 0.59 0.74 0.25 0.45 0.18

51.77 0.87 3.74 0.28 13.32 0.36 13.24 0.54 0.17 0.01 4.53 0.35 8.19 0.49 2.80 0.29 1.71 0.33 0.54 0.06

42 45 238 33 166 14 388

25 21 44 6 32 3 87

52 27 191 34 131 10 240

18 13 41 6 31 3 69

77 17 288 29 135 11 337

types have relatively low Ti abundances, and the Urubici, Pitanga and Paranapanema have relatively high Ti. The Pitanga and the Urubici have the highest TiO2 contents, typically > 3 wt%, and they also have the highest rare earth element (REE) and high field strength element (HFSE) abundances. The trace element signatures of the Urubici and Pitanga are very similar, with the Urubici pattern tending to be at slightly higher abundances, particularly for St, which is arguably the single most diagnostic feature to distinguish these two magma types (Figs 2 & 3). The third 'high-Ti' magma type, the Paranapanema, has more intermediate TiO2 contents, generally in the range 2-3 wt%. Nonetheless, it is grouped with the 'high-Ti' magmas because it shares many of the trace element characteristics of the Pitanga, albeit at lower abundances, although the two have similar Y and Yb contents. The trace element signature of the Ribeira (the rare low-Ti lavas of the northern Paran~i described by Petrini et al. 1987) contrasts markedly with the other 'low-Ti' magma types and it shows more similarities with the abundance patterns of the 'high-Ti' magma types, particularly that of the Paranapanema. The 'low-Ti' Gramado magma type has a distinctive trace element signature relative to the 'high-Ti' types. It shows a marked enrichment of the large ion lithophile (LIL) elements over HFS and LRE elements which is not seen in the 'high-Ti' magmas, and it also displays prominent troughs at Ti, P and Sr. The Esmeralda pattern is similar to that of the

14 5 49 4 17 3 76

51 21 306 35 174 15 290

17 9 47 7 33 3 74

30 32 466 41 268 25 484

15 12 56 7 43 4 93

54 37 768 38 307 28 636

s.d.

13 10 105 3 24 3 86

Gramado, except with generally lower trace element abundances, and less well developed troughs at Ti and P. A key feature of the trace element patterns of all these Paranfi magma types is the depletion of Nb and Ta relative to La and, to a lesser extent K. This is not a characteristic of oceanic basalts (MORB or OIB), and together with their 'enriched' Sr and Nd isotope signature of positive esr and negative end, these have been the basis for models invoking mobilization of continental mantle lithosphere in the generation of the Paran~i flood basalts (e.g. Hawkesworth et al. 1988). A s s o d a t e d dykes Within the Paran~i-Etendeka CFB province three main areas of dyke emplacement have been studied in some detail: (1) in and around the Etendcka lavas in Namibia, (2) in the Ponta Gross Arch in southern Brazil; and (3) along the coast between Santos and Rio de Janerio (Fig. 1). Many of the dykes in Namibia and in the coastal swarm in Santos-Rio de Janeiro were sub-parallel to their respective coastlines at the time of emplacement, whereas the Ponta Grossa dykes are at a high angle to both the main axis of the Paranfi Basin, and to the present coastline. Such dyke swarms are extremely important in the development of quantitative models for the generation of the Paran~i basalts, and their links with the opening of the South Atlantic, since

PARANA MAGMATISM AND SOUTH ATLANTIC OPENING

225

Table 2. Cont. Dykes Tristan da Cunha PGla (n = 25) av. 51.11 2.58 13.36 15.11 0.22 5.08 9.17 2.52 1.17 0.34 49 31 374 40 191 16 442

PGlb (n = 21)

s.d.

av.

0.59 0.42 0.43 0.82 0.03 0.67 0.71 0.13 0.25 0.09

s.d.

51.08 3.29 13.20 15.11 0.20 4.55 8.39 2.69 1.56 0.52

15 9 104 6 34 3 110

SRJI (n = 29) av.

1.52 0.34 0.45 1.02 0.02 0.79 1.11 0.20 0.48 0.22

40 44 513 44 263 22 563

s.d.

55.00 2.95 13.87 11.03 0.16 4.01 6.18 3.11 3.34 0.59

16 18 148 10 95 5 168

385 45 1214

av.

2.08 0.26 0.65 0.61 0.05 1.14 1.54 0.41 0.35 0.14

136 64 1102 34 290 73 1269

1

d'/N~,~ ~" " \ ~ ' , ,

30 8 218 4 36 6 241

25 130 1147 29 375 99 798

^ /~

o

/

Gramado Esmeralda Ribeira

~ I

I

27 203 405 4 161 48 352

0.7048-0.7057II

Umbici

/-o-L o

Pitanga 0.7055-0.7060II Paranapanema 0.7055-0.707y I

I I !

:Low-Ti magmatypes 87Sr/86Sr130Ma~

/

6.16 1.18 2.44 3.98 0.02 2.72 3.38 1.26 1.63 0.32

In Namibia three major suites of dolerites have been recognized, and all of them are low-Ti in that they have < 2 wt TiOz (Erlank et al. 1984). The dominant magma type is termed Tafelberg, which is also well represented in the low-Ti lavas of the Etendeka and is equivalent to the Gramado of the Paramt. The Horingbaai doler-

lOq

5--4/"

47.53 3.06 16.50 11.32 0.18 4.80 8.74 4.36 3.26 0.68

s.d.

~High_Timagmatypes 87Sr/86Sr130M~a]

V.\_ _~ u50--I

(n = 17)

s.d.

46.13 2.46 13.55 11.72 0.20 8.08 11.81 2.87 2.16 0.79

19 33 106 7 70 13 298

" 40

/~

av.

2.73 0.65 0.52 1.90 0.02 0.71 1.19 0.27 0.98 0.13

46 96 863

they indicate the prevalent stress fields, and perhaps constrain the amount of extension, during the emplacement of particular magma types. There is also some suggestion that the Ponta Gross dykes, and those in Namibia and Santos-Rio de Janeiro were emplaced at c. 120" to another (Fig. 1).

lO0--q

SRJII (n = 14)

I

I

I

1

~

II~

I

I I

0.7075-0.7160 I 0.7046-0.7076 I 0.7055-0.7060,,,/ I

1

I

I

I

I

[ I

I

I

I

I

I

I

RbBaTh K T a N b L a C e S r Nd P H f Z r S m E u T i T b Y Yb

I

J

Fig. 2. Average compositions of five Paranfi basalt magma types, normalized to primitive mantle using the abundances of Sun & McDonough (1989). The negative Nb-Ta anomalies, plus exceptionally low Ti/Y and Ti/Zr of the Grammado and Esmeralda magma types, are features not commonly observed in oceanic basalts and suggest involvement of lithospheric mantle material (after Peate et al. 1990).

226

C.J. HAWKESWORTH E T A L .

(a)

(b~

m

.Pitanga

Pitanga Urubici

700'

TiN

700 -

2

TiN

600'

~

• "

500

""

panema ofl~Z • 0

400"

.

v

tCF~l-~

°oo.

" ,,

•

,'-

Parana- p ~ pane~ru~. ~ F @

•

A

p.¢," .

Pa'='a"

A~

A•

600-

500'

Urubici

AA

z~A I~

•

~

A

400-

~A 3O0 -

: ~ ~w-=4) "

°

A

300

r~ O P A R A N A 20O

LAVAS

200

Gramado & ~ Esmeralda •

Esme~alda 100 0

200

400

' 600

' 800

' 1000

Sr ppm

100 0

I hA A • 200 400

•

PARANA DYKES

, , 600 800 Sr ppm

, 1000

Fig. 3. Ti/Y and Sr variations in (a) the lavas, and (b) the dykes of the Paran~i. The diagrams are subdivided to illustrate the compositions of the different magma types recognized in the lavas (see also Table 1), and to demonstrate the extent to which similarmagma types may be recognized in the associated dykes. The filled and the open triangles in (b) represent the dykes from the Ponta Grossa Arch and the Santos-Rio de Janeiro section, respectively. The Gramado and Esmeralda rocks are unusual in that many samples have Ti/Y ratios < N-type MORB (Ti/Y = 271 in average N-MORB, Sun & McDonough 1989). ites are very distinctive in that they have strong isotope and trace element similarities to TMORB (Erlank et al. 1984; Duncan et al. 1990), and the Huab dolerites share some of the compositional features of the Horingbaai, but with somewhat different Ti/Zr and Ba/Nb and markedly higher Sr isotope ratios. The Ponta Grossa Arch in Brazil is a tectonic swell trending NW-SE (Fig. 1). Its formation is thought to have begun in the Devonian and culminated during the Triassic-Jurassic, just before the Paran~i CFB event (Fulfaro et al. 1982), and it now includes hundreds of NW-SE tholeiitic dykes. These dykes have yielded a range of K-Ar ages similar to the range reported from ~he Paran~i lavas, 132+10 and 131+9 Ma, respectively (Pinese, unpubl.; Rocha-Campos et al. 1988), and they have been studied previously by Piccirillo etal. (1990). The dykes of both the Ponta Grossa Arch and the Santos-Rio de Janerio section (Fig. 1) are the subject of a detailed geochemical and isotope study which is currently in progress (Regelous, in prep.). Over 90 samples have been analysed for major and trace elements, and the results for selected representative samples are presented in Table 3. The first step was to evaluate how the dyke compositions compare

with those of the overlying lavas, and it is clear from Tables 1 and 2 that Ti/Y ratios and Sr contents are two of the most diagnostic features for distinguishing the different basalt magma types in the Paran~i. Sr has a particular role in the study of tholeiitic rocks because it is often buffered by low pressure fractionation of a gabbroic mineral assemblage, and so the measured Sr contents may be similar to those in the parental magmas. The overwhelming majority of Paran~i basalts are aphyric, and thus there are no problems with significant plagioclase accumulation. Incompatible element ratios, such as Ti/Y, are also relatively insensitive to fractional crystallization processes and, in addition, the Paran~i low-Ti magma types (Gramado and Esmeralda) are characterized by distinctive negative Ti anomalies on mantle normalized diagrams (Fig. 2). Thus Fig. 3 summarizes the variations in Ti/Y and Sr in the major basalt magma types, and then compares the variations in the extrusive basalts with the available analyses of dyke samples from Ponta Grossa and the Santos-Rio de Janeiro section (data from Piccirillo et al. 1990; Peate 1990; Regelous unpubl.). The main features to note are: (1) few dykes have compositions comparable with those of the major low-Ti magma types of Gramado and Esmer-

*NBS987 0.710242+ 12 (12 anal). tJ&M 0.511851+10 (8).

0.70548 0.70521 0.51266 0.51255

58 81 291 56 795 265 415 26 470 17 131 32.1

Zn Cu Ni Co Cr V Ba Rb Sr Y Zr Nb

S7Sr/S6Srm* S7Sr/a6Sri 143Nd/144Ndm']" 143Nd/144Ndi

45.53 1.85 8.59 9.89 0.15 14.22 17.62 0.91 0..88 0.31 100.11

MR8967

109 18 206 58 266 238 1407 62 1041 33 324 74.9

93 132 46 46 20 310 977 56 972 33 271 77.5 0.70525 0.70494 0.51266 0.51256

41.31 3.54 12.11 13.25 0.19 11.88 11.93 2.91 2.24 0.97 100.37

45.22 2.62 16.92 12.91 0.21 4.7 10.66 3.34 2.29 0.52 99.37

0.70513 0.70483 0.51265 0.51255

MR8981

MR8976

SJII

Chemicalanalysis of selected dyke samples

SiO2 TiO2 A1202 Fe202 MnO MgO CaO Na20 K20 P205 TOTAL

Table 3.

0.70544 0.70516 0.51271 0.51261

64 60 376 58 812 209 713 46 874 28 177 49.2

43.65 1.3 12.57 11.07 0.19 13.15 14.66 1.48 1.25 0.67 100.13

MR8974

0.70828 0.70757 0.51222 0.51213

91 34 37 31 128 172 1400 116 858 33 412 51.5

56.89 2.52 13.66 9.77 0.14 4.2 5.9 3.15 3.84 0.48 100.59

MR8985

131 144 73 49 52 281 865 79 868 43 319 30.2

96 32 51 37 127 171 1302 111 871 38 404 52.8 0.70786 0.70739 0.51231 0.51219

51.83 3.67 13.5 12.72 0.17 4.63 6.88 3.08 2.58 0.55 99.62

56.79 2.5 13.71 9.74 0.15 4.24 5.83 3.19 3.78 0.47 100.41

0.70838 0.70771 0.51217 0.51208

MR8963

MR8964

SJ1

0.70849 0.70809 0.51223 0.51213

106 54 54 48 23 244 1064 70 911 42 350 37.1

52.84 3.38 13.76 12.68 0.18 4.19 6.7 3.04 2.7 0.88 100.36

MR8982

0.70596 0.70568

119.8 125.7 31.5 39.6 11 367.7 613.9 37.5 721.6 74.1 238.6 22.1

50.2 3.81 14.4 14.33 0.19 4.23 8.33 2.93 1.57 0.62 100.6

MR898

0.70628 0.70593 0.51239 0.51227

93.6 205.9 78.6 49.8 145.6 406.8 318.9 20.6 309.4 38.5 130.5 11.5

51.21 1.83 14.11 13.88 0.26 6.51 10.29 2.33 0.84 0.21 101.48

MR899

0.70712 0.70651

124.7 262.2 39.8 47.5 35.6 436.2 685.9 52.6 452.1 43.9 274 24.6

52.1 3.22 12.98 15.26 0.21 4.4 7.05 2.93 1.91 0.42 100.51

MR8914

Ponta Grossa

0.70619 0.70574

104.9 139 39.6 43.9 26.9 453.6 489.6 35.5 415.2 42.9 231.4 20.9

50.12 3.11 13.08 16.55 0.29 4.74 7.87 2.57 1.46 0.48 100.26

MR8915

228

C.J. HAWKESWORTH E T A L .

alda; (2) the majority of the Ponta Grossa dykes are similar to the higher Ti magma types of the Paranapanema and the Pitanga; and (3) there are a number of dyke samples which appear to have compositions not represented in the present data base of extrusive Paran~i basalts, and many of these dykes are from the Santos-Rio de Janeiro section. As indicated above, all the Paran~i magma types have negative Nb and Ta anomalies on mantle normalized diagrams (Fig. 2), and such anomalies are not a feature of oceanic basalts generated either along MORs or on oceanic islands. Figure 4 therefore compares the variations in Nb/Ti and Zr/Ti in the Paran6 lavas and dykes with those from Tristan da Cunha (Le Roex et al. 1990) and the South Atlantic MOR (Humphris et al. 1985). All the Paran6 lavas have relatively low Nb/Ti and Nb/Zr, consistent with their relatively low Nb abundances, and in marked contrast to the high Nb/Zr of the Tristan OIB. The basalts of the Walvis Ridge (not 0.006

PARANA DYKES

Nb/1-i

SANTOS- RIO DE JANEIRO

0.005

0.004

0.00.3Tristan 0.002 -

da Cunha

0.001

'~Parana

lavas

PONTA GROSSA MORB

0 0

I 0.01

I 0.02

I 0.03

0.04

Zr/Ti Fig. 4. A diagram of Nb-Ti against Zr/Ti to illustrate

the variations in the three main dyke groups relative to those of the Parana lavas and selected basalts from the S. Atlantic. The filled triangles are Ponta Grossa dykes, and open triangles are dykes from the SantosRio de Janeiro section. Both the high-Ti and low-Ti magma types of the Parana lavas have low Nb/Zr ratios, reflecting their relatively low Nb abundances (Fig. 2), and the Ponta Grossa dykes have low Nb/Ti and Zr/Ti ratios. In contrast, the dykes from the Santos-Rio de Janeiro section have a bimodal distribution of Nb/Zr: many are similarto those of the main Paran,'i magma types, even though some of the samples have relatively low Ti contents, and hence high Zr/Ti and Nb/Ti, and then the youngest dykes have high Nb/Zr similar to the recent, hotspot related basalts on Tristan da Cunha (Le Roex et al. 1990).

shown) have Nb/Zr ratios which are transitional between the high Nb/Zr of the Tristan basalts and the low Nb/Zr of MORB and the Paramt lavas (Humphris & Thompson 1983). Many of the available dyke analyses do not include Nb contents, but for those that do (Peate 1990; Regelous in prep.) the dykes can be broadly sub-divided into a number of groups. Most of the dykes from the Ponta Grossa have compositions similar to those of the Paranapanema and Pitanga magma types recognized within the Paran~i lavas (see also Fig. 3). However, those from the Santos-Rio de Janeiro section either have Nb/ Zr similar to the main Paranti lavas, albeit often with relatively low Ti contents, or they are strikingly similar to the recent basic rocks on Tristan da Cunha. (These two dyke groups are termed SRJ I and SRJ II respectively in Tables 2 and 3.) Igneous activity on Tristan da Cunha is linked to the mantle plume that is inferred to have been present at the time of Paran~i magmatism, and these high Nb/Zr dykes appear to be the first direct evidence for the involvement of typical plume-related OIB in the Paran6 province. Regional s t r a t i g r a p h y The salient features of the Paran~i lava field stratigraphy, and the magma types observed in the dyke rocks, are summarized in Fig. 5. This is a schematic profile through the lava pile along a north-south section at longitude 52°W, constructed by integrating results from borehole samples with data on the surface lavas, and the depth to the base of the lavas (Peate et al. 1990, in press). T h e lithostratigraphical pattern of units defined by specific magma types has several implications for the development of this part of the magmatic province. First, it indicates that the dominant, magma type has evolved from Gramado (low-Ti) to Esmeralda (10w-Ti) to Pitanga (high-Ti) to Paranapanema (intermediate-Ti) with time. Second, the overlapping sequence of units dipping towards the north suggest a northward-migrating source for the Paran~i magmatism in this region. The relatively exaggerated dips of the unit boundaries as well as the lava/sediment interface, could be partly a post-eruptional feature imposed by subsequent epeirogenic uplift, a process that appears to have been restricted to near the coastal margin (Gallagher et al. 1991). However, the northward migration of volcanism is indicated not just by the northerly dip of unit boundaries, but more explicitly by the fact that units thin down-dip in this direction and pinch-out against the basal contact, as seen in the Gramado unit, as well as thinning out up-dip as shown by the Pitanga unil

PARANA MAGMATISM AND SOUTH ATLANTIC OPENING ESMERALOA / ~

South

to_

r lkm

~

.

o

North

RIBEIRA

_

•

.....

1

.....

I

229

,

.:iiiiiiiiii!i"

..,

qb'~/~l ETENDEKA

.

......

F'O.TA GROSSA

'~ '

:..=

iii ~ o oE " "~ "

JANEIRO

Fig. 5. Schematic south-north section along 52°W illustrating the internal stratigraphy of the Paranfi lava pile, based primarily on borehole data (after Peate et al. 1990, in press). The position of the three areas of associated dykes discussed in the text are also projected on to the south-north section: (i) most of the dykes in the Etendeka are very similar to the Gramado magma type of the Paran~i, but there is also a late stage dyke swarm, the Horingbaai, with compositions similar to MORB (Ertank et al. 1984; Duncan et al. 1990); (ii) the dykes of the Ponta Grossa Arch are similar to the magma types of the Pitanga and the Paranapanema, and (iii) the youngest, cross cutting dykes of the Santos-Rio de Janeiro section have minor and trace element compositions strikingly similar to those of the recent basalt on Tristan da Cunha (Le Roex etal. 1990). Both the overstepping relation of units defined by each magma type, and the south-north variations in the associated dyke swarms, suggest northward migration in the site of magmatism. The shaded unit is the Palmas rhyolite, the speckled units are later sediments, and the samples chosen for Ar isotope analysis are from the top and bottom of the southern escarpment, labelled I ~ . Although there are as yet no precise age determinations on the dyke rocks, a relative chronology can be inferred from cross cutting relationships in the field and by comparing their compositions with the stratigraphy of the lava pile. The predominant coast-parallel dykes in Namibia are similar to the low-Ti Gramado basalts, and they are presumably older than the Ponta Grossa dykes which are compositionally equivalent to the younger Pitanga and Paranapanema lavas (Fig. 5). The SRJ II dykes have no equivalents in the extrusive lava pile, and since these dykes are the youngest in the Santos-Rio de Janeiro section it may be further inferred that they represent a magma type which is younger than the Paranapanema. Thus even within the intrusive rocks there is compelling evidence that magmatism migrated northwards with time.

Laser 4°Ar/39Ar geochronology Three samples of Gramado low Ti basalts have been analysed as a pilot study using the 4°Ar/39Ar laser technique. The aim was to understand the reasons for the apparently wide range (most fall between 100 and 160 Ma) and complexity of Paranfi K-Ar ages (Rocha-Campos et al. 1988, and references therein), when the available evidence from other well constrained CFB provinces, such as the Deccan, indicate very short

eruption histories (Duncan & Pyle 1988; Baksi & Farrar 1991). Baksi etal. (1991) reported stepped heating release spectra on a small selection of Paran~i basalts in abstract form, and cautioned that there were complexities in their data including both excess argon and recoil loss of 39Ar. Their preferred ages ranged from 130 to 135 Ma. Thus, the likely explanation for the range of apparent K-Ar ages reported from the Paranfi is that many of the rocks exhibit either excess argon or argon loss. In such situations it is normal to utilize stepped heating studies of whole rock or feldspar separates. The risk of pursuing such a path is that by crushing and in some cases separating minerals, information about the distribution of argon within and between different elements of the rock is lost. Additionaly, stepped heating 'bulk' samples averages the signals from all grains and sub-grains, which is valid if the argon concentrations are homogeneous or the distributions are symmetrical about grain centres. However, stepped heating can not easily decipher distributions in situations of mixed populations of single minerals, a fact illustrated well by recent combined laser and stepped heating 4°A/39Ar studies (Lo BeUo et al. 1987). Three samples were selected from a much larger sample set (Peate 1990; Peate et al. in press); one coarse grained basalt (MV6), one fine grained basalt (MV3) and one fine grained

230

C.J. HAWKESWORTH E T A L . (a) o~os~ 0.002

0.001

I 0.~

Fig. 6. A plot of 36Ar/4°Aragainst 39Ar/4°Arillustrating .mixtures of radiogenic, atmospheric and excess argon components. A simple age is represented as a mixing line between the radiogenic end member and atmospheric argon (whether modern or ancient). An excess argon component causes points to be displaced towards the orion. Commonly, in situations where excess argon is present, it is not pure, parentless 4°Ar, but a mixture of 4°Ar and argon with the atmospheric 36Ar/4°Arratio of 0.003384. This results in a mixing line between the radiogenic member and a point between the orion and the atmospheric ratio. basalt with plagioclase phenocrysts up to 2 mm in length (MG 1). The sample localities are given in Hawkesworth et al. (1988), and their relative stratigraphic positions are illustrated on Fig. 5. The analytical procedures are summarized in the Appendix. The samples were selected for their lack of alteration products and this has resulted in the exclusion of ages less than 130 Ma, suggesting that some of the rocks analysed in earlier studies were altered. The data are presented on argon correlation diagrams, a plot of a6ArPOAr against 39Ar-4°Ar (Roddick 1978; Heizler & Harrison 1988), which illustrates the mixture of radiogenic, atmospheric and excess argon components (Fig. 6). For samples MG1 and MV6 (Fig. 7a and b) we were able to analyse the different mineral phases, but MV3 proved to be too fine grained and so the individual laser analyses are better considered as total fusions. This is reflected in the lack of a linear data array in Fig. 7c. The overall range in apparent ages for samples MG1 and MV6 was 132-164 Ma, and the preferred isochron ages are 132.4_+1.4 and 132.9__.2.8 Ma respectively (for more detailed discussion, see Appendix). MV3 did not yield a sensible isochron age due to the lack of spread in the data, but apparent ages ranged from 147 to 188 Ma. In both MG1 and MV6, plagioclase analyses plotted closer to the atmospheric end point, partly as a result of low potassium contents, but also because they contained higher concentrationsof the atmospheric/excess argon mixture. The degree of cracking in the plagioclase appears to be correlated with the proportions of the excess Ar component. The same effect was seen in sample MV6, although it was not as marked. In summary, the laser technique is able to see

0.~

0.00

~.00

0.06

"0.00

(b) 0.003 --

"" ~ ' ,~. ,,, ,**~.

o.oo2 --

g

~

0~001 I 0.02

(c)

0.04 i~t3

0.003

g 0.001

o

i ox~

+"~ ~

"'~ ~"' ~' " " " " "

0.00

I

3.00

~Ar/4OAr

Fig. 7. 4°Ar/~Ar correlation diagrams for the Paran~i samples. (a) MG1, an isochron through all data yields an age of 134.8+ 4.2 Ma and a 4°Ar/39Arintercept of 319. Neglecting vein material and analyses close to veins, yields an age of 132.4+_1.4 Ma with an intercept of 299. (b) MV6, an isochron throughall data (solid line) yields an age of 132.9+2.8 Ma and a"Ar/~"Arintercept of 320 (dashed line indicates age/atmospbere mixing line). (e) MV3, no isochron, apparent ages range from 147 to 188 Ma (dashed line indicates age/atmospheric mixing line for 132 Ma). Closed squares represent cpx analyses; open squares represent plagioclase; closed circles represent analyses of vesicle filling material and closed triangles represent effectivelywhole rock analysesdue to the fine grained nature of MV3. through the excess argon component for MV6 and MG1 indicating a short eruption history (probably < 1 Ma) for at least the Gramado magma type of the Paran~i CFB, and that mag: matism took place several million years after the species extinction in the Tithonian (c. 141 Ma). The source of the excess argon appears to have been post-eruption percolation of low salinity fluids containing up to 20% parentless 4°Ar, which resulted in a complex mixed population of ages (for more detailed discussion, see Appendix). Contrary to normally held belief the groundmass of MG1, rich in small plagioclase and pyroxene grains which are traditionally thought to be susceptible to excess argon, contained a lower proportion of excess argon than

PARAN/k MAGMATISM AND SOUTH ATLANTIC OPENING the larger plagioelase phenoerysts. In MV6, the effect was less marked, but there was no indication of a greater susceptibility of pyroxene to the excess argon.

The role o f mantle plumes and the continental mantle Hthosphere Recent studies on the causes and nature of partial melting processes have focused the debate on the role of mantle plumes and the extent to which material from the continental mantle lithosphere contributes in the generation of CFB (McKenzie & Bickle 1988; White & McKenzie 1989; Arndt & Christensen in press; Gallagher & Hawkesworth 1992). Not all CFB provinces are the same, and in the Deccan, for example, there is good evidence for a major contribution from the sub-lithospheric mantle in that some of the basalt magma types are very similar to those erupted recently on the island of Reunion (Mahoney 1988). However, in the Gondwana CFB provinces of the Paran~i, the Karoo and the Ferrar, the overwhelming majority of samples analysed have distinctive trace element ratios (low Nb/La and Nb/Ba) and relatively enriched Sr, Nd and Pb isotope compositions (Duncan et al. 1984; Erlank et al. 1984; Hawkesworth et al. 1984, 1986, 1988; Petrini et al. 1987; Ellam & Cox 1989, 1991; Hergt et al. 1991; Peate et al. 1990, in press). Since such features are not commonly observed in oceanic basalts, and they occur in CFB samples which have been screened for the effects of contamination as the magmas passed through the continental crust (e.g. Mantovani & Hawkesworth 1990), they would appear to reflect distinctive source regions in the sub-continental mantle. Such source regions must be old, in order to develop the enriched radiogenic isotope ratios, and so most observers have concluded that they are situated within the continental mantle lithosphere. CFB provinces represent large volumes of magma generated in comparatively short periods of time, for example, a minimum of 1 x 106 km 3 in 2-3 Ma in the Paran~i-Etendeka province. Such rapid rates of magma generation indicate that partial melting took place in response to extension and decompression (White & McKenzie 1989). Moreover, in the case of the Paran~i it may be further inferred that the upper mantle was anomalously hot, both because of the presence of the mantle plume which is currently beneath Tristan da Cunha, and because similar volumes of magma are not associated with the opening of the South Atlantic to the north and south of the Paran~i basin. Most models have assumed that partial melting took place at the dry

231

peridotite solidus, and thus concluded that during continental extension most (> 95%) of the magmas were generated within the sub-lithospheric upper mantle (McKenzie & Bickle 1988; Arndt & Christensen in press). Such conclusions appear to contrast sharply with the isotope and trace element data on, in particular, the Gondwana CFB which suggest that at least theminor and trace element contents of these CFBs were largely derived from lithospheric source regions. In the following discussion two models are considered; one, in which the minor and trace elements are scavenged in small degree melts from the mantle lithosphere by asthenospherederived magmas, and a second in which melting takes place within the continental mantle lithosphere in the presence of small amounts of water. Figures 8 and 9 summarize the variations in Nd isotopes and selected trace element ratios in the Parand 'basalts'. As illustrated in Fig. 2, the various basaltic magma types recognized in the Parand lavas all have negative Nb anomalies on mantle normalized trace element diagrams, and so they have lower Nb/La than average MORB and OIB (Fig. 8b). The trend to lower end and Nb/La is also accompanied by a shift to higher Zr/Ti ratios than those commonly observed in oceanic basalts (Fig. 8a). On the eNd-Nb/La diagram simple two component mixing results in straight line arrays. Thus it may be concluded: (i) that if the isotope and trace element signatures of the Paranti basalts were generated by mixing between a lithosphere component with negative er~d, and an asthenospheric component with positive er~d, the latter had relatively depleted trace element characteristics similar to those in MORB; (ii) neither the recent Tristan da Cunha basalts, nor the late stage SRJ II Paranti dykes from the Santos-Rio de Janeiro section, plot on the main array of the Paran~i basalts in Fig. 8b. Thus, such trace element enriched, typically plume-related magmas do not appear to have contributed significantly in the generation of the main Paramt magma types. Ellam & Cox (1991) recently reinterpreted the positive correlation between end and Sm/Nd observed in the Nuanetsi high-MgO CFB in terms of mixing between lithospheric and asthenospheric components. The former was regarded as small degree melts, similar to lamproites, and the asthenospheric component was inferred to have had high Sm/Nd ratios and very low incompatible element contents. The lamproite-asthenosphere mixing line is reproduced in Fig. 9a, and while no well defined linear arrays are observed within the Paran~i basalts, the data might

232

C.J. HAWKESWORTH E T A L . 10"

(a)

MORB [] OIB

HB 0

(b.)

m MORB

o HB

m OIB

5"

ENd ~O-

0"

•

Tristan

00/~

Tristan

O •O'

-5-

-10 0

I

I

I

0.01

0.02

0.03

0

I

I

I

I

I

0.3

0.6

0.9

1.2

1.5

Zr/Ti

Nb/La

Fig. 8. Variations of end against Zr/Ti and Nb/La in selected basaits and basaltic andesites from the paran~i CFB. Open diamonds, Esmeralda; filled diamonds, Gramado; open squares, Urubici; filled triangles, SRJ II dykes; open triangles, remaining dykes from both the Ponta Grossa and Santos-Rio de Janeiro section. The element ratios for average MORB and OIB are from Sun & McDonough (1989), and HB is an analysis of a late stage Horingbaai dyke from the Etendeka (Duncan et a/. 1984; Hawkesworth et al. 1984). The fields for the recent basalts from Tristan da Cunha are from Le Roex et al. (1990).

10 (a)

[] OIB

5-

e

HB

R

MORB

e

HB

~ OIB

/ /

ENd O"

cb)

i

MORB

- -- --A

.

-0

--~

/ o

-5-

\ -10 -

[] iamprolte

-15 0.1

shale

I

t

0.2

0.3

Sm/Nd

0

iamprolte

I

I

I

200

400

6O0

T1/Y

800

Fig. 9. Variations of end against Sm/Nd and T'dY in selected basalts and basaltic andesites from the Parana CFB. Open diamonds, Esmeralda; filled diamonds, Gramado; open squares, Urubici; filled triangles, SR2 II dykes; open triangles, remaining dykes from both the Ponta Grossa and Santos-Rio de Janeiro section. The element ratios for average MORB and OIB are from Sun & McDonough (1989), and HB is an analysis of a late stage Horingbaai dyke from the Etendeka (Duncan et al. 1984; Hawkesworth et aL 1984). The element ratios for#verage shale and lamproite are from Taylor & McLennan (1985) and Bergman (1987), and they are plotted with e ~ = -11, since that is the value used by Ellam & Cox (1991) in their lamproite model for the petrogenesis of the Nuanetsi picrites. The dashed line in (a) is from Ellam & Cox (1991). Part (b) simply illustrates the results from the Parana lavas, and for reference the Tristan da Cunha rocks with > 6% MgO have Ti/Y = 7.50-1200 (Le Roex et al. 1990).

PARANA MAGMATISM AND SOUTH ATLANTIC OPENING be taken to be broadly consistent with the Ellam & Cox (1991) mixing line. The Gondwana CFB can be subdivided into .high- and low-Ti provinces (Bellieni et al. 1984; Cox 1988), and since lamproites have high incompatible element contents, and high Ti/Y ratios, they are much better suited as the lithospheric end member for the high-Ti than for the low-Ti magma types (Fig. 9b). The latter tend to have negative Ti anomalies on mantle normalized diagrams (Fig. 2), and such negative anomalies are rare in known small degree melts of the upper manthe. Rather, the combination of low Ti/Y and high Rb/Ba (see the Gramado and Esmeralda averages on Fig. 2) is a feature of upper crustal sediments, and this has encouraged models in which such low-Ti CFB were derived from source regions which contain a significant contribution from subducted sediment (e.g. Hergt et al. 1991). In summary, models can be set up in which most of the major elements in CFBs are derived from the asthenosphere, and most of the minor and trace elements are introduced in small degree melts (lamproites) from mantle lithosphere. However, two points should be noted. (i) Such models require that the asthenosphere derived end-member has very low incompatible element contents, and high Sm/Nd (Fig. 9a), and so it is much more depleted in trace elements than typical OIB (Ellam & Cox 1991). In the Paran~i such an agthenospheric end-member would have to be much more depleted than the plume-related basalts of Tristan da Cunha, or the stage SRJ II dykes (e.g. Fig. 8b), which in turn implies that two asthenospheric components (one enriched, and one depleted) were involved in the generation of the Paranfi CFB. (ii) The lamproite model is consistent with the data from the high-, but not the low-Ti CFB. No small degree melts have been identified with suitable trace element ratios to be the small degree melt component in low Ti/Y CFB, and so in recent models for the generation of the low-Ti Gondwana CFB it has been argued that they were derived from the continental mantle lithosphere, with little or no contribution from the underlying asthenosphere (e.g. Hergt et al. 1991). In practice lamproite-type models in which most of the major elements in CFB are derived from the asthenosphere, and the distinctive isotrope and trace element signatures are scavenged in small volume melts from the lithosphere, were initially developed in response to geophysical arguments that the continental mantle lithosphere is too cold and infertile to generate sufficient basalt. However, such arguments are based on calculations for melting at

233

the dry peridotite solidus, and yet it is well known that the addition of water to peridotite dramatically lowers its solidus temperature and promotes the crystallization of hydrous phases such as amphibole and phologopite (Green 1973; Mysen & Boettcher 1975; Olafsson & Eggler 1983). An alternative approach, therefore, is to consider the effects of small amounts of water on models of partial melting during lithospheric extension, although it should be emphasized that the way in which melt compositions vary with P, 7", and Xa2o, and how water behaves during partial melting, are not well constrained. Fertile mantle can accommodate up to c. 0.4% H20 in the form of amphibole produced by the hydration of clinopyroxene, with the amount of water accommodated in this way depending on the modal abundance of clinopyroxene (Olafsson & Eggler 1983). Melts produced from such hydrated peridotite would therefore have compositions controlled by amphibole, rather than clinopyroxene and, at least at higher degrees of melting, the partial melts will be silica saturated (e.g. Kushiro 1990). In fertile mantle, in which 10-20% clinopyroxene has been converted to amphibole, it is clear that considerable volumes of melt could be produced at the relatively low temperatures of the amphibole peridotite solidus. However, the mantle lithosphere is generally thought to be relatively depleted in major element consitutents of basalt, such as Ca and AI (Maaloe & Aoki 1977; Boyd & Mertzman 1987; Hawkesworth et al. 1990), and so it will contain less clinopyroxene, or less amphibole (in the presence of H20). Melting a depleted peridotite composition also produces silica-saturated melts because the melts are dominated by the orthopyroxene contribution. Thus, the combined effect of the presence of water in amphibole, and a depleted major element composition, indicate that hydrous melting of depleted peridotite, as expected in the lithosphere, will produce silica-saturated basaltic melts. The depleted nature of the continental mantle lithosphere will obviously limit the modal abundance of amphibole, but even with 5-10% amphibole, the amount of melt produced could still be 10-15%, with the remaining major element contribution being derived from orthopyroxene and olivine. Thus, the production of a melt thickness of 2 km may require a source region that is equivalent to only a 15-20 km thick layer within the continental lithosphere. Finally, there is the question of whether up to 0.4% H20 in the mantle lithosphere is reasonable. This is clearly difficult to establish, but Jambon & Zimmermann (1990) reported

234

C.J. HAWKESWORTH E T A L .

H 2 0 data on MORB #asses, and argued that the H 2 0 / K 2 0 ratio in the source of N-MORB was

probably c. 4. Moreover, if the sub-continental mantle contains a contribution from previous subduction episodes, as has been invoked to account for the relative depletion in Nb and Ta in many continental basalts, then the H 2 0 / K 2 0 ratio will be even higher (see disscussion in Saunders & Tamey 1991). If H20/K20 ratios of -> 4 are applicable to the continental lithosphere, they are consistent with a H 2 0 contents of -> 0.4%, given previous estimates of 0.10.15% K20 (e.g. Hawkesworth et al. 1990).

Melt generation in the presence of small m o u n t s of water during continental extension Gallagher & Hawkesworth (1992) recently explored the potential of dehydration melting as a mechanism for generating CFB. They assumed that the mantle lithosphere discussed by petrologists and geochemists is effectively equivalent to the mechanical boundary layer (MBL) and so the calculations were designed to evaluate the contribution of melt from this region, relative to that from the underlying asthenosphere. In this approach, the MBL is specified a priori and a horizontally averaged geotherm is calculated ac-

Temperature(°C) 0

500

1000

1500

2000

0 1

3 W

_

_

4

8 (L 5

6

Fig. 10. The hydrous solidus (Olafsson & Eggler 1983) and the anhydrous solidus (McKenzie & Biclde 1988) used in this work. Also shown are geotherms for potential temperatures of 1480°C,and MBL of (a) 100 kin, and (b) 200 km. The geotherm was calculatedusing the methodology and basic parameters of McKenzie & Bickle (1988), with a mantle viscosityof 4x l 0 ts m 2 s - t . Note that the hydrous solidus is intersected in the lower part of the MBL in both cases and so melting is predicted in this region.

cording to McKenzie & Bickle (1988) for a mantle potential temperature of 1480°C, and this geotherm is shown on Fig. 10. The occurrence of CFB is limited compared with the total length of the associated rift margins implying a local control such as a mantle plume. The hydrous peridotite solidus was taken from Olafsson & Eggler (1983), which has c. 0.3% water (and c. 0.7% CO2) held in amphibole or phlogopite, and this is also shown in Fig. 10, together with the anhydrous peridotite solidus from McKenzie & Bickle (1988). The hydrous solidus was considered appropriate over the depth interval from 50 km to the base of the MBL, and the depth of the base of the MBL was varied. Above and below this interval the anhydrous peridotite solidus of McKenzie & Bickle (1988) was used, and so there is a window of hydrous, and therefore lower temperature, melting. The choice of 50 km for the upper boundary is arbitrary, but unimportant as no melting is predicted near such depths in any of the models considered. The potential temperature of 1480°C is appropriate for a plume (Watson & McKenzie 1991; White & McKenzie 1989; McKenzie & Bickle 1988) and is equivalent to mantle c. 200°C hotter than normal. As can be seen from Fig. 10, the lower part of the geotherm intersects the hydrous solidus, and so melting would occur in this region. For a potential temperature of 1280°C, or normal oceanic mantle temperatures, the geotherm still intersects the lowermost region of the hydrous solidus, while for 1180°C, the geotherm is always less than the solidus. Therefore, the amount of melt generated in the presence of a plume will also depend on the pre-plume solidus, although, as cratonic regions are generally relatively cool at depth, a lower effective mantle potential temperature may be appropriate for calculating the geotherm. In this context, the question also arises as to how long it takes to heat up the lithosphere in order to achieve melting. If hot asthenosphere is advected into the MBL, then the lithosphere may be heated quickly and then melt. However, in this situation melting will generally also occur in the upwelling asthenospheric material. In the end member of melting only in the lithosphere, or MBL, heating will primarily be by conduction, although it is only the lower part of the MBL that needs to be at significantly elevated temperatures. In this case, appropriate timescales are of the order of 20-40 Ma to achieve 30-50% of the total melt generation possible with the steady state 1480°C potential temperature geotherm for MBL = 150 km. These timescales are consistent with those required to move over a hotspot with a diameter between 1000 and 2000 km, using average plate velocities of 5 cm per year (50 km Ma -1) and the

PARANA MAGMATISM AND SOUTH A'I"LANTIC OPENING implication is that to melt the lithosphere, a plume needs to be insulated on timescales of 10100 Ma. In the following discussion it is assumed that the pre-plume geotherm is always less than the solidus, the geotherm is in steady state equilibrium, and so the predicted melt thicknesses may be regarded as upper limits. In order to calculate the degree of melting as a function of temperature, Gallagher & Hawkesworth (1992) used the very limited experimental data of Olafsson & Eggler (1983) to constrain a pressure-independent melt fraction-temperature function of the same form to that used by McKenzie & Bickle (1988). The peridotite liquidus of McKenzie & Bickle (1988) was used in all calculations, although the details of the liquidus position are not significant as the predicted melt fraction never exceeded the upper limit of the experimental observations (c. 30%). The solidus and melt function together allow the melt fraction as a function of depth to be calculated and also the total melt thickness to be

50%

10o% 110-

10%

701.0

11.2 1~1 1.11

1~

~O

I 2.2

I ;I.4

I 2.6

I 2.8

I 3.0

235

predicted. A final point to note is that the MBL derived melt may progressively freeze because adiabatic cooling can result in the material cutting across lines of constant melt fraction. This freezing effect may be allowed for by monitoring the melt temperature relative to an appropriate melt solidus and liquidus (see Gallagher & Hawkesworth 1992). The results of the model calculations, with and without freezing, are summarized in Fig. 11 in terms of total melt thickness generated and the fraction of the total melt derived from the MBL, as a function of lithospheric extension factor, ~. In the absence of extension (/3 = 1.0), the generation of larger quantities of melt with no asthenosphere contribution is favoured by a thicker MBL because more material is above the solidus as a result of the impinging hot plume. For the model considered here, > 5 km of melt (or > 7 km if no melt freezes) is predicted for a MBL thickness of 200 km with a maximum of 25% melting at the base of the MBL. The melt thickness generated for MBL of 150 and 100 km, values which are probably more appropriate to real geological situations, are about 3.6 and 1.2 km respectively. These estimates increase to 5.0 and 3.2 km if no melt freezes. In the situation where extension of the lithosphere occurs over anomalously hot mantle, additional melt is generated through adiabatic decompression. As the amount of lithospheric extension increases, so too does the asthenosphere contribution to the total melt generated. In order to estimate quantitatively the relative

1.0

100%

~4) ~QO_.o., 0.8

t~

~ 1so

04) E "~--'0

\\

L

.o~

1so 7

1.0

1.2

1,4

1.6

1.8

2.0

2.2

S0%

2.4

2.6

2.8

o.e

~

~q~\ 200km ', ~"\ 150km ~ ' \, ,

,

,,

.

•

1.0 Fig. 11. (a) Calculated melt thickness as a function of MBL thickness and extension factor, 8, using a geotherm with a potential temperature of 14800C(see Fig. 10). The contours labelled 100, 50 and 10% represent the percentage contribution of melt derived from the MBL to the total melt thickness, and to the left of 100% contour all the melt is generated in MBL. (b) As (a) but the effects of freezing are not included.

1.5

2.0

2.5

3.0

Fig. 12. The fraction of the total melt derived entirely from within the MBL as a function of extension factor,/3 for 3 values of the MBL thickness. The solid lines include the effects of melt freezing while the dashed lines do not. The actual melt thicknesses can be determined from Fig. lla and b for the appropriate MBL thickness.

236

C.J. HAWKESWORTH E T A L .

contributions of asthenosphere and lithosphere melting during rifting, Gallagher & Hawkesworth (1992) closely follow the methodology of McKenzie (1984) and McKenzie & Bickle (1988), assuming instantaneous rifting with a steady state geotherm and mantle potential temperature of 1480°C. A value of 250 J kg -1 eC -~ was used for the entropy change on melting and entropy was conserved to 1 part in 105 during the calculation. Figures 11 and 12 illustrate that a thicker MBL favours a larger proportion of dehydration melt for a given amount of lithospheric extension, although a small amount of MBL derived melt is predicted in all cases. Extension factors greater than 1.2-1.3 are sufficient to allow asthenosphere melts rapidly to dominate the magmatic signature when the MBL is 100 km. However, for a value of 200 km, the lithosphere may be thinned by over 200% before the orginal MBL signature is lost (Fig. 12). Thus, the general trend is that thicker MBL may undergo a greater amount of extension before the underlying asthenosphere begins to melt. Provided that the melt can be extracted quickly enough, a time dependent chemical stratigraphy will develop in a progressively extending region, with the initial melts being lithosphere dominated and the later melts having more contribution from the asthenosphere. A striking feature of CFB is their large volumes, and simple geometrical arguments imply that an average melt thickness of 320 m is required over the total area of a mantle plume of radius 1000 km to produce 106 km 3, approximately the volume of the Paran~i province. Alternatively, this volume is consistent with an average 3 km of melt thickness produced over a region 150-200 km wide across the centre of the plume, or a circular region of radius 320 km. This range of melt thicknesses is possible with the dehydration model, although a requirement of no asthenospheric melting implies that the total extension factor (/3) would be at most 2.0 and more likely 1.5. In summary, the behaviour of water during partial melting in the MBL, and the composition of subsequent melts are very poorly constrained. However large volumes of CFB are a feature of the geological record, and in many areas their major and trace elements and radiogenic isotope ratios indicate they were derived from source regions different to those sampled by oceanic basalts. Lamproite type models in which the major elements of such CFB are derived from the asthenosphere and most of the miflor and trace element abundances are introduced in small degree melts scavanged from the mantle lithosphere, may be appropriate in some areas.

However, these models require the presence of a very incompatible element depleted asthenospheric end member and yet in those flood basalt provinces where asthenosphere derived melts can be readily identified they tend to have high incompatible element contents, similar to OIB. In addition there is general agreement that lamproite type models are not appropriate for lowTi CFB. A valid alternative may therefore be that those CFB with major and trace element features different from common oceanic basalts are simply derived by partial melting within the MBL, in the presence of very small amounts of water. The results of model calculations indicate that for a MBL between 100 and 200 km, up to 6 km of melt may be generated entirely from within the MBL, for/3 values of less than 1.2 (Gallagher & Hawkesworth 1992).

Conclusions (i) Six magma types are now recognized within the extrusive basalts and basaltiv andesites of the Paran~i CFB (Peate et al. 1990, in press), and data from borehole and surface samples have been used to demonstrate that the Paran,'i lavas comprise an overlapping series of units dipping towards the north. It is concluded that the dominant magma type evolved from low Ti, to high Ti to intermediate Ti with time, and that the source for the Parami magmas migrated northwards. (ii) New analyses on the associated dyke rocks of Ponta Grossa Arch and Santos-Rio de Janeiro section, indicate that while the former are similar to the Pitanga and Paranapanema magma types, the latter includes dykes with compositions not observed in the Paran~i lavas. Rather their minor and trace elements are strikingly similar to basalts recently erupted on Tristan da Cunha, and thus these late stage dykes may represent the first direct evidence for the involvement of typical plume-related OIB in the Paran~i province. (iii) Preliminary laser 4°Ar/39Ar studies of selected Gramado low Ti basalts have yielded preferred isochron ages of 132.4+1.4 a n d 132.9+2.8 Ma. These indicate a short eruption time for at least the Gramado magma type, and that magmatism took place several millionyears after the species extinction in the Tithonian (c. 141 Ma). (iv) The majority of basalts and basaltic andesites in the Paran~i CFB have distinctive trace element ratios (low Nb/La and Nb/Ba), and relatively enriched Sr, Nd, and Pb isotope compositions (Hawkesworth et al. 1986, 1988; Petrini et al. 1987; Cordani etal. 1988; Peate etal. 1990, in

PARANA MAGMATISM AND SOUTH ATLANTIC OPENING press). Since such features are not commonly observed in oceanic basalts, and they occur in CFBs which have been screened for the effects of crustal contamination, they are typically attributed to old, incompatible element enriched source regions in the continental mantle lithosphere. (v) Most models of partial melting in response to extension and decompression have assumed that partial melting took place at the dry peridotite solidus, and consequently concluded that most ( > 95%) CFB type magmas were generated in the sub-lithospheric upper mantle (McKenzie & Bickle 1988; Arndt & Christensen in press). Thus, it has been argued that the minor and trace element 'mantle lithosphere' component of, for example, the Gondwana CFBs was introduced in small degree melts (lamproites) added to asthenosphere derived magmas (Ellam & Cox 1991). However, such a model appears to be inconsistent with the data from low Ti CFB, and it also requires-that the asthenosphere derived magmas have very low incompatible element contents, in marked contrast to the high Nb/La late stagedykes in the Paranfi. (vi) A n alternative approach is to investigate the amounts of melt generated from the mechanical boundary layer in the presence of small amounts of water. In practice, the behaviour of water during partial melting in the MBL, and the compositions of the subsequent melts are very poorly constrained. Nonetheless, the results of preliminary calculations indicate that for a MBL between 100 and 200 km, in the presence of a mantle plume, up to 5 km of melt may be generated entirely from within the MBL, for/3 values of less than 1.2 (Gallagher & Hawkesworth 1992). As extension progresses (higher /3), the proportion of melt from the underlying asthenosphere increases rapidly as observed for example in both the Paranfi and the opening of the south Altantic, and the Basin and Range Province in the western USA. (vii) In the currently preferred model, the Paranfi CFB were generated in response to the northwards rifting of the south Atlantic, over an area of anomalously hot mantle presently associated with magmatism on Tristan da Cunha. Until the generation of the late stage high Nb/La dykes, the mantle plume would appeared to have contributed heat, rather than significant volumes of asthenosphere derived magmas to the generation of the Paranfi CFB. This project is part of a joint IAG-USP and Open University programme. We thank K. G. Cox, A. J. Erlank, J. Hergt, P. R. Hooper, and R. S. White for many discussions on the origin of these enigmatic continental flood basalts, and A. D. Saunders and J.

237

Pearce for their detailed and constructive reviews of this manuscript.

Appendix: Laser 4°/39Ara n a l y s e s Rock slices 1 cm2 and around 500/,m thick, were polished on one side and irradiated at the Ford reactor, Michigan, where they received l x 10TM fast n cm -2. The resulting J values for the three samples were calculated using the mmhbl and hb3gr standards, and were, 0.00595(MV6), 0.00605(MV3) and 0.00602(MG1), with errors of around 0.5%. When returned, the samples were loaded into the laser port and baked using heating tape and a heat lamp. Blank levels during the analysis were 2, 0.06, 0.03, 0.06, and 0.04× 10-12 ClTI3 STP for 4°Ar, 39Ar, 38Ar, 3TArand 36Ar respectively. Argon was extracted from the rock slices by firing short pulses of a continuous Nd-YAG laser beam (TEM00, wavelength 1064 nm), focused at the sample surface. Typical powers of 10 to 17 W were used and pulse lengths of 2 to 100 ms. Argon was released by melting single grains or several adjacent grains of the same mineral. Resulting isotope abundances were corrected for reactor interferences, mass spectrometer discrimination, and decay of 3TAr. A range of apparent ages from c. 130 to 188 Ma was obtained from the three samples, which is similar to that for the published bulk K-Ar ages on over 200 samples from the Paran~, except that none of the ages reported here are < c. 130 Ma. A least squares fitted isochron age of all the data from MG1 yielded an isochron of 134.8+ 4.2 Ma (lo-), with a 36Ar/4°Arintercept of 0.003138+0.00017 (4°Ar/36Ar = 319) and an MSWD of 0.7. MV6 yielded an age of 132.9+2.8 Ma (lcr), with a 36Ar~Ar intercept of 0.003121+0.00009 (4°Ar/36Ar = 320) and an MSWD of 0.5. MV3 did not yield a sensible isochron age due to the lack of spread in the data, but apparent ages ranged from 147 to 188 Ma. In both MG1 and MV6, plagioclase analyses plotted closer to the atmospheric end point, partly as a result of lower potassium contents, but also because they contained higher concentrations of the atmospheric/ excess argon mixture. Superficially, the laser has simply separated minerals and allowed samples to be analysed by an isochron technique. However, features of the laser probe other than its high spatial resolution can be used to extract additional information by, for example, varying the laser pulse. Varying the pulse can be viewed as a crude form of stepped heating; more sophisticated laser stepped heating analysis can be achieved, though minerals have to be separated beforehand. In this case we performed two-step analyses of plagioclase grains in situ in MG1. The first step consisting of short 1 to 5 ms pulses across the surface of the grain, causing no visible melting, the second using pulse lengths up to 100 ms, causing the grain to melt. In all cases, the plagioclase yielded to low temperature step poor in radiogenic argon but relatively rich in the atmospheric/ excess argon end member (Fig. 7a). Subsequent melting produced higher proportions of radiogenic argon though still generally less than the groundmass. The degree of cracking in the plagioclase seemed to be correlated with the proportions of the excess Ar

238

C.J. HAgrKESWORTH ETAL.

component. The same effect was seen in sample MV6, although it was not as marked. There are several analyses of MG1 which give little or no indication of any excess argon component. These either interacted with a fluid containing negligible excess argon or they were sited in areas unaffected by these late stage fluids. Neglecting vein material, low temperature plagioclase analyses and groundmass points close to veinlets; 7 points remain (out of a total of 14). An isochron through these points yields an age of 132.4+1.4 Ma and a 36Ar/4°-Ar intercept of 0.003344+0.00003 (4°Ar/36Ar= 299+3) with an MSWD of 0.8. Moreover, the closeness of the intercept to the true atmospheric Ar isotope ratio gives us confidence that the result is a true age. Another advantage of analysing rock sections is that all the features present can be analysed and their importance assessed. Two veinlets of a white carbonate rich material in the MG1 section contained measurable quantities of both 39Ar (from potassium) and 37At (from calcium) and some of the highest proportions of atmospheric/excess argon measured (Fig. 7a). Additionally, analyses of groundmass close to the vein yielded higher proportions of the excess Ar signature. In the light of this irregular distribution of the atmospheric/ excess argon, it seems most likely that it originated in fluids percolating through the volcanic pile subsequent to eruption, rather than argon derived from depth as a component of the magma. Such fluids were most probably associated with the renowned agates of the Paran~i CFB, and the low 3SAr contents (derived from chlorine during irradiation) of the vein material relative to other analyses indicate that the fluids had low salinity. Further, by assuming the age for MG1 we may estimate the 4°Ar/36Ar ratio of the fluid end member, which varys from 370 to 295. This indicates a mixture varying from one with 80% atmospheric argon and 20% excess argon to a fluid with negligible excess argon. Variation in the fluid end member is responsible for the high errors on the ages calculated using all the data points, since it affects the goodness of fit of the isochron (Fig. 7a). More importantly, the variation of the 4°Ar/36Ar ratio in the fluid end member creates the analytical problem of mixed populations within single minerals. In the case of MG1, separating and analysing the plagioclase penocrysts (the standard technique), would not have been the best course for stepped heating analysis, though this may not be common to all phenocryst rich volcanics.

References ALT, D., SEARS,J. W. • HYNDMAN,D. W. 1988. Terrestrial Maria: the origins of large basalt plateaus, hotspot tracks, and spreading ridges. Journal of Geology, 96, 647-662. ARNDT,N. T. & CHRISTENSEN,U. In press. The role of lithospheric mantle in continental, flood volcanism. Journal of Geophysical Research. BAKSI, A. K. & FARRAR,E. 1991.4°Ar/39Ar dating of the Siberian Traps, USSR: Evaluation of the ages of the two major extinction events relative to episodes of flood-basalt volcanism in the

USSR and the Deccan Traps, India. Geology, 19, 461-464. ...... , FODOR, R. V. & FARRAR,E. 1991. Preliminary results of 4°Ar/39Ardating studies on rocks from the Serra Geral flood-basalt province and the Brazilian Continental Margin. Abstracts, EOS Spring

Meeting, 300. BEANE, J. E., TURNER, C. A., HOOPER, P. R., SUBBARAO,K. V. & WALSH,J. N. 1986. Stratigraphy, composition and form of the Deccan basalts, western Ghats, India. Bulletin of Volcanology, 48, 61-83. BELLIENI, G., COMIN-CHIARAMONTI,P., M~.~OUES,L. S., ]VIELFI,A. J., PICCIRILLO,E. M., NARL"', A. J. R. & ROISENBERO, A. 1984. High- and h. v-Ti flood basaits from the Paran~i plateau (Braz,'X: petrology and geochemical aspects bearing on their mantle origin. Neues Jahrbuch Mineralogische Abhandelungen, 150, 273-306. BERGMAN, S. C. 1987. Lamproites and other K-rich igneous rocks: a review of their occurrence, mineralogy and geochemistry. In: FrrroN, J. G. & UPTON, G. B. J. (eds) Alkaline Igneous Rocks. Geological Society, London, Special Publication, 30, 103-190. BOYD, F. R. & MERTZMAN,S. A. 1987. Composition and structure of the Kaapvaal lithosphere, south Africa. In: MYSEN, B. O. (ed.) Magmatic Processes: Physiochemical Principles. Geochemical Society Special Publication, 1, 13-24. CORDANI,U. G., CIVETrA, L., MANTOVANI,MI S. M., PETRINI, R., KAWASH1TA,K., HAWKESWORTH,C. J., TAYLOR,P., LONGINELLI,A., CAVAZZINI,G. & PICCIRILLO, E. "M. 1988. Isotope geochemistry of flood volcanics from the Paran~i basin (Brazil). In" PICCIRILLO, E. M. ,g" MELFI, A. J. (eds) The

Mesozoic flood volcanism of the Paran~ basin: petrogenetic and geophysical aspects. IAG-University of Sao Paulo Press, 157-178. COUR~LLCrr, V. E. & OsowsKI, C. 1987. The Cretaceous-Tertiary boundary events: external or in, ternal causes? EOS, Transactions of the American Geophysical Union, 68, 193"200. Cox, K. G. 1988. The Karoo province. In: MACDOUOALL, J. D. (ed.) Continental flood basalts. Kluwer Academic Publishers, 239-271. & HAWKESWORTH, C. J. 1985. Geochemical stratigraphy of the Deccan traps at Mahabaleshwar, Western Ghats, India, with implications for open system magmatic processes. Journal of Petrology, 26, 355-377. DEVEY, C. W. & LIGHTFOOT,P. C. 1986. Volcanological and tectonic control of stratigraphy and structure in the western Deccan traps. Bulletin of Volcanology, 48, 195-207. DUNCAN,A. R., ERLANK,A. J. & MARSH, J. S. 1984. Regional Geochemistry of the Karoo Igneous Province. In: ERLANK, A. J. (ed.) Petrogenesis

of the Volcanic Rocks of the Karoo Province. Special Publication, Geological Society of Africa, 13, 355-388. - - 8,~ARMSTRONG,R. A. 1990. MORBrelated dolerites associated with the Etendeka Volcanics, Northwestern Namibia. In: M~LNER, S. C., DUNCAN,A. R., MARSH, J. S., MILLER,

PARANA MAGMATISM AND SOUTH ATLANTIC OPENING R.McG. & BAILEY,M. M. (compilers) Field Excursion Guide to the Etendeka Volcanic Rocks and associated intrusions, N. W. Namibia. DUNCAN, R. A. 1984. Age progressive volcanism in the New England seamounts and the opening of the central Atlantic Ocean. Journal of Geophysical Research, 89, 9980-9990. & PYLE, D. G. 1988. Rapid eruption of the Deccan flood basalts at the Cretaceous/Tertiary boundary. Nature, 333, 841-843. ELLAM, R. M., & Cox, K. G. 1989. A Proterozoic lithospheric source for Karoo magmatism: evidence from the Nuanetsi picrites. Earth and Planetary Science Letters, 92, 207-218. & 1991. An interpretation of Karoo picrite basalts in terms of interaction between asthenospheric magmas and the mantle lithosphere. Earth and Planetary Science Letters, 105, 330342. ERLANK,A. J., MARSH,J. S., DUNCAN,A. R., MILLER, R.McG., HAWrmSWORTH,C. J., BETrON, P. J. REx, D. C. 1984. Geochemistry and petrogenesis of the Etendeka volcanic rocks from SWA/ Namibia. In: ERLANK, A. J. (ed.) Petrogenesis

of the Volcanic Rocks of the Karoo Province. Special Publication, Geological Society of South Africa, 13, 195-246. FULFARO, V. J., SAAD, A. R., SAMOS, M. V. & VIANNA,R. B. 1982. Compartimenta~o e evolu~ao tectonica da Bacia do Parawt. Revista Braseira de Geoci@ncias, 12, 590-611. GALLAGHER, K. & HAWKESWORTH, C. J. 1992. Dehydration melting and the generation of continental flood basalts. Nature, 358, 57-59. , - - , LEWIS, C. & MANTOVANI,M. S. M. 1991. The evolution of rift-margin topography: a preliminary model for the onshore continental margin of S.E. Brazil. IAVCEI, X X General

Assembly, IUGG Vienna, l l - 2 4 August 1991. GREEN, D. H. 1973. Experimental melting studies on a model of upper mantle compositonal high pressures under water-satuarated and water unsaturated conditions. Earth and Planetary Science Letters, 9, 37-53. HAWKESWORTH,C. J., KEMPTON, P. D., ROGERS, N. W., ELLAM, R. M. & VAN CARLSTEN,P. 1990. Continental mantle lithosphere, and shallow level enrichment processes in the Earth,s Mantle. Earth and Planetary Science Letters, 96, 256-268. ~, MANTOVANI,M. S. M. & PEATE, D. W. 1988. Lithosphere remobilisatien during Paran~i CFB magmatism. In: MENSES, M. A. & COX, K. G. (eds) Oceanic and continental lithosphere: similarities and differences. Journal of Petrology Special Volume, 205-223. - - , - - , TAYLOR P. N. & PALACZ, Z. 1986. Evidence from the Parami of south Brazil for a continental contribution to Dupal basalts. Nature, 322, 356-359. ..... , MARSH, J. S., DUNCAN,A. R., ERLANK,A. J. & NORRV, M. J. 1984. The role of continental lithosphere in the generation of the Karoo volcanic rocks: evidence from combined Nd- and Sr-isotope studies. In: ERLANK, A. J. (ed.) Pet-

239

rogenesis of the Volcanic Rocks of the Karoo Province. Special Publication, Geological Society of South Africa, 13, 341-354. HEIZLER, M. T. & HARRISON,T. M. 1988. Multiple trapped argon isotope components revealed by 4°Ar/39Ar isochron analysis. Geochimica et Cosmochimca Acta, 52, 1295-1303. FIERGT, J., PEATE, D. & HAWKESWORTH,C. J. 1991. The petrogenesis of Mesozoic Gondwana low-Ti flood basalts. Earth and Planetary Science Letters, 105, 134-148. HOOPER, P. R. 1982. The Columbia River Basalts. Science, 215, 1463-1466. HUMPHRIS,S. E. & THOMPSON,G. 1983. Geochemistry of REE in basalts from the Walvis Ridge: implications for its origin and evolution. Earth and Planetary Science Letters, 66, 223-242. - - , SCHILLING,J.-G. & KINGSLEY,R. H. 1985. Petrological and geochemical variations along the Mid-Atlantic Ridge between 46°S and 32°S: Influence of the Tristan da Cunha mantle plume. Geochimica et Cosmochimica Acta, 49, 1445-1464. JAMBON, A. & ZIMMERMANN,J. L . 1990. Water in oceanic basalts: evidence for dehydration of recycled crust. Earth and Planetary Science Letters, 101,323-331. KUSmRO, I. 1990. Partial melting of mantle wedge and evolution of island arc crust. Journal of Geophysical Research, 95, 15929-15939. LE ROEX, A. P., CLIFF, R. A. & ADAIR, B. J. I. 1990. Tristan da Cunha, South Atlantic: Geochemistry and petrogenesis of a basanite-phonolite lava series. Journal of Petrology, 31, 779-812. LIGHTFOOT, P. & HAWKESWORTH,C. J. 1988. Origin of Deccan Trap Lavas: evidence from combined trace element and Sr-, Nd- and Pb-isotope studies. Earth and Planetary Science Letters, 91, 89-104. Lo BELLO,PH., FI~RAUD,G., HALL, C. M., YORK, D., LAVIRA, P. & BERNAT, M. 1987.4°Ar/3°Ar stepheating and laser fusion dating of a quaternary pumice from Neschers, Massif Central, France: The defeat of xenocrystic contamination. Chemical Geology (Isotope Geoscience), 66, 61-71. MAALOE, S. & Aora, K. 1977. The major element composition of the upper mantle estimated from the composition of lherzolites. Contributions to Mineralogy and Petrology, 63, 161-173. MAnONEV, J. J. 1988. Deccan Traps. In: MACDOUGALL,J. D. (ed.) Continental Flood Basalts. Kluwer Academic Publishers, 151-194. MANGAN, M. T., WRIGHT, T. L., SWANSON,D. A. & BYERLY, G. R. 1986. Regional correlation of Grande Ronde flows, Columbia River basalt group, Washington, Oregon and Idaho. Geological Society of America Bulletin, 97, 1300-1318. MAm'OVANI, M. S. M. & HAWKESWORTH,C. J. 1990. An inversion approach to assimilation and fractional crystallisation processes. Contributions to Mineralogy and Petrology, 105, 289-302. , MARQUES,L. S., DE SOUSA,M. A., CIVETrA,L., ATALLA, L. & INNOCENTI,F. 1985. Trace element and strontium isotope constraints on the origin and

240

C. J. HAWKESWORTH E T A L .

evolution of Paramt continental flood basalts of Santa Catarina State (southern Brazil). Journal of Petrology, 26, 187-209. McKENZIE, D. P. 1984. The generation and composition of partially molten rock. Journal of Petrology, 25, 713-765. & BICKLE,M. J. 1988. The volume and composition of melt generated by extension of the lithosphere. Journal of Petrology, 29, 625-679. MYSEN, B. O. & BOETTCHER,A. L. 1975. Melting of a hydrous mantle: II. Geochemistry of crystalds and liquids formed by anatexis of mantle peridotite at higher pressures and high temperatures as a function of controlled activities of water hydrogen and carbon dioxide. Journal of Petrology, 16, 549-593. OFFICER, C. B. & DRAKE, C. L. 1985. Terminal Cretaceous environmental events. Science, 227, 1161-1167. OLAFSSON,M. & EOOLER,D. H. 1983. Phase relations of amphibole, amphibole-carbonate and phlogopite-carbonate peridotite: petrologic constraints on the asthenosphere. Earth and Planetary Science Letters, 64, 305-315. PEATE, D. W. 1990. Stratigraphy and petrogenesis of the Parand, continental flood basalts, southern Brazil. PhD thesis, Open University. - - , HAWKESWORTH,C. J. & MANTOVANI,M. S. M. (In press). Chemical stratigraphy of the Paran~ lavas (South America): classification of magma types and their spatial distribution. Bulletin of Volcanology. & SHUKOWSK~',W. 1990. Mantle plumes and flood basalt stratigraphy in the Paranfi, South America. Geology, 18,1223-1226. PETRINI, R., CIVET]A, L., PICCIRILIX),E. M., BELLIENI, G., COMIN-CHIARAMONTI,P., MARQUES, L. S. & MELVa,A. J. 1987. Mantle heterogeneity and crustal contamination in the genesis of low-Ti continental flood basalts from the Parami plateau (Brazil): Sr-Nd isotope and geochemical evidence, Journal of Petrology, 28, 701-726. PICCIRILLO, E. M., BELLIENI, G., CAVAZZINI, G., COMIN-CHIARAMONTI,P., PETRINI,R., MELH, A. J., PINESE, J. P. P., ZAWrADESCHI,P. & DE MIN, A. 1990. Lower Cretaceous tholeiitic dyke

swarms from the Ponta Grossa Arch (southeast Brazil): petrology, Sr-Nd isotopes and genetic relationships with the Paran~ flood volcanics. Chemical Geology, 89, 19-48. RAMPmO, M. R. & STOTHERS,R. B. 1988. Flood basalt volcanism during the Past 250 million years. Science, 241,663-668.

ROCHA-CAMPOS, A. C., CORDANI, U. G., KAWASHITA, K., SONOKI, H. M. & SONOZI, I. K. 1988. Age of the Parand flood volcanism. In: PICOmLLO, E. M. & MELW,A. J. (eds) The Mesozoic flood vol-

canism of the Pararut basin: petrogenetic and geophysical aspects, lAG-University of Sao Paulo press, 25-46. RODmCK, J. C. 1978. The application of isochron diagrams in 4°Ar-39Ar dating: A discussion. Earth and Planetary Science Letters, 41, 233-244. Sus, S. S. & MeDONOUGH, W. M. 1989. Chemical and isotopic systematics of oceanic basalts: implications for mantle compositions and processes. In: SAUNDERS,A. D. & NORRY,M. J. (eds) Magmatism in Ocean Basins. Geological Society, London, Special Publication, 42, 313-345. SwAlqSO~, D. A., WRIOWr, T. L., HOOVER, P. R. & BENTLEY, R. O. 1979. Revisions in stratigraphic nomenclature of the Columbia River Basalt Group. US Geological Survey Bulletin, 14Lq7-G, G1-G59. TAYLOR, S. R. & MCLENIqAU,S. M. 1985. The Continental Crust: its composition and evolution. Blackwell Scientific Publications. WATSON, S. & MCKENZIE, D. 1991. Melt generation by plumes: A study of Hawaiian Volcanism. Journal of Petrology, 32, 501-537. WHITE, R, S. & MCKENZlE, D. P. 1989. Magmatism at rift zones: the generation of volcanic continental margins and flood basalts. Journal of Geophysical Research 94, 7685-7730. ZALAN, P. V., WOLF, S., CONCEICAO,J. C. J., AsTom, M. A. M., VIEIRA, l. S., APPI, V. T., ZANOTrO, O. A. & MARQUES,A. 1991. Tectonics and sedimentation of the Paranfi Basin. In: ULBmCH, H. & ROCHA-CAMPOS,A. C. (eds). Gondwana Seven Proceedings. lnstituto de Geociencias-University of Sao Paulo, Sao Paulo, 83-117.

Magmatism and continental rifting during the opening of the South Atlantic Ocean: a consequence of Lower Cretaceous super-plume activity? MARJORIE WILSON

Department o f Earth Sciences, Leeds University, Leeds, LS2 9JT, UK

Abstract: Two large-scale mantle plumes, whose present-day foci are dose to the oceanic islands of Tristan da Cunha and St Helena, appear to have played a significant role in the initial stages of rifting between Africa and South America during the Early Cretaceous opening of the South Atlantic Ocean. They may represent the initialburst of a super-plume event which generated extensive oceanic plateaux in the Pacific and Indian oceans. The recent volcanic products of Tristan da Cunha and St Helena have near endmember Sr-Nd-Pb isotopic characteristics (EM I and HIMU) in the spectrum of ocean basalt isotopic compositions. These isotopic signatures are recognised for more than 100 Ma in the plume-related magmatic products and therefore appear to be a long-lived feature of the plume source. The history of rifting and magmatism in West and Central Africa/NE Brazil and in southern Brazil, above the broad heads of the initial starting plumes between 145 Ma and 130 Ma, strongly suggests that there are different physical differences between the two plumes in addition to chemical ones. The St Helena plume appears to have been much weaker and cooler, with a smaller buoyancy flux. The hotter Tristan plume has generated voluminous flood basalts volcanism in the Paran~i basin of Brazil and appears to be associated with continental break-up within a few million years of the plume head impinging on the base of the lithosphere. In contrast, in West and Central Africa, rifting above the St Helena plume, associated with small volumes of alkaline-transitionalmagmatism, spans an extended period of 30-40 Ma before break-up occurs in the Equatoiial Atlantic.

The opening of the South Atlantic Ocean during the late Jurassic and early Cretaceous resulted in the divergence of the African and South American plates along the line of Mid-Atlantic ridge, with the development of passive continental margins on either side. This last phase of the break-up of the Gondwana supercontinent is associated with a complex history of rifting and magmatism in West and Central Africa and Brazil. Whilst the main driving force for continental splitting must undoubtedly be ascribed to plate boundary forces (e.g. Hill 1991), at least two major mantle plumes, St Helena and Tristan da Cunha, appear to have been influential in weakening the continental lithosphere along the line of the developing South Atlantic rift. In the area overlying the Tristan plume head rifting is associated with voluminous tholeiitic flood basalt volcanism (Richards et al. 1989; White & McKenzie 1989; O'Connor & Duncan 1990), emplaced over a comparatively short time span (135-130 Ma) during the Early Cretaceous (Peate et al. 1990). This has generated one of the largest continental flood basalt provinces on Earth ( > 800000 km 3) in the Paran~i basin of Brazil. In contrast, in West and Central Africa

lithospheric extension is expressed as a broad zone of Late Jurassic-Early Tertiary rifting, extending for over 4000 km from Nigeria, northwards into Niger and Libya and eastwards, through southern Chad, into Sudan and Kenya (Wilson & Guiraud 1992). Many of these rifts have associated alkaline-transitional magmatie activity but the total volume is difficult to estimate because of the generally poor exposure. However, in comparison with the Parand, these rifts are clearly only weakly magmatic. Unlike the Paran~i province, in which rifting, magmatism and continental break-up were essentially contemporaneous, in West and Central Africa extension and magmatism occurred for some 3040 Ma before the onset of sea-floor spreading in the Equatorial Atlantic (Wilson & Guiraud 1992). The contrasting styles of lithospheric extension and magmatism in the vicinity of the Tristan and St Helena plume heads provides a unique opportunity to investigate the response of the continental lithosphere to mantle plume activity. Additionally, it may allow greater insight into the dynamics of the plumes themselves. The oceanic islands of Tristan da Cunha and St

From STORey,B. C., ALABASTER,T. & PANKHURST,R. J. (eds), 1992, Magmatismand the Causes of ContinentalBreak-up, Geological Society Special Publication No. 68, pp. 241-255.

241

242

M. WILSON

Helena, which represent the most recent subaerial activity of the plumes, define extrema within the spectrum of oceanic basalt (OIB) Sr-Nb-Pb isotopic compositions (Zindler & Hart 1986). A fundamental question therefore, is whether these geochemical differences between the two plumes are also correlated with physical differences, for example in buoyancy flux or temperature. Available data suggest that Tristan is currently a fairly strong plume, in terms of its buoyancy flux, whereas St Helena is very weak (Sleep 1990, 1991; Schilling 1991). These differences appear to have persisted throughout the opening of South Atlantic Ocean. The earliest initiation of magmatic activity during the Early Cretaceous break-up phase of Gondwana may allow us to place approximate constraints on the time that the starting plume heads (using the terminology of Campbell & Griffiths 1990) impinged on the base of the continental lithosphere. This is c. 135 Ma for Tristan and 145 Ma for St Helena. Clearly these are minimum estimates as there may be a lag of several Ma between the two events. On a global scale, the Paramt flood basalt volcanism just precedes a major phase of mantle plume activity beneath the Pacific basin (Larson 1991) which has generated extensive basaltic oceanic plateau such as Ontong-Java, Manihiki and the Darwin Rise. The Kerguelen plateau in the Indian ocean also formed about this time (Weis et al. 1989). These plateau forming events have been correlated with a major period of stabilisation in the polarity of the Earth's magnetic field, the Cretaceous Long Normal superchron, which has been attributed to modification of core dynamics by the uplift of large-scale plume heads (termed here super-plumes) from the core-mantle boundary (Larson 1991). The ascent of the St. Helena and Tristan starting plume heads beneath Gondwana may represent the initial stages of this super-plume event, which has been speculated to involve dispruption of the D " thermal boundary layer at the base of the lower mantle (Larson & Olson 1991).

History of opening of the South Atlantic Ocean Continental break-up along the line of the protoSouth Atlantic Ocean was intiated in the late Jurassic to early Cretaceous and progressed from south to north in a step-wise manner (Ni~rnberg & Miiller 1991). From 150-130 Ma rifting propagated from the southern-most tip of South America to about 38"S in the vicinity of the Salado Basin. Fig. 1, causing continental

stretching and minor dextral strike-slip motion within the Colorado and Salado basins. At about 130 Ma, rifting combined with dextral strike-slip motion started along the Parami-Chacos basin deformation zone, associated with further northward propagation of rifting up to 28°S. This was accompanied by a major phase of flood basalt volcanism from 135-130 Ma (Peat et al. 1990; Hawkesworth et al. this volume). Between 126 Ma and 119 Ma rifting propagated further northward into the region of the Benue Trough of Nigeria. In the Equatorial region, seafloor spreading commenced in the Early Albian (c.

120 Ma

Helena Martin Vas 0

bE~ndeka I Tristan

Shona • Bouvet 0

I

km

I000 I

Fig. 1. Reconstruction of Africa and South America at 120 Ma during the initial stages of opening of the South Atlantic Ocean, showing the location of hotspots (diaganol shading) presumed to be active at the time (O'Connor & Duncan 1990) and the Paran,'iEtendeka flood basalt province. After Niirnberg & Miiller (1991). S, Salado basin; C, Colorado basin; B, Benue Trough. Heavy dashed lines represent postulated zones of strike-slip motion. Circles indicate approximate diameters of the Tristan and St Helena starting plume heads.

S ATLANTIC OPENING AND CRETACEOUS SUPER-PLUMES

MALl v

NIGER

"~'i:. :-i

BRAZIL

~

243

Ir~

CHAD

I

J

500 km

Mesozoic rift basins

Fig. 2. The location of flit basinsin West and Central Africa and NE Brazil at c. 115Ma. After Wilson& Guiraud (1992). M, Maranhio Basin; B, Benue Trough; Mu, Muglad basin. 112 Ma) in the southern part and Late Albian (c. 107 Ma) in the northern part (Uchupi 1989). However not until Late Albian to Early Cenomanian (c. 100 Ma) were small oceanic basins created, establishing the final breach between South America and Africa (Mastic et al. 1988). The step-wise opening of the South Atlantic from south to north resulted in considerable stress build-up in the equatorial region (Fairhead & Binks, 1991) which was taken up by continental stretching and sinistral strike-slip motion in the Benue Trough/Niger rift system of West and Central Africa (Wilson & Guiraud 1992; Fig. 2). Here a complex' network of extensional sedimentary basins preserve a record of rifting and associated alkaline-transitional magmatism spanning a period of 30-40 Ma before the onset of sea-floor spreading. In this region, localised rifting and alkaline magmatism continued long after the final separation of Africa from South America (Wilson & Guiraud 1992). Sedimentary basins also developed in NE Brazil during the Late Jurassic-Early Cretaceous (Castro 1987), concurrent with the development of rift basins in West and Central Africa.

Far less information is available concerning the initiation of extensional tectonics along the South American margin, particularly in the Paran~i basin of Brazil where an extensive cover of early Cretaceous flood basalts largely obscures older sequences. Rifting in the Salado and Colorado basins, related to the initial stages of South Atlantic opening, probably commenced in the Middle to Late Jurassic (Urien & Zambrano 1973) but had ceased by 126 Ma (Niirnberg & Muller 1991). Unternehr et al. (1988) proposed that intra-plate deformation in the Paranfi--Chacos basin was taken up along a major NW-SE-trending dextral strike-slip zone extending across South America to the Andes (Fig. 1). This extensional phase had effectively ceased by 118 Ma (Niirnberg & Miiller 1991). Prior to this the Parami basin had a long history of rift-related sedimentation, with major extensional phases in the Silurian-Devonian, Carboniferous--Permian and Jurassic--Cretaceous (Zalan et al. 1987). This may have fundamental implications for the strength of the lithosphere beneath the basin. Several authors (Morgan 1972; Anderson

244

M. WILSON

O

•

~

Fernando ~ Ascensio~n

St Helena M a r t i n Vas

Gough Discovery Fig. 3. The location of plume trails (shaded) and oceanic islands in the South Atlantic Ocean. After O'Connor & Duncan (1990). Reconstructed hotspot tracks (lines) connect the currently active locus of the plume (circles) with the initial locus of the starting plume axis (squares). Heavy solid line marks the axis of the Mid Atlantic Ridge.

1982; White & McKenzie 1989; O'Connor & Duncan 1990; Wilson & Guiraud 1992) have considered that mantle plume activity was important during the initial opening and subsequent evolution of the South Atlantic Ocean. This has given rise to V-shaped hotspot trails of oceanic islands and seamounts on the oceanic parts of the African and South American plates and flood basalt magmatism onshore (Fig. 3). One of the earliest signs of activity of such plumes may be continental flood basalts of early Cretaceous age in Brazil (Paran~i) and Namibia (Etendeka). These flood basalt provinces were shown to be connected on reconstructions of the South Atlantic for early Cretaceous time (Rabinowitz & La Breque 1979) and related to the activity of the Tristan da Cunha mantle plume. Richards et al. (1989) suggested that these flood basalts may have been erupted in response to rapid partial melting in the large diapiric head the Tristan starting plume as it impinged on the base of the continental lithosphere. Alternatively, as suggested by White & McKenzie (1989), the flood basalts may have formed by pressure release melting of the upper mantle as continental rifting occurred above a region warmed by a hotspot, which was previ-

ously unable to penetrate the lithosphere until plate boundary forces caused extension and rifting. Fleitout et al. (1989) have detected numerous small wavelength elongated features on filtered geoid and topographic maps of the South Atlantic. A number of these features are orientated at N50 E for the African plate and N65 W for the South American plate and record the directions of Africa and South American absolute plate motion. These authors attribute these elongated features to magmatic traces generated by mantle plume activity.

Tristan plume: opening o f the South Central Atlantic The most prominent bathymetric features in the South Atlantic, apart from the mid-Atlantic spreading axis, are the paired volcanic lineaments of the Walvis Ridge-Rio Grande Rise and the St Helena seamount chain (Fig. 3). O'Connor & Duncan (1990) have documented that the age of the Walvis Ridge basement increases approximately linearly with increasing distance northeastwards from Tristan da Cunha, linking recent plume-related volcanism on Tristan da Cunha (Le Roex et al. 1990) via the Walvis

S ATLANTIC OPENING AND CRETACEOUS SUPER-PLUMES Ridge to the earliest expression of the plume activity on the Namibian coast, the Etendeka flood basalt province. Geochemical and isotopic data also suggest a common source for Paran,'iEtendeka and Tristan-Walvis Ridge basalts (Hawkesworth et al. 1986). The Tristan hotspot was apparently ridge centred until about 70 Ma (O'Connor & Duncan 1990) when the spreading axis began to migrate westward away from the hotspot resulting in termination of magmatic activity along the Rio Grande Rise. The currently active focus of the hotspot is located some 550 km to the east of the ridge axis in the vicinity of the island of Tristan da Cunha. In a refit of the continents at c. 120 Ma, Niirnberg & Miiller (1991) show significant overlap in the vicinity of the postulated palaeoposition of the Tristan hotspot (Fig. 1) which may suggest that crustal stretching had occurred in response to doming of the continental lithosphere above the upwelling plume. The Parami and Etendeka flood basalts appear to have been erupted at the same time that the northward propagating South Atlantic rift reached the latitude of Namibia, suggesting a close association between plume-related volcanism and continental rifting. Erlank e t al. (1984) have determined 4°Ar/39Ar plateau ages of 125-130 Ma for late stage dolerite dykes intruding the Etendeka basalts, which constrains the minimum age of the flood basalt volcanism. Previously, a wide spectrum of K-At ages had been obtained for the magmatism, many of which' are probably suspect due to A r gain (Niirnberg & Mfiller 1991). In Brazil the main phase of Paran~i volcanism appears to have occurred between 135 Ma and 130 Ma (Peate et al. 1990). According to Sibuet et al. (1984) magmatism in the Paran~i basin may be related to the failed arm of a triple junction on the South American plate that was active as a dextral strike-slip zone during the late Jurassic and early Cretaceous. There is a distinctive regional dyke swarm emplaced along this direction which may suggest that rifting has reactivitated older basement structures, ~though there is little direct evidence to support this. Such reactivation appears to be a common feature of extensional tectonics throughout West and Central Africa (Wilson & Guiraud 1992). A question of some debate concerns the exact location of the Tristan plume at 130-120 Ma during the initial rifting of the South Atlantic Ocean at this latitude. O'Connor & Duncan (1990) have constrained their plate reconstruction models such that it lay beneath the Paran~i flood basalt province of Brazil at this time. This seems logical as the Paranfi is by far the greatest surface

245

expression of volcanism and thus may be expected to be located above the plume axis. However Thompson & Gibson (1991) have argued that in fact the plume was located further east, in a position more closely approximating that shown in Fig. 1, and that the greater expression of magmatic activity in the Paran~i basin is due to rifting of an area which had already been thinned during a long history of crustal extension. Harry & Sawyer (1992) have suggested that a horizontal pressure gradient may have developed in the lower crust beneath the basin, during the early stages of extension, as a consequence of the dynamic interaction of pre-existing weaknesses in the middle crust and upper mantle. This could provide a mechanism for transporting magma generated beneath the incipient sea-floor spreading axis laterally into the Paran~i province some 100-200 km distant. An alternative explanation for the asymmetry of the flood basalt province with respect to the proto- South Atlantic rift could be that between 130 Ma and 120 Ma there was signficant E - W motion of Gondwana, which would be impossible to detect palaeomagnetically. This would allow the axis of the plume to lie beneath the Paran~i basin at 135-130 Ma but to have been closer to the present ocean-continent boundary by 125 Ma due to the westward migration of Gondwana. O'Connor & Le Roex (1992) suggest that the diameter of Tristan plume has been of the order of 500 km throughout the history of opening of the South Atlantic. This is much smaller than the diameter of the starting plume head postulated by White & Mckenzie (1989) based upon the lateral extent of the Paran~i -Etendeka flood basalt province (c. 2000 km, Fig. 1) and gives an idea of the diameter of the plume tail (using the terminology of Campbell & Griffiths 1990).

St Helena plume: opening o f the Equatorial Atlantic The St Helena seamount chain (Fig. 3) appears to have been generated by the migration of the African plate over the St Helena plume (O'Connor et al. 1992), which was also instrumental in the break-up of Africa from South America during the Early Cretaceous. Seamounts along the chain become progressively older towards the continental margin, in a similar manner to the age progression along the Walvis Ridge (O'Connor & Le Roex 1992). The seamounts terminate at the seaward edge of the TertiaryRecent Cameroon Volcanic line (CVL) which represents, at least in part, reactivation of magmatic activity along the line of the former hotspot track (Wilson & Guiraud 1992; Halliday

246

M. WILSON

et al. 1990). Halliday et aL (1990) have suggested

that fossil plume components in the base of the lithosphere may have been remobilized to provide a source for the Tertiary-Recent magmatism of the CVL. A distinctive Pb isotopic anomaly in the CVL volcanics, centred over the ocean-continent boundary at Mt Cameroon, decreases over a distance of 400 km both oceanwards and into continental Africa. This may indicate the location of the plume axis 130-120 Ma ago and suggests that the diameter of the initial starting plume head was at least 800-1000 km. This contrasts with a proposed diameter of 500 km for the plume tail during the entire opening of the South Atlantic (O'Connor & Le Roex 1992). O'Connor & Duncan (1990) suggest that the St Helena plume was ridge centred up until about 70 Ma when the ridge migrated to the west, following a similar pattern to the Tristan plume. The oceanic island of St Helena is located on the African plate 800 km east of the Mid-Atlantic Ridge. It had a history of subaerial volcanism between 9-7 Ma (Chaffey et al. 1989) and is characterised by the eruption of basalts with a high ~spb/e°4pb isotopic signature which define an end-member for the spectrum of OIB isotopic compositions known as HIMU (Zindler & Hart 1986). Basalts dredged from the axis of the MidAtlantic ridge to the west of St Helena exhibit a similar HIMU signature suggesting lateral flow of plume material from the currently active hotspot towards the ridge (Hanan et ai. 1986). Brozena & White (1990) place the currently active focus of the hotspot some 200 km west of St Helena itself and suggest that the diameter of the anomalously hot asthenosphere currently associated with the plume is a c. 1000 km. Widespread magmatic activity in West and Central Africa and in NE Brazil from Triassic to Cretaceous times may be related to the activity of mantle plumes (particularly St Helena) which acted to pre-weaken the lithosphere and enhance the effects of deviatoric stresses on the subsequent rifting of Gondwana to form the proto-Atlantic Ocean (Wilson & Guiraud 1992). However, plume activity did not, in general, lead to voluminous tholeiitic flood basalt volcanism as it did above the Tristan da Cunha plume, active at broadly the same time beneath the newly developing plate boundary to the south. The only area in which tholeiitic flood basalts do occur is in the Maranh~o basin of NE Brazil (Fig. 2). Low Ti Triassic-Jurassic (189154 Ma) flood basalts occur in the western part of the basin while high Ti Cretaceous (122-115 Ma) hypabyssal intrusions occur in the eastern part (Fodor et al. 1990). The low Ti basalts could

be related to the opening of the southern part of the Central Atlantic, whereas the high Ti basalts could relate to the initial stages of opening of the Equatorial Atlantic at c. 120 Ma (Wilson & Guiraud 1992). The Triassic-Jurassic flood basalts are the nearest comparator to the flood basalt volcanism of the Paramt-Etendeka. These could be plume related, although it is difficult to attribute them to a particular plume. Morgan (1983) places the St Helena hotspot near Maranh~o at 180 Ma, although there is little hard evidence to support this. O'Connor & Duncan (1990) locate the Martin Vas plume beneath NE Brazil at 130-120 Ma (Fig. 1) and it is possible that the thermal effects of this plume were responsible for triggering the Cretaceous magrnatism. The Trindade-Columbia seamount chain and a trend of alkaline igneous intrusions (including kimberlites) ranging in age from 12251 Ma (Hartnady & Le Roex 1985; Crough et al. 1980) represent the postulated trail of the South American plate over this plume (O'Connor & Duncan 1990). During the opening of the equatorial Atlantic the St Helena plume does not appear to have generated anything like the volume of magma that the Tristan da Cunha plume did further to the south. This raises the question as to whether there are any fundamental differences between the St Helena and Tristan da Cunha plumes which may account for this contrasting behaviour. Global events at the Jurassic--Cretaceous boundary: evidence for super-plume activity It has been shown in the previous sections that seamount chains, oceanic islands and continental flood basalts may all be the surface expression of mantle plume activity. Several authors (e.g. Richards et al. 1989; Coffin 1991) have also suggested that the formation of large oceanic plateaux occurs cataclysmically when the inflated heads of starting plumes impinge on the base of the oceanic lithosphere. Larson (1991) has estimated the volume production for all possible plume-related features over the past 150 Ma and has demonstrated that there is an apparent burst in volcanic activity coincident with the onset of the long Cretaceous normal polarity interval in the early Cretaceous (c. 125 Ma). This he associates with the upwelling of one or more super-plumes from the core-mantle boundary beneath the Pacific basin, generating vast oceanic plateaux such as Ontong-Java and Manihild. Such super-plume activity may actually trigger large-scale variations in magnetic fever-

S ATLANTIC OPENING AND CRETACEOUS SUPER-PLUMES sal frequency by thinning the D" thermal boundary later at the core-mantle boundary (Larson & Olson 1991), and may result in increased global temperatures and eustatic sea-level changes. Super-plume events may be so large that they represent first order modifications of the convection pattern of the Earth's mantle. As shown in Fig. 4 many oceanic plateaux, including Ontong-Java and Manihiki in the Pacific and Kerguelen in the Indian ocean, were emplaced within 10 Ma after the onset of the Cretaceous Long Normal Polarity superchron at 125 Ma. The polarity reversal frequency had already started to decline by about 130 Ma which Larson (1991) attributes to the lift-off of the super-plume (or plumes) from the core-mantle boundary. The time gap between the onset of the decline in polarity reversal frequency and oceanic plateau magmatism suggests a rise time for the plume of some 5-15 Ma, assuming that there is an instantaneous response in the core to thinning of D" layer by plume formation. However, the ages of most oceanic plateaux are not well constrained (Coffin & Eldholm 1991) and therefore this value should not be taken too literally. In addition, the magmatism of the Paranfi/ Etendeka flood basalt province of Brazil, which may represent the ascent of a similar superplume beneath Gondwana, predates the start of the superchron by c. 10 Ma, (Fig. 4). This may indicate that the relationship between superplume activity and polarity reversal of the

247

Earth's magnetic field is rather more complex than envisaged bv Larson (1991). Estimated volumes for the Kerguelen and Ontong-Java plateaux are 2x107 km a and 6x107 km 3 respectively (.Coffin 1991). This compares with 1-2x10 ~' km ° for the Paranfi continental flood basalt province (Rampino & Stothers 1988). Thus, the volume of magma emplaced in the largest oceanic plateaux appears to be much greater than that in the largest CFB provinces, although there are considerable uncertainties attached to such volume estimates. Assuming that starting plume heads have similar dimensions beneath both oceans and continents, this volume difference may indicate greater degrees of partial melting in the oceanic case, possibly related to differing plate thickness in the two environments. Several authors have suggested that the emplacement of oceanic plateaux and continental flood basalts occurs very rapidly, with perhaps 90% of the magmas erupted in less than 2 Ma (Courtillot et al. 1988; Duncan & Pyle 1988; Rampino & Stothers 1988). This implies tremendously high eruption rates. For example, if the Ontong-Java plateau had been emplaced in 3 Ma, the annual magma production rate would exceed that of the present day 50 000 km long mid-ocean ridge system, including Iceland (Coffin & Eldholm 1991). This confirms that such super-plume events are indeed unusual phenomena.

Ma _ 80 _ 90

OCEANICPLATEAU

-- 100

Darwin Rise

--110 _ 120

~

_ 130

Ontong-Java Manihiki

-- 140 _ 150

South Atlantic rifting phases Maranhao Parana Kerguelen N ~

I Ill

I

I

CFB's

Fig. 4. Ages of Lower Cretaceous continental flood basalt provinces (CFB's) and oceanicplateaux in relation to polarity reversal frequency and the main rifting phases in the South Atlantic Ocean. Data sources: Niirnberg & Miiller (1991); Peate et aL (1990); Fodor et al. (1990); Weis et al. (1989); McNutt et al. (1990);Tarduno et al. (1991). N.B. The polarity reversal frequency preceding the Cretaceous Long Normal is shown schematically.

248

M. WILSON Geochemical characteristics of South Atlantic mantle plumes Oceanic island basalt (OIB) suites are extremely diverse in terms of their major element, trace element and Sr-Nd-Pb isotopic characteristics. Three end-member source compositions (HIMU, EM I and EMII) are generally invoked to explain the array of isotopic compositions (e.g. Weaver 1991; Zindler & Hart 1986; Fig. 5). However, relatively few OIB have end-member isotopic compositions, the majority representing mixtures between these end-members and the depleted upper mantle source of N-MORB (DMM). Such mixing is inferred to occur in mantle plumes during their ascent from deep levels in the mantle, perhaps as deep as the coremantle boundary (e.g. Griffiths & Campbell 1990). HIMU OIB are distinctive in terms of their anomalously radiogenic Pb isotope composition

Whilst large oceanic plateaux and continental flood basalt provinces may form cataclysmically when the inflated heads of mantle plumes reach the base of the lithosphere (Richards et al. 1989; Campbell & Griffiths 1990), seamount chains or elongated aseismic ridges probably reflect the much narrower but long lived plume tails that follow. For example, the Louisville Ridge, in the South Pacific, has been proposed as a hotspot trail originating at the Ontong-Java hotspot (Pringle, 1991); The Kerguelen plateau, in the Indian Ocean, is linked with the Ninetyeast Ridge and the extensive Parami-Etendeka continental flood basalt province with the Walvis Ridge-Rio Grande Rise. However, not all plateaux building/flood basalt events necessarily establish a stable plume which feeds a long-lived hotspot chain (Pringle 1991). This is possible if for some reason supply of material into the tail of the plume is cut off after the initial lift off of the buoyant diapir.

0.5134

0.5130

-

i'

~ 0.51:~ EM II 0.5122

................

17

,

18

I

19

I

~ I

20

SOPITA i

I

21

I

I

22

206Pb/204Pb ]Fig. 5. mNd/l~Nd versus 2°6pb/2°4Pbfor Cretaceous-Recent basalts from oceanic islands, aseismic ridges, seamounts chains and continental basalts in and bounding the South Atlantic. Also shown for comparison are data for the Ontong-Java and Manihiki (O-J-M) and Kerguelen (K) oceanic plateaux and the SOPITA superswell. End-member mantle components (EM I, EMII, HIMU and DMM) are from Zindler & Hart (1986). Abbreviations: WR, Walvis Ridge; M, Maranhio; CVL, Cameroon Volcanic Line; SHS, St Helena seamounts; SH, St Helena; T & G, Tristan da Cunha/Gough; HPT, high P-Ti Parana flood basalts; LPT, low P-Ti flood basalts. Data sources: Parana, Hawkesworth et al. (1986); Walvis Ridge, Richardson et al. (1982); Tristan da Cunha, Le Roex et al. (1990); S. Atlantic MORB, Hanan et al. (1986); St Helena, Chaffey et al. (1989); St Helena seamounts, O'Connor etal. (1992); DMM, HIMU, EM I, EMII, Zindler & Hart (1986); Maranhao, Fodor etal. (1990); CVL, Hailiday et al. (1990).

S ATLANTIC OPENING AND CRETACEOUS SUPER-PLUMES (2°6Pb/Z°4Pb > 20.5), which has been attributed to derivation from mantle sources involving an end-member derived from recycled, subduction zone processed, ancient (1-2 Ga) oceanic crust (e.g. Weaver 1991). The oceanic island of St Helena is close in composition (isotopically) to the HIMU end-member (Fig. 5) which should make magmatic activity related to the St Helena plume easy to track due to its distinctive Pb isotopic characteristics, if we assume that the plume source has remained isotopically relatively homogeneous over the past 130 Ma. The EM I OIB end-member is most closely approximated isotopically by basalts from the Walvis Ridge. OIB from Tristan da Cunha and Gough in the South Atlantic (Fig. 5) also contain a signficant EM I component. Weaver (1991) considers that 5% ancient pelagic sediment mixed with a HIMU source could explain the isotopic characteristics of EM I OIB. EMII OIB are considered by Weaver (1991) to reflect mixing of the basic HIMU component with terrigenous sediments. However, as shown in Fig. 5, EM II is not a signficant component in the petrogenesis of South Atlantic OIB, which can be described isotopically in terms of three component mixtures of DMM, HIMU and EM I. Such mixing is shown simplistically by the heavy dashed lines in Fig. 5., assuming that the endmembers have broadly similar abundances of Nd and Pb. Weaver (1991) notes that the HIMU component seems to be the primary mixing component in most OIB suites. Taking this to an extreme we might suppose that all mantle plumes are fed by ancient (1-2 Ga), recycled, subducted oceanic crust with variable proportions of subducted sediment. As demonstrated by Weaver (1991) only very small proportions of sedimentary components ( < 5%) subducted along with the basaltic crust are necessary to dominate completely the isotopic characteristics of the resultant partial melts. Whilst not all workers would agree with such a simplistic interpretation of OIB, this is a useful starting hypothesis for the interpretation of the isotopic characteristics of the magmatic rocks under discussion here. Essentially we are making a presumption that the main source for mantle plumes is subducted mafic oceanic crust+sediment. Thus, if we compare and contract the St Helena and Tristan da Cunha plumes we might predict a source for both involving recycled subducted ancient oceanic crust, which in the Tristan case had incorporated some pelagic sediment at the time of subduction. Both plumes could originate from similar depths in the mantle, from thermal boundary layers such as the 670 km seismic discontinuity or the core-mantle boundary.

249

Figure 5 shows the variation of 143Nd/144Nd versus 2°6pb/2°4pb for basalts from oceanic islands and seamounts in the South Atlantic and continental flood basalts from South America. In theory, as considered by Staudigel et al. (1991), comparison of the radiogenic isotope characteristics of Cretaceous basalts with those of recent OIB requires calculation of a 'presentday' equivalent for the isotope ratios of the Cretaceous basalts. This must take into account the fact that they were removed from their mantle sources up to 130 Ma before their recent counterparts and thus, during this time the radiogenic growth of Sr-Nd and Pb in the Cretaceous basalts differs from the growth that would have occurred in their mantle sources if they had not melted during the Cretaceous. To do this we should really calculate the initial ratios of the Cretaceous basalts and then add to these the hypothetical radiogenic growth in the source since 130 Ma. In practice however, the latter is probably small and, particularly for Nd-Pb, we can probably compare Cretaceous initial ratios with measured ratios for young OIB without introducing significant error. From Fig. 5 we can see that the oceanic islands of St Helena and T r i s t a n da Cunha/Gough have distinctive isotopic compositions, close to the proposed HIMU and EM I end-members of Zindler & Hart (1986). In general, aseismic ridges and seamounts related to the activity of these two plumes (St Helena-Bahia-Pernambuco seamounts; Walvis Ridge-Rio Grande Rise) have broadly similar isotopic characterstics to the respective oceanic island, suggesting a common origin. However there are exceptions (O'Connor et al. 1992) which suggest that both EM I and HIMU components were available in both the Tristan and St Helena plumes throughout their history. It is notable that the St Helena seamount (SHS) field in Fig. 5 defines an elongate array trending towards EM I. This is interpreted by O'Connor et al. (1992) as strong evidence for two component mixing between HIMU and EM I components in the ascending plume. Such mixing is inevitable if, as proposed by Weaver (1991), all deep mantle plumes potentially incorporate both HIMU and EM (EMI and E M I I ) components. If we accept the hypothesis of Campbell & Griffiths (1990) that continental flood basalts represent the arrival of a starting plume head beneath a continental plate then we might expect to see the geochemical signature of the plume in the flood basalts. In the case of the Tristan plume the continental flood basalts of the Paranfi do indeed have a strong EM I signature like Tristan and the Walvis Ridge (Fig. 5). However, there

250

M. WILSON

has been much debate about the interpretation than 120 Ma. However, in detail, as shown by of these data, in particular whether the EM I sig- O'Connor et al. (1992) both HIMU and EM I nature is introduced by crustal contamination or components appear to have been available whether it is derived from ancient enriched man- throughout the lifespan of these plumes, productle sources within the continental lithosphere ing localized isotopic heterogeneities. A further (Hawkesworth et aI. 1986, 1988, 1990). As we observation from Fig. 5, which may be imporhave considered previously, it is much harder to tant, is the suggestion that the present St Helena find flood basalts unequivocally associated with plume may involve significant mixing with dethe early stages of activity of the St Helena pleted mantle (DMM). However, this only holds hotspot. The flood tholeiites of the Maranh~o true if we assume that the HIMU end-member basin of NE Brazil could be related to either the isotopic composition is defined by the isotopiSt Helena or to the Martin Vas plume. In terms cally more extreme Pacific oceanic island of of Fig. 5 the Maranh~o tholeiites (Fodor et al. Mangaia (Zindler & Hart 1986). If correct, this 1990) have similar isotopic characteristics to may indicate that the St Helena plume has enTristan da Cunha, which might eliminate the St trained shallow asthenospheric mantle (DMM) Helena plume if we accept that flood basalts do which would cool it significantly. Also shown in Fig. 5 is the range of isotopic indeed retain some of the isotopic characteristics of their plume source. Thus, one possible in- composition of tholeiitic basalts from the Onterpretation of the data would be that the tong-Java and Manihiki oceanic plateaux Maranhfio basalts are related to the Martin Vas (Mahoney & Spencer 1991), which Larson plume and not to St Helena. The alternative in- (1991) has associated with Cretaceous superterpretation would be that all continental flood plume activity beneath the Pacific Ocean basin. tholeiites are extensively contaminated by the These define an elongate array extending from continental lithosphere (both crust and mantle) EM I towards the Atlantic MORB field. This through which they pass and rarely record the strongly suggests that a Tristan-like EM I plume, isotopic signature of the mantle plume which which enlarged its head by entrainment of depleted mantle (DMM), was responsible for the triggered their generation. Also plotted in Fig. 5 is the isotopic composi- activity. Also shown is a field for basalts from the tion of alkaline and transitional basalts from the Kergnelen plateau in the Indian Ocean (Weis et ocean--continent boundary of the Cameroon al. 1989) which have similar characteristics. volcanic line (CVL). Halliday et al. (1990) and Many Cretaceous seamounts in the Central and Wilson & Guiraud (1992) have suggested that Western Pacific have been related to the activity these represent Tertiary-Quaternary remobili- of the SOPITA superswell (South Pacific zation of the fossil head of the St Helena plume Isotopic and Thermal Anomaly, Staudigel et al. that enriched the base of both oceanic and conti- 1991). These exhibit a wide range of Nb-Pb nental lithosphere during the early Cretaceous. isotopic compositions falling within the range of The data overlap with the field of St Helena sea- present day SOPITA OIB, (Fig. 5). If we regard mounts and are therefore consistent with such a the extremely heterogeneous SOPITA supermodel. If so, these data may provide us with an swell as yet another relict of Cretaceous superadditional constraint as to what the St Helena plume activity we can regard such super-plumes starting plume was like isotopically. The CVL as having both EM and HIMU characteristics. data, in combination with that from the St This is important because without the SOPITA Helena seamounts suggest that the starting control we might be tempted to infer that all plume incorporated both HIMU and EM I com- super-plumes were EM I. ponents, but that the more recent activity of the plume no longer samples EM I. Campbell & Griffiths (1990) have postulated that mantle plumes may be chemically zoned and that major Plume dynamics and continental break-up changes in the OIB components of a hotspot Although models for the structure and temporal could occur quite commonly, particularly bet- evolution of mantle plumes vary considerably, a ween the initial plume-head phase and the ensu- common factor is the capacity of a plume to gening plume-tail stage. However, while major tem- erate large quantities of partial melt by adiabatic poral changes in the isotopic composition of decompression (e.g. White & McKenzie 1989). plumes are certainly possible they appear not to When the thermal anomaly is associated with be common. For example, in Fig. 5 the Tristan continental break-up (e.g. Duncan & Richards and St Helena plumes broadly appear to have 1991; Hill 1991) the plume head may create a maintained their isotopic signatures for more flood basalt province and related volcanic conti-

S ATLANTIC OPENING AND CRETACEOUS SUPER-PLUMES nental margin, while the plume tail may be manifested as a submarine ridge or seamount chain on the oceanic lithosphere. Griffiths & Campbell (1990) and Campbell & Griffiths (1990) have argued on the basis of the size of flood basalt provinces that the strong starting plumes which produce these features must originate deep within the lower mantle, almost certainly at the core-mantle boundary. Griffiths & Campbell (1991) predict that long lived plumes responsible for continental flood basalts and linear volcanic chains will become deflected from the vertical as they penetrate mantle which is overturning on much larger length scales in response to surface cooling. They suggest that plume inclination leads to entrainment of surrounding cooler mantle such that plume temperature, as reflected in the volcanic products of the hotspot track, may decrease with time. The effects of entrainment depend primarily on the buoyancy flux of the plume, with greater cooling for smaller fluxes. Such entrainment may contribute to the compositional heterogeneity of hotspot melts and should generate isotopic characteristics indicative of involvement of greater amounts of depleted mantle (DMM in Fig. 5). Sleep (1990, 1991) estimates that the buoyancy flux of the present day St Helena and Tristan hotspots is 0.5 and 3.4 Mg s -~ respectively. St Helena has no obvious associated topographic swell or geoid anomaly, consistent with it being a weak plume, and therefore its buoyancy flux is poorly constrained. In contrast, the Tristan plume is still comparatively strong. Sleep (1991) considers that its buoyancy flux appears to have waned with time and may have been as high as 4.5 Mg s -~ in the Cretaceous. Griffiths & Campbell (1990) suggest that weak plumes with buoyancy fluxes less than 1 Mg swould take greater than 100 Ma to reach the surface from the core mantle boundary and would tend to be deflected by larger scale convective motions in the mantle. Such weak plumes may break up into a series of smaller diapirs instead of ascending as one large head and, if they reach the base of the lithosphere at all, there may be no major event at the beginning of the consequent hotspot track. In contrast, for buoyancy fluxes as high as 10 Mg s -~ they predict that a plume may traverse the mantle in around 50 Ma. These plumes will not be deflected by large scale convective motions in the mantle associated with plate motion and could remain connected to their source by a stable feeder conduit. Therefore, it is predicted that only strong plumes with high buoyancy fluxes will give rise to continental flood basalt provinces. Plume heads that reach

251

the base of the lithosphere while still receiving a constant influx from their source region (D") are predicted by Griffiths & Campbell (1990) to have diameters in the range 800-1200 km which are almost independent of buoyancy flux. These grow predominantly by entrainment of lower mantle material and upon reaching the base of the lithosphere they flatten and almost double in diameter. The mean temperature of the head decreases most rapidly soon after the head detaches from the source. Whilst uplift of the surface above plumes undoubtedly results in horizontal deviatoric stresses within lithosphere, the magnitude of these stresses is probably incapable of actually ir~tiating continental break-up (Hill 1991). Instead, the uprise of a new plume may lead to the local reorganisation of plate-scale motions, or provide sufficient extra force to drive a weak plate-scale system from slow extension through to continental rifting, with the resultant formation of a new ocean basin. The latter is the situation envisaged here for the opening of the South Atlantic Ocean during the Early Cretaceous. Although mantle plumes did not provide the ultimate driving force for continental break-up, the extra gravitational potential they imposed means that they may have played an important role in determining where and when continental break-up occurred. Additionally, super-plume activity, such as that envisaged in the formation of oceanic plateaux and possibly in the Early Cretaceous opening of the South Atlantic, may even contribute to plate-scale convective motions in the mantle. The opening of the South Atlantic Ocean during the early Cretaceous appears to have resuited from the rapid propagation of existing spreading ridges into the tensional environment created by the rise of the Tristan da Cunha and St Helena starting plumes. In general, the pattern of rifting reflects local variations in the prerift strength of the continental lithosphere. For example, Wilson & Guiraud (1992) have demonstrated that reactivation of ancient basement linements within the Pan-African lithosphere of west and central Africa has exerted a profound control on the location of rift basins. Thompson & Gibson (1991) suggest that if a plume impacts beneath a structureless lithospheric plate of uniform thickness then the surface expression of the hotspot (i.e. uplift and volcanism) should overlie the axis of the plume head. However, this may not always be the case if the lithosphere is already locally thinned due to earlier rifting and basin development. They suggest that while a thick region of a plate may experience uplift, with little associated mag-

252

M. W I L S O N

matism, above the axis of a plume, basaltic magmas may be emplaced locally, hundreds of kilometres away, in lithospheric 'thin spots'. This may lead to the axis of the plume being incorrectly located beneath the zone of thinned lithosphere, rather than beneath the topographic anomaly. Thompson & Gibson (1991) show that the time gap between the arrival of the mantle plume and the previous rifting event is critical because the thermal time constant of the lithosphere is c. 60 Ma and thinned lithosphere re-thickens by conductive cooling. This geometric situation appears to have operated during the formation of the Paranfi flood basalt province where it appears likely that the axis of the Tristan plume lay closer to the Etendeka area of Namibia, on the edge of the once continuous flood basalt province. Hill (1991) suggests that initiation of extension leading to rifting is critically dependent upon the temperature of the lithosphere. If this is hot and therefore weak, extension may occur before the onset of basaltic volcanism. In contrast, if a plume rises beneath a stable craton, with thick cold lithosphere, it may take some time for sufficient heat to be transported by conduction from the top of the plume into the lithosphere to weaken the crust and upper mantle sufficiently for extension to begin. In such a case, extension may post-date the onset of basaltic magmatism by 10-20 Ma. Watson & McKenzie (1991) have shown that for steady state plumes impinging on the base of a lithospheric plate, the melt production rate is a function of the thickness of the mechanical boundary layer (MBL; Fig. 6). Above a MBL thickness of 125 km (corresponding to old cratonic lithosphere) all melt production stops. Conversely, as the MBL thickness approaches zero, melt production increases to about 1 km 3 a -1. If the entire volume of the Ontong-Java plateau (6x 107 km 3) were emplaced in less than 3 Ma as proposed by Tarduno et al. (1991) this would imply a melt production rate of 20 km 3 a -~. In contrast, for the Parami, if we assume a total volume of 2× 10 6 km 3 emplaced over 5 Ma we get an eruption rate of 0.4 km~ a -1. From Fig. 6 this would predict a MBL thickness of 40 km for a potential temperature of 1550°C. This might be quite reasonable for a region of continental lithosphere which had experienceda long history of extension prior to the arrival of the plume (e.g. Thompson & Gibson 1991). On the basis of Fig. 6 we would predict that plumes impinging on the base of young oceanic lithosphere (thin MBL) would generate much larger volcanic edifices than plumes impinging on the base of the continental lithosphere. This could explain the much larger volumes of oceanic plateaux

1.2

'~ E

1.0

T - 1550 C P

._~ 0.6 ~

.4

~

.2 0

0

30

60

90

120

150

MBL thickness, km F i g . 6. Melt production rate as a function of

mechanical boundary layer thickness for a potential temperature (Tp) of 1550°C. After Watson & McKenzie (1991). compared to continental flood basalt provinces. However, it is not capable of explaining the enormous magma production rates, of the order of 20 km 3 a -1, necessary to produce the OntongJava plateau in 3 Ma. This would suggest that melt production rates are much higher during the initial non-steady state stage when the plume head first impinges on the base of the lithosphere. Summary

During the Early Cretaceous opening of the South Atlantic Ocean two deep mantle plumes, St Helena and Tristan da Cunha, appear to have exerted a fundamental control on the process of continental break-up. The St Helena plume appears to have been cool and relatively weak with a low buoyancy flux. It may have cooled by entrainment of depleted mantle material within the upper mantle. Extension across the starting plume head generated a broad zone of rifting and scattered alkalic-transitional basaltic magmatism in West and Central Africa and NE Brazil, active for 30-40 Ma before break-up occurred in the Equatorial Atlantic. In contrast, the Tristan plume appears to have been hotter and more vigorous (relatively high buoyancy flux) and appears to have triggered the eruption of voluminous tholeiitic flood basalts in the Paranfi basin of Brazil shortly before continental break-up. Whilst it cannot be proven, it seems reasonable to suppose that both the Tristan and St Helena plumes originated by destabilization of the D" layer at the core-mantle boundary. The different characteristics of the two plumes then reflect initial heterogeneities in D" and different amounts of entrainment of cooler mantle material as the plume heads rose towards the surface. The St Helena plume appears to sample both EM I and HIMU components initially but not later. This may reflect the involvement of small amounts of pelagic sediment (EM I component)

S ATLANTIC OPENING AND CRETACEOUS SUPER-PLUMES

in the plume source which rapidly became exhausted. In contrast, the isotopic characteristics of the Tristan plume appear to be dominated by the EM I component throughout its history, suggesting involvement of a larger sedimentary component in the plume source. On a global scale, the flood basalt volcanism of the Paran~i precedes, or is concurrent with the start of the Long Cretaceous normal polarity superchron, envisaged by Larson (1991) to reflect the uprise of super-plume heads from the core-mantle boundary. Major oceanic plateaux volcanism in the Pacific and Indian oceans commences some 5-10 Ma later. Thus we can envisage the Early Cretaceous as a period of great upheaval in the Earth's mantle with progressive uprise of plume heads from D". The first superplume to lift off may well have been St Helena although this is not well constrained. Whilst rifting and magmatism initiated in West and Central Africa at c. 150-145 Ma, these may initially have been the passive response to plate boundary forces and not necessarily related to the arrival of the St Helena plume at the base of the lithosphere. The timing of extensive flood basalt volcanism in the Paranfi is, however, somewhat better constrained and clearly precedes the Pacific/ Indian Ocean super-plume activity. I would like to thank J. M. O'Connor for allowing me to see pre-prints of O'Connor etal. (1992) and O'Connor & Le Roex (1992) which stimulated several of the ideas expressed in this work.

References ANDERSON, D. L. 1982. Hot spots, polar wander, Mesozoic convection and the geoid. Nature, 297, 391-393. BROZENA, J. M. & WHITE, R. S. 1990. Ridge jumps and propagations in South Atlantic Ocean. Nature, 348, 149-152. CAMPBELL, I. H. & GRI~ITHS, R. W. 1990. Implications of mantle plume structure for evolution of flood basalts. Earth and Planetary Science Letters, 99, 79-93. CASTROJr., A. C. M. 1987. The northeastern Brazil and Gabon Basins: a double rifting system associated with multiple crustal detachment surfaces. Tectonics, 6, 727-738. CHAFfy, D. J., CLIFF,R. A. & WILSON,B. M. 1989. Characterization of the St Helena magma source. In: SAUNDERS,A. D. & NORRY,M. J. (eds) Magmatism in Ocean Basins. Geological Society, London, Special Publication, 42, 257-276. COFFIn, M. F. 1991. Crustal structure and emplacement rate of the Kerguelen and Ontong-Java plateaus. Caltech Mantle Plume Symposium Abstracts. Pasadena, California. & ELDHOLM,O. (eds) 1991, Large Igneous Pro-

253

vinces: JOI/USSAC Workshop Report. The University of Texas at Austin Institute for Geophysics Technical Report No. 114. COURTILLOT, V., FERAUD, G., MALUSKI, H., VANDAMME, n . , MOREAU, i . G. & BESSE, J. 1988. Deccan flood basalts and the Cretaceous/Tertiary boundary. Nature, 333, 843-846. CROUGH, S. T., MORGAN,W. J. & HARGRAVES,R. B. 1980. Kimberlites: Their relations to mantle hot spots. Earth & Planetary Science Letters, 50, 260274. DUNCAN,R. A. & PYLE,D. G. 1988. Rapid eruption of the Deccan flood basalts at the Cretaceous/Tertiary boundary. Nature, 333, 841-843. & RICHAROS, M. A. 1991. Hotspots, mantle plumes, flood basalts and true polar wander. Reviews of Geophysics, 29, 31-50. ERLANK,A. J., MARSH,J. S., DUNCAN,A. R., MILLER, R. M., HAWKESWORTH,C. J., BETrON, P. J. & REX, D. C. 1984. Geochemistry and petrogenesis of the Etendeka volcanic rocks from SWA/ Namibia. Special Publication of the Geological Society of South Africa, 13, 195-245. FAIRHEAD,J. D. & BINKS,R. 1991. Differential opening of the Central and South Atlantic Oceans and the opening of the West African Rift System. Tectonophysics, 187, 191-203. FLEITOUT,L., DALLOUBEIX,C. & MORICEAU,C. 1989. Small-wavelength geoid and topography anomalies in the South Atlantic Ocean: A clue to new hot spot tracks and lithospheric deformation. Geophysical Research Letters, 16, 637-640. FODOR, R. V., SIAL,A. N., MUKASA,S. B. & MCKEE, E. H. 1990. Petrology, isotope characteristics, and K-Ar ages of the Maranhao, northern Brazil, Mesozoic basalt province. Contributions to Mineralogy and Petrology, 104, 555-567. GRI~ITHS, R. W. & CAMPELL,I. H. 1990. Stirring and structure in mantle starting plumes. Earth and Planetary Science Letters, 99, 66-78. & 1991. On the dynamics of long-lived plume conduits in the convecting mantle. Earth and Planetary Science Letters, 103, 214-227. HALLIDAY, A. N., DAVIDSON,J. P., HOLDEN, P. DE WOLF, C., LEE, D.-C. & FITI'ON, J.-G. 1990. Trace element fractionation in plumes and the origin of HIMU mantle beneath the Cameroon line. Nature, 347, 523-528. HANNAN, B. B., KINGSLEY,R. H. & SCHILLING,J.-G. 1986. Pb isotope evidence in the South Atlantic for migrating ridge-hotspot interactions. Nature, 322, 137-144. HARRY, D. L. & SAWYER,D. S. 1992. Basaltic volcanism, mantle plumes, and the mechanics of rifting: The Parana flood basalt province of South America. Geology, 20, 207-210. HARTNADY,C. J. H. & LE ROEX,A. P. 1985. Southern Ocean hot spot tracks and the Cenozoic absolute motion of the African, Antarctic and South American plates. Earth and Planetary Science Letters, 75, 245-257. HAWKESWORTH,C. J. & GALLAGHER,K., KELLY, S., MANTOVANI,M., PEATE,D. W. REGELOWS,R. & ROGERS,N. W. 1992. Parami magmatism and the opening of the South Atlantic. This Volume.

254

M. WILSON

, KEMPTON,P. D., ROGERS,N. W., ELLAM,R. M. & VAN CALSTEREN, P. W. 1990. Continental mantle lithosphere, and shallow level enrichment processes in the Earth's mantle. Earth and Planetary Science Letters, 96, 256-268. - - , MANTOVANI,M. S. M. & PLATE, D. W. 1988. Lithosphere remobilisation during Parand CFB magmatism. In: Cox, K. G. & MENZIES, M. A. (eds) Oceanic and continental lithosphere: Similarities and differences. Journal of Petrology Special Volume, 205-223. , - - , TAYLOR P. N. & PALACZ, Z. 1986. Evidence from the Parami of south Brazil for a continental contribution to Dupal basalts. Nature, 322,356-359. HILL, R. I. 1991. Starting plumes and continental break-up. Earth and Planetary Science Letters, 104, 398-416. LARSON, R. L. 1991. Latest pulse of Earth: Evidence for a mid-Cretaceous superplume. Geology, 19, 547-550. & OLSON,P. 1991. Mantle plumes control magnetic reversal frequency. Earth and Planetary Science Letters, 107, 437-447. LE ROEX, A. P., CLIFF, R. A. & ADAIR, B. J. I. 1990. Tristan da Cunha, South Atlantic: Geochemistry and petrogenesis of a basanite-phonolite lava series. Journal of Petrology, 31,779-812. MAHONEY, J. J. & SPENCER,K. J. 1991. Isotopic evidence for the the origin of the Manihiki and Ontong Java oceanic plateaus. Earth and Planetary Science Letters, 104, 196-210. MASCLE, J., BLAREZ, E. & MARINHO, M. 1988. The shallow structures of the Guinea and Ivory CoastGhana transform margins: their bearing on the Equatorial Atlantic Mesozoic evolution. Tectonophysics, 155, 193-209. McNuTr, M. K., WINTERER, E. L., SAGER, W. W., NATLANO, J. H. & ITO, G. 1990. The Darwin Rise: A Cretaceous Superswell? Geophysical Research Letters, 17, 1101-1104. MORGAN, W. J. 1972. Deep mantle convection: Plumes and plate motions. Nature, 230, 42-43. 1983.Hot-spot tracks and the early rifting of the Atlantic. Tectonophysics, 94, 123-139. NURNBERG, D. & MULLER, R. D. 1991. The tectonic evolution of the South Atlantic from Late Jurassic to present. Tectonophysics, 191, 27-53. O'CONNOR, J. M, & DUNCAN,R. A. 1990. Evolution of the Walvis Ridge-Rio Grande Rise hot spot system: Implications for African and South American plate motions over plumes. Journal of Geophysical Research, 95, 17475-17502. & LE ROEX, A. P. 1992. South Atlantic hot spot volcanism: 1. Time-space constraints on African plate motion. Earth & Plantary Science Letters, (in press). • WHITE, W. M., CHRISTIE,D. M. & SICHEL,S. E. 1992. South Atlantic hot spot volcanism: 2. Isotopic variability in response to mixing of three mantle components. Earth and Planetary Science Letters, (in press). PEATE, D. W., HAWKESWORTH,C. J. MANTOVANI,M. S. M. & SHUKOWSKY,W. 1990. Mantle plumes

and flood basalt stratigraphy in the Parant, South America. Geology, 18, 1223-1226. PRINGLE, M. S. 1991. Early Cretaceous plateau basalts, EM I from the lower mantle and the origin of the Dupal anomaly. Extended Abstract.

Caltech Plume Symposium, Pasadena. RABINOWITZ, P. D. & LA BRECQUE, J. L. 1979. The Mesozoic South Atlantic Ocean and evolution of its continental margin. Journal of Geophysical Research, 84, 5973-6002. RAMPINO,M. R. & STOTHERS,R. B. 1988. Flood basalt volcanism during the past 250 million years. Science, 241, 663-668. RICHARDS,M. A., DUNCAN,R. A. & COURTILLOT,V. E. 1989. Flood basalts and hotspot tracks: plume heads and trails. Science, 246, 103-107. RICHARDSON,S. H., ERLANK,A. J., DUNCAN,A. R. & REID, D. L. 1982. Correlated Nd, Sr and Pb isotope variation in Walvis Ridge basalts and implications for the evolution of their mantle source. Earth and Planetary Science Letters, 59, 327-342. SCHILLING,J. G. 1991. Fluxes and excess temperatures of mantle plumes inferred from their interaction with migrating mid-ocean ridges. Nature, 352, 397-403. SIBUET, J.-C., HAY, W. W., PRUNIER, A., MONTADERT, L. HINZ, K. & FRITSCH, J. 1984. Early evolution of the South Atlantic: Role of the rifting phase. Initial Reports of the Deep Sea Drilling Project, 75, 469-481. SLEEP, N. H. 1990. Hotspots and mantle plumes. Some phenomenology. Journal of Geophysical Research, 9S, 6715-6736. 1991. Time dependence of mantle plumes. Ex-

tended Abstract. Caltech Plume Symposium, Pasadena. STAUDIGEL, H., PARK, K.-H., PmNOLE, M., RUBENSTONE,J. L., SMITH,W. H. F. & ZINDLER, A. 1991. The longevity of the South Pacific isotopic and thermal anomaly. Earth and Planetary Science Letters, 102, 24-44. TARDUNO, J. A., SLITER, W. V. & MAYER, H. 1991. Mantle plume volcanism and the formation of the Ontong Java plateau. Extended Abstract. Clatech

Plume Symposium. Pasadena.

THOMPSON,R. N. & GIBSON,S. A. 1991. Subcontinental mantle plumes, hotspots and pre-existing thinspots. Journal of the Geological Society, London, 147, 973-977. UCHUPI, E. 1989. The tectonic style of the Atlantic Mesozoic rift system. In: ROSENDHAL,B. R., ROGERS, J. J. W. & RACH,N. M. (eds) African Rifting. Journal of African Earth Sciences, 8, Special Issue, 143-164. UNTERNHER,P., ComE, D., OLIVET,J. L., GOSL~, J. & BEUZART,P. 1988. South Atlantic fits and intraplate boundaries in Africa and South America. Tectonophysics, 155, 169-179. URIEN, C. M. & ZAMaRANO,J. J. 1973. The geology of the basins of the Argentine continental margin and Malvinas Plateau. In: NAIm~, A. E. M. & STEHLI, F. G. (eds) The Ocean Basins and its Margins, Volume 1, The South Atlantic. Plenum,

S ATLANTIC OPENING AND CRETACEOUS SUPER-PLUMES New York, 135-169. WATSON, S. & MCKENZIE, D. 1991. Melt generation by plumes: A study of Hawaiian volcanism. Journal of Petrology, 32, 501-537. WEAVER,B. L. 1991. The origin of ocean island basalts end-member compositions: trace element and isotopic constraints. Earth and Planetary Science Letters, 104, 381-397. WEIS, D., BASSIAS,Y., GAUTIER,I. & MENNESSIER,J.P. 1989. Dupal anomaly in existence 115 Ma ago: Evidence from isotopic study of the Kerguelen Plateau (South Indian Ocean). Geochimica et Cosmochimica Acta, 53, 2125-2131. WHITE, R. & MCKENZm, D. 1989. Magmatism at rift zones: The generation of volcanic continental

255

margins and flood basalts. Journal of Geophysical Research, 94, 7685-7729. WmsoN, M. & GUmAUD, R. 1992. Magmatism and rifting in Western and Central Africa from Late Jurassic to Recent times. Tectonophysics 213, 123. ZALAN, P. V., WOLFF, S., CONCEICAO,J. C. J., ASTOLFI, M. A. M., VIERA, I. S., APPI, V. T. & ZAt4OTrO, O. A. 1987. Tectonica e sedimentacao da bacia do Parana. Atas do Ill simposio sulbrasilerio de geologica, Curitiba, 1,441-477. ZINDLER, A. & HART, S. R. 1986. Chemical geodynamics. Annual Reviews of Earth & Planetary Science, 14, 493-571.

New geophysical evidence for extensional tectonics on the divergent margin offshore Namibia M. P. R. L I G H T , M. P. M A S L A N Y J & N. L. B A N K S

Intera Information Technologies Ltd., Exploration Division, Highlands Farm, Greys Road, Henley-on-Thames, Oxon R G 9 4PS, UK

Abstract: Over 14 000 km of high-resolution multifold seismic data together with gravity and

magnetic data provide the opportunity to examine the mechanism and history of rifting on the Namibian continental margin. The region is a completely developed divergent margin containing a pre-rift megasequence of interior cratonic sag origin; two rift basin megasequences dominated by siliclastic deposition; a transitional megasequence and thermal sag megasequences dominated by overlapping progradational wedges. In the Orange Basin acid to intermediate volcanic rocks erupted at the end of the pre-rift phase in Mid- to Late Jurassic times and this was followed by regional uplift. Basaltic volcanic activity was associated with the synrift phases and widespread volcanic rocks developed during the second synrift event, related to the Tristan da Cunha-Walvis Ridge mantle plume. Regional seismic mapping indicates that rifting migrated from south to north with time. The geophysical data enable the recognition of major structural elements which include the Eastern Graben Province, the Medial Hinge Line, the Central Half Graben and the Marginal Ridge. The geometry of the rift basin is asymmetric and can be explained by a simple shear mechanism. South of Walvis Ridge extension was accommodated by movement along a major normal fault which is listric at depth. The basin depocentre represented by the Central Half Graben lies landward of this major fault and is offset from the area of thinnest crust as interpreted from the gravity data.

The Namibian offshore region occupies some 310000 sq km out to the 2000 m isobath (Fig. 1). It is part of the classic Atlantic type passive margin yet until recently very little was known of the offshore structure. Between 1989 and 1991 over 14 300 km of multifold high resolution seismic, gravity and magnetic data were acquired by Intera Information Technologies/Halliburton Geophysical Services on behalf of Namcor, the Namibian National Oil Company. The survey was designed to provide regional coverage of the entire offshore area out to the 2000 m bathymetric contour, between latitudes 17° and 30°30'S (Fig. 1). Two major end-members of crustal extension models have been established: the model of pure shear (McKenzie 1978) and the model of simple shear (Wernicke 1985). Refinements to the models have been developed in order to explain other features of modern rift valleys and sedimentary basins (Kusznir et al. 1988; Buck 1988). In the South Atlantic two different simple shear models have been proposed. The first model, based on a stratigraphical and gravity study of onshore and offshore Brazilian basins, proposes Wernicke-style extension producing

an eastward-inclined crustal shear between Brazil and Gabon (Ussami et al. 1986). Etheridge et al. (1989) on the other hand, have argued that the Walvis Ridge represents a transfer zone separating a major eastward-inclined crustal shear in the south from a westwardinclined crustal shear to the north. The major escarpment along the African coast juxtaposed against wide rifted Argentine basins is mirrored by an escarpment in Brazil juxtaposed against wide basins offshore Angola (Etheridge et al. 1989). Alternatively the model for the Campos Basin offshore Rio de Janeiro suggests regional lithospheric stretching and crustal thinning with Moho uplift compensating for sediment accumulation (Mohriak et al. 1990). The new data provide an opportunity to examine the mechanism and history of rifting on the Namibian margin and to test the above extension models. Synrift volcanic sequences are imaged well by these data and enable an assessment of the relationship in time between rifting and volcanism. This paper summarizes geophysical and geological characteristics of the offshore area that support passive rifting and a simple shear crustal extension model.

From STOREY,B. C., ALABASTER,T. & PANKHURST,R. J. (eds), 1992, Magmatism and the Causes of Continental Break-up, Geological Society Special Publication No. 00, pp. 000-000.

257

258

M . P . R . LIGHT

ETAL.

~.,.,,,."-" \ .

NAMIBE BASIN

ETOSHA

BASIN o

o /

DSDP

WALVIS RIDGE

DAMARA OROGENIC BELT

\ \

WALVIS BASIN

• Windhoek Walvis

\ \ \

LUDERITZ ARCH

:::::::::::::::::::

\

:~:::

::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::: : ...

w~,,s]~ i ~i~i? .

LUDERITZ " BASIN

~ u o u

~ \~

\='~1 5)

T

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

--

9A-1 & 9 A - 3 ~'~.:.:.J v

'

'

f

;

\

.

.J

: \ \

CONTINENT / ' : , ,

A-J

/

1

OCEAN BOUNDARY

i" 0 I

' su"vEYI°"A"GE 1991 SURVEY

I

I

I

400kin ,,J

BASIN

\

\

9

Cape TOWN

\

Fig. 1. The Atlantic margin, offshore Namibia, location map and Intera/Halliburton Geophysical Services seismic survey.

EXTENSIONAL TECTONICS OFFSHORE NAMIBIA

Geophysical data Data acquisition Seismic data were acquired using a TSROO 1 recording system. A 3600 m digital streamer and 2180 cubic inch airgun array were towed at depths of 8 and 6 m respectively. Data were captured at 30 m intervals on 240 channels with a seven second record length. These were resampled to 60 m during processing. Pre-and post-stack deconvolution and FK migration were the main features of the processing sequence. •The gravity data were acquired using a Lacoste and Romberg stable platform air-sea gravity meter (meter ID S-10"/). Dockside gravity readings were recorded on a portable Lacoste and Romberg land gravimeter and facilitated an international network tie and monitoring of meter drift. The magnetic data were acquired using a Geometrics G-801 magnetometer. A Simrad EA deep water bathometer system was used to measure and record water depths up to 700 m. Seismic data were used to determine water depths in excess of this. Navigation was determined using Syledis, an ultra-high frequency (405-450 mhz) radio positioning system. Gravity and magnetic data were processed in a conventional manner using LCT processing software. Misties were removed using a line weighted correction prorated between adjacent intersections. After systematic network adjustments the average rms misties for the gravity and magnetic data were less than 1 mgal and 5 nT respectively.

Seismic data interpretation The offshore region consists of four basins (Fig. 1): (1) the Orange Basin; (2) the Luderitz Basin; (3) the Walvis Basin; (4) the Namibe Basin. The main structural elements imaged by seismic data south of Walvis Ridge are: (1) an Eastern Graben Province on the platform area where grabens and half grabens developed along reactivated coastward-verging Pan African thrusts (Fig. 2); (2) a Medial Hinge Line where rift sequence boundaries are truncated and which can be traced with confidence from the Orange Basin to the Walvis Ridge (Fig. 3); (3) a Central Half Graben developed west of the Medial Hinge Line where the •dominant rotation of strata is down to the east (Fig. 3); (4) a Marginal Basement Ridge developed further west (Fig. 4); (5) a major sedimentary wedge which presumably developed during thermal subsidence. Principal sequence boundaries (Fig. 5) were identified and mapped in the offshore area to define the major depositional sequences (Maslanyj et al. 1992), and extend the work of Gerrard & Smith (1982). The interpretation that follows was constrained by the K ~ e l l s (9A-1, and 9A-3) and A-J1 in the Orange Basin, DSDP

259

wells near Walvis Ridge and regional onshore geology (Fig. 1). The main tectono-stratigraphical megasequences can be sub-divided into five phases of rift development (Fig. 5); (1) Thermal Sag (Horizon P-Sea Floor); (2) Transitional (Horizon Q-P); (3) Synrift II (Horizon R - Q ) ; (4) Synrift I (Horizon T - R ) ; (5) Pre-rift (Horizon W-T). Horizon T is the Late Jurassic angular unconformity at the top of the pre-rift section. It is marked by a very high amplitude continuous reflector. Horizon R has previously been given two different ages; Valanginian by Gerrard & Smith (1982) and Hauterivian by McLachlan & McMiUan (1979). In the present study, two separate unconformities, Horizon Q (Hauterivian) and Horizon R (Valanginian), are proposed. Horizon P is the mid-Aptian unconformity, which is the break-up unconformity north of the Walvis Ridge. It forms a continuous acoustic marker of moderate to high amplitude, overlies extensive Lower Aptian shales and, to the east of the Medial Hinge Line, truncates Horizons R and T. It represents the top of a prograding sequence, and there is downlap of overlying beds. The Turonian unconformity, Horizon N, occurs at the top of a prograding sequence and is continuous to the east. In the west it deteriorates into a zone of slumping. Overlying beds downlap in the west, and onlap in the east. The base of the Tertiary, Horizon L, is a continuous high amplitude reflector at the top of a prograding sequence and is usually disconformable. There is onlap and downlap of overlying beds, and the underlying sequence exhibits toplap/erosion. North of the Walvis Ridge, Horizon G (Oligocene) and Horizon A (Mid-Miocene) form erosional unconformities that are dissected by several transverse channels.

Pre-rift phase. The Pre-rift megasequence consists essentially of the Karoo formations as defined within the main Karoo Basin and southern Namibia (Kent 1980). It is difficult, however, to interpret the seismic facies of this unit with confidence due to the general deterioration in seismic signature with increasing depth, and there is almost total character loss in places, particularly beneath overlying highly reflective intervals. Furthermore, there is no well control to confirm sub-Horizon T facies interpretation except in the South African portion of the Orange Basin. Along a broad N-S-trending belt, west of the Medial Hinge Line, seismic reflectors display a complex pattern of high amplitude, discontinuous, irregular and • hummocky events. They probably represent arid continental deposits with aeolian dunes, fluvial sands, lower energy

260

M . P . R . LIGHT E T A L .

A,

Fig. 2. An East West seismic profile (A-A') across the Eastern Graben Province (see Fig. 8 for position).

B

B'

Fig. 3. "An East West seismic profile (B-B') across the Medial Hinge Line and Central Half Graben (see Fig. 8 for position).

261

EXTENSIONAL TECTONICS OFFSHORE NAMIBIA

C'

o

15

I

I

KM

Fig. 4. An East West seismic profile (C-C') across the Marginal Basement Ridge (see Fig. 8 for position). braidplain silts and shales, and extensive tracts of subaerially extruded lavas. In the western part of the Orange Basin, seismic reflection character suggests a rather more uniform sequence, where reflectors are generally more continuous. This aspect might indicate a shallow water-lain sand and shale sequence, possibly of shallow marine origin. A series of major deltaic systems formed in the Permian, building out westwards from Central Africa (Nossob, Early Auob and Late Wankie Sandstones) into the marine environment in Namibia, where they were reworked by marine processes. Formations equivalent to the marine reworked deltaic complexes of the Nossob and Auob (Early and Mid-Ecca) may be present within the Central Half Graben, probably as sand-shale sequences. These units are believed to undergo a series of facies changes to the west of the Medial Hinge Line, passing from a clastic apron (alluvial fans) through fluvial systems, reworked deltaic facies and shallow marine units to deep water sand-shale sequences in the core of the Central Half Graben. Similar facies changes may occur east of the Medial Hinge Line in smaller graben that are rotated on listric detachments located on older Pan-African

thrusts. Seismic reflection character implies that these are infilled with marginal alluvial fans grading laterally and vertically into fluvial and lacustrine sediments. The period of aridity preceded and coincided with extensive volcanicity in the Mid- and Late Jurassic, when acid to intermediate lavas erupted in the Orange Basin. The period of volcanism was terminated by regional uplift, block rotation and the development of a regional unconformity in the KimmeridgianOxfordian (+155.5 Ma). Uplift exceeded 1 km in the region of the Great Escarpment, east of the modern Namib Desert.

Synrift I phase. The Synrift I interval represents a wedge-shaped, generally westwards thickening sequence, which pinches out at the Medial Hinge Line in the east, usually due to erosion. No wells in the Namibian offshore have penetrated this sequence, therefore facies interpretations are speculative. E a s t of the Medial Hinge Line, narrow elongate thrust ramp graben and half graben occur, which were syntectonically filled with alluvial and fluvial sediments during the waning phases of rifting. Progradational and chaotic reflection patterns along the margins

M . P . R . LIGHT E T A L .

262

TECTONO STRATIGRAPHY

FACIES QUAT

~

~

~---

>" rr

'~~ "


30 Ma) uplift despite the commonly held belief that the Afar plume existed beneath the region 30 Ma ago. Geological data point to an episode of uplift that occurred after the initiation of magmatism. Fission track data indicate that uplift related exhumation postdates magmatism by some 10-15 Ma, perhaps the amount of time needed to change the thermal character of the Pan-African lithosphere "above the Afar plume. A sequence of magmatism followed by synchronous crustal extension and uplift for Yemen does not fit with the traditional categories of active (uplift-magmatism-rifting) and passive (riftinguplift-magmatism) rifting. Clearly such end-member models do not simply apply to the Red Sea or the Great Basin of the western USA where a period of tectonic quiescence, followed by post-volcanic extension and uplift (1 km), post-dated the Oligo-Miocene ignimbrite flare-up.

The relative timing of surface uplift, magmatism and extension is thought to be pivotal to understanding whether the mantle was a passive or active participant in rift formation (Seng6r & Burke 1978). The passive model requires that extension predates any uplift and magmatism, while in the active model uplift predates magmatism and extension. However, observations of rift margin sequences, particularly in the Red Sea, show that rift formation is in practice more complex. It has long been recognized that the triple-junction structure of active rifts is strongly associated with domal surface uplift and volcanism (Cloos 1939). Receiit theoretical considerations (McKenzie & Bickle 1988; White & McKenzie 1989; Houseman 1990; Farnetani & Richards 1991) indicate that mantle, plumes or hot spots are inextricably linked to the rapid

effusion of continental flood basalts. The generation of large volumes (c. 2 x 106 km 3) of magma involved in flood volcanism requires superposition of rifting on anomalously hot mantle (i.e. plumes, > 1380°C). According to these models, based on swells in oceanic lithosphere, considerable amounts of uplift of the order of 1-2 km at the plume centre are expected to pre-date magmatism and rifting, and because of the lateral dimensions of plumes, uplift is expected to have some effect up to 1500-2000 km radius from the plume centre. Houseman (1990) further suggests that, at triple-junctions where extension is followed by the separation of a failed arm and two-armed passive margin (e.g. Ethiopia and southern Red Sea/Gulf of Aden), the point of inflection between the two arms of the passive margin (e.g. Yemen) ' . . . should be associated

FromSTOREY,B. C., ALABASTER,T. & PANKHURST,R. J. (eds), 1992, Magmatismand the Causes of ContinentalBreak-up, Geological Society Special Publication No. 68, pp. 293-304.

293

294

M.A. MENZIES E T A L .

with the most intense pre-break-up uplift and volcanism and the earliest initiation of rifting'. Such theoretical considerations have important implications for the relative timing of surface uplift, magmatism and extension in flood basalt provinces as they imply that plume involvement will trigger significant surface uplift prior to magmatism and finally extension. This paper summarizes recent geological observations from Yemen and utilizes these data to constrain the timing and amount of uplift, magmatism and crustal extension. Preliminary fission track data from Yemen are compared with the fission track data from the Sinai Peninsula, eastern Egypt and Saudi Arabia in an attempt to evaluate the appropriateness of active and passive models of tiffing. 20,

30'

Yemen Excellent exposure exists along the Yemen riftflank of the southern Red Sea (Fig. 1) due to 3.5 km of relief. This offers a unique opportunity to study the detailed geological relationships between magmatism, sedimentation and tectonics within the framework of uplift and subsidence of the Arabian rift margin. The exposed lithologies bracket a considerable period of time from PanAfrican (c. 500-900 Ma) basement through Mesozoic to Tertiary sediments and > 2 km thickness of Tertiary to Recent flood volcanism. The inter-relationships of these lithologies record distinct phases of tectonics, sedimentation and volcanicity which may be used to elucidate the particular nature of the rift process. In conjunction with an on-going programme of K-Ar

40.

So °

e0.

EURASIAN PLATE

TURKISH

40"

PLATE

:::::::Mediterrar •~::::

Sea

20"

30'

ARABIAN PLATE

/ 20"

20"

YEMEN AFRICAN PLATE

10"

r

--:-'-'.'.'.'.--'-'-',',','.'.:," • ":'-':':':': A r a b / a n ,':

!:!:!:!:!:i:!:i: ~ .,..o°...,°,o.¢ . . . . . . . . . .

0 Ma

PRESENT

=

| 40"

/

3

,..

~ /..:i!i!i!!iii!~iii~i~i~!~!ii!i!!!i!!?!!!!ii~!iii!~!i ::iiiiiiiiii!!iiiiiiii!:iiii:iiiililili!iiiiiiii!iiii:ii:iiiiii:i 50 °

Fig. 1. Tectonic setting of the Red Sea. Note the northward movement of the African and Arabian plates and the possible 'sphere of influence' of the Afar plume (White & McKenzie 1989). The Yemen and Ethiopian flood volcanics are located above the postulated plume.

MAGMATISM, UPLIFT AND EXTENSION age determinations in the Yemen volcanics (e.g. Menzies et al. 1990; Huchon et al, 1992), the existence of pre-, syn-, and post-volcanic structure can help constrain the temporal relation between tiffing and volcanism. Sedimentologieal studies (Al'Subbary & NiehoUs 1991; Bosenee et al. 1992) indicate that the Kholan Formation (of unknown age), which immediately overlies the Pan-African basement, is up to 200 m thick. These sediments represent a transition from continental sedimentation to a shallow-marine environment and pass upwards into a succession of carbonates that reach a thickness of 400 m (Amran Formation of CaUovian to Kimmeridgian age). The Amran carbonates eventually became emergent and the siliciclastic Tawilah Formation, of Cretaceous to Palaeocene age, was deposited on an eroded Amran surface. This is thought to be a major sequence boundary where shallow marine sediments onlap a shoreline of cemented Amran limestones. Within the Tawilah sandstones there is evidence for shallow-marine sedimentation but the bulk of the formation is a sequence of braided fluvial channel deposits interbedded with palaeosols developed on overbank deposits. The thickness of the formation appears to be relatively constant at around 400 m across the traverse. Eventually a transition occurs between the Tawilah Formation and the overlying volcaniclastic units. In this transition a variety of lithologies occur including terrestrial or shallow-marine sandstones, lateritic palaeosols, shallow-marine sediments and volcanic rocks. Here gastropodrich horizons, presumed to be of shallow marine origin, are exposed at different altitudes ranging from 900-2400 m, implying 1.0-2.5 km of uplift after the onset of the overlying magmatism (30 Ma). Lateritic disconformities also occur at the sediment-basalt contact and are widespread in Yemen as they are in Saudi Arabia (Camp & Roobol 1989) and elsewhere. The presence of marine sediments above these laterites and the lack of evidence for erosion indicates that at this time (30 Ma) uplift was minimal. Also, the lack of angular unconformities between the sediments and overlying basalts can be used as evidence that crustal extension and possibly uplift did not significantly pre-date magmatism. However, it should be remembered that if uplift is distributed over a broad area (1000-2000 km) around the plume head without concomitant extension it will not necessarily be associated with significant breaks in the sedimentary succession. In contrast, if uplift had occurred synchronously with magmatism and extension then clastic sediments shed from any uplifted region should appear in the sedimentary record and palaeocur-

295

rent directions may be expected to radiate from the updomed region. Theoretically, the unroofing sequence preserved in the sediments should record the inverse of the present stratigraphy. This was not observed anywhere within the traverse and only one basaltic pebble (Tertiary?) was found within a conglomeratic horizon in the Tawilah Formation. Consequently there is no evidence in the Jurassic to Tertiary sedimentary record for pre-volcanic uplift with associated erosional unroofing of older rocks and their involvement in sedimentary processes. Any elevation change within these sedimentary units can be measured a t most in a few tens of metres, and not in kilometres as would be required with significant pre-volcanic uplift and extension above a plume. It is vital to determine the timing of crustal extension and to ascertain whether it predates, is synchronous with, or postdates the volcanic rocks. While structural investigations (McClay et al. 1991) in the plaform stratigraphy indicate little or no evidence for widespread pre-volcanic structure ( > 30 Ma) there is evidence for minor amounts of crustal extension, in the form of block faulting, within the upper part of the volcanic sequence (25-20 Ma) and considerable amounts of crustal extension that post-dates the eruption of the volcanic rock units ( < 20 Ma). Pre-volcanic extensional structures (> 30 Ma) as a result of rifting or uplift would be apparent as angular unconformities between basal volcanic units and underlying lithologies, basal conglomerates or breccias, and control of basement fault blocks on the distribution of volcanic units. No marked angular conformities were observed at the sediment-volcanic contact throughout the study area in Yemen (see also Menzies et al. 1990) and no evidence was found, in the underlying lithologies, for faulting that dies out upwards. All of the pre-volcanic sediments have been rotated by the same amount indicating that they were deposited before the main episode of extensional faulting. Syn-volcanic crustal extension (30-20 Ma) would result in angular unconformities between volcanic units, fanning dips, sedimentary deposits within the volcanic pile, significant lateral variations in the thickness of ash flows within and between adjacent fault blocks and limitations on the lateral extent of ash flows due to topographic highs. Possible synvolcanic extension was observed within the Yemen volcanics at one locality where an angular unconformity may occur within the uppermost volcanic units ( 4 25 Ma). Here, the upper volcanic units may have been rotated 20 degrees less than the underlying sediments suggesting that the upper volcanics were erupted during ex-

296

M.A. MENZIES E T A L .

tensional faulting. However, this is difficult to evaluate since the section studied is in close proximity to a granitic intrusion which may have disrupted the adjacent volcanic rocks. If late syn-volcanic extension did occur it may be related to the reported change in the extensional stress field from E - W ( > 22 Ma) to N - S ( < 22 Ma) (Huchon etal. 1992). It is important to point out that Huchon et al. (1992) dated dyke intrusion which does not necessarily date the episode of crustal extension particularly if it occurred by block faulting. Post-volcanic crustal extension (< 20 Ma) generates structures that are the most widespread and best developed throughout the region. Rotated fault blocks contain hundreds of metres of volcanic rocks resting on platform sediments which in turn rest on Pan-African basement. Assessment of the presence or absence Of pre-, syn- and post-volcanic structure indicates no pre-volcanic uplift ( > 30 Ma). Uplift appears to have occurred during or immediately after most of the Tertiary volcanism. Not only is this conclusion consistent with what was deduced from the nature of the pre-volcanic sediments in Yemen but it is also consistent with the lack of pre-voleanic structure to the north in Saudi Arabia (Bohannon et al. 1989). Volcanological studies indicate that present exposures of sub-aerial volcanic rocks are some 2500 m thick. However, this may not constitute the true thickness of the volcanic rocks because contemporaneous Tertiary granites ( < 24 Ma) intruding the volcanic rocks are now exposed, unroofed, at 3015 m altitude. Age data indicate that minor volcanism may have begun around 45 Ma but reached a peak at 30-19 Ma (Civetta et al. 1978; Chiesa et al. 1983, 1989; Capaldi et al. 1987; Menzies et al. 1990; Huchon et al. 1992). Since the 30-19 Ma range is determined on the erosional remnants of the volcanic pile, erosional unroofing may have removed as much as 1-2 km of the volcanics. For example, peak basaltrhyolite volcanism may have lasted in total for longer than 11 million years (30-19 Ma) with a significant amount of the volcanic pile having been removed by erosion. Age determinations on dyke rocks (Huchon et al. 1992) indicate a possible major structural change around 22 Ma when a dominant E - W extensional stress regime was replaced by a N - S extensional system. This structural change appears to be synchronous with the major period of granite emplacement and also marks the onset of syn-volcanic extension. Four important observations can be made by considering the sedimentological, structural and volcanological evolution of the region. Firstly, the palaeoenvironmental record within the

Mesozoic to Tertiary sediments indicates n o marked (> 100 m) n0n-eustatic sea-level change as one might expect with significant pre-volcanic uplift. Secondly, the sedimentary rocks record no marked erosional periods that would be caused by erosional unroofing during doming or uplift. A disconformity occurs between the Tawilah Formation and the flood volcanics. Thirdly, the lack of pre-volcanic crustal extension and the presence of late syn-volcanic and post-volcanic structure indicates that most crustal extension postdated the onset of magmatism at 30 Ma. Fourthly, erosional unroofing after volcanism may have removed a significant amount of the flood volcanics. One can deduce from these geological observations that significant uplift did not occur prior to 30 Ma and extension began to affect the region at ,~ 22 Ma, some 8 million years after the onset of peak magmatism and at a time of major structural change. Alternatively, if uplift did happen synchronously with the onset of volcanism at 30 Ma it had little or no affect on the geological record. If one accepts that the Pan-African lithosphere beneath Yemen had a pre-rift thickness of c. 180 km (McGuire & Bohannon 1989) then it follows that > 10 Ma may be required, from the time of plume impingement, before uplift is registered in such thick lithosphere (Spohn & Schubert 1983). This would require that the plume had been under the region for several million years prior to the onset of uplift.

Preliminary fission track data Fission track (VI') research in Yemen is facilitated by three major advantages: (a) the geodynamic position of the rift margin, close to the centre of the Afar thermal anomaly and triplejunction (Fig. 1), should ensure a maximum crustal response to the thermo-tectonic processes of rifting; (b) excellent exposure and completeness of section, encompassing Proterozoic basement (c. 900 Ma) to Quaternary extrusives, allow detailed stratigraphic relationships to be resolved; and (c) a rigorous geological framework which can be integrated with quantitative FT estimates of exhumation and shallow crustal cooling. Since FT dating of apatites indicates the age at which the rock cooled below 120-125°C care was taken to sample basement rocks at some distance from dyke swarms and other intrusives which may have reset the apatite FT ages. Dixon et al. (1989)pointed out that FT ages may record the age of local magmatic pulses rather than the beginning of exhumation. In a

2080

1710

1380

970

Elevation m apatite 20 apatite 6 apatite 9 apatite 9

0.550 (202) 0.171 (18) 0.154 (81) 2.736 (360)

Spontaneous ~ (Ns) 8.610 (3160) 2.556 (269) 2.216 (1163) 1.733 (228)

Induced pi (Ni)

11%

75%

7%

30%

PX2 1.309 (9068) 1.309 (9068) 1.309 (9068) 1.309 (9068)

Dosimeter pd (Nd)

379+_41

17+_2

16_+4

16+1

FT Central age Ma (_+lor)

Track densities (p) are as measured and are (x 106 tr cm-2); numbers of tracks counted (N) shown in brackets. Analyses by external detector method using 0.5 for the 4~2¢r geometry correction factor. See Hurford & Carter (1991). Ages calculated using dosimeter glass CN-5 for apatite with ~c~5 = 374+9. PX2 is probability for obtaining X2 value for v degrees of freedom, where v = no. crystals- 1.

Yem969 F4 Yem 970 F12 Yem971 F13 Yem973 F28

Sample No. Field No.

Mineral and no. crystals

12.83_+0.12 (68)

--

13.85+0.34 (36) D

Apatite mean track length 0zm)

1.00

m

2.01

Length standard deviation (/tin)

Table 1. Fission track ages and length data for apatites from Pan-A~can basement rocks of Yemen. Samples 969, 970 and 971 are amphibolites and 973 a gneiss

298

M.A. MENZIES E T A L .

rift environment this cooling can be brought about by (a) surface uplift and subsequent erosional exhumation, (b) exhumation without surface uplift due to the competing effects of isostatic compensation and erosion, or (c) crustal thinning and unroofing due to extensional block faulting. FF dates were determined on samples of Pan-African basement from Yemen (Table 1). Apatite FT ages of c. 16 Ma with long mean track lengths indicate rapid crustal cooling and exhumation of the proto-Red Sea rift-flanks. This period of cooling and exhumation occurred approximately 14 million years after the onset of significant flood volcanism, presumed to have begun around 30 Ma (Fig. 2). In Yemen the large thickness (> 3 km) of erupted volcanics (30-20 Ma) and the associated high geothermal gradient will have annealed all apatites in the Pan-African basement such that any pre-volcanic uplift will not have been recorded in the FT ages. Approximately 3-4 km of erosion at < 20 Ma can be demonstrated using FT data. This may have important implications for the possible removal of younger volcanics. The presence of a sample with a partially reset, apparent apatite FT age of c. 380 Ma, outside the extended area, indicates slower exhumation from shallower crustal levels. There is a general increase of sample FT age with elevation, and the base of the uplifted partial annealing zone (once at c. 3000-4000 m depth) in north Yemen is currently believed to be located at 900-1700 m elevation. It is possible that Fir' ages in Yemen may record the cessation of volcanism and subsequent erosion coupled with the development of late synvolcanic and post-volcanic extensional structures rather than post-volcanic surface uplift. However, geological evidence points to surface uplift initiating exhumation. Marine sediments near the base of the flood volcanics are found at a regional elevation of 2400 m which attests to this amount of surface uplift at some point during the past 28 Ma. Theoretically, eruption of 3-4 km of flood volcanics in Yemen would produce an increase in elevation of approximately 600 m (assuming Airy isostasy), which is much less than the average 2000 m elevation seen today in the Yemen plateau where there is no observable extension. In fact a thickness of 11 km of magmatic underplating would be required to produce this observed average surface elevation. This is theoretically possible (McKenzie & Bickle 1988) if we assume that the potential temperature in the mantle beneath this extended region (for fl = 2) is elevated relative t o normal asthenosphere. Such a scenario would generate 10-15 km of melt of which c. 5 km was erupted. Studies

of xenoliths from Saudi Arabia (McGuire 1988) point to relatively hot shallow mantle (1020°C at 36 km). Garnet pyroxenite xenoliths which crystallized over the depth range 40-50 km at 9001000°(2 (McGuire & Bohannon 1989) may have formed as a result of magmatic underplating near the crust-mantle boundary. However recent gravity modelling indicates that the crust underneath the Yemen rift mountains is 35 km thick or less (Makris et al. 1991). The gravity data therefore suggests that magmatic underplating (with crustal densities) is not the reason for the present uplift and that it maybe a transient phenomenon due to thermal expansion above the plume. This is supported by the presence of numerous hot springs and young volcanic cones throughout the Yemen highlands. Moreover evidence exists in Saudi Arabia for anomalously high temperatures at the base of the crust (McGuire & Bohannon 1989) almost twice as hot as would be expected from surface heat flow data. The role of underplating and/or thermal expansion close to the Afar plume need to be resolved if the cause of uplift is to be fully understood.

Sinai, eastern Egypt and Saudi Arabia One of the most contentious issues in the Red Sea is the relative timing of uplift, extension and magmatism. Gass (1970a, b) drew attention to the temporal and spatial coincidence of volcanism and surface uplift associated with the formation of the Afro-Arabian dome. He proposed that the causal mechanism was localized thermal disturbances in the mantle (i.e plumes), an idea that has recently gained wide acceptance (e.g. White & McKenize 1989; Fig. 1). For some time it has been accepted that uplift and formation of broad domal structures predated magmatism (Gass 1970a, b; Kohn & Eyal 1981) but more recently this has been questioned (Almond 1986; Bohannon et al. 1989). The presence of distinct domes and resultant differential uplift is supported by fission track studies for the Sinai Peninsula and the southeastern desert of Egypt (Kohn & Eyal 1981; Omar et al. 1987; Garfunkel 1988). It is apparent that uplift at c. 27 Ma predates rifting and the 20 Ma peak of magmatism (23-17 Ma) in the Sinai Peninsula (Baldridge et al. 1991; Fig. 2). Kohn & Eyal (1981) estimate up to 3 km of erosion to have occurred since c. 9 Ma and Omar et al. (1987) proposed that distinct domes existed in the southeastern desert of Egypt, and that these produced variable uplift induced erosion that occurred some 5 million years prior to magmatism. A detailed study of the basement of eastern

MAGMATISM, UPLIFT AND EXTENSION

Egypt (Omar e t al. 1989) concluded, however, that the length distributions of fission tracks were vital in understanding the uplift age. Of the three distinct groupings only one group characterized by unimodal, narrow negatively skewed track length distributions and long mean lengths gave the best 'cooling ages'. These authors

299

concluded that rift-flank uplift began around 21-23 Ma and that is was contemporaneous with extension and subsidence. In an investigation of the Pan-African basement in Saudi Arabia, Bohannon e t al. (1989) concluded that doming cannot have occurred at any time between the late Cretaceous to early

AGE OF INITIAL EXTENSION, IGNEOUS ACTIVITY AND EXHUMATION AROUND THE RED SEA NORTH CENTRAL Sinai

so

~

tit:[]

-

JO

'r~ "~

Z

28

I

40

TIME

WEST 28

I

I

30

I

20 (Ma)

10

EAST I

I

rmD O~ UJ

• TmT ii mTlTr

~1 -24

24

< "O

i"

¢.,,

IITITITffl ill

ill

a

ittttt

-20

i_~o3 20

nTIT~ 16

: : ? : : : : ' n r m l lit

-

13

-

16

Q.

.o_

w

,g=

m

IIIIIIIIIIIIIIIIIIIIIIII

IIII I I I I I I I I I I I I I

I

I

I

40

30

20

10

50

I

I

I

40

30

20

12

10

TIME (Ma)

TIME (Ma) LEGEND APPROXIMATE BEGINNING OF EXTENSION

/

I

AGEOF

RIFT-RELATED ITITITITI IGNEOUS ACTIVITY

BEGINNING

OF UPLIFT?/EXHUMATION

/

Fig. 2. Relative timing of initial extension, igneous activity and exhumation from north to south along the margins of the Red Sea (after Dixon et al. 1989). Note that volcanism in Yemen and Saudi Arabia predates extension and exhumation. Data are taken from several sources (Sinai, Omar et al. 1989 and Baldridge et al. 1991; Saudi Arabia, Bohannon et al. 1989; Yemen, Civetta et al. 1978 and Chiesa et al. 1989; Ethiopia, Hart et al. 1989 and Mohr & Zanettin 1988; eastern Egypt, Omar et al. 1987 and Ressetar et al. 1981).

300

M.A. MENZIES E T A L .

Oligocene in the central and northern Red Sea because of complete sections of Upper Cretaceous-Eocene (c. 73-45 Ma) marine rocks in Egypt and Saudi Arabia (Fig. 1). Bohannon et al. (1989) thought that the axial part of a dome cannot have been uplifted at any time between late Cretaceous to early Oligocene times in the northern and central Red Sea. Similarly, the existence of thin continuous marine sediments, coastal zone non-marine rocks and thick lateritic soils of late Cretaceous to mid-Oligocene age over a large part of Arabia argue strongly against late Cretaceous to early Tertiary domal structures. This is because these sediments indicate that the entire Afro-Arabian continent was at a low elevation or below sea level for 45 million years prior to Red Sea rifting. It is important to note that all the surrounding marine rocks are fine grained so presumably there was little difference in elevation between the areas of marine deposition and of soil development. Consequently there is little evidence for Cretaceous doming in the northern and central Red Sea but late Oligocene to early Miocene uplift appears to have been active in the northern Red Sea (Egypt and Sinai). Geological observations and a detailed fission track investigation of the Pan-African basement in Saudi Arabia (Almond 1986; Bohannon 1986; Bohannon et al. 1989; Camp & Roobol 1989; Camp et al. 1991) have provided evidence that exhumation presumably related to doming or uplift, postdates magmatism. In contrast, Almond (1986) provided evidence that early extension was related to subsidence not uplift and doming and that the uplift which eventually produced the Afro-Arabian dome occurred around 10 Ma. However, it should be noted that all of the flood volcanics studied in Yemen are subaerial with no known submarine eruptives. This would argue against significant subsidence during their formation (30-20 Ma). Most of the faulting that formed the Red Sea rift occurred during late Oligocene (Bohannon 1986) with a peak of crustal extension around 25 Ma. In a later study Bohannon et al. (1989) stressed the lack of evidence for pre-volcanic rifting or crustal extension. This is similar to observations in Yemen. Early Oligocene volcanic rocks in Saudi Arabia conformably overlie sedimentary rocks deposited in marine and coastal zone environments and the oldest angular unconformity is beneath 15-18 Ma old flows. These geological observations and FT results (Bohannon et al. 1989) indicate 2.5-4 km of uplift in early to middle Miocene times (Fig. 2). The geographical distribution of FT ages across the western Saudi

Arabian escarpment (Bohannon et al. 1989) show trends in common with other rift-flanks throughout the world. The youngest FT ages generally occur at the lowest elevations along the base of the escarpment and the older ages occur along, and to the east, of the escarpment crest. Overall the FT data for Saudi Arabia tentatively suggest exhumation marginal to the central Red Sea, beginning at 20 Ma and accelerating at < 14 Ma. A significant phase of erosion postdates rifting and magmatism by 1015 million years. On the basis of these data Bohannon et al. (1989) invoked a passive rifting model for the Red Sea, contrasting with the active, doming (early uplift) models proposed elsewhere (Gass 1970a, b). Several aspects of the geological and FT data have important implications for the timing of uplift, magmatism and extension. Firstly, uplift cannot significantly predate magmatism (c. 30 Ma) due to the lack of any evidence in the sedimentary record for major changes in sea level. From geological observations one can constrain the beginning of uplift to be around early to mid-Miocene in Sinai, eastern Egypt and Saudi Arabia. The lack of pre-volcanic structure cannot be used as an indicator of the lack of uplift as uplift may occur without significant extension. Secondly, some 2.5-4.0 km of exhumation marginal to the central Red Sea postdated rifting in Saudi Arabia and Egypt by 5-10 Ma (Fig. 1). In Sinai recent data indicate that uplift, rifting and magmatism were broadly synchronous (Fig. 2). Thirdly, plume impingement on a moving plate may result in a systematic increase in the age of volcanism, and possibly uplift, away from the plume head such that regions in the northern Red Sea may have been uplifted earlier than those in the southern Red Sea as the latter are within the present-day 'sphere of influence' of the Afar plume. In Sinai, eastern Egypt and the northern Red Sea, uplift is believed to have started around 25-20 Ma whereas in Saudi Arabia and Yemen, in the southern Red Sea, exhumation appears to have started around 2015 Ma (Fig. 2). With regard to systematic changes in the age of the volcanic rocks Dixon et al. (1989) reported temporal changes that are the opposite of that produced by a stationary plume and a 'northward'-moving plate. However, a recent evaluation of all available age data does not support any systematic regional variations in volcanism around the margins of the Red Sea (Menzies et al. 1990). A detailed investigation of the timing of initiation of volcanism and the temporal and spatial evolution of the flood volcanism in Yemen is underway at present,

MAGMATISM, UPLIFT AND EXTENSION

Discussion In an overview of the Red Sea, Dixon et al. (1989) stated that the timing of magmatism, uplift and extension support neither a purely active or passive rifting model but that the early volcanism implied a causal association between upwelling mantle and rift initiation. In contrast, White & McKenzie (1989) suggested that the Afar plume must have existed under the southern Red Sea at 30 Ma coincident with the initiation of a period of major flood volcanism. Implicit in this model is c. 1-2 km of surface uplift synchronous with, or just prior to, magmatism. White & McKenzie (1989) stress that such uplift is spread over an area of 1000-2000 km, with a maximum directly above the plume head. Although the dynamic uplift associated with initial plume impingement may effectively cause instantaneous uplift, several tens of millions of years may be required before the conductive thermal-uplift effects are registered in 150 km thick lithosphere (Spohn & Schubert 1983). Following the work of Houseman (1990), the proximity of Yemen towards the centre of the proposed plume, suggests that we might expect significant amounts of dynamic uplift prior to magmatism. While the White & McKenzie (1989) model requires pre-volcanic uplift, most authors report no widespread pre-volcanic uplift or structure in Yemen or Saudi Arabia with only one report of pre-volcanic faulting (Hempton 1987). Most of the lowermost volcanic units in Saudi Arabia, Yemen and Ethiopia are erupted disconformably onto palaeosol horizons or fluviatile sediments without an intervening angular unconformity. Some of the lower volcanic units conformably overlie gastropod-rich horizons or fluviatile sediments. Recently, passive and active models for the evolution of the Red Sea and elsewhere have come under scrutiny. Pallister (1987) and Dixon et al. (1989) invoked active rifting models mainly because magmatism predated extension and uplift. In contrast, Bohannon et al. (1989) invoked a passive rifting model for the Red Sea on the basis of FT data and Menzies et al. (1990) supported such a model primarily on the presence of synsedimentary structures in the Tawilah sandstones underlying the Yemen volcanics. This was interpreted as evidence for pre-volcanic crustal extension. A recent investigation of these features revealed that they are deeply weathered igneous dykes and consequently there now appears to be no unequivocal evidence f o r prevolcanic structure in the sediments underlying the volcanic rocks in Yemen. Moreover crustal

extension in Yemen has primarily been accommodated by domino-style block faulting rather

301

than dilation due to dyke intrusion. As such, passive and active models do not adequately explain the Yemen data and Red Sea rifting. The Great Basin of the western USA offers an interesting comparison with Yemen. The Great Basin is also a region of Oligo-Miocene basaltrhyolite magmatism in a region of crustal extension and uplift. In the Great Basin, Taylor et al. (1989) demonstrated the existence of pre- (> 32 Ma), syn- (30-27 Ma) and post-volcanic (16-14 Ma; < 5.3 Ma) structure but stressed that only period of faulting was synchronous with magmatism. Therefore magmatism and faulting need not be closely related in space and time. Although support can be found for a passive rifting model (i.e. extension began prior to volcanism) in the Great Basin, the genetic relationship between volcanism and extension is not simple and direct. This is similar to the southern Red Sea where most of the structure tends to be late synvolcanic or post-volcanic and the eentres of volcanism do not always coincide with extended areas. In both the Red Sea and the western USA evidence exists for only local extension prior to the main episode of volcanism. Little or no evidence exists for significant regional extension during the peak of volcanism when the greatest volume of magma was erupted. Extension after peak volcanism is apparent both in the western USA (Best & Christensen 1991) and the Red Sea where faults cut the entire volcanic sequence and older rocks. Best & Christensen (1991) coneluded that regional extension did not occur in the Basin and Range and that it was episodic. Basal angular unconformities are not widespread in the Great Basin (Best & Christensen 1991) or the Red Sea and faulted angular discordances are limited in distribution. Bohannon et al. (1989) reported angular unconformities under 18 Ma flows in Saudi Arabia indicating that crustal extension had begun by that time. Angular unconformities have only been reported in Yemen at Jabal an Nar and Jabal Khariz. At Jabal an Nar, late Miocene (10 Ma) basalts unconformably rest on early Miocene silicified rhyolites (18 Ma) (Capaldi et al. 1987; Huchon et al. 1992). This constrains extension to have occurred between 20 and 10 Ma. A similar angular unconformity is apparently located at Jabal Khariz on the southern coast of Yemen west of Aden. Here late Miocene volcanics (9.6 Ma) rest unconformably on block faulted Yemen Volcanics of inferred early Miocene age (Cox et al. 1969). It is interesting to note that the FI" data indicate that the initiation of exhumation (17 Ma) coincided with the age of the angular unconformities in Yemen (20-10 Ma) and Saudi Arabia (18 Ma). This may point to a similar evolutionary history for both regions. Other

302

M.A. MENZIES E T A L .

evidence for syn-volcanic crustal extension in the form of intervolcanic sedimentary deposits are volumetrically rather limited. In the case of the western USA they amount to 1% of the cumulative thickness of Tertiary. sediments (Best & Christensen 1991). In the case of Yemen no accurate assessment of the amount of sedimentary material has been undertaken but preliminary results of traverse work indicate that in parts of northern Yemen there is an east (10%) to west ( < 1%) variation in the amount of sediments. Best & Christensen (1991) concluded that simple models of active and passive rifting (Sengor & Burke 1978) are inappropriate when applied to complexly evolving terrains like the Great Basin of the western USA. This equally well applies to Yemen and Saudi Arabia and elsewhere (Brown et al. 1991) where magmatism frequently pre-dates crustal extension and uplift/ exhumation, a sequence that is neither passive nor active. Perhaps the reason no uplift occurred before or during early magmatism is that magmas were efficiently transported to the surface via narrow conduits in which magma velocity greatly exceeded conductive heat transfer. The crust would take ten million or more years to respond to the heat perturbation caused by the arrival of the Afar plume under the lithospheric plate (Spohn & Schubert 1983). This is consistent with a gradual change in the lithological character of the Yemen Volcanics. The early volcanism was predominantly mafic perhaps the result of efficient magma transfer with magma velocities exceeding heat transfer into the lithosphere. In contrast, later volcanism was more silicic (granites and rhyolitic ignimbrites) indicating storage and differentiation of mafic magmas in crustal magma chambers thus enhancing conductive heat transfer into the lithosphere.

Summary The relationship between magmatism, crustal extension and uplift in continental rifts can only be properly evaluated with an integration of geological field observations, age determinations and fission track analysis. In Yemen, adjacent to the southern Red Sea, geological and preliminary fission track data indicates that the onset of flood volcanism (c. 3-4 km) predates significant crustal extension and uplift/exhumation. This is an area that is frequently cited as a classic example of the opposite phenomenon where plumedriven uplift precedes magmatism. It is becoming increasingly apparent in the southern Red Sea, the Basin and Range of the western USA, southeastern Africa and elsewhere that the development of volcanic and non-volcanic

margins cannot be adequately explained by traditional active and passive models. We suggest that a plume was responsible for flood volcanism, but the earliest expression of continental break-up was magmatism rather than domal uplift. This work was funded by an expedition grant from the Royal Society which made fieldwork possible. Additional support from British Petroleum and the Industrial Association of the department of Geology RHBNC is gratefully acknowledged. The British Council is thanked for supporting postgraduate field studies in Yemen (J.B., M.A. and A.A.). The University of Sana'a is thanked for its continued hospitality and for provision of vehicular support. S. Muir is thanked for the diagrams. R. G. Bohannon and N. Harris are thanked for their comments on an earlier version of this manuscript.

References ALMOND, D. C. 1986. Geological evolution of the Afro-Arabian dome. Tectonophysics, 131, 301-332. BALDRIDGE,W. S., EYAL, Y., BARTOR,Y., STEI~ITZ, G. & EYAL, M. 1991. Miocene magmatism of Sinai related to the opening of the Red Sea. Tectonophysics, 197, 181-201. BEST, M. & CHmSTENSEN,E. H. 1991. Limited extension during peak Tertiary volcanism, Great Basin of Nevada and Utah. Journal of Geophysical Research, 96, 13509-13528. BOHANNON,R. G. 1986. Tectonic configuration of the western Arabian continental margin, southern Red Sea. Tectonics, 5, 477-499. ~, NAESER,C. W., SCHMIDT,D. L. & ZIMMERMAN, R. A. 1989. The timing of uplift, volcanism and rifling peripheral to the Red Sea: A case for passive firing? Journal of Geophysical Research, 94, 1683-1701. BOSENCE, D. W. J., DAVISON, I., MENZIES, M. A., MCCLAY, K. R. & NICHOLS, G. 1992. Lithospheric extension, sedimentation and magmatism in the southern Red Sea. lnternal report on Field Expedition to Yemen, RHBNC, Department of Geology. BROWN, R. W., GLEADOW,A. J. W. & SUMMERFIELD, M. A. 1991. The tectonic and geomorphic evolution of the continental margins of southern Africa: evidence from apatite fission track analysis. Geological Society programme and abstracts. Magmatism and the causes of continental breakup. October 1991. CAMP,V. E. & ROOBOL,M. J. 1989. The Arabian continental alkaline basalt province. Part I Evolution of Harrat Rahat, Kingdom of Saudi Arabia. Geological Society of America Bulletin, 101, 71-95. - - & HOOPER, P. R. 1991. The Arabian continental alkaline basalt province. Part II Evolution of Harrats Khaybar, Ithnayn and Kura,

MAGMATISM, UPLIFT AND EXTENSION Kingdom of Saudi Arabia. Geological Society of America Bulletin, 103, 363-391. CAPALDI, G., MANETrt, P., PICCARDO,G. B. & PuN, G. 1987. Nature and geodynamic significance of the Miocene dyke swarm in the North Yemen.

Neues Jahrbuch fl~r Mineralogie Abhandlungen, 156, 207-229. CHIESA, S., LA VOLPE,L., LIRER, L. & ORSI, F. 1983. Geological and structural outline of Yemen Plateau: Yemen Arab Republic. Neues Jahrbuch far Geologic und Paliiontologie Monatshefle, 11, 641-656. , CIVE'VrA,L., DEFINO, M., LA VOLPE,L. & ORSI, G. 1989. The Yemen Trap Series: Genesis and evolution of a continental flood basalt province.

Journal of Volcanology and Geothermal Research, 36, 337-350. CIVE'I'TA, L., LA VOLPE, L. & LIRER, L. 1978. K-Ar ages of the Yemen Plateau. Journal of Volcanology and Geothermal Research, 4, 307-314. CLOOS, H.

1939. Hebung-Spaltung-Vulkanismus.

Geologische Rundschau, 30, 405-527. Cox K. G., GASS, I. G. & MALLICK,D. I. J. 1969. The evolution of the volcanoes of Aden and little Aden, south Arabia. Journal of the Geological Society, London, 124, 283-308. DIXON, T. H., IVINS, E. R. & FRANKLIN,B. J. 1989. Topographic and volcanic asymmetry around the Red Sea: constraints on rift models. Tectonics, 8, 1193-1216. FARNETANI, C. & RICHARDS,M. A. 1991. Numerical Modelling of flood basalt events: timing, uplift, melting and mantle structure. Transactionsof the American Geophysical Union, 72, 579. GARFUNKEL, Z. 1988.-Relation between continental rifting and uplifting: evidence from the Suez Rift and northern Red Sea. Tectonophysics, 150, 33-49. GASS, 1. G. 1970a. Evolution of volcanism in the junction area of the Red Sea, Gulf of Aden and Ethiopian rifts. Philosophical Transactions of the Royal Society of London, A267, 369-381. 1970b. Tectonic and magmatic evolution of the Afro-Arabian dome. In: CLWFORD, T. N. & GASS, I. G. (eds) African Magmatism and Tectonics. Oliver and Boyd, 285-300. HART, W. K., WOLDEGABRIEL,G., WALTER, R. C. & MERTZMAN, S. A. 1989. Basaltic volcanism in Ethiopia: constraints on continental rifting and mantle interactions. Journal of Geophysical Research, 94, 7749-7755. HEM~rON, M. R. 1987. Constraints on Arabian Plate Motion and extensional history of the Red Sea. Tectonics, 6, 687-705. HOUSEMAN, G. A. 1990. The thermal structure of mantle plumes: axisymmetric or triple junction? Geophysical Journal International, 102, 15-24. HUCHON, P., JESTIN, F., CANTAGREL,J. M., AL KHIRBASH, S. & GAFENEH, A. 1992. Extensional deformations in Yemen since Oligocene and the Africa-Arabia-Somalia triple junction. Annales Tectonicae (in press). HURFORD, A. J. & CARTER, A. 1991. The role of fission track dating in discrimination of prov-

303

enance. In: MORTON, A. C., TODD, S. P. & HAUGHTON, P. D. W. (eds) Developments in Sedimentary Provenance Studies. Geological Society, London, Special Publication, 57, 67-78. KOHN, B. P. & EYAL,M. 1981. History of uplift of the crystalline basement of Sinai and its relations to opening of the Red Sea as revealed by fission track dating of apatites. Earth and Planetary Science Letters, 52, 129-141. MAKRIS,J., HENKE,C. H., EGLOFF,F. & AKAMALUK, T. 1991. The gravity field of the Red Sea and east Africa. Tectonophysics, 198, 369-382. MCCLAY, K., AL-KADASI, M., AL-SUBBARY, A., BAKER,J., BOSENCE,D., DART, C., DAVISON,I., MENZIES,i . & NICHOLS,G. 1991. Extension and magmatism, eastern margin of the Red Sea rift, Yemen. Tectonics Studies Group (abs.). McGuIRE, A. V. 1988. Petrology of mantle xenoliths from Harrat al Kishb: The mantle beneath western Saudi Arabia. Journal of Petrology 29, 73-92. & BOHANNON,R. G. 1989. Timing of Mantle Upwelling: Evidence for a passive origin for the Red Sea Rift. Journal of Geophysical Research, 94, 1677-1682. MCKENZIE, D. & BICKLE, M. 1988. The volume and composition of melt generated by extension of the lithosphere. Journal of Petrology, 29, 625-679. MENZIES, i . A., BOSENCE, D., EL-NAKHAL,H. A., AL-KHIRBASH, A., AL-KADASI, i . & AL SUBBARY, A. 1990. Lithospheric extension and the opening of the Red Sea: basalt-sediment relationships in Yemen. Terra Nova, 2, 340-350. MOHR, P. & ZANETnN, B. 1988. The Ethiopian Flood Basalt Province. In: McDOUGALL, J. D. (ed.) Continental Flood Basalts. Kluwer Academic Publishers, Holland, 63-110. OMAR, G. I., KOHN, B. P., LUTZ, T. M. & FAUL, H. 1987. The cooling history of Silurian to Cretaceous alkaline ring complexes south eastern deseft, Egypt, as revealed by fission track analysis. Earth and Planetary Science Letters, 83, 94-108. ~, STECKLER,M. S., BUCK,W. R. & KOHN, B. P. 1989. Fission track analysis of basement apatites at the western margin of the Gulf of Suez rift, Egypt: evidence for synchroneity of uplift and subsidence. Earth and Planetary Science Letters, 94, 316-328. PALLISTER, J. S. 1987. Magmatic history of Red Sea rifting: perspective from the central Saudi Arabia coastal plain. Geological Society of America Bulletin, 98, 400-417. RESSETAR, R., NAIRN, A. E. M. & MONRAD, J. R. 1981. TWO phases of Cretaceous-Tertiary magmatism in the eastern desert of Egypt: paleomagnetic, chemical and K-Ar evidence. Tectonophysics, 73, 169-193. TAYLOR, W. J., BARTLEY,J. M., LUx, D. R. & AXEN, G. J. 1989. Timing of Tertiary extension in the Railroad VaUe),-Pioche Transect Nevada: Constraints from'~"Ar/39Ar ages of volcanic rocks. Journal of Geophysical Research, 94, 7757-7774. ~ENGOR, A. i . C. & BURKE,K. 1978. Relative timing

304

M. A. MENZIES ET AL.

of rifting and volcanism and its tectonic implications. Geophysical Research Letters, $, 419-421. SPOHN,T. & Scm~nERT, G. 1983. Convective thinning of the lithosphere: A mechanism for rifling and mid-late volcanism on earth. Tectonophysics, 94, 67-90. AL-SUBBARY,A. & NxcHou.s,. G. N. 1991. Creta-

ceous.-carly Tertiary pre-rifi sediments Yemen. British Sedimentological Research Group Meeting, Edinburgh (abs.). WHITE, R. & MCKENZIE, D. 1989. Magmatism at rift zones: the generation of volcanic continental margins and flood basalts. Journal of Geophysical Research, 94, 7685-7729.

Tectonism and magmatism during NE Atlantic continental break-up: the Vering Margin J A K O B S K O G S E I D 1, T O M P E D E R S E N 1, O L A V E L D H O L M 1 & B J O R N T. L A R S E N 2

1Department o f Geology, University o f Oslo, PO Box 1047 Blindern, N-0316 Oslo, Norway 2Statoil, Forushagen, PO Box 300, N-4001 Stavanger, Norway

Abstract: The temporal and spatial relationships of tectonic and magmatic features on the Vcring volcanic margin show that continental break-up occurred in association with significant magmatic activity about 18 Ma after initiation of lithospheric extension. From the distribution of extension across the margin and the volumes of melt produced, a thermal anomaly of 50-80°C is estimated, in agreement with predictions from recent plume models. A tectono-magmatic model is proposed in which the ascending proto-Iceland plume released the rifting, over a > 300 km wide zone, by uplift-induced extension of the NE Atlantic lithosphere already affected by tensional stresses. Initial rifting took place without decompressional melting when the plume, carrying the thermal anomaly, was still 600-700 km beneath the lithosphere. Subsequently, widespread magmatism occurred when the plume impinged on, spread out beneath and infilled the rift-defined relief at the base of the lithosphere. Break-up is suggested to be a consequence of melt-induced weakening of the lithosphere, whereas the anomalously thick igneous crust at the continent-ocean transition along the NE Atlantic margins is explained by melt focusing towards the uplifted break-up axis.

Recently, the formation of flood basalt provinces, oceanic plateaux and volcanic rifted margins have been linked to mantle plumes impinging at the base of the lithosphere (Richards et al. 1989; White & McKenzie 1989; Griffiths & Campbell 1990; Loper 1991; Coffin & Eldholm this volume). To understand the formation of volcanic rifted margins it is, however, necessary to resolve the relationships between rifting, magmatism and break-up in space and time. Most evolutionary models are based on seismic data outlining the lateral extent and geometry of igneous complexes, and on petrological/geochemical data from a limited number of drill holes. With volcanic margins, however, in contrast to non-volcanic rifted margins, there is relatively little observational data available to study the lateral extent, mode and amount of lithospheric deformation leading to margin formation, and the relative timing of tectonism and magmatism. The lack of structural and stratigraphic data is primarily due to the extrusion of flood basalts over large areas of the rift zone during break-up. On the V~ring volcanic margin off mid-Norway (Fig. 1), rifting occurred within, or at the flank of, an intracratonic sedimentary basin. There is a discernible record of both the intrus-

ive and extrusive igneous activity during rifting, break-up and initial seafloor spreading which allows us to resolve the tectonic-magmatic development. The aim of this paper is to use this information to study the Late CretaceousPalaeocene rift episode that led to continental break-up between Europe and Greenland near the Palaeocene-Eocene transition. We address the width of the rift zone by incorporating seismic data from the conjugate margin off NE Greenland, and discuss the character and origin of the asthenospheric thermal anomaly associated with the observed melt volumes. The work is based on a regional grid of highquality multichannel seismic (MCS) profiles on the V~ring margin, as well as results from the Deep Sea Drilling Project (DSDP), the Ocean Drilling Program (ODP) and commercial wells (Fig. 2). On the conjugate NE Greenland margin only a few seismic profiles exist due to data acquisition problems on the perennially ice-covered shelf, and the tectonic development of this margin is less well understood. Crustal velocitydepth profiles derived from seismic refraction data (Mutter et al. 1984, 1988; I-Iinz et al. 1987; Planke et al. 1991) are used for depth conversion and to establish the crustal configuration of the

margins.

From STOREY,B. C., ALASAS'mR, T. & PAmmuPar, R. J. (eds), 1992, Magmatismand the Causes of ContinentalBreak-up, Geological Society Special Publication No. 68, pp. 305-320.

305

306

MANTLE PLUMES AND NE ATLANTIC RIFTING

~

So. 2°"

/

I':

.... i I,I¸

~ii:i¸ ~

~:/

o"

":':-

|

\4

i [~!!ii!

l

10"

O"

i

:i:

"~0"

Fig. 1. North Atlantic Volcanic Province in a reconstruction to anomaly 23 time, showing seaward dipping reflectorwedges, onshore early Tertiary igneousrocks (black) and areas covered by flowsand sill/flowcomplexes (dark-shaded). Note that areas affected by intrusives, e.g. sills and low-angledykes, exist also landward of the sill/flow complex along the eastern margin. The location of Iceland hotspot (White & Mckenzie 1989)(double circle) and the conjugate profilesacrossthe VOtingand NE Greenland margins(Fig. 10) (solidline) are shown. GB: Great Britain; VM: V~ring margin.

Geological setting The Vcring margin comprises three main geological provinces: The Tr0ndelag Platform, the V0ring Basin and the VOting marginal high (Figs 3 A & 4). The Trendelag Platform, an area up to 160 km wide between the Norwegian mainland and the V0ring Basin, was relatively stable during late Mesozoic times. The V0ring Basin is a large sedimentary basin that was formed during and subsequent to a major Late Jurassic-Early Cretaceous extensional episode. The Fulla Ridge separates deep Cretaceous depocentres (e.g. the R~s Subbasin) adjacent to the Trendelag Platform from the Fenris and Hel graben in the western basin. It is also possible to define structures at deeper stratigraphical levels (Planke et al. 1991; Skogseid et al. 1992), although the stratigraphical control is relatively poor. The western basin is dominated by the Cenozoic subsidence of the

continental margin, resulting from Late Cretaceous-Palaeocene extension and subsequent break-up (Skogseid & Eldholm 1989). The breakup took place within chron 24R, and is dated to 57.5 Ma by Eldholm et al. (1989a) using the Berggren et al. (1985) timescale. A late pulse of continental uplift and increased erosion resulted in rapid outbuilding of the shelf leaving a huge Pliocene wedge in the eastern V0ring Basin. The V0ring Basin is separated from the V0ring marginal high by the V0ring Escarpment (Fig. 4). The high was formed during the early Cenozoic break-up by massive emplacement o f flood basalts. The volcanic event also left a 1040 km wide sill/flow complex east of the V0ring Escarpment which inhibits seismic resolution to deeper levels. On the inner high, there are sequences of relatively flat-lying lavas above a continental block, which is presumably strongly intruded (Skogseid & Eldholm 1989; Zehnder et al. 1990), whereas the outer high is characterized

J. SKOGSEID E T A L . 69°---

2

e

a

1~

307

gional rift unconformity (Rift Unc.; Fig. 5). Despite its diachronous character, the unconformity documents stratigraphical relations that allow us to date the tectonic and magmatic events. The eastern faults are important for dating the initiation of the tectonic activity. Well 6607/5-1, off the northeastern flank of the Fulla Ridge (Fig. 3A), is located in an analogue setting to that shown on Fig. 6, and additional seismic 67 ties may be made to other commercial wells on the continental shelf. In fact, the rift unconformity lies just below the base Tertiary reflector, whereas the initiation of movement on the in66 dividual faults again slightly pre-dates the unconformity. The base Tertiary reflector itself marks a regional hiatus in the NE Atlantic realm, and many wells show absence of Late MaastrichI t 2 4 6 8 tian to early Palaeocene sediments with renewed Commercial drill site • DSDP/OOP drill siW deposition in late Palaeocene, about 63-62 Ma ,,,,,,,,,.,,., ConUnentOcean l~uncU~ 0 1 ~ Km (Fjaerran et al. 1990). This implies that the initiation of faulting pre-dates mid-Maasttichtian (c. Fig. 2. Multichannel seismic data used in this study. 70 Ma), whereas there is seismic and drillhole Commercial exploration area with high data coverage evidence that it post-dates mid-Campanian is marked in lower right comer. Identification of times (c. 80 Ma; Fig. 6). Thus, we tentatively structural lineaments in Fig. 3A. suggest onset of faulting at the CampanianMaastrictian transition, 75 Ma. The exact duraby thick units of seaward dipping volcanic layers tion and extension rate of the faulting activity cannot be resolved from the existing data. gradually changing into 'normal' oceanic crust. The main fault complex demonstrates the The VOting margin and East Greenland onshore areas document several Palaeozoic and temporal relationship between the faulting and Mesozoic tectonic episodes dominated by exten- the subsequent break-up. The results from ODP sion since the end of the Caledonian orogeny. Site 642 are consistent with break-up being During this evolution towards continental sep- coeval with the deposition of a regional tuff aration, the Late Jurassic-Early Cretaceous and marker in the adjacent sedimentary basins the Late Cretaceous-Palaeocene rifting epis- (Knox & Morton 1988; Eldholm et aL 1989a). odes represent a late and final stage, respectively The tuff marker bounds a relatively thick Palaeocene sequence, which onlaps the rift un(Skogseid et al. 1992). conformity towards the west. Taking into account that the rift unconformity represents a well Break-up related tectonism and magmatism defined erosional unconformity, we suggest that The Late Cretaceous-Palaeocene extension a large part of the Palaeocene sequence consists preceding the break-up near the Palaeocene- of sediments eroded from an uplifted western Eocene transition is manifested by shallow-de- V0ring Basin (Pedersen & Skogseid 1989; Skogtachment listric faults along the flanks of, and seid et al. 1992). The fact that the Palaeocene within, the Cretaceous graben west of the Fulla sediments are not offset by the faults indicates, Ridge. The most prominent tectonism, within however, a change in tectonic development, the 'main fault complex', is found close to the where active faulting was separated from breakmarginal high (Fig. 5). However, contempor- up by an episode of uplift and erosion. We may, aneous but less prominent faulting occurred also however, argue for the probability that faulting east of the ridge, as far as 150 km from the V0r- continued in the areas now masked by the lavas, ing Escarpment, or about 200 km east of the i.e. narrowing the active rift. The extrusive edifice constituting the marginal continent--ocean boundary (COB) (Fig. 3B). On the other hand, the Fulla Ridge region appears high is an expression of intense magmatic acas a relatively undisturbed and decoupled crustal tivity during break-up. The composition and block with only minor evidence of faulting emplacement history of the sub-basement features is discussed elsewhere (e.g. Eldholm et al. activity. Both western and eastern regions of faulting 1989a), but we note that the apex of the seawardhave been eroded and are truncated by a re- dipping wedges presumably overlies continental ,

4

308

MAN'I1,E PLUMES AND NE ATLANTIC RIFLING 6~' /

2 ,

6

~

%

S T R U C T U F I ~ ~ • ~

~)

~

v •

'

?

,

/

,

:

2

;

~

'

?

,

8.

. ~069

XY~

8

~

,

~

t0

~

~

~'

'~4'

~

~

~,o~

69o

~ , ~

,

/

'

Dipping reflector wedge

'~ ~

~

8 ~11065

.2

,

,f.

,

?

.

?

CENOZOIC TEcoCmuO~r: To nC e> aUIBS/DE" ~

468

'"

4

~

Commercialddll site

tD 0

,

/ "PLATFORMI

EARLYT E R T I A R / ~

681-

,2

LATESTCRETACEOUS , E C T O / / ~ ~ ~ ~

/~

'6

drill site DSDP/ODP

~__~

,

'

2 Cont[nent.O£~a n

,

i ~

I.

A

69_

'

8

10

650

i . .-, 30, while N-type MORB and may have eNd as low as 7 and Zr/Nb ~ 20 (cf. Le Roex 1987). At Svartenhuk, in the northern part of the province, the erupted compositions were less depleted and overlap with the most enriched early as well as with later Disko picrites. The MORBlike character in early Disko lavas and at Baffin Island is thus suggested to have been caused by the incorporation of magma derived from the depleted asthenosphere in the magmas that reached the surface. Evidence of passive interaction with asthenosphere is suggested by the complex mineralogy of olivine xenocrysts in some of the picritic melts (Pedersen pers. comm.). For early Disko lavas the effect of the plume was primarily an enhancement of the melting process in the rising asthenosphere, i.e. picrite generation. Extension in the Baffin Bay-Davis Strait region may have accelerated due to the arrival of plume material as suggested by White & McKenzie (1989). This would have increased the

360

GREENLAND LAVAS AND THE ICELAND PLUME

speed of asthenospheric rise and melting. We therefore propose that the compositional gradient from N-type MORB in the southern part of the province to the Iceland plume types in the northern area (Holm et al. in press) is similar to the plume-asthenosphere mixing relationship displayed on the transition from SW Icelan.d to the MAR. The general lack of a lithospheric component in the early West Greenland lavas is a strong indication that the magma had a swift travel through the lithosphere and that the plume arrived after rifting had commenced. Another way to explain the contrast in lithospheric interaction is that, processes were similar in East and West Greenland but the composition of the lithosphere is different. If percolation of plume-head melts interacted with enriched Archaean litho: sphere in East Greenland but with younger depleted lithospheric mantle and the rising asthenosphere in West Greenland it is possible that quite large differences might result in the erupted magmas. However, as the timing seems very different in West and East Greenland and as no N-MORB-type melts have been observed in East Greenland, we find it most likely that the magmas interacted only slightly with the lithosphere in West Greenland. In the White & McKenzie (1989) model occurrence of extensive picrite volcanism would indicate a position of West Greenland over the plume-axis. However, if the plume was centred under West Greenland at c. 62 Ma, instead of under East Greenland, as suggested by White & McKenzie (1989) and others, the East Greenland picrites still require a Kangerlugssuaq centered plume at 58 Ma. The uplift predicted by the White & McKenzie (1989) model and the Campbell & Griffiths (1990) model did not take place, in West Greenland. Thus, although it seems correct that Icelandic plume material was erupted in both West and East Greenland, the proposed mechanism for the evolution of the plume, plume-head and plume-derived melts cannot be correct in all detail. East G r e e n l a n d rifting

At around 58 Ma the main rifting activity shifted from between West Greenland and Canada to between East Greenland and Europe (Srivastava & Tapscott 1989). On a regional scale rifting in the NE Atlantic propagated from the south towards Kangerlussuaq at c. 1 m a -1 (Larsen 1988). Srivastava and Tapscott (1989) estimated a separation rate of c. 2.5 em a -1 for c. 2 Ma along 5000 km between Greenland and Europe and along the Lomonosov Ridge. While

initial, opening occurred close to the SE and NE Greenland coast from magnetic chron 25R-24R times, continental rifting at that time was unsuccessful between Kangerlussuaq and Scoresby Sund, possibly because of the robust lithosphere in this area (Larsen 1988). It has been suggested that the Scoresby Sund basalts reflect several aborted attempts at continental rifting under extensional forces (Larsen & Watt 1985). The direction of rifting was instead deflected far towards the east for some time until chron 24N times when a short-lived spreading axis became active near to the Blosseville Kyst (Larsen 1988). The coherence of the lithosphere may have been due to the thickness of and/or lack of old sutures in the Archaean continent at Kangerlussuaq, or that rifting was incapable of splitting the structure of the Caledonian belt that is supposed to underlie part of the KangerlussuaqScoresby Sund area (Brooks et al. 1981; Larsen 1988). From Cretaceous times onwards, the Kangerlussuaq area in East Greenland was subjected to slow subsidence. Conglomerates and coarse clastic sediments of Eocene age probably deposited into a SSW-ENE graben-like zone preceded the Eocene initiation of magmatic activity. The change in sedimentation may be due to uplift of the margins of the zone, but in general the Lower Lavas record extrusion during subsidence (Nielsen et al. 1981; Nielsen 1987). The subsiding graben-like zone appears to be the continuation of the ENE propagating rift along the coast south of Kangerlussuaq. Rifting as defined by dyke intrusion and onset of volcanism begin at 58 Ma. The Scoresby Sund basalts include several horizons of sediments and hyaloclastites, especially in the lower formations (Larsen etal. 1989) and the Lower Lavas indicate a subsidence of c. 2 km. There are thus no indications of doming prior to the volcanic period. As described in the introduction, the present high elevation of the top of the CFBs is due to both doming and general uplift of the continental margin post-dating the formation of the CFBs. The regional uplift probably is as young as 33-37 Ma as indicated by apatite fission-track ages (Gleadow & Brooks 1979). Brooks (1979) and Brooks & Nielsen (1981) concluded, based on field evidence, that the doming in the Kangerlussuaq area took place immediately after the extrusion of the basalts and contemporaneous with formation of the coastal flexure, but before the regional uplift in the Oligocene. If plume impingement on the East Greenland lithosphere was followed in just 4 Ma by continental break-up, time was too short to generate a bulge by heat conduction

P. M. HOLM ETAL. until long after the initiation of continental rifling. Late uplift of the continental margin is also argued by Larsen & Marcussen (this volume). Volcanism followed by doming has also been recorded from the Yemen volcanic rocks (Menzies et al. this volume). It is worth noting that the upfir of the East Greenland margin is not indicated to be due to heating, as no thermal relaxation has been detected since uplift. The relatively high Ti in East Greenland pricrites cannot be ascribed to variation in the degree of partial melting, because the analysed East Greenland picrites have as high or higher CaO/ A1203 ratios, and thus degree of melting, than the West Greenland picrites. The high Ti must be evidence of an enriched source region for the East Greenland lavas. The positive correlation between eSr and Ti/Y (Fig. 5) is evidence that the source cannot be the crust but more probably Ti-rich domains in the lithospheric mantle. Such domains would not be generated by subducted slab-derived materials, which are low in Ti, but rather by asthenosphere or deeper-derived high Ti-melts. These rose into the lithospheric mantle, and were later incorporated into the plumederived magmas. Thus, the relative high Ti seen in both West Greenland and Icelandic lavas compared to MORB is consistent with a derivation from relatively Ti-rich plume material. On the other hand the even higher Ti-level observed in the East Greenland Lower Lavas most probably was incorporated from the subcontinental lithosphere.

The plume under East Greenland and continental break-up In the various models of plumes impinging upon the lithosphere, the thermal effect is swelling. In the case of the present Icelandic plume, White & McKenzie (1989) estimate a residual depth anomaly of ~- 2000 m. The model of Campbell & Griffiths (1990) suggest that the uplift caused by the plume may cause firing. However, there is no clear indication of doming in the Kangerlussuaq area before or at the time of volcanism and rifting. In the Scoresby Sund area, regional subsidence of 1-2 km accompanied flood volcanism (Larsen & Marcussen this volume). Also on the eastern margin of the proto-Atlantic there is an apparent lack of heat input before rifting (Joy & Cartwright 1991). As the plume had caused no thermal doming of the lithosphere, rifting in East Greenland was not generated by a gravitational slide of the lithosphere off a plume generated thermal bulge as suggested by Campbell & Griffiths (1990), The sequence of events in East Greenland are in strong contrast to the correla-

361

tion of plume and rifting in the Campbell & Griffiths (1990) model, as rifting was ongoing both north and south of Kangerlussuaq before volcanism and eventually rifting was initiated above the plume-axis at Kangerlussuaq. One of the major geochemical differences between West and East Greenland picrites is the abundance of Fe. The relatively high Fe in the East Greenland picrites reflects their high pressure derivation (e.g. Falloon & Green 1988). This is consistent with the Archaean lithospheric age of the Kangerlussuaq area and thus the greater thickness of the lithosphere under which the magma generation took place. Other differences in the chemistry of the lavas in East and West Greenland are probably also caused by a major contrast in geological setting. One explanation would be that the hot plume material arrived approximately simultaneously in East and West Greenland, but the continent did not rift in East Greenland. The potential temperatures of the West Greenland melts were around 1480°C (Gill et al. this volume) and, if it is correct that the plume was centred under East Greenland before rifting, similarly hot material impinged on the lithospheric mantle and enhanced the geotherm. During the 4 Ma period between onset of magmatism in West Greenland and East Greenland hot material was constantly supplied to the base of the East Greenland lithosphere possibly leading to an upwards percolation of small amounts of melts derived either from decompression melting in the most shallow hottest parts of the mantle plume or in (enriched) domains with relatively low solidus temperatures near the base of the lithosphere. Such melts may have mobilized or almost mobilized the incompatible elements in the Archaean East Greenland lithosphere and thus facilitated their incorporation into the early CFB magmas in East Greenland. Because the incompatible elements (such as Rb, Sr, Sm and Nd) of the East Greenland lithosphere had been isolated since Archaean times, their addition to the CFB magmas had a gross effect on the low concentration inventory of these elements and the radiogenic isotopes of Sr and Nd in the plumeand asthenosphere-derived melts as proposed by Holm (1988). The exclusive occurrence of the lithosphereenriched basalts and picrites in the Kangerlussuaq area strengthens the case for the focusing of the plume activity under this area. The fact that a major deflection of spreading-axis took place near Kangerlussuaq at the time of initial rifting along the present coast-line may not be a consequence of plume activity but of the robust nature of the lithosphere. Thus, another important

362

GREENLAND LAVAS AND THE ICELAND PLUME

feature of the plume model of Campbell & Griffiths (1990), the plume-generated thinning, may not apply universally. Instead it seems most likely that it was the rifting event at the continental break-up, which allowed plume-axis derived melts to be generated and rise at Kangerlussuaq. Time is probably required for the pre-treatment of the lithospheric mantle, by small amounts of melt probably derived from the plume, in order to have the incompatible elements mobilized and included into the rising CFB magmas. Therefore, the plume would have been focused at Kangerlussuaq some time before the CFBs were erupted. The most probable time was when the activity of the plume was initiated in West Greenland. Thus, as in the case of West Greenland, it is suggested that the activation of the Icelandic mantle plume took place without a major thermal effect (i.e. doming) on the lithosphere before lavas were emplaced, but with small scale melt infiltration of the lithosphere. The time span of c. 4-5 Ma between onset of volcanism in West and East Greenland conflicts with the Campbell & Griffiths (1990) plume model, which predicts sudden onset of volcanism over the total area of the plume-head. The rapid contraction of volcanism, in their model did not take place in Greenland either. Interestingly, it can be noted that picrite volcanism in West Greenland extended for 250 km along the firing margin, while in East Greenland it was focussed narrowly at Kangerlussuaq.

Conclusion The overall conclusion is that the initial rise of the pl:oto-Icelandic plume took place at c. 62 Ma under Kangerlussuaq, East Greenland, as advocated by White & McKenzie (1989). However, unlike Campbell & Griffiths' (1990) proposal, the flood basalt volcanism is suggested to have been initiated by rifting and not generated through doming. The envisaged scenario is that, at time of plume impingement, mobilization of the incompatible elements of the East G r e e n land Archaean lithospheric mantle commenced, probably by means of percolating small-scale melts, and the plume spread out laterally. At the already rifting West Greenland margin the rising asthenosphere soon received extra uplift from the arrival (by lateral flow) of the very hot plume material, with the result that extraordinarily primitive and depleted flood lavas were produced in the southern part of the province at the early stage of volcanism primarily by the melting

of asthenosphere. In the northern part, the plume-derived melts rose without asthenospheric but with some lithospheric interaction. This difference may be correlated with the decrease in amount of plate separation from south towards north along West Greenland. Alter, natively, the requirement for very hot source material for the West Greenland picrites would be met by a temporary second plume axis centred under the Svartenhuk-area for c. 4 Ma. When, due to major plate-motion rearrangements around Greenland at c. 58 Ma, rifting shifted to the East Greenland continental margin. West Greenland volcanism waned and CFB volcanism was initated over the (principal) the plume centre can be identified as the area where continued hot material had been supplied from below for c. 4 Ma, and the lithosphere above permeated by small amounts of melt which indirectly led to the enrichment of the subsequently erupted anomalously enriched lavas. The lack of doming in this model requires that no massive heat transfer to the East Greenland lithosphere took place, probably due to the limited time of 4 Ma. The erupted CFB compositions may be modelled as having the Icelandic mantle plume as a common component. Compositional constraints of this common component are approximately: eSr = -18, eNd = +7, Ti/Y = 450, Zr/Nb = 13, Zr/Y = 4.5 and Ce/Nb = 2.7. We thank G. Marriner for advice and assistance with the analytical work and R. Madsen and Trine Heegaard for drafting the diagrams. Reviews by B. Upton, M. Thirlwall & R. Pankhurst on an early version of the manuscript is highly acknowledged. The West Greenland project is supported by NATO grant number 04-0771/86 and the Carlsberg Foundation, the East Greenland by the Danish Nature Science Research Council. This contribution is published with the authorization of the Geological Survey of Greenland.

References Boyr, M. H. P. 1988. A new look at the causes and consequences of the Icelandic hot-spot, ln: MORTON,A. C. & PARSON,L. M. (eds) Early Tertiary Volcanism and the Opening of the NE Atlantic. Geological Society of London, Special Publication, 39, 15-23. & GU~ARSSON,K. 1980. Crustal structure of the Iceland-Faroe Ridge. Journal of Geophysics, 47, 221-227. BROOKS,C. K. 1973. Tertiary of Greenland-a volcanic and plutonic record of continental break-up. Memoir of the American Association of Petroleum Geologists, 19, 150-160. 1979. Geomorphological observations at Kangerdlussuaq, East Greenland. Meddelses. Gr0nland, Geoscience 1.

P. M. HOLM E T A L . &~NIELSEN,T. F. D. 1982. The Phanerozoic development of the Kangerlussuaq area, East Greenland. Meddelelses Grenland, Geoscience,

363

GLEADOW, A. J. W. & BROOKS, C. K., 1979. Fission track dating thermal histories and tectonics of igneous intrusions in East Greenland. Contribu9. tions to Mineralogy and Petrology, 71, 45-60. ~ , FAWCETI',J. J., GrrrtNs, J. & RUCKLIDGE,J. C. GRIFFITHS, R. W. & CAMPBELL,I. H., 1989. Stirring 1981. The Batbjerg complex, east Greenland: a and structure in mantle plumes. Earth and unique ultrapotassic Caledonian intrusion. Planetary Science Letters, 99, 66-78. Canadian Journal of Earth Sciences, 1 8 • 274-285. HALD, N. 1976. Early Tertiary flood basalts from ~, NIELSEN,T. F. D. & PETERSEN,T. S. 1976. The Hareeen and western N~gssuaq, West Gre'enland. Blosseville Coast basalts of East Greenland: Bulletin GrOnlands Geoiogiske Undersgelse, 120. Composition and temporal variation. Contribu& PEDERSEN,A. K. 1975. Lithosstratigraphy of tions to Mineralogy and Petrology, 5, 279-292. CAMPBELL, I. H. & GmFFn'HS, R. W. 1990. Implicathe Early Tertiary volcanic rocks of central West tions of mantle plume structure for the evolution Greenland. Rapport, Grenlands Geologiske Undersogelse, 69, 17-24. of flood basalts. Earth and Planetary Science Letters, 99, 79-93. HEMONO, C., CONDOMINES, M., FOURCADE, D. ALLEGRE, C. J., OSKARSSON, N. & JAVOY, i . CHALMERS,J. A. 1991. New evidence on the structure of the Labrador Sea/Greenland continental mar1988. Thorium, strontium and oxygen isotopic geochemistry in recent tholeiites from Iceland: gin. Journal of the Geological Society, London, 148, 899-908. crustal influence on mantle-derived magmas. & LAURSEN,K. H. 1991. The eastern margin of Earth and Planetary Science Letters, 87,273-285. HILL, R. I. 1991. Starting plumes and continental the Labrador Sea. Geological Society of London, meeting: Magmatism and the causes of continental break-up. Earth and Planetary Science Letters, break-up (Abstract volume) 8. 1 0 4 , 398-416. CLARKE, D. B. 1970. The Tertiary basalts of Baffin HOLM, P. M. 1988. Nd, Sr and Pb isotope geochemisB a y : possibly primary magma from the mantle. try of the Lower Lavas, E Greenland Tertiary Contributions to Mineralogy and Petrology, 25, Igneous Province. In: MORTON,A. C. & PARSON, 203-224. L. M. (eds) Early Tertiary Volcanism and the Opening of the NE Atlantic. Geological Society, & PEDERSEN,A. K. 1976. Tertiary volcanic province of West Greenland. In: ESOtER, A. & London, Special Publication, 39, 181-195. , GILL, R. C. O., PEDERSEN,A. K., LARSEN,J. G., WAa'r, W. S. (eds) Geology of Greenland. HALO, N., NIELSEN,T. F. n . , & THIRLWALL,i . Geological Survey of Greenland, Copenhagen, 364-385. F. 1988. Tertiary picrites of West Greenland: petrogenetic implications of trace elements & Sr-Nd & UPTON, B. G. 1971. Tertiary basalts of Baffin isotope geochemistry. Chemical Geology, 70, 49. Island: field relations and tectonic setting. Cana, , , , , - &~ In press. dian Journal of Earth Sciences, 8, 248-258. The Tertiary picrites of West Greenland: conCONDOMINES, M.• GRONVOLD, K., HOOKER, P. J., tributions from 'Icelandic' and other sources. MUEHLENBACHS, K., O'NIONS, R. K., OSKAREarth and Planetory Science Letters. SSON, N. & OXBURGH, E. R. 1983. Helium, oxygen, strontium and neodymium isotopic re- JoY, A. M. & CARTWRIGHT,J. 1991. The pattern of lationships in Icelandic volcanics. Earth and heat loss during continental break-up: evidence Planetary Science Letters, 66, 125-136. from sedimentary basins adjacent to the NorweCOURTILLOT, V., BESSE, J., VANDAMME, D., MONgian-Greenland Sea. Geological Society, London TIGNY• R., JAEGER) J.-J. & CAPETrA, H. 1986. meeting: Magmatism and the causes of continental Deccan flood basalts at the Cretaceous/Tertiary break-up, (Abstract volume) 15. boundary. Earth and Planetary Science Letters, LARSEN, H. C., 1980. Geological perspective of the 8 0 , 361-374. East Greenland continental margin. Bulletin of ELLIOTr, T. R., HAWKSWORTH,C. J. & GRONVOLD,K., the Geological Society of Denmark, 29, 77-101. 1990, The East Greenland Shelf. In: GRANTZ,A., 1991. Dynamic melting of the Iceland plume. Nature, 321,201-206. JOHNSON,A. & SWEENEY,J. F. (eds) The geology of North America Vol. L. The Arctic Ocean RegFALLOON, T. J. & GREEN, D. H. 1988. Anhydrous ion. Geological Society of America, 185-210. partial melting of peridotite from 8 to 35 kb and the petrogenesis of MORB. Journal of Petrology 1988. A multiple and propagating rift model for special lithosphere vol., 379-414. the NE Atlantic. In: MORTON,A. C. & PARSON, GILL, R. C. O., NIELSEN,T. F. D., BROOKS,C. K. & L. M. (eds). Early Tertiary Volcanism and Opening of the NE Atlantic. Geological Society, LonINGRAM, G. A. 1988. Tertiary volcanism in the don, Special Publication, 39, 157-158. Kangerlussuaq region , East Greenland: trace& MARCUSSEN, C. 1992. Sill-intrusion, flood elements geochemistry of the Lower Basalts and basalt emplacement and deep crustal structure of tholeiitic dyke swarms. In: MORTON,A. C. & PARthe Jameson Land basin, East Greenland. This SON,L. M. (eds) Early Tertiary Volcanism and the volume. Opening of the NE Atlantic. Geological Society• LARSEN, J. G. 1977. Transition from potassium olive London, Special Publication, 39• 161-179. tholeiites to alkali basalts on Ubekendt Efland, the • PEDERSEN,A. K. & LARSEN,J. G. 1992. Tertiary Tertiary volcanic province of West Greenland. picrites in West Greenland: melting at the periphery of a plume? This volume. Meddelelser om Gr0nland, 2 0 0 .

GREENLAND LAVAS AND THE ICELAND PLUME

364

LARSEN, L. M. & WATr, W. S. 1985. Episodic volcanism during break-up of the North Atlantic: evidence from the east Greenland plateau basalts. Earth and Planetary Science Letters, 73, 105-116. , PEDERSEN,A. K., PEDERSEN,G. K. & PIASECKI, S. 1992. Timing and duration of the volcanism during break-up of the North Atlantic in the Tertiary: New evidence from West Greenland. This

volume. , WAYr, W. S. & WATr, M. 1989. Geology and

petrology of the Lower Tertiary plateau busalts of the Scoresby Sund region, East Greenland. Bulletin Gronlands Geologiske Undersogelse, 157. LEEMAN,W. P. & DASCH,E. J. 1976.2°Tpb/2°6pbwhole rock age of gneisses from the Kangerlussuaq area, eastern Greenland. Nature, 263, 469-471. & HAWKESWORTH,C. J. 1986. Open magma systems: trace element and isotopic constraints. Journal of Geophysical Research, 91, 5901-5912. LEROEX, A. P. 1987. Source regions of mid-ocean ridge basalts: evidence for enrichment processes. In: MENZIES,M. A. & HAWKESWORTH,C. J. (eds) Mantle Metusomatism, Academic Press, London, 389-422. LOUDEN, K., LEVESQUE,S., OSOLER,J., CHIAN, D. & SRIVASTAVA, S. 1991. Hot spot activity and the formation of the Labrador Sea. Geological Society London meeting: Magmatism and the causes of continental break-up, (abstracts volume), 18. MORGAN, W. J. 1971. Convection plumes in the lower mantle. Nature, 230, 42-43. MENZIES, M. A., BAKER,J. E. & 9 OTHERS. 1992. The timing of magmatism, uplift and crustal extension: preliminary observations from Yemen. This

volume. NIELSEN, T. F. D. 1987. Tertiary alkaline magmatism in E. Greenland: a review. In: FIYroN, J. G. & UPTON, B. G. J. (eds) Alkaline Igneous Rocks. Geological Society London, Special Publication, 30, 489-515. ~, SOPER, N. J., BROOKS, C. K., FALLER, A. M. HIGGINS, A. C. & MATIHEWS, n . W. 1981. The

pre-basaltic sediments and the Lower Basalts at Kangerlussuaq, East Greenland, their stratigraphy, lithology, palaeomagnetism and petrology. Meddelelser om Gr0nland Geosciences, 6, 25 pp. OSKARSSON, N., STE1NTHORSSON,S. ,(z SIGVALDASON, G. E. 1985. Iceland geochemical anomaly: ori-

gin, volcanotectonics, chemical fractionation and isotope evolution of the crust. Journal of Geophysical Research, 90, B12, 10 011-10 0"25, PEDERSEN, A. K. 1985. Reaction between picrite

magma and continental crust: Early Tertiary silicic basalts and magnesian andesites from Disko, West Greenland. Bulletin Grcmlands Geologiske Undersogelse, 152. RICHADS, M. A., DUNCAN,R. A. & COURTILLOT,V. E. 1989. Flood basalts and hot-spot tracks: plume heads and tails. American Association for the Advancement of Science, 246, 103-107. ROEST, W. R. & SRIVASTAVA,S. P. 1989. Sea-floor spreading in the Labrador Sea: A new reconstruction. Geology, 17, 1000-1003, SAUNDERS, A. D., NORRY, M. J. & TARNEY,J. 1988. Origin of MORB and Chemically-Depleted Mantle Resevoirs: Trace Element Constraints. Journal of Petrology, Special Lithosphere Issue, 415-445. SRIVASTAVA, S. P. • TAPSCOTr, C. R. 1989. Plate kinematics of the North Atlantic. In: VOGT,P. R. & TUCHOLKE,B. E. (eds) The Geology of North

America, Vol. M, the Western North Atlantic Region. Geological Society of America, 379-404. SOPER, N. J., DOWNIE,C., HIGGINS,A. C. & Cos~rA,L. I. 1976. Biostratigraphic ages of Tertiary basalts on the East Greenland continental margin and their relationship to plate separation in the Northeast Atlantic. Earth and Planetary Science Letters, 32, 149-157. UPTON, B. G. J., EMELEUS,C. H. & BECKINSALE,R. D. 1984. Petrology of the northern East Greenland Tertiary flood basaits: evidence from Hold with Hope and Wollaston Forland. Journal of Petrology, 25, 151-184. VINK, G. E. 1984. A hotsport model for Iceland and the Vcring Plateau. Journal of Geophysical Research, 89, 9949-9959. WHITE, R. S. 1988. A hot-spot model for early tertiary volcanism in the NE Atlantic. In: MORTON,A. C. & PARSON,L. M. (eds)Early Tertiary volcanism and the opening of the NE Atlantic. Geological Society, London, Special Publication 39, 3-14. & MCKENzm, D. 1989. Magmatism at rift zones: the generation of volcanic continental margins and flood basalts. Journal of Geophysical Research, 94, B6, 7685-7729. WILSON, M. 1988. Igneous Petrogenesis--a global tectonic approach. Unwin Hymann, London.

Sill-intrusion, flood basalt emplacement and deep crustal structure of the Scoresby Sund region, East Greenland H. C. L A R S E N & C. M A R C U S S E N

Geological Survey o f Greenland, Ostervoldgade I0, DK-1350, Copenhagen, Denmark

Abstract: The late Palaeozoic-Mesozoic Jameson Land basin in East Greenland was situated during early Tertiary break-up 300-400 km north of a presumed plume centre and 100-200 km landward of the initial line of North East Atlantic opening. The basin continues southward below Scoresby Sund fiord and the Scoresby Sund plateau basalts south of the fiord. By combining surface mapping with marine reflection and refraction seismics from Scoresby Sund and deep (6-12s) reflection seismics from the exposed part of the basin, it is possible to construct a composite image of the entire crustal structure in this region from the details of basalt stratigraphy at the top, through the basin fill, the foundation of the basin, and into the uppermost mantle. At the basin centre there is a very thin crystalline crust (6-8 kin?) below as much as 16-18 km of basin fill. Virtually all crustal extension took place prior to mid-Permian time, and the post-rift Upper Permian to Cretaceous section contributes to only one third or less of the total thickness of the succession. During Tertiary break-up the basin was intruded by basaltic sills and dykes, and basaltic flood basalts flowed over the basin with an apparently decreasing thickness to the north. The sills are mainly exposed along the basin margins as rather thin (10-15 m) layers, but apparently increase in intensity and thickness at depth and towards the basin centre. It seems that magmas were intruded as sills up to 300 m thick in the deep (10-15 km) central parts of the basin, from where they ascended towards the basin margins and to much younger stratigraphic levels. Their geometry and possible volume make them potential candidates as mid-crustal magma chambers and crustal magma pathways for the flood basalts which show low-pressure fractionation. However, extension surface geological data and deep crustal reflection seismic data show no faulting or crustal extension associated with this intense break-up volcanism. In Scoresby Sund there is a general rather conformable relationship between the basin stratigraphy and the gross stratigraphy of the flood basalts, suggesting limited or no initial uplift prior to flood basalt volcanism. The present-day high elevation of the basalts is considered part of large regional margin uplift post-dating N Atlantic break-up by 20 Ma or more. The apparent guidance exerted by the basin on the break-up magmatic activity without renewed rifting of the basin itself, the apparent lack of a broad initial uplift during break-up, and the late regional margin uplift, all seem at odds with several current plume models.

Continental break-up may be associated with either relatively little, or very intense, volcanism. The two types of rifted margins are often referred to as non-volcanic rifted margins and volcanic rifted margins: they seem to constitute two distinctly different developments. The temperature regime within the upper mantle during break-up seems to be an important controlling factor in the rifted-margin development and the possible role of hot spots and deep-seated mantle plumes in the formation of volcanic rifted margin and continental flood basalts is vividly discussed (White & McKenzie 1989; Duncan & Richards 1991; Griffiths & Campbell 1990; Hill 1991). The Jameson Land basin is situated near the East Greenland volcanic rifted margin (Fig. 1) occupying a position within the central landward part of Zone I ('Sediment Basin') in the volcanic

margin model of Roberts et al. (1991). It formed within crust deformed by the Caledonian orogeny, and is bounded towards both the east and the west by high-grade metamorphic rocks and granites of Caledonian and older age (Fig. 2). The up to perhaps 16"18 km thick basin comprises sediments of Devonian to Cretaceous age and intrusive and extrusive igneous rocks of early to mid-Tertiary age (H. C. Larsen et al. 1989; Marcussen & Larsen 1991; Surlyk et al. 1986). It is about 70 km wide and at least 200 km long (Fig. 2). At the present level of exposure it forms a slightly southward-plunging (1-2 °) synclinical structure with late Palaeozoic to early Mesozoic rocks outcropping along the western, eastern and northern basin margins and progressively younger strata towards the centre and south. In the very south it is covered by the early Tertiary basalts of the East Greenland flood

From STOREY,B. C., ALABASTER,T. & PANKHURST,R. J. (eds), 1992, Magmatism and the Causes of ContinentalBreak-up, Geological Society Special Publication No. 68, pp. 365-386.

365

366

H.C. LARSEN & C. MARCUSSEN

Fig. 1. Location of the Jameson Land basin and Scoresby Sund study area in relation to the supposed plume geometry at break-up time. Black, seawarddipping reflector sequences; cross-hatching, plateau basalts. Note that a straight line between the southern and northern areas with seaward dipping reflectors would cross over the Jameson Land basin and very close to the supposed plume centre. Volcanism is found along this line (Nielsen 1987). Modified from White & McKenzie (1989). KA, Kangerdlugssuaq.

basalt province, the so-called Scoresby Sund basalts, which also form a slight southward-dipping synclinal structure. The basin is intruded in the north by early to mid-Tertiary intrusive rocks (Fig. 2). During initial break-up in the Palaeocene, the. Jameson Land basin was in a position approximately 300 to 400 km north of the supposed plume centre (Fig. 1). A line connecting the simple straight segments of the southern and northern parts of the North-east Atlantic break-up rift passes through both the plume centre and the Jameson Land basin, and break-up volcanism is found all along this line (Fig. 1 and Larsen & Watt 1985). Break-up within this central part of the Northeast Atlantic, however, was deflected to the east and followed a partly sinuous trend 100 to 200 km east of the Jameson Land basin (H. C. Larsen 1988). The ocean--continent transition later migrated closer to the Jameson Land basin and is now only 75-100 km east of the basin (H. C. Larsen 1988, 1990). The basin remained tectonically undisturbed during the initial break-up, but was intruded by basaltic

dykes and sills and for a large part covered by up to two kilometers of the thick flood basalts that overflowed the East Greenland margin. Later uplift and erosion exposed both the basin and the overlying volcanic rocks and provides a unique opportunity to study the crustal structure below the flood basalts of a volcanic rifted margin in a position proximal to the centre of the supposed mantle plume and hot spot. Some of the key questions to be addressed are the vertical crustal movements prior to, during and after break-up, the role of pre-existing rift structures in generating and locating the anomalous amounts of magmatic melt, the process of lithospheric thinning and the nature of the supposed shallow (crustal?) magma chambers in which the primary magma for the low-pressure fractionated flood basalts apparently reequiliberated. Predictions made from present plume models include initial, pre-volcanism uplift in the order of 1 to 2 km, stretching and thinning of the lithosphere by reactivation of the older rift structures crossing the plume-generated regional dome structure, followed by adiabatic rise and extensive melting of anomalously hot asthenosphere. A sub-vertical magma conduit to the surface and likewise a sub-vertical sheet- or dyke-like shallow magma chamber is often supposed or implied. As documented in the following, these pre,dictions or assumptions seem to fail within the Jameson Land basin.

Data background The Scoresby Sund and the Jameson Land region was geologically mapped during the late 1960s and early 1970s (Henriksen 1986) including the Caledonian basement, the mainly Mesozoic sedimentary outcrops of the Jameson Land basin, and the Tertiary flood basalts south of Scoresby Sund. The detailed basalt stratigraphy and geochemistry were recently published (L. M. Larsen et al. 1989). The region has been covered by various geophysical studies, including an aeromagnetic survey (H. C. Larsen 1975; H. C. Larsen et al. 1988), regional gravity stations (Forsberg 1986), marine single channel seismics (B. Larsen 1980) and multi-chanel seismics (Andersen et al. 1981). Onshore seismic studies within the Jameson Land basin were conducted by Atlantic Richfield Company (ARCO) in connection with oil exploration from 1986 to 1989. In a joint study between the Geological Survey of Greenland and ARCO, approximately 550 km of seismic data were acquired and processed to 12 s TWT in order to provide information on the deeper part of the basin and the

BASALT AND DEEP CRUSTAL STRUCTURE, EAST GREENLAND

367

Tertiary sediments ~]

Tertiary intrusives Tertiary b a s a l t s - S u c c e s s i o n I Tertiary basalts - S u c c e s s i o n II L. C r e t a c e o u s (LC) U. Jurassic (UJ) Jurassic (J)

r~ r~

Triassic U. P a l e o z o i c (Devonian to Permian) Basement i

25 km

•

1° -

Milne Land)

UJ

S Upper Cretaceous ,~ sediments

--

70

° _

Fig. 2. Simplified geological map of the Scoresby Sund and Jameson Land region. The informal sucessions I and II within the Scoresby Sund basalts correspond to the regional lower and upper sequences of L. M. Larsen et al. (1989). Eocene and younger, marine sediments are found within the down-faulted coastal area. The regional structure shown by the successions I and II is a gentle, 1-2 ° southward-dipping synclinal structure overlying a similar structure within the late Mesozoic sediments of southern Jameson Land and Scoresby Sund. Crosssections A B and C are shown in Figs 3, 4 and 7 respectively. Sills and dykes within Jameson Land are shown in Fig. 3. Cp youngest pre-basaltic sequence in the fiord of late Cretaceous to possible Palaeocene age.

underlying crust and mantle (Table 1). Finally, deep seismic refraction studies were conducted in 1988 and 1990 in Scoresby Sund by a German group from the Alfred-Wegener-Institut of Bremerhaven. Thus the last two decades have brought about a comprehensive database within this key area. We have analysed the mapping and reflection seismic data in particular, in order to portray the complete crustal structure of this segment of a volcanic margin.

Tertiary intrusives seen in outcrop The Jameson Land basin exhibits three different kinds of intrusive rock: (1) dykes of roughly E W to N W - S E orientation, cutting vertically or

sub-vertically through the entire sequence of basin fill, (2) sills mainly intruding the shaly parts of the basin fill, and (3) plutonic to sub-volcanic intrusions of roughly circular shape in the far north (Fig. 3). The dykes are generally rather thin (metresized) and occur infrequently also outside the basin. The sills are much thicker (typically 10 to 50 m, and locally 75 m), are semi-concordant to basin fill, and are not observed outside the basin or within the flood basalts south of the Jameson Land basin (W. S. Watt, pers. comm.). The plutonic intrusions in the north attain dimensions of 20 km in diameter or more. At the present exposure level, sills are most frequent within lower and upper Jurassic shales,

368

H.C. LARSEN & C. MARCUSSEN

Table 1. Acquisition and processing parameters used for seismic data in this study Acquisition parameters

Dynamite lines

Vibroseis lines Source:

4 vibrators inline 16 sweeps 10-70 I-lznon-linear sweep 10 s. sweep length, 6 s listening time

Charge size: Charge depth:

Receiver:

3.6 km spread 2000% coverage

as for vibroseis line

25 kilograms 15-20 metres

Processing parameters 6 s data

1. 2. 3. 4. 5. 6. 7.

Demultiplex F-K filter Deconvolution Staticcorrections DMO stack RPF filter Migration

8. Time variant filter Dynamite 0-1.5s 12-72Hz 1.5-2.0 10-45 2.0-3.0 10-45 3.0-3.5 8-40 3.5-6.0 8-40

Vibroseis 0-0.6s 0.6-1.0 1.0-3.0 3.0-4.0 4.0-6.0

14-70 Hz 10-70 10-60 10-45 10-35

12 s data

1. 2. 3. 4. 5.

Demultiplex Static corrections F-K filter Seismic array forming Deconvolution

but are also found within Triassic and Permian strata in Jameson Land. North of Jameson Land, similar sill systems up to 160 m thick are found in sediments of Devonian to Cretaceous age (Hailer 1970; N. Hald pers. comm.). South of the Scoresby Sund region, sills within the Upper Cretaceous to Palaeocene sediments below the basalts are found in the Kangerdlugssuaq region (Fig. 1, close to the hotspot centre). They here attain thickness of more than 300 m and in places constitute a very large proportion of the pre-basaltic section (Nielsen et al. 1981; T. F. D. Nielsen, pers. comm. 1992). The geological setting of the sills and dykes strongly suggests a Tertiary age and close relationship to the flood basalt volcanism which in turn was dated by Soper etal. (1976) to the latest Palaeocene to earliest Eocene age (within the magnetic chron 24R or 55-57 Ma according to the Harland et al. 1989 time scale; see also L. M. Larsen et al. this volume). Noble et al. (1988) suggests on the basis of K-Ar dating that the main extrusive activity took place between 54 and 57 Ma. However, very little published data exist on the age and geochemistry of the Jameson Land sill and dyke complex. 4°K-4°Ar dating on whole rock and mineral separate samples carded out by A R C O on 8 samples of basaltic intrusive rocks gave ages from 47.0 Ma to 55.3 Ma

6. NMO stack 7. RPF filter 8. Time variant filter 0.5 s 8-32 Hz 5-12 6-24

with six ages within the interval 52.4+3.2 Ma to 55+3.3 Ma, in good agreement with the inferred age of the lavas (Bergman & Henk 1990). This study also confirms the sparse geochemical data reported by Noe-Nyegaard (1976), namely that the sill and dyke intrusives are mainly quartz tholeiitic dolerites comparable to the flood basalts, but that alkaline compositions also occur. The interval 52.4-55.3 Ma according to the Harland etal. (1989) time scale is predominantly of the normal magnetic polarity (magnetic chrons 23N and 24N). However, preliminary investigations of the magnetic polarity of the Jameson Land sills and dykes like the lavas all show a reverse magnetic polarity with the remanent magnetization largely balancing the present-day induced magnetization (Fuller 1986). This potential misfit is entirely within the uncertainty of the 4°K-4°Ar datings. Thus, the radiometric, paleomagnetic and data geochemical data available suggest that the sills and the basalts are cosanguineous, and emplaced primarily or perhaps entirely within magnetic chron 24R. The circular plutonic intrusions in the northern part of the basin are predominantly alkaline and vary from mafic cumulates to granitic and syenitic compositions and the ages range from

BASALT AND DEEP CRUSTAL STRUCTURE, EAST GREENLAND 3-, ~'-~

- \

72°oo._

u

:

.--- v

o ,.¢ -~

.--- .'~ -

~

--

I 20

.~

.~..¢..,

', ', ,,

;:

L 40

Intrusive centres, m a i n l y syenites and granites

i ,, ,~'

I 60 km .~-

Dolerlte dykes

Sills

Fig. 3. Sills, dykes and plutons in the Jameson Land basin. Modifiedfrom Noc-Nygaard (1976) to highlight sills. approximately 40 to 30 Ma (Rex et al. 1979). K-Ar dating of supposed Permian lamprophyric dykes from the western basin margin shows ages from 35 Ma to 48 Ma and these dykes probably belong to this later and partly alkaline magmatic episode following the main tholeiitic phase (P. H. Larsen et al. 1990). This late phase of volcanism was reviewed by Nielsen (1987) and is not discussed further in the present paper.

Present and past volcanic cover E m p l a c e m e n t o f the East G r e e n l a n d f l o o d basalts

The inferred time interval of the main flood basalt volcanism in East Greenland (within

369

magnetic chron 24R) is generally agreed upon as the time of final break-up, regional formation of seaward-dipping reflectors along the margins and the start of sea-floor spreading in the NE Atlantic (Talwani & Eldholm 1977; Eldholm et al. 1989; Larsen & Jakobsd6ttir 1988; L. M. Larsen et al. 1989). The East Greenland flood basalts that cover the southern extension of the Jameson Land basin south of Scoresby Sund can be divided into two major and regionally distributed successions and a third (youngest) and more limited succession (Figs 2 & 4). This division is based on extensive and detailed field mapping and geochemistry (L. M. Larsen et al. 1989). The two older and major successions (lower and upper sequences of L. M. Larsen et al. 1989) can further be divided into formations and units that, because of the exceptional mapping and geochemical control, also can be followed regionally. This allows an examination of the basalt stratigraphy with the same order of resolution and lateral extension as a regional seismic stratigraphic study. Paired with the seismic coverage of the prebasaltic basin, direct comparison of the regional basalt stratigraphy and the stratigraphy of the underlying basin is possible. Since the two oldest basalt successions (called successions I and II in this paper) volumetrically and areally are far the most important, we limit our discussion to these. The third succession only occurs along the faulted (post-basalt faulting) coastal zone, a deformation process not considered by the present paper. Successions I and II are true chronostratigraphic (and lithostratigraphic) sequences, with the younger succession II partly overlying the older succession I, but they do not show a simple layer parallel, concordant stratigraphic structure (Fig. 4). Sucession I was derived from a western source close to the southward-extrapolated western boundary of the Jameson Land basin (L. M. Larsen et al. 1989). Although the lower part of sucession I is thickest in the west, it overflowed the basin towards the east in a 'progradationallike' fashion with onlap onto the basement along the western basin margin and downlap towards and across the basin centre (Fig. 4). A N N E SSW-trending dyke swarm is considered the source by L. M. Larsen et al. (1989), but no actual feeder dyke relationships have been firmly established in spite of fairly detailed mapping, and very few dykes seem to be present within the central part of the basalt region. Onlap relations of succession I onto the basinward-sloping basement surface along the western basin margin make it unlikely that significant parts of succes-

370

H.C. LARSEN & C. MARCUSSEN C-C'

B

~

2 kin-

B'

?

-

"

,,~ll

~ ,-

o . , . ~~ ,~,.,',,-

i

~ ..~._ .

-2

,?--

km

~:~`~:~`:~#~.~i :,.,;.'....,-..'~:.* ~.!;~,;:::?~`~:~:~:~::~::.;~4!~::?:~:;~:~;~:!~;~' ".........*.,...,,-..:.:'..-" ...... ~.-."..,'.~.--ot.... i:;:~;:~'~.

. . . . . . .

o-

o T E R T I A R Y 9• ) SHORE LINE

-1

"

~"":':+:'::::':':':':'" ;'::'::'::':"::::':'::f:::':'::::':?"""

FACIES

'"

' ..............

-2

...........

I

60

km

:::¢:~":.........

/

-

.

: ' : : ; : i $ ~ ?

I

1

-2

IBASEMENT

?

I I ¢

PRE-EXISTING

BASIN

w

E

Fig. 4. Simplified cross-secion BB' (located in Fig. 2) along the northern end of the Scoresby Sund flood basalts. Based on extensive mapping (see L. M. Larsen et al. 1989 for references) and the seismic data shown in Fig. 8. The lowermost basalt unit may actually continue farther east, but as a very thin unit. Eastern basement constitutes the possible southward extension of the Liverpool Land basement ridge (see also Fig. 2). J, Jurassic; C, Cretaceous; Cp, upper Cretaceous to possibly Palaeocene. w e a k synclinal structure shown by the two com-

sion I transgressed widely over the western basement (Fig. 4). Succession II was derived from a source east of the present coastline, but also overflowed the basin, in this case towards the western boundary (L. M. Larsen et al. 1989; Fig. 4). Thus, there was a change in the location of the source region of around 100 km or more across the pre-existing basin between the two volcanic episodes. However, the general geochemistry does not suggest that a fundamental change in the primary magma source was associated with this displacement of eruption site (L. M. Larsen et al. 1989). The resultant 'depo-centre' of the two overlapping successions is close to the centre of the NW 0

bined successions, and this in turn is a southward continuation of the synclinal structure shown by the Jameson Land basin (Figs 2, 4 & 5). The synclinal structure of the basalt plateau is more pronounced at lower stratigraphic levels than at the upper levels. Moreover, the younger part of the lava pile seems to have spread over relatively larger areas and with relatively less thickness variation compared to the lower lavas. Since no major viscosity differences between the magmas of sucession I and II can be expected on the basis of their geochemical composition (L. M. Larsen et al. 1989), a change in 'basin geometry' is more likely.

JAMESON LAND

SE

. ~-~ ,, < - ~ . . ~ . , . _ ~ ~ ~ _ __ - ~ _ . ~ __~ •j >,.. , (;',,'," ~, - ~ . ~ . . . . . . .~. ~ ........ • " ~ - .--.".----.~--:-~---------~.~__-:~-.-~~x-, o

O

10

~

10

20-

A

~

~

o

~'~o

: . .,'2: : . "-.,:. M 2'0 km Basement

4'0 ~

oo

2O _,

gO Devonian

"

, , "

-,

k r.

" ,

80 ~

1;0 Carboniferous_ L. Permian

li0

li0 ~

1;0

JurassicU" Permian-

]Fig. 5. Generalized basin structure with Devonian to Carboniferous block faulting covered by undeformed late Permian and Mesozoic sediments. Cross section A A ' in Fig. 2. Based on H. C. Larsen etal. (1989); Marcussen & Lateen (1991) and unpublished exploration seismic data (see seismic track map in Fig. 10) M, Moho.

BASALT AND DEEP CRUSTAL STRUCTURE, EAST GREENLAND A subaerial depositional environment dominated throughout the entire period of flood basalt volcanism. Initial conditions close to sea level are indicated by underlying marine to shoreline facies sediments and some pillow breccias and hyaloclastites within the otherwise dominantly subaerial flow units; pillow breccias and hyaloclastites decrease in occurrence upwards and are not reported from the younger succession II. The pillow breccias and hyaloclastites have been interpreted as having formed within a low-land with numerous lakes and watercourses including possible marine embayments (L. M. Larsen etal. 1989; Henriksen 1986; W. S. Watt pers. comm.).

371

Considerable circumstantial evidence suggests that the plateau, now approximately 2000 m hi~h, was originally much lower, and that the present elevation is the result of later regional uplift of the East Greenland margin (Larsen 1990; Christiansen et al. 1992). In particular, the presence of succession I lavas (including hyaloclastites and pillow breccias) at around 1200 to 1800 m altitude clearly demonstrates uplift of this order (Figs 4 & 6). The lack of any significant erosional unconformity within or between the two lava successions I and II suggests that this uplift post-dated succession II. In addition, the local presence of marine lower Eocene sediments on top of succession II (Fig. 2) does not suggests that a huge regional, highly elevated plateau existed at that time, but corroborate post-eruption uplift. On the other hand, the restriction of overlying marine sediments to tectonic depressions does indicate some regional elevation of the lava plateau, perhaps in the order of a few hundred metres. In conclusion, volcanism started close to sealevel with depositional space being created by relative strong differential subsidence over the 'old' basin, passed through less differential subsidence and eventually started to build up a plateau perhaps several hundred metres high relative to sea-level.

Erosion o f flood basalts from the Jameson Land basin

Fig. 6. Typical incised plateau morphology within the Scoresby Sund flood basalt indicating limited 'vertical erosion'. The view is from the western part of profile B-B' (Figs 2 & 4) looking west over Ghseland. Height of plateau is up to 2000 m above sea level with the basalt to basement contact around 1000m. The basalts here belong to the lower succession I which locally include pillow lavas and hyaloclastites. Shore line facies sediments are locally found between basement and basalts east of this locallity. Thus a large postbasaltic uplift is strongly indicated. Note that the basalt plateau is similar or slightly higher than the old basement peneplain. The basalts overstepped the basement toward the west and north-west. (Copyright, Kort- og Matrikelstyrelsen, Denmark.)

Neogene erosional exposure of the partly late Paleogene plutonic and sub-volcanic complexes of northern Jameson Land and fission track dating of pre-basaltic sediments suggest.considerable late Tertiary uplift and accelerated erosion of the Jameson Land and Scoresby Sund region (Bergman & Henk 1990; H. C. Larsen 1990; Cristiansen et al. 1992). The southern shore of Scoresby Sund is a major erosional escarpment within the basalts, whereas the northern shore is a low-relief area of Mesozoic sediments dipping below the fiord and the basalts in the south (Fig. 7). Seismic and aeromagnetic data show that the basalts have been completely eroded away from the bottom of the fiord. Reconstruction of a N-S profile integrating reflection seismic data, shows that the basalts most probably covered most of the Jameson Land area to the north of Scoresby Sund, although with decreasing thickness (Fig. 7). It is suggested that undercutting by river erosion of the northern basalt palaeo-edge took place in Tertiary times. Regional uplift caused a southward migration of this erosional sensitive zone to its present position in Scoresby Sund, where it

372

H.C. LARSEN & C. MARCUSSEN C

C' BASALT I~SCARPMENT

3 km-

J

2

Z'c_--_--_-_

c

s

~':"~:

1

:.:;:f, ~.:,.'

~..~...~_. t,,~. _w.l~"

"~... %.~., ..:

"""" ' " ' : " : ' " ' : ° " ' : ' " ' " ' " ' " " ' " ' " " " ' " ' " : ' : ' ~ ' ~ ' ~ ' ~ % %

" ~ 0 ~ -2

kn

2

1

0

-3

T

-

':

-

"

~""'~""~"

0

"".~:'-:.::..'.::"-:.::::.:-:-:'~ :....:.:. _

?:ii

--

'~- ~ : . . . .

¢1.

JAMESON

LAND

N

,,a., S C O R E S B Y "'" SUND

,,j "'

S

Fig. 7. Simplified N-S cross-section C--C' (located in Fig. 2) from the flood basalts south of Scoresby Sund to northern Jameson Land. The flood basalts originally covered Scroesby Sund and most if not all of Jameson Land. Unless considerable northward thinning (depositional or erosional) of the Cretaceous sediments occurred, the flood basalts must have shown marked northward thinning. It is unlikely that they exceeded 1 km in thickness in central to northern Jameson Land and may only have been a few hundred metres thick. P, Permian; T, Triassic; C, Cretaeeous; Cp, upper Cretaceous to possibly Palaeocene. has been overdeepened by Quarternary glaciation (see Fig. 10 in H. C. Larsen 1990). The result of this erosional process is a southward progressive and 'lateral' erosion of the flood basalt plateau rathern than a 'vertical' erosion starting from the top of the plateau. This is in excellent agreement with the rather immature, pre-glacial erosional relief of the flood basalts and the widespread preservation of a 'near the original top plateau' (Fig. 6).

Stratigraphical relationship between basin and flood basalts The weak southward-plunging and weak synclinal structure of the Jameson Land basin is largely mimicked by the overlying flood basalts, suggesting a roughly concordant relationship between flood basalts and the underlying basin. An

E - W seismic profile from just north of the basalt escarpment is shown in Fig. 8A, which clearly shows the synclinal basin structure. Along the basalt escarpment the sediment/basalt contact varies from around sea level in the east to perhaps as much as 500 m water depth near the centre of the profile, the contact is not preserved at the seismic profile location within the fiord (Figs 4 & 8). The youngest seismic stratigraphic sequence present in the fiord (except for thin and erratic Quaternary sediments) is the sequence Cp (Fig. 8, Cp: Cretaceous to possibly Palaeocene). Sequence Cp shows onlap and infill into a pre-existing weak synclinal structure indicating local deepening of the basin around the 'old' basin axis and relative uplift of basin flanks prior to flood basalt volcanism. This sequence is also readily seen in a N-S seismic profile (Fig. 8B) showing northward-directed onlap of sequence Cp onto the southward-dipping basal

BASALT AND DEEP CRUSTAL STRUCTURE, EAST GREENLAND unconformity. In a broader context a reconstructed N-S profile also shows northward-directed onlap of the flood basalts onto the former Cretaceous rocks of Jameson Land (Fig. 7). Sequence Cp must be considerably younger than the early Lower Cretaceous rocks of southern Jameson Land which it overlies by as much as 700 m. The maximum thickness of Cp is around 500 m. A total of around 1.2 km of Cretaceous sediments below Scoresby Sund (approximately 20 m/Ma) would be consistent with the fairly slow average subsidence rate exhibited by the Jameson Land basin through midMesozoic time (approximately 25 m/Ma; Marcussen & Larsen 1991) and makes it possible that a fairly complete Cretaceous section is present in the fiord. However, only a very limited prebasaltic exposure of sediments at the southern Scoresby Sund shore more directly constrains the age of sequence Cp. It is located to the east of the profile in Fig. 8A and could predate or be part of the sequence Cp. It is tentatively dated as late Cretaceous with reworked lower Cretaceous fossils (S. Piasecki & H. N. Hansen pers. comm. 1991). Since younger deposits are likely to be present towards the basin centre in Fig. 8, also Palaeocene sediments may underlie the basalts in the Scoresby Sund, as they do in the Kangerdlugssuaq region further south and close to the hot-spot centre (Nielsen et al. 1981; N¢rgaard-Pedersen 1991; Hoch 1991). Thus continued late Mesozoic and pre-basaltic Subsidence took place around the basin axis, possibly with southward deepening of the basin. The unconformity underlying Cp shows only limited erosion but could reflect limited uplift in the order of 100-200 m in the late Cretaceous well ahead of break-up. The unfaulted nature of the Cretaceous to possibly Palaeocene sediments underlying Scoresby Sund and the flood basalts (Fig. 8) is remarkable and emphasizes the generally conformable relationship between the pre-existing basin and the flood basalts. However, because the Jameson Land basin synclinal structure and southward tilt were augmented somewhat prior to, or perhaps along with, the early volcanism, the fle~,~d basalts overstep older basin units toward the basin margins and toward the north.

Deep crustal structure The White & McKenzie (1989) model for volcanic margin and flood basalt formation implies significant crustal and lithospheric thinning over the plume head in order to create the excessive amounts of magmatic melts observed and to ease their access to the surface. White &

373

McKenzie (1989) further suggested that this process preferentially develops as a reactivation of pre-existing rift basins with a favourable setting in relation to the plume head and the regional extensional stress regime. Evidence from the Jameson Land basin is now considered in this context.

Amount and age o f crustal thinning The interpretation of the deep reflection seismic data from the Jameson Land basin (Fig. 5) shows that a very thick sedimentary basin overlies a locally very thin crystalline crust (H. C. Larsen et al. 1989; Marcussen & Larsen 1991). Interpretation of seismic refraction data from the Scoresby Sund region strongly confirms the reflection seismic image (J. Hepper 1991; E. Fifth pers. comm. 1991). These studies both show that the total crustal thickness is around 30 km along the basin margins and decreases to 22 to 24 km at the basin centre. Since there may be as much as 16 to 18 km of basin fill at the basin centre, intense crustal thinning with the maximum thinning factor around four must have taken place (Fig. 9). However, on the basis of geological mapping, the main rifting and faulting in Jameson Land is evidently of pre-Late Permian age (Surlyk et al. 1986; Surlyk 1990), and the reflection seismic survey over the basin strongly suggests that the main faulting and crustal stretching took place during mid-Devonian to early Carboniferous time (Figs 5 & 10; H. C. Larsen etal. 1989; Marcussen & Larsen 1991). The crustal stretching thus is 200-250 Ma too old to be related to the break-up volcanism in the early Tertiary. Two possible explanations for this enigmatic relationship offer themselves; (a) The basin was flooded only passively by basalts and did not play any role in guiding magmas to the surface; or (b) crustal and lithospheric thinning not involving the upper crust took place. These two possibilities are examined in the following sections.

Emplacement o f sills: magma conduits and possible crustal magma chambers The concentration of intrusive rocks (sills) seen at the surface within the Jameson Land basin make it likely that the basin guided the magmatic melts to the surface, at least partially. The relatively thin sills we can identify in outcrop can be discerned in the seismic data only with difficulty, due to lack of seismic resolution of the tops and the bottoms and the low brightness of the seismic reflections from sills at near outcrop location. However, wherever sills are

~'"

100

I

~'~.---t~x...:'.

. 6.7 km/s, Batzle & Landisman 1988). Apparent sill thicknesses at depth is thus up to approximately 300 m as opposed to the typical 10-20 m thickness at outcrop level (in central to northern Jameson Land up to 50-75 m). However, several factors can affect the apparent thickness of sills. Firstly, the time-varying filtering applied during processing of the seismic line (largely designed to follow the natural depth-varying filtering in the subsurface) reduces resolution with depth and can result in apparently larger thicknesses at depth. Secondly, the thickness of individual sills at depth may be overestimated if they have a complex layered structure resulting in a complex polarity relationship at the tops and the bottoms of the sills. Clearly, if multiple sill intrusion at the same stratigraphic level, perhaps with inclusions of thin layers of sediments, or complex deformation zones around the sill margins occurred, such complexities would not be uniquely resolved by

ds.

d,"

30km

~;'~

375

b'5 ,r, ,

~ 2

.5 km

/

i

i

IX ,

"~

,'

i ' ~ . 5

215=

!>3o

,~

! km(/i '~< km

24 °

Fig. 9. Crustal thinning factor below central and southern Jameson Land. Prepared from the 6 and 12 sec. reflection seismic data onshore and refraction seismic data in the fiord (Hepper 1991; E. Fliih pers. comm. Kiel 1991). The thinning factor is calculated from the directly observed total crustal thickness minus the mapped thickness of basin fill. An original minimum crustal thickness of 30 km is assumed. Hatched area: crustal thinning factor > 4. the seismic data. Such complexities are not seen, however, in the Jameson Land region (G. Dam & N. Hald pers. comm. 1992), but could nevertheless occur at depth. Regarding the question of natural filtering and narrowing of the frequency content of the seismic pulse with depth, comparative analysis of the broad-band processed 6 s exploration data and the narrowband processed 12 s data (Table 1) suggest recovery of at least 40 Hz energy down to 4 s T r W (approximately 11 kin). Assuming a dominant frequency of say 30 Hz at 3 s TWT (Fig. 14) and a P-wave velocity of 6.6 km/s, one-quarter of a wave-length resolution corresponds to a bed thickness of 55 m, well below the 200-300 m under discussion.

Fig. 8. Seismic sections (unmigrated MCS data) of the Scoresby Sund fiord. The sequence Cp (see also Figs 2, 4 & 7) is the youngest sequence in the Scoresby Sund below the basalts (Quaternary deposits are very sparse in Scoresby Sund). Note the absence of post-Jurassic faulting. The pre-upper Jurassic section is very noisy and difficult to interpret. Some minor faulting is indicated. Star indicates exposure of the marine upper Cretaceous below the basalts.

;P

W

i

200

i

LINE J L 8 8 - 0 4 D

3 km

I

3P 6 0 0 |

JL 8 7 - 0 1 D ~P 1 0 0 0 J

I

JL 8 8 - 1 3 D

I

.

~P 1400

BASALT AND DEEP CRUSTAL STRUCTURE, EAST GREENLAND The deep sill reflections from Jameson Land are compared in Fig. 14 with drilling constrained, deep sill reflections (broad-band data) from the Siljan area in central Sweden and clearly show the relatively large seismic thickness of the deep Jameson Land sills. A similar large seismic thickness of sills is seen within the Faeroe-Shetland sill complex where drilling recovered intrusives with thicknesses from 2 to 340 m (Gibb & Kanaris-Sotiriou 1988). In summary we conclude that the deep, strong, discordant to semi-concordant reflectors in Jameson Land are caused by thick sill complexes which towards the basin centre and at depth may attain thicknesses up to 300 m. These could be complex systems or single sills similar to those exposed elsewhere in the region. Using their mapped minimum areal extent, the present volume at depth of one of the major sill complexes may be up to 3000 kin3; around three, deep-lying and thick, sill systems can be discerned at 10-15 km depth providing an integrated sill volume potentially as large as c. 104 km 3 at mid-crustal level below Jameson Land. About three to five additional, thinner and more shallow sill systems are present. Total integrated thickness of sills in Jameson Land thus may be up to about 10% (or roughly 1.5 km) of the total basin fill and is comparable to the thickness of the extrusive cover. The presence and geometry of large sill systems at large depth is interesting not only in terms of a magmatic pathway through the crust. The deep sills may in fact also have functioned as mid-crustal magma chambers. They potentially have the volume and the depth envisaged by L. M. Larsen et al. (1989) on the basis of the lowpressure fractionation and individual flow volumes seen within the flood basalts. Likewise, if the sill systems acted as the main magmatic pathway, the sill-geometry within the basin would tend to favour eruption sites along the basin margins rather than at the basin centre, and would in fact provide a potential mechanism for lateral migration of eruption across and along the basin without a significant change in the location of the primary magma source and magma plumbing system. Our geophysical data do not provide constraints on the location and structure of the man-

377

tle to mid-crustal magma pathway, and hence, cannot demonstrate whether or not large scale, mid-crustal lateral injection of the suggested order (170 k m + ) along the basin occurred. Likewise, the limited geochemical data from the Jameson Land intrusives cannot distinguish whether or not several primary magma sources were involved. However, in general the geochemical data from the flood basalts to the south of Jameson Land would support involvement of different magmatic mantle reservoirs along strike and radially away from the hot-spot centre, although occasionally magmas with a clear southern (i.e. plume proximal) character have been erupted far to the north and an unknown magma conduit has been hypothesized in order to explain these observations (L. M. Larsen et al. 1989). A recent study of the regional geochemical variations of the Paran~i flood basalts and the underlying sill system also demonstrates the presence of coherent and regional variations (Peate et al. 1990) indicating multiple mantle sources rather than large lateral flow of the same magma body away from a common centre. In conclusion, the Jameson Land basin seems to have played a major role in guiding magmas to the surface. The deep sill systems we have observed within the basin both volumetrically, structurally and in distribution provide perfect possibilities for mid-crustal magma chamber reservoirs, magma ascent from mid-crustal level to the surface, and for large scale injection of magma along and across strike of the pre-existing rift structure. One consequence of the model would be a concentration of eruptions along the basin margins fed by sills reaching the surface in these areas. The Jameson Land basin could be interpreted as an example of a 'thin-spot' effect proposed in the model by Thompson & Gibson (1991).

Did thinning take place in the lower crust? The lower crust below the western half of the Jameson Land basin shows a distinct horizontal to sub-horizontal seismic layering (Fig. 12), a feature that is also observed in many other deep seismic studies. Of particular interest is a recent deep reflection seismic study of the Scottish and

Fig. 10. Seismic example (unmigrated data) of deep sill system stepping up toward the south-eastern basin margin. The sills can be followed seismicallyalmost to outcrop along the eastern basin margin (see Fig. 3), but are seismicallybrighter and more distinct towards the basin centre and at depth. The outcrop controlled MesozoicPermian boundary (M/P) and the inferred Permian-Carboniferous boundary (P/C) are shown. The early basin nil (Devonian?) is exposed along the eastern basin margin (see Figs 2 & 5), where it is highly faulted (Co¢ 1971). The position of basement is constructed on the basis of all available data, but cannot be precisely defined.

378

H.C. LARSEN & C. MARCUSSEN km 0

10

20

30

40-

40

~" "~ TOP OF TERTIARY ~ MANTLE MAGMA RESERVOIR OR A LATE PALEOZOIC STRUCTURE ?

5O

50-

W o

E '

'

'

4'5

'

/o

'

¢5

90 km

Fig. 11. Generalized crustal cross-section of the Jameson Land basin showing basin structure and simplified sill geometry. Note how sills mimicbasin structure and step up the stratigraphy toward basin margins. This pattern, although in a slightly larger scale, is very similarto that described by Francis (1982) from the Carboniferous basin of northern Britain.

Irish Caledonides (Klemperer et al. 1991); an area originally close to our study area. It shows a very similar lower crustal layering and a crustal thickness almost identical (around 30 kin) to that below the basin margins in Jameson Land. There is no generally agreed interpretation of this type of lower crustal layering, and the question in Jameson Land is whether it could represent lower crustal deformation associated with early Tertiary break-up. Deformation could be either thinning through, say sub-horizontal shearing or conversely it could be crustal underplating of the sort suggested to take place at volcanic margins (White et al. 1987). However, the lower crustal layering in Jameson Land is only developed below the western (lower plate) part of the basin and not below the eastern part (Fig. 12 and H. C. Larsen et al. 1989). Similarly the Moho reflection is much more clearly developed in the west. The simple-

shear, lower and upper plate model suggested for the late Palaeozoic basin formation easily explains this asymmetry (Fig. 11) provided that the lower crustal layering existed prior to completion of the simple-shear stretching. A mode of origin and asymmetrical distribution of the lower crustal layering after basin formation, however, seems most difficult to explain, and furthermore would not explain the wide regional occurrence of lower crustal layering outside volcanic rifted margins. For this and for other reasons (seismic velocities, general crustal thickness considerations, apparent preservation of simple-shear detachment in the lower crust/ upper mantle and basin history) we do not believe that any regionally significant lower crustal thinning or thickening took place since the late Palaeozoic, apart from that provided by the progressive filling of the basin. On the basis of the apparent continuation (and preservation) of

Fig. 12. Deep seismic line (unmigrated 12 s vibroseis data) from the western basin margin to the basin centre. The general basin structure is synthesized in Figs 5 and 11. The main detacthment outcrops west of profile (P. H. Larsen 1988). Direct stratigraphical outcrop control exists down to the lower Permian. Supposed Devonian sediments are found on basement along the opposite (eastern) margin (Figs 2 & 5). Interpretation based on all data transferred to this profile. Note the mid-crustal and lower crustal layering and their relationship to the main detachment. This kind of crustal layering concordant to a well developed Moho is not found within the eastern part of the basin, strongly indicating a fundamental crustal asymmetry. Main detachment seems to extend into the upper mantle (Fig. 13). S, sills; B, basement; UP, Upper Permian; LP, Lower Permain.

n

NW

LINE JL 8 6 - 0 5 V

JL 8 ~ - IOV

30

J, 88,;o8v

4o

JLgt-o6v

~0 k m

0

30 km i

SE

m ~n Z > Z

>

C

>

C

> Z

> >

r2

10

2"

:~/C

~ec. 0-

.[

JL 8 8 - 0 8 V

SP 2750

NW

LINE JL 8 7 - 0 1 D

~P 2500

J

~P 2250

6 km .LsP 2000

JL 8 6 - 0 6 V 3P 1750

J.

sp 15oo

IL 8 8 - 0 4 D ~P 1250

SE

1.0

17

SP 1000 J, Cm 3

JL 8 7 - 2 0 D

BASALT AND DEEP CRUSTAL STRUCTURE, EAST GREENLAND the main detachment into the upper mantle (Figs 11 & 13) we also hypothesize that this uppermost part of the mantle lithosphere has in general maintained its structure since late Palaeozoic times.

Mantle reflections Local Tertiary deformation of the upper mantle lithosphere and the lowermost crust directly below the original basin centre cannot be completely excluded. The lower crustal layering is locally disturbed in this region and strong mantle reflections are seen from 10 to 15 km sub-crustal depths (35-40 km subsurface; Figs 11 & 13). Theres is no specific evidence in favour of these features being of Tertiary age rather than of the late Palaeozoic age, but it could be hypothesized that magma ascended sub-vertically from mantle reservoirs at approximately 35-40 km depth through the 'old' basin centre, from where it established the mid-crustal sill systems.

Discussion We will limit our discussion to such direct or inferred observations that are particularly pertinent to plume models for volcanic rifted margins.

381

Two possibilities for such lithospheric weakening might be considered consistent with our deep crustal data: (a) mechanical erosion and thinning of the lower continental lithosphere by say lateral flow within the supposed plume head; (b) pure thermal thinning of the continental lithosphere by the excessively hot plume head. If (a) occurred with any significance it would be likely that (b) also occurred. In terms of surface manifestations, pure type (b) lithospheric thinning would tend to create very significant uplift, exceeding the 1-2 km predicted by White & McKenzie (1989), since in this case, there would be no subsidence related to a mechanical thinning of the lithosphere to counteract the thermal expansion and uplift of the lithosphere. Likewise a similar strong postbasaltic subsidence should in this case be expected following the removal of plume support and thermal relaxation. If process (a) acted on its own, the surface uplift within a specific area would depend much on whether there was a net flux of mantle material laterally away from below this area. If so, and if the net flux of mantle material was fast compared to thermal transmission, very reduced uplift or even subsidence might occur. Subsidence would probably tend to occur early in the process, before substantial heating of the continental lithosphere took place.

Thinning of the lithosphere Our data suggest that the pre-existing Jameson Land basin acted as a crustal pathwa.y for the subcrustally formed magmatic melts during break-up. However, we can exclude any significant tectonic stretching of the crust and possibly also of the uppermost mantle below the Jameson Land basin during the period of flood basalt volcanism. On the other hand we do see indications for a reduction of the mechanical thickness of the lithosphere in the form of a lowered flexuraI strength of the lithosphere prior to and during volcanism. Our present data do not allow for more precise quantification of this apparent lithosphere weakening, but the deformation associated with the deposition of the lower lava succession I indicates significant weakening and indeed high strain rates.

Uplift and subsidence Since uplift is naturally associated with erosion, it is often difficult to demonstrate and in particular to date. Accordingly some of the most striking evidence for regional uplift of flood basalt regions such as the Deccan, Karroo and Paran~i, is the (palaeo-) river drainage pattern (Cox 1989). The Tertiary river drainage pattern of East Greenland has been strongly modified by glacial erosion, overdeepening and possible reversal of flow direction. It has been used, however, for construction of the erosional history of the northern part o f the central East Greenland flood basalt area (see H. C. Larsen 1990, fig. 10). From regional field geological relationships and seismic data H. C. Larsen (1990) constructed a regional uplift and subsidence map of

Fig. 13. Deep seismic profile (unmigrated 12 s dynamite data) over the centre of the Jameson Land basin. Strong reflections from the upper mantle occur below the basin centre and are not observed outside this area with the present data base. The Moho configuration is verified by four tie lines marked with arrows on top of section (part of line JL88-04D is shown in Fig. 10). The centre of the profile is over the basin centre. Note how sills even at shallow levels step up toward basin margins from the basin centre. This profile is the only one showing strong upper mantle reflections. The position of basement and geometry of fault blocks is poorly constrained on this line; the interpretations shown are transferred from the systematic grid of 6 s migrated data and are only approximate.

11

km

tV

I

2 km

m

E

~ec.

B

S

i

=

N

Section expanded at drill s i t e G r a v b e r g - 1 at Siljan, Sweden

2 km

Z

¢3

Z R=

t11

=:

BASALT AND DEEP CRUSTAL STRUCTURE, EAST GREENLAND the East Greenland margin following pioneering work by Brooks (1979). The general pattern is of regional coastal uplift in the order of 1.5 to 2.5 km coupled with similar, or locally much larger, subsidence off the coast. This amount of margin uplift is similar to those reported by Cox (1989) and interpreted by White & McKenzie (1989) as initial plume uplift. However, the regional vertical crustal movements along the East Greenland margin mainly are post-eruptive. The fairly unique setting and database around Scoresby Sund and the Jameson Land basin provide a possibility to describe the vertical crustal movement history in this region in further detail. From the seismic and field geological data presented above and from the fission track and thermal maturity data presented by Christiansen et al. (1992) and Bergman & Henk (1990), we can recognize the following significant events within the region: (1) a possible limited uplift up to about a few hundered metres about 30 to 10 Ma prior to volcanism; (2) continued subsidence and sedimentation at an average rate around 20 m/Ma at the basin centre and towards the supposed plume centre following (1); (3) start of volcanism close to sea level; (4) rapid, strong and differential subsidence (around 1-2 km in 1-2 Ma) of the base of volcanic succession I maintaining volcanic accretion close to sea level; (5) perhaps less strong and less differential subsidence during deposition of volcanic succession II (0.5-1 km?) allowing volcanic accretion to rise somewhat above sea level; (6) faulting along the coast and marine deposition within tectonic lows; (7) regional uplift of the order of 1-2 km largely post-dating volcanism with about 20-30 Ma. The limitations of this data set are: (a) the restricted observation window for a process that is expected to involve very large regions; (b) lack of resolution. Concerning (a) we are partly helped by recent work from the area in the south, close to the supposed plume centre (Fig. 1). A late Early Cretaceous to Palaeocene marine section is present here below the first volcanics and shallow marine tO coastal Palaeocene sediments, and have been reported to have an

383

interdigitating relationship with the early volcanism (Higgins & Soper 1981; Nielsen et al. 1981; Nergaard-Pedersen 1991). The late Cretaceous to Palaeocene deposits have been uplifted to 1400-1650 m altitude (Hoch 1991). These observations, at least in the N-S direction, expand our observation window to more than 300 km and strongly re-affirm the existence of a significant post-basaltic uplift. It is particularly remarkable that the marine upper Cretaceous to Palaeocene was deposited and preserved below the basalts at the supposed hot spot centre (Fig. 1).

Consistency with p l u m e models In spite of our limited observation window compared to the size of the supposed plume head, the obvious and important inconsistencies between the observation from the central East Greenland flood basalt region and various plume models are too marked to be completely neglected as local variations. The inconsistencies primarily pertain to the lack of crustal stretching in the rift basin guiding magma to the surface and, compared to model predictions, to the largely reversed uplift-subsidence history within the region. If the magmas originated from below the basin, the lack of crustal stretching in connection with the magmatism would be in direct conflict with the plume model of White & McKenzie (1989) which requires significant crustal and subcrustal lithospheric stretching in order to generate large amounts of magmatic melts. However, since lateral injection from the supposed plume centre cannot be excluded, a modified model with large amounts of melt generation concentrated near the plume centre could be maintained. Alternatively, a plume model with higher temperature and not requiring lithospheric stretching in order to generate large amounts of melts could be inferred (Duncan & Richards 1991; Griffiths & Campbell 1990). A similar suggestion was made by Hooper (1990) for the Columbia River, the Paran~i and the Deccan flood basalts, but was refuted by Peate et al. (1990) for the Paran~i basalts.

Fig. 14. (A) Detailed seismic view of sills at various depths. Note that top and bottom of the sills apparently can be distinguished. With a seismic velocity of 6.6 km/s a sill thickness around 300 m is suggested. Since three, deep and thick sills can be distinguished, total integrated thickness of the deep sills may be close to 1 km. The more shallow sillsmay add up to around 400 m making total intregrated sill thickness between 1 and 1.5 km. This estimate applies to large parts of the basin. Migrated vibroseis data (See text for discussion of seismic resolution). (B) Seismic and drilling data from the Gravberg well, Siljan area, Sweden for comparison (note: same time scale). Sills (heavy lines) between 4 and 60 m thick were penetrated by the well. The host rock (granite) was found strongly fractured around the sills and may have effected a broader seismic appearance of the sill structure with apparent negative (black) reflections from above and below the actual sill itself. From Juhlin (1988).

384

H.C. LARSEN & C. MARCUSSEN

The vertical tectonic history associated with central East Greenland flood basalts seems to be in enigmatic contrast with most published plume models. If we accept the hypothesis proposed above, however, that lateral and radial flow within the plume head and away from the plume centre takes place with the effect of a small net flux of mantle material away from the central parts of the plume, the observed sequence of events actually makes sense. We briefly review such a hypothetical development in the following: (1) The plume head impinges on the base of the continental lithosphere and effects a small dynamic uplift prior to any mechanical or thermal change of the continental lithosphere and prior to flood basalt volcanism; (2) Mechanical erosion and thinning of the lower continental lithosphere starts and initially dominates over the thermal influence on the lithosphere, leading to net subsidence; (3) The combined mechanical and thermal erosion of the lithosphere lead to initial plume volcanism preferrably through zones of pre-existing weakness such as older rift structures, and with volcanic accretion balanced by subsidence; (4) Heating and, hence, thermal erosion of the continental lithosphere accelerates, helped by thermal convection associated with the rise of magmatic melts so that volcanic accretion may exceed the subsidence of the volcanic basin. While these four hypothetical stages would all be consistent with our observations, the late and significant uplift we see in East Greenland is not explained; nor is it explained by other plume models. In general an initial thermal uplift should be followed by subsidence after say 1520 Ma and continuing for another 20-40 Ma, quite different from the sequence of events actually observed in East Greenland. Uplift of other volcanic margins (Cox 1989) is not well dated and could in fact often post-date rifting in a similar fashion to the Northeast Atlantic margins. What the uplift of older volcanic margins shows, however, is that the uplift, once established, is maintained for very long geological time, far exceeding normal cooling periods. The East Greenland uplift is so young that it is hard to tell whether the uplift is 'permanent' like other volcanic rifted-margin uplifts. Its delayed appearance in relation to the initial thermal pulse suggests that it is non-thermal, and hence, might have a similar origin and duration as other long-lived volcanic margin uplifts. The origin of such late margin uplift is poorly understood. Flexural margin uplift caused by margin erosion has been recently proposed (Gilchrist & Summerfield 1990). Another cause of volcanic-margin uplift may be found within

some sort of flexural coupling between the subsiding oceanic lithosphere and the continental margin. In this case margin uplift could be considered a kind of peripheral-bulge uplift. Both processes may well act together and they both will favour a 'permanent' uplift post-dating break-up. It is conceivable that ridge-push stress could further amplify flexural deformation across rifted margins.

Conclusion Our findings stress the importance of the preexisting inhomogeneity within the continental lithosphere in controlling the surface manifestations of hot-mantle plume material impinging onto the base of the lithosphere. We further conclude that there is a need for modified plume models. We favour modifications involving active participation by the plume head in a mechanical thinning of the lower lithosphere in addition to the thermal thinning thereby reducing the need for large initial uplift and reducing the need for lithospheric thinning to involve crustal thinning. We would further suggest post break-up, flexural margin uplift in response to the waning effect of the plume and associated subsidence of the spreading centre as an instigator for the late, regional margin uplift observed. Such flexural response is likely to be further amplified by margin erosion and ridge-push stress. We thank Atlantic Richfield Exploration Company, exploration manager D. Ehman and programme director B. Lanni in particular, for continued support for the deep crustal sounding study carried out in parallel with the oil exploration seismic programme. We thank our colleagues within the Geological Survey of Greenland for numerous discussions on the Jameson Land basin geology, and we thank our colleagues S. Watt, L. M. Larsen and T. Dahi-Jensen for providing discussions on important field geological, geophysical and geochemical data pertaining to the flood basalt volcanism. J. Halskov and J. Lautrop are thanked for technical assistance. We thank D. Smythe, R. Pankhunt, L. M. Larsen and one anonymous reviewer for very useful comments on an earlier version of this paper. We thank the Danish energy research programme for financial support of the deep seismic study (grant no. 1313/88-3). The paper is published with permission of the Geological Survey of Greenland.

References ANDERSEN, M. S., LARSEN,H. C., Risu~a, J. B. & THOaNmO, L. 1981. Geophysical investigations offshore East Greenland---Project NAD. Report of Activities. 1980 Rapport GrCnlands Geologiske UndersCgelse, 105, 56-60.

BASALT AND DEEP CRUSTAL STRUCTURE, EAST GREENLAND BATZLE,M. & LANDISMAN,C. 1988. Greenland sample velocities. Unpublished exploration report by ARCO International. BERGMAN,S. C. & HENK,F. H. B. 1990. K-Ar and fission track thermo chronology data on sedimentary and igneous rocks in Jameson Land, East Greenland: Importance of Eocene magmatism on the thermal evolution of the basin. Unpublished exploration report by ARCO International. BROOKS, C. K. 1979. Geomorphological observations at Kangerdluggssuaq, East Greenland. Meddelelser on Grenland. Geoscience, 1, 1-21. CHRISTIANSEN,F. G., LARSEN,H. C., I~L~RCUSSEN,C., HANSEN, K., KRABBE, H., LARSEN, L. M., PIASECKI, S., STEMMERIK,L. & WATT, W. S. 1992. Uplift study of the Jameson Land basin, East Greenland. Norsk Geologisk Tidsskrifl, 72, 291-294. Co~, K. 1971. Faulting in the western part of Liverpool Land, East Greenland. Bulletin Geological Society of Denmark, 20, 260-264. Cox, K. G. 1989. The role of mantle plumes in the development of continental drainage patterns. Nature, 342, 873-877. DUNCAN, R. A. & RICSARDS,M. A. 1991. Hotspots, mantle plumes, flood basalts and true polar wander. Reviews of Geophysics, 29, 31-50. ELDHOLM, O., TmEDE, J. & TAYLOR,E. (eds) 1989. Proceedings, Scientzfic results, Leg 104. College Stations, Texas, Ocean Drilling Program. FRANCIS, E. H. 1982. Magma and sediment--l. Emplacement mechanism of late Carboniferous tholefite sills in northern Britain. Journal of the Geological Society, London, 139, 1-20. FORSBERO, R. 1986. Gravity measurements in Jameson Land and neighbouring parts of East Greenland. Meddelelser om Gr¢nland. Geoscience, 15, 1-21. FULLER, M. 1986. Paleomagnetism of samples from Greenland. Unpublished exploration report by ARCO International. GIBB, F. G. F. & KANARIS-SOTIRIOU,R. 1988. The geochemistry and origin of the Faeroe-Shetland sill complex. In: MORTON,A. C. & PARSON, L. M. (eds) Early Tertiary Volcanism and the Opening of the NE Atlantic. Geological Society, London, Special Publication, 39, 241-252. Gn.CHmST, A. R. & SUMMERFIELD,M. A. 1990. Differential denudation and flexural isostasy in formation of rifted-margin upwarps. Nature, 346, 739-742. GRIFFITHS, R. W. & CAMPBELL,I. H. 1990. Stirring and structure in mantle starting plumes. Earth and Planetary Science Letters, 99, 66-78. HALL~R, J. 1970. Tectonic map of East Greenland (1:500,000). An account of,tectonism, plutonism, and volcanism in East Greenland. Meddelelser om Grenland, 171, 5. HARLAND,W. B., ARMSTRONG,R. L., Cox, A. V., CRAIG,L. E., SMITH,A. G. & SMrrH, D. O. 1989. A geological time scale 1989. Cambridge University Press, Cambridge. HEPPER, J. 1991. Refraktionsseismische Untersuchungen im Scoresby Sund, Ostgr6nland. Thesis, University of Kiel.

385

HENRIKSEN,N. 1986. Descriptive text to 1:500 000 sheet 12, Scoresby Sund. Grenlands Geologiske Undersegelse, Copenhagen. HU.L, R. I. 1991. Starting plumes and continental break-up. Earth and Planetary Science Letters, 104, 398-416. I-hoor~s, A. C. & SOPER, N. J. 1981. CretaceousPalaeogene sub-basaltic and intrabasaltic sediments of the Kangerdlugssuaq area, Central East Greenland. Geological Magazine, 188, 337-354. HocH, E. 1991. Palaeontological data on life in the intra-cratonic Kangerdlugssuaq basin. In: BROOKS, C. K. & STAERMOSE, T. (eds) Kangerdlussuaq Studies: Processes at a Rifled Continental Margin. Geological Institute, University of Copenhagen, 45-52. HOOPER, P. R. 1990. The timing of crustal extension and the eruption of continental flood basalts. Nature, 345, 246-249. JUHLnq, C. 1988. Interpretation of the seismic reflectors in the Gravberg-1 well. In: BODES, A. & EgIKSSON,K. G. (eds) Deep Drilling in crystalline bedrock, 1, Springer, Berlin, 113-121. KLEMPERER, S. L., RYAN, P. D. & SW/DER, D. B. 1991. A deep seismic reflection transect across the Irish Caledonides. Journal of the Geological Society, London, 148, 149-164. LARSEN,B. 1980. A marine geophysical survey of the East Greenlafid continental shelf between latitudes 60° and 70°N--Project DANA 79. Report of Activities, 1979. Rapport Grenlands Geologiske Undersegelse, 100, 94-98. LARS~N, H. C. 1975. Aeromagnetic investigations in East Greenland. Report of Activities, 1974. Rapport Grenlands Geologiske Undersegeise, 75, 88-91. 1988. A multiple and propagating rift model for the NE Atlantic. In: MORTON,A. C. & PARSON, L. M. (eds) Early Tertiary Volcanism and the Opening of the NE Atlantic. Geological Society, London, Special Publication, 39, 15%158. 1990. The East Greenland Shelf. In: GRAmZ, A., JOHNSON, L. & SWEENE¥,J. F. (eds) The Arctic Ocean region. The Geology of North America, L. Geological Society of America, 185-210. & JAKOnSDCYrnR,S. 1988. Distribution, crustal properties and significance of seawards-dipping sub-basement reflectors off E Greenland. In: MORTON, A. C. & PARSON, L. M. (eds) Early Tertiary Volcanics and the Opening of the NE Atlantic. Geological Society, London, Special Publication, 39, 95-114. , ARMSTRONG,G., MARCUSSEN,C., MOORE, S. & STEMMEmg, L. 1989. Deep seismic data from the Jameson Land basin, East Greenland Terra Abstracts, 1, 10-11. , MARCUSSEN, C. & THORNING, L. 1988. Aeromagnetic anomaly map of Greenland, shelf,4 (69°-74°N, 19°-27°W, scale 1:1,000,000). Copenhagen Gr~nlands Geologiske Undersegelse. LARSEN, L. M. & WATT, W. S. 1985. Episodic volcanism during break-up of the North .~tlantic: evidence from the East Greenland plateau basalts. Earth and Planetary Science Letters, 73, 105-116.

386

H. C. LARSEN & C. MARCUSSEN

, PEDe~SEN, A. K., PEDERSEN,G. K. & PIASECKI, S. 1992. Timing and duration of Early Tertiary volcanism in the North Atlantic: new evidence from West Greenland. This volume. , WATr, W. S. & WATT, M. 1989. Geology and petrology of the Lower Tertiary plateau basalts of the Seoresby Sund region, East Greenland. Bulletin Grcnland Geologiske Underscgelse, 157, 1-164. LARSEN, P.-H. 1988. Relay structures in a Lower Permian basement-involved extension system, East Greenland. Journal o.fStructural Geology, 10, 38. , STEMMeR]K,L., NIELSEN,T. F. D. & RF.X,D. C. 1990. Lamprophyric dykes in Revdal, Scoresby Land, East Greenland: conflicting field observations and K-Ar determinations. Bulletin of the Geological Society of Denmark, 38, 1-9. MARCUSSEN,C. & LARSEN,H. C. 1991. Project 'DYBSE1S' : Deep seismic studies in the Jameson Land basin. Unpublished final report. GrCnlands Geologiske UndersCgelse. Nn~LSEN,T. D. F. 1987. Tertiary alkaline magmatism in East Greenland: a review. In: FrrroN, J. G. & UPTON, B. G. J. (eds) Alkaline Igneous Rocks. Geology Society, London, Special Publication, 30, 489-515. , SOPER, N. J., BROOKS,C. K., FALLER, A. M., HIGGINS, A. C. & MATrHEWS, D. W. 1981. The pre-basaltic sediments and the lower lavus at Kangerdlugssuaq East Greenland. Their stratigraphy, lithology, palaeomagnetism and petrology. Meddelelser om Grenland, Geoscience, 6. NOBLE, R. H., MAc.~rYRE, R. M. & BROWN, P. E. 1988. Age constraints on Atlantic evolution: timing of magmatic activity along the E Greenland continental margin In: MORTON, A. C. & PARSON,L. M. (eds) Early Tertiary Volcanism and the Opening of the NE Atlantic. Geological Society, London, Special Publication, 39, 201-214. NOE-NYGAARD, A. 1976. Tertiary igneous rocks between Shannon and Scoresby Sund East Greenland. In: ESCrmR, A. & WArr, W. S. (eds) Geology of Greenland. GrCnlands Geologiske Under~gelse, Copenhagen, 386-402. Nt~3AARD-I~DmtSEN, N. 1991. A Sedimentological approach to the Paleocene Coastal Environments at Kulh¢je, East Greenland. In: BROOKS, C. K. & ST~RMOSE, J. (eds) Kangerdlussuaq Studies: Processes at a Rifted Continental Margin. Geological Institute, Unversity of Copenhagen, 53-58.

-

l~A'm, D. W., HAWKESWORTH,C. J., MANTOVANI,M. S. M. & SHUKOWSKI,W. 1990. Mantle plumes and flood-basalt stratigraphy in the Parana, South America. Geology, 18, 1223-1226. Rex, D. C., GLEDHILL, A. R., BROOKS', C. K. & Saa~ENrELT, A. 1979. Radiometric ages of Tertiary salic intrusions near Kong Oscar Fjord, East Greenland. Rapport GrCnlands Geologiske Underscgelse, 95, 106--109. ROBERTS, D., COFFIN, M. F., CRANE, K., ELDHOLM, O., HARRY, D., LARSEN,H. C., McNuTr, M., OKAY, N., PEDERSEN, T., SKOGSEID, J. 8£ TUCHOLKE,B. 1991: Conjugate volcanic passive margin and oceanic plateau development. In: COFr'aN, M. F. & ELOHOLM, O. (eds) Large igneous provinces. JOIIUSSAC Workshop Report. The University of Texas at Austin, Institute for Geophysics, Technical Report, 114, 29-46. SO~R, N. J., HmGms, A. C., DowNm, C., MArmEWS, D. W. & BROW~, P. E. 1976. Late Cretaceous-early Tertiary stratigraphy of the Kangerdlugssuaq area, East Greenland, and the age of opening of the north-east Atlantic. Journal of the Geological Society, London, 132, 85-104. SUnLW, F. 1990. Timing, style and sedimentary evolution of Late Palaeozoic---Mesozoic extensional basins of East Greenland. In: HAnDMAN,R. F. P. & BROOKS,J. (eds) Tectonic Events Responsible for Britain's Oil and Gas Reserves. Geological Society, London, Special Publication, 55, 107125. , HURST,J. M., PIASECKI,S., ROLLE,F., SCHOLLE, P. A., STEMMERIK,L. & THOMSEN,E. 1986. The Permian of the western margin of the Greenland Sea--A future exploration target. In: HALBOUTY,M. T. (ed.) Future petroleum provinces of the world. American Associations of Petroleum Geologists Memoir, 40, 629-659. TALWAm, M., & ELDHOLM,O. 1977. Evolution of the Norwegian-Greenland Sea. Geological Society of America Bulletin, 88, 969-999. THOMPSON,R. N. & GmsoN, S. A. 1991. Subcontinental mantle plumes, hotspots and pre-existing thinspots. Journal of the Geological Society, London, 148, 973-978. Wrna~, R. S. & MCI~Nzm, D. 1989. Magmatism at Rift Zones: The generation of volcanic continental margins and flood basalts. Journal of Geophysical Research, B94, 7685-7729. , SPENCE, G. D., FOWLER,S. R., McKENzm, D. P., Wr~'TBROOK,G. K. & BOW~N, A. N. 1987. Magmatism at rifted continental margins. Nature, 330, 4 3 9 - ~ .

Right place, wrong time: anomalous post-rift subsidence in sedimentary basins around the North Atlantic Ocean A I D A N M. J O Y

Department of Geology, Imperial College of Science, Technology and Medicine, London SW7 2BP, UK

Abstract: The central and northern North Sea Basin was formed as a result of at least two extensional episodes, the latest of which took place during the Late Jurassic. During the Early Tertiary, some 80 Ma later, the basin experienced a dramatic acceleration in subsidence. Subsidence acceleration of Early Tertiary age has also been reported from a number of other extensional basins, on both sides of the North Atlantic, in which stretching had ceased several tens of millions of years previously. This phenomenon is not in agreement with the published models of extensional basin formation, which predict that the rate of thermal subsidence should progressively decline following extension. It is possible that the predicted pattern of thermal subsidence was obscured by the effects of a mantle hot spot beneath the North Atlantic lithosphere during the Cretaceous to Early Tertiary. However, it is difficult to explain the subsidence phenomena observed in basins around the North Atlantic Ocean using simple hot spot models. It is concluded that the relationships between hot spot activity, continental break-up and vertical crustal movements in the North Atlantic region are subtle and complex, but that subsidence history analysis can yield valuable insights into these relationships.

The purpose of this paper is to discuss some features of subsidence patterns observed in extensional basins around the margins of the North Atlantic Ocean. Many of these basins experienced an acceleration in subsidence during the Early Tertiary, and in several basins this subsidence acceleration post-dated the cessation of extension by a significant period (up to 80 Ma). Post-rift subsidence therefore occurred in the 'right' place--namely, in these extensional basins--but at the 'wrong' time, long after extension had ceased. Figure 1 shows a reconstruction of the northern North Atlantic region at magnetic anomaly 24 time, shortly after the commencement of spreading in the Norwegian-Greenland Sea (approximately end Palaeocene). The Palaeocene was a time of widespread vulcanism in and around the Norwegian-Greenland Sea. White (1988) and White & McKenzie (1989) suggested that this resulted from extension of the lithosphere above a hot spot, a region of anomalously hot mantle possibly fed by a mantle plume. In the vicinity of such a hot spot the earth's surface is supposedly supported both dynamically and isostatically (Bott 1988; White & McKenzie 1989). It has been suggested that this hot spot, which at the present day lies beneath Iceland, affected the progress of thermal subsidence in at

least some of those basins from which anomalous subsidence patterns have been reported (White & Latin in press). Evidence for Early Tertiary subsidence acceleration in the North Sea Basin is presented below, followed by quoted evidence for a similar phenomenon from other basins around the North Atlantic Ocean. The possibility that these anomalous subsidence patterns were caused by hot spot activity is then discussed. A n o m a l o u s post-rift subsidence in the North Sea Basin The structure and stratigraphy of the central and northern North Sea Basin are now fairly well known as a result of almost 30 years of petroleum exploration and related activities. It is generally agreed that the latest extensional episode in these basins took place during the Late Jurassic, lasting into the Ryazanian (-- Berriasian), the earliest stage of Cretaceous (Boote & Gustav 1987; Badley et al. 1988; Bertram & Milton 1989; Roberts et al. 1990), though there is good evidence for at least one earlier extensional episode (Barton & Wood 1984; Hellinger & Sclater 1989). The post-rift phase has therefore lasted from the Late Ryazanian to the present day.

From STOREV,B. C., ALABASTER,T. & PANKHURST, R. J. (eds), 1992, Magmatismand the Causes of ContinentalBreak-up, Geological Society Special Publication No. 68, pp. 387-393.

387

388

A.M. JOY LEGEND

Q

~

Plume & Plume Head (White & McKenzie 1989) Early Tertiary Igneous Activity

~NNS

. W3 t t

I

(off

'l P B ~ I !

tj

'

, 4

Fig. 1. Reconstruction of the northern North Atlantic region at magnetic anomaly 24 time (end Palaeocene), redrawn after White & McKenzie (1989). 2000 m bathymetric contour indicates approximate position of continent-ocean boundary. Rectangle indicatesposition of Fig. 2. Abbreviations: CNS, Central North Sea Basin; LS, Labrador Shelf; NENS, Northeast Newfoundland Shelf; NNS, northern North Sea Basin; OB, Orphan Basin; PB, Porcupine Basin; RT, Rockall Trough; SS, Scotian Shelf. In this basin, sedimentation took place mainly beneath wave base during the post-rift period. Diagnostic water depth indicators are therefore rare (Barton & Wood 1984, White & Latin in press) and, as a result, subsidence history analyses are typically poorly constrained. Bertram & Milton (1989) tackled this problem by constraining water depths using the coals of the Moray/Ninian delta top, deposited at sea level at the end of the Palaeocene (Fig. 2). In Fig. 3, Well A drilled these coals; its subsidence history plot is therefore 'pegged' at the end of the Palaeocene. Over the remainder of the postrift phase the palaeobathymetric maps of Barton & Wood 0984), which are based on palaeontological studies, were used to assign water depths. Wells A and D on Fig. 3 drilled the Moray/Ninian delta top. The only palaeobathymetric data used in the other six plots were taken from Barton & Wood (1984). Standard methods of data reduction were used; they are described in Joy (in press).

The subsidence history analyses in Fig. 3 show an increase in subsidence of approximately Palaeocene age. (Due mainly to uncertainties concerning water depth it is not possible to be more precise about timing than this, though ongoing work is aimed at obtaining better water depth constraints using seismic stratigraphical methods). This episode was closely connected with an important change in the morphology of the central and northern North Sea Basin: the commencement of subsidence of the previously supported rift shoulders and adjacent platform areas (Joy in press). Anomalous post-rift subsidence in a regional

context The North Sea is not the only area in which an increase in subsidence took place during the Early Tertiary. To the west, in the Porcupine Basin, subsidence rates accelerated during the Palaeocene (White & Latin in press; White et al.

RIGHT PLACE, WRONG TIME

o¢ n

~cP

.7,,G

LEGEND A

xXX'~.~

------4-------I-

Well illustrated in Fig. 3.

I

Top of Moray/Ninian Delta (after Rochow 1981)

'

Major Faults

/-4- ........ ]

,.,

i o I

\ SCALE lOO 200 km I

J

Fig. 2. North Sea Basin location map showing the position of the wells in Fig. 3. Note that these eight wells occur in a wide variety of structural positions within the basin.

1992). The most recent extensional episode had ceased during or immediately before the Early Cretaceous (Croker & Shannon 1987; MacDonald et al. 1987), and ocean floor spreading along the adjacent Biscay-Labrador trend commenced during the early Late Cretaceous (Keen et aL 1990). There was a dramatic increase in sedimentation rates in the Faeroe-Shetland Channel and in the Rockall Trough during the latest Cretaceous, and this was interpreted by Megson (1987) as evidence for an increase in subsidence rate. Uruski (1987) and Larsen & Marcussen (this volume) both report an increase in subsidence along the East Greenland continental margin around the end of the Palaeocene, resulting in the oceanward downwarping of the shelf. This pattern is also seen on the Norwegian margin. On the other side of the North Atlantic, Keen et al. (1987, 1990) reported evidence for an acceleration in subsidence during the PalaeoceneEarly Eocene in the Orphan Basin region , offshore Canada. These authors observed a similar though less pronounced phenomenon in the Labrador Shelf area, while the Scotian Shelf yielded no evidence for an increase in subsi-

389

dence rates during the post-rift period. In the Orphan Basin and Labrador Shelf areas rifting took place during the Early Cretaceous, leading to the creation of oceanic lithosphere in the adjacent NW Atlantic Ocean by the early Late Cretaceous. A critical distinction must be made when considering the subsidence histories of these areas. The rapid Late Palaeocene subsidence reported from the East Greenland and Norweg!an continental margins can be explained in terms of extension leading to the opening of the Norwegian--Greenland Sea at the end of the Palaeocene. The same applies to the Faeroe-Shetland Channel and the Rockall Trough, in which a latest Cretaceous subsidence acceleration could be explained in terms of an episode of extension of this age (Megson 1987). By contrast the accelerated subsidence which commenced during the Palaeocene in the central and northern North Sea Basin, in the Porcupine Basin, in the Orphan Basin area and on the Labrador Shelf was not directly associated with extension. In these areas rifting had ceased several tens of millions of years previously. The chronology of these events is summarized on Fig. 4. While there is no empirical evidence for a connection between anomalous Palaeocene subsidence patterns in the different areas where they have been observed, it seems unlikely that they are completely unrelated. The geographical proximity of these areas at the time of subsidence acceleration argues for a common cause for these phenomena.

Agreement with published models Extensional models o f basin formation Evidence has been presented and quoted which shows that the rate of subsidence accelerated during the post-rift phase in a number of basins around the North Atlantic. Extensional models of basin formation indicate that rifting is both accompanied and followed by thermal subsidence, which reflects the cooling of the extension-induced thermal anomaly beneath the rift (McKenzie 1978; Jarvis & McKenzie 1980). The rate at which thermal subsidence occurs is critically dependent upon the thermal time constant of the lithosphere. Whatever the value of this variable, however, the rate of subsidence should decrease gradually with time. The observed Palaeocene acceleration in subsidence in the North Sea Basin and elsewhere is therefore not in agreement with published models of extensional basin formation.

390

A.M. JOY AGE (Ma) 300 ,

2.50 ~

200 d

150 ~

-

300 ZS0 ~0 150 100 50 0].-~'~0 100 ~

50

0

~

-500

•.

~

-~ .

"/2

0 ~RESI~rrDAY SEALEVE~

WEL

500

F.ndc/last I

~,~t_ \ h

1000

l WELL D Tectom'c Subsidence correctedI -~~

~-

3000

Sediment-loaded subsidence [

3500

Thermal Subsidence curve for Beta=2.0, t=140 Ma

4000

I WELL G I

Fig. 3. Subsidence history plots for eight North Sea Basin wells. The locations of the wells are shown in Fig. 2. For illustrative purposes a series of variables has been plotted for well A, while only compaction-corrected tectonic subsidence is shown for wells B to H. The filled symbols on the Well A plot indicate sedimentological water depth data; these symbols occur at 54 Ma, indicating Late Palaeocene coals (see text). Wells B to H are displayed at the same scales; the arrows indicate the timing of the latest extensional episode (140 Ma).

The observed acceleration in subsidence can be modelled assuming a Palaeocene extensional episode (White & Latin in press). However the seismic data shows no evidence that such an episode actually occurred (Bertram & Milton 1989).

Hot spot~plume models for the North Atlantic The possibility of an Early Tertiary period of uplift interfering with thermal subsidence in the North Sea Basin was first suggested by Bertram & Milton (1989). Though they did not mention hot spot-related uplift specifically, they noted that uplift was common in the areas around the North Sea during the Palaeocene, and assumed

that it had occurred in the North Sea as well. In other words, normal thermal subsidence had been going on since the Late Jurassic extensional episode; this was interrupted by an ephemeral phase of uplift, and subsidence then accelerated to 'catch up' with the underlying thermal subsidence curve (Bertram & Milton 1989, Fig. 8). White & Latin (in press) suggested that the uplift phenomenon might have been due to the arrival of the Icelandic hot spot, and that the acceleration in subsidence took place as the North Sea Basin drifted away from the hot spot at the end of the Palaeocene. There is, however, no compelling evidence for such a period of uplift. Bertram & Milton (1989) suggested that, had thermal subsidence been taking place as predicted by published models,

RIGHT PLACE, WRONG TIME

ognize where deposition is taking place below wave base, but such a large amount of uplift as 300 m could scarcely occur without leaving some trace. There is no significant unconformity (neither angular discordance nor missing sec-

300 m of uplift must have occurred in the Viking Graben in order to bring the sediment surface to sea level during the Palaeocene, even discounting eustatic sea-level fluctuations. Admittedly minor amounts of uplift may be difficult to recr

ACe ea~ 0

SS

GB

NENS

LS

391

FRSTC/

PB

CNS

NNS

lo-2o-3O ~

EG

STRAT, CHRON(~

too.

-

V

OLIG,

J

eo.

"

,0--

V v V VVV

V v ¢ PAL

70~

~m

9o--

,~"

~

~/V~V V

~®1'°----~oW 'V z~o.~

I[VVV i~

Subsidence l

~.

I Iv'vv1

Accelerati°l'r I I Continental L.-../ I

"reak-ul:' z~o

"

g VVV

I~D~I"~ ,._-

~

"

/ l

L L

l

i

Rifting Vulcanis

Fig. 4. Chronology of rifting, vulcanism, continental break-up and accelerated subsidence in the meas mentioned in the text. Abbreviations and references: SS, Scotian Shelf; GB, Grand Banks; NENS, Northeast Newfoundland Shelf (Orphan Basin); LS, Labrador Shelf; (Keen etal. 1987,1990); PB, Porcupine Basin (Croker & Shannon 1987; MacDonald et al. 1987; Tate & Dobson 1988; White et al. 1992); RT/FSC, Rockall Trough/ Faeroe--Shetland Channel (Megson 1987); CNS, Central North Sea Basin (Ziegler 1982; Joy in press; White & Latin in press); NNS, Northern North Sea Basin (Ziegler 1982; Joy in press); EG, East Greenland Margin (Uruski 1987; Larsen & Marcussen, this volume).

392

A.M. JOY

tion) in the North Sea Basin between the Late Cretaceous and the Miocene, except in the extreme west of the area where it is clearly related to the weU-documented Early Tertiary uplift of Scotland. Had thermal subsidence been interrrupted by uplift associated with the arrival of a hot spot, this would no doubt have produced an unconformity which would have been particularly noticeable in wells drilled on tectonic highs and on the basin margins. The absence of such an unconformity suggests that, as the subsidence history plots in Fig. 3 indicate, the Palaeocene witnessed a true acceleration in subsidence and not merely a resumption of thermal subsidence following temporary uplift. If interference by the arriving Iceland hot spot cannot account for the anomalous subsidence patterns of the North Sea Basin, this is doubly true of the Porcupine Basin, Orphan Basin and Labrador Shelf. This is because they all lie outside the area of the plume head on the reconstruction of White & McKenzie (1989). The plume head might in fact have extended somewhat further southwest, underlying the areas mentioned above. In this case, however, we would expect to see abnormally thick Tertiary oceanic crust in the Biscay-Labrador Ocean as a result of spreading above anomalously hot mantie. There is no evidence for thick oceanic crust of any age in this region (d. White & McKenzie 1989, Fig. 8). Another possibility is that the anomalous basins were underlain by a hot spot throughout the Cretaceous. This could have supported the basins, isostatically and/or dynamically, during the early post-rift phase. At the time of continental break-up, however, the basins and the underlying thermal anomaly would have become physically separated as the hot spot moved toward its present position beneath Iceland. This might have resulted in accelerated subsidence, since the basins would then have resulted in accelerated subsidence, since the basins would then have moved away from an area of hotter than normal mantle to relatively cooler areas. Unfortunately this idea also has its drawbacks. If such a hot spot had been situated beneath the North Sea Basin at the beginning of the Cretaceous, movement of the European continental plate in the hot spot reference frame would have shifted the hot spot to a position beneath the Alps by 80 Ma (Latin 1990, Fig 7.2). Moreover a hot spot underlying all of the anomalous basins mentioned in this paper during the Cretaceous would have resulted in the generation of abnormally thick oceanic crust adjacent to the Orphan Basin, Porcupine Basin etc.

Summary The anomalous Palaeocene subsidence documented from the central and northern North Sea basin (Joy in press; White & Latin in press) has also been observed in the Porcupine Basin (White et al. 1992), in the Orphan Basin area and on the Labrador Shelf (Keen et al. 1987, 1990). In each of these areas the subsidence acceleration occurred tens of millions of years after the latest extensional episode. These subsidence acceleration phenomena are not in agreement with published models of basin formation. Moreover they cannot be explained in terms of simple hot spot models. If the Iceland hot spot arrived at the base of the lithosphere at the beginning of the Tertiary an episode of uplift should have resulted; there is no evidence of this, at least in the North Sea Basin. If, on the other hand, the hot spot was supporting the regions of anomalous subsidence throughout the Cretaceous, we would expect to see anomalously thick oceanic crust in the southern Labrador Sea region. The subsidence acceleration is approximately synchronous with the opening of the Norwegian-Greenland Sea at the end of the Palaeocene, but it is difficult to see what precisely the nature of the physical connection between these events might be (Joy in press). Practically the only conclusion about these Palaeocene phenomena that can be reached with confidence is that, if they are related, their cause is a truly regional one. This work is funded by Fina Exploration Limited, whose support is gratefully acknowledged. I would also like to thank M.-C. Williamson of the Atlantic Geoscience Centre for some helpful input from the Canadian perspective, and M. Badley, B. Storey and N. White for their thoughtful reviews of this paper.

References

BArLeY, M. E., PRICE, J. D., ~ E C H DAm. C. & AGDESTEIN,T. 1988. The structural evolution of the northern Viking Graben and its bearing upon extensional modes of basin formation. Journal of the Geological Society, London, 145, 455-472. BARTON,P. & WOOD,R. 1984. Tectonic evolution of the North Sea Basin: Crustal stretching and subsidence. Geophysical Journal of the Royal Astronomical Society, 79, 987-1022. BERTRAM,G. T. & MILTON,N. J. 1989. Reconstructing basin evolution from sedimentary thickness; the importance of bathymetric control with reference to the North Sea. Basin Research, 1, 247257. BOOTE,D. R. D. & GUSTAV,S. H. 1987. Evolving depositional systems within an active riR, Witch Ground Graben, North Sea. In: BgooKs, J. &

RIGHT PLACE, WRONG TIME GLE~IE, K. (eds) Petroleum Geology of North West Europe. Graham & Trotman, London, 819-833. Bo'rr, M. H. P. 1988. A new look at causes and consequences of the Icelandic hot-spot. In: MORTON, A. C. & PARSON,L. M. (eds) Early Tertiary Volcanism and the Opening of the NE Atlantic. Geological Society, London, Special Publication, 39, 15-23• CROrd~R, P. F. & SHANNON,P. M. 1987. The evolution and hydrocarbon prospectivity of the Porcupine Basin, offshore Ireland. In: BROOKS,J. & GLEN~m~, K. (eds). Petroleum Geology of North West Europe. Graham & Trotman, London, 633-642. HELLrNCER, S. J., SCLATER,J. G. & GILTNER,J. 1989. Mid-Jurassic through mid-Cretaceous extension in the Central Graben of the North Sea--part 1: estimates from subsidence. Basin Research, 1, 191-200. JARVlS, G. T. & MCKENZm, D. P. 1980. The development of sedimentary basins with finite extension rates. Earth & Planetary Science Letters, 48, 4252. Jot, A. M. in press. Comments on the pattern of postrift subsidence in the Central and Northern North Sea Basin. In: WILLIAMS,G. D. & DOBBS, A. (eds) Tectonics and Seismic Sequence Stratigraphy. Geological Society, London, Special Publication. KEEN, C. E., STOCKMAL,G. S., WELSINK, H., Qun~LAN, G. & MUDFORD, B. 1987. Deep crustal structure and evolution of the rifted margin northeast of Newfoundland: results from LITHOPROBE East. Canadian Journal of Earth Sciences, 24, 1537-1549. • LONCAREVIC,B. D., REID, I., WOODSIDE,J., HAWORTH,R. T. & WILLIAMS,H. 1990. Tectonic and geophysical overview. In: KEEN, M. J. & WILLIAMS,G. L. (eds) Geology of the Continental Margin of Eastern Canada. Geological Survey of Canada, Geology of Canada, 2, 31-85. LARSEN,H. C. & MARCUSSEN,C. 1992. Sill-intrusion, flood basalt emplacement and deep crustal structure of the Scoresby Sund Region, East Greenland. This Volume. LATIn, D. M. 1990. The relationship between extension and magmatism in the North Sea Basin. PhD thesis, University of Edinburgh. ]~ACDONALD, H., ALLAN,P. M. & LOVELL,J. P. B. 1987. Geology of oil accumulation in Block 26/ 28, Porcupine Basin, offshore Ireland. In: Bgoogs, J. & GLENNm,K. (eds) Petroleum Geology of North West Europe. Graham & Trotman,

393

London, 643-651. MCKENZIE, D. 1978. Some remarks on the development of sedimentary basin. Earth & Planetary Science Letters, 40, 25-32. MEGSON, J. B. 1987. The evolution of the Rockall Trough and implications for the Faeroe-Shetland Trough In: BROOKS,J. & GLEr,n~IE, K. (eds) Petroleum Geology of North West Europe• Graham & Trotman, London, 653-665. ROBERTS,A. M., YIELDING,G. & BADLEY,M. 1990. A kinematic model for the orthogonal opening of the late Jurassic North Sea rift system, DenmarkMid Norway. In: BLUNDELL,D. J. &.GIBBSA. D. (eds). Tectonic evolution of the North Sea Rifts. Oxford University Press, 180-199. ROCHOW, K. A. 1981. Seismic stratigraphy of the North Sea "Palaeocene" deposits. In: ILLING,L. V. & HOBSON,G. D. (eds). Petroleum geology of the continental shelf of North.West Europe. Institute of Petroleum, London, 255-266. TATE, M. P. & DOBSON, 1988. Syn- and post-rift igneous activity in the Porcupine Seabight Basin and adjacent continental margin W. of Ireland. In: MORTON, A. C. & PARSON, L. M. (eds) Early Tertiary Volcanism and the Opening of the NE Atlantic. Geology Society, London, Special Publication 39, 309-334. URUSKI, C. I. 1987. East Greenland: the connection with the North Sea. In: BROOKSJ. & GLENNIE,K. (eds) Petroleum Geology of North West Europe. Graham & Trotman, London, 1189-1200. WHITE, N. & LATIN,D. in press. Lithospheric thinning from subsidence analyses in the North Sea 'triple junction'. Journal of the Geological Society, London. - , TATE, M. & CONROY, J. J. 1992. Lithosperic stretching in the Porcupine Basin, west of Ireland. In: PARNELL,J. (ed.) Basins of the Atlantic Seaboard: Petroleum Geology, Sedimentology and Basin Evolution. Geological Society, London, Special Publication, 62, 327-333, WHITE, R. S. 1988. A hot-spot model for early Tertiary volcanism in the N Atlantic. In: MORTON,A. C. & PARSON,L. M. (eds). Early Tertiary Volcanism and the Opening of the NE Atlantic. Geology Society, London, Special Publication, 39, 3-13. & McKENzm, D. P. 1989. Magmatism at rift zones: the generation of volcanic continental margins and flood basalts. Journal of Geophysical Research, 94, 7685-7729. ZIEGLER, P. A. 1982. Geological Atlas of Western and Central Europe. Elsevier, Amsterdam.

395

Index Abraham Member, 324 accretionary complexes, Antarctic Peninsula, 153 active plume heads, 24--5 Aden, 19, 301 aeolian deposits, offshore Namibia, 263 Afar plume, 32, 36, 300--2 rift, 96 African plate, 84 Agatdal Formation, 323-4 Agulhas Fracture Zone, 152, 210, 213, 217, 269 Ahlmann Ridge, 171-2 Albuquerque Basin, 76 alkaline magmatism, Africa, 91-8 Allan Hills, 169 Alpine orogeny, 95 Ambenali, 49, 279 Amirantes Arc, 273 amphibole, 34-5, 37, 51,54, 233--4 Amran Formation, 295 andesites Antarctic Peninsula, 153, 217 Antarctica, 159 high-Mg, 216 Paran~t, 223 Rio Grande rift, 64 Taos, 78 Angolan Basin, 269 ankaramites, East Greenland, 353 Antarctic Peninsula, 152-6, 174--5, 212, 216-17 Antarctica, 150--1 low velocity anomalies, 101 separation from Africa, 213 Ant6nio Enes, 145 apatite, Yemen, 296 Appalachian/Caledonian orogeny, 157 Ar-Ar ages Deccan, 273 Etendeka, 245 Paranfi, 223,229--30, 236-7 Seychelles, 274 Arabian Sea, 283 opening, 275 Arabian shield, 103 arc magmatism, Antarctic Peninsula, 153 Arica deflection, 213 Ascension, 115 aseismic ridges, 41, 44, 55, 248-9 asthenosphere, 108 partial melting, 80--1 asthenosphere derivation, 61-89, 100, 231,233-4, 236-7, 362 Asuk Member, 323 Atanikerdluk Formation, 324 Atlantic opening, 4, 10, 55, 95, 117, 141,315-18, 321,360, 366, 389 subsidence, 387-90 Auob Sandstone, 263 Aussivik Member, 324 Australia-Banda Arc collision, 131

Australian-Antarctic Discordance, 107 Azores, 103, 108, 115 back-arc basin Antarctic Peninsula, 154 Antarctica, 153, 187 Neo-Tethys, 157 Pacific, 104, 118 Siberia, 117 back-arc spreading Antarctic Peninsula, 216 Antarctica, 152 Pacific, 101 Baffin Bay, 344, 347, 350, 359 Baffin Island, 6, 36, 317, 329, 335,337, 342-4, 350, 359 Bailey Ice Stream, 188 Baltimore Canyon Trough, 13 basanites Eifel, 83 Espanola Basin, 67 Sardinia, 83 basement reactivation, 245 Basin and Range, 51, 82-3, 157, 237, 301-2 Beacon Supergroup, 165-8 Beardmore Glacier, 167-8, 170, 174, 176 bending stress, 127 Benue Trough, 84, 94, 242-3 Bermuda plume, 5 Biscay-Labrador Ocean, 389, 392 Blosseville Kyst, 351,354, 360 Bombay, 274, 283 boundary-layer instability, 5, 9 Bounty Island, 173 Brazil alkaline complex, 94 dykes, 9 British Tertiary Igneous Province, 31, 36, 44, 315, 344 Broken Ridge, 22-3 Bubye Coalfield, 142-3 Bumbeni complex, 146 buoyancy flux, plumes, 242, 251 Bushe magmas, 279 Caledonian orogeny, 307, 343, 351,360, 365 Cameroon Line, 5, 83-4, 92, 245-6, 250 Campbell Plateau, 173-4 Campos Basin, 257 Canary Islands, 93, 110 Cape Fold Belt, 142, 157 Cape Verde islands, 110, 115 Cape Verde plume, 342 carbonatites, African rift, 9 I, 93--4 Caribbean basalts, 23, 25 Caribbean Plateau, 45 Carlsberg ridge, 274 Carlsberg rift, 271 Carolina Trough, 13 Caucasus, flood basalts, 117 Central Graben, I0 Central Indian Ridge, 51,279 Cerillos, 80

396 Cerro Colorado, 76 Cerros del Rio lavas, 61, 64, 67-9 Chacos Basin, 242 Chagos-Laccadive Ridge, 22, 273-4 Cheyenne Belt, 79 Chile, 155 Chilwa Igneous Province, 94 Chon Aike Formation, 155, 174, 215-16 Clarens Formation, 142 Coastal Cordillera, 211 Coats Land, 185-207 Nd isotope ratios, 203-4 Cocos plate, 104 cold slab effects, 101,102, 110 Colorado Basin, 242-3 Columbia River basalts, 7, 10-12, 19, 25, 44-6, 50, 54 duration, 45 plume association, 44 volume, 7 Comores Islands, 273 Comores plume, 273 conduction asthenosphere, 53 lithosphere, 50 conductive cooling, 3-4 Congo craton, 94, 96 continental collisions, 95 continental extension, 62 continental flood basalts, 31-9, 44, 103-10 crustal component, 50 mantle component, 50-1 origins, 31-9, 49 and plume heads, 108, 112 convection asthenosphere, 50, 53 small scale, 104 uppermantle, 114 Coppermine River basalt, 8 Cordillera Darwin, 216 core--mantle boundary, 7, 113,242, 249, 252 crack propagation, 36-7 cratonic lithosphere, 119 cratonic mantle, 103 Cretaceous activity African rift, 93-4 Pacific, 24 Crozet hotspot, 187 crust, Archaean, 37 crustal contamination, 51, 62 Antarctica, 198 British Tertiary Province, 36 Columbia River, 50 Deccan, 50 Espanola Basin, 75 Ferrar province, 171,202 Karoo, 32 Kerguelen, 22 Madagascar, 50 Paranfi, 50, 231,237 Rio Grande rift, 83 South Atlantic, 250 Taos, 78 Crustal shear, 257 crustal thickening, Antarctica, 157-8

INDEX crustal thickness, 3-4, 33 Atlantic, 4 Yemen, 298 crustal thinning, 4, 127 Atlantic, 4 and CFB, 12 Columbia River, 11 Jameson Land, 373 West Antarctica, 167 Cuvier Plateau, 19 dacites, 9 Antarctic Peninsula, 153 Tans, 78 Voting margin, 310, 312 Darwin Rise, 242 Davis Strait, 6, 344-7, 350, 359 Deccan associated plume, 115-16 asthenospheric source, 36 duration, 45-6, 272 extrusion locus, 12 first phase, 10 India-Seychelles breakup, 19 K-Ar ages, 273 melt flow, 5 pictites, 337 plume association, 45 Sr isotope ratios, 50 velocity anomalies, 108 volume, 7 xenoliths, 35 Deccan province, 231,271--4 Deccan Traps, 22, 221 low velocity anomalies, 115-16 Deccan--Chagos-Laccadive Ridge, 44 dehydration reactions, descending slabs, 109 Del Carlo Rise, 23 delaminated lithosphere, 113 density gradients, 110 Deseado Massif, 173-4, 215 diamonds, Tanzania, 93 dinoflagellate cysts, Greenland, 324-5, 336 Disko, 322-5, 337, 343, 345, 359 dolerite sills Karoo, 144 Transantarctic Mountains, 168 dolerites Coats Land, 185-208 Etendeka, 225 Ferrar, 140-1 Karoo, 137-8 Lesotho, 140 Seychelles, 279 doming, 44--6 East Greenland, 360-1 Karoo, 139 Tristan hotspot, 245 Yemen, 300, 361 doughnut plume model, 344, 346 Dronning Maud Land, 138, 140-1,143-4, 158, 165, 16972, 174, 176-7, 185--8, 204-5 AFC processes, 201 associated plume, 187-8

INDEX geochemistry, 197 REE ratios, 202-3 dry lithosphere, 33-4 DSDP, 259, 273, 305 Dufek intrusion, 141,174, 176, 186, 201 Dupal basalts, 51 Dwyka tillite, 142 dyke injection, Rio Grande rift, 81 dyke swarms Jameson Land, 369 paran,,i 223 dykes Albuquerque Basin, 78 associated with CFB, 8 Etendeka, 224-5 Jameson Land, 368 North Atlantic Province, 6 Paran~i, 224-5 protection from contamination, 49 Transantarctic Mountains, 169 Yemen, 296 East African Rift, 104, 109-10 East Greenland, 349-50, 353 Sr isotope ratios, 50 East Greenland basalts, 36 East Greenland plume, 327 East Mariana Basalt, 23 East Pacific Rise, 101,104, 111 Ecca Group turbidites, 142 Egypt, 294,298-300 Eifel, 83, 109 Ekalulia basalt, 8 Ellsworth-Whitmore Mountains, 157, 159, 173, 176-7, 216 enriched hotspots, 118 enriched mantle, 34-5 Albuquerque Basin, 78 Espanola Basin, 74 Esmeralda basalts, 223,228 Espanola Basin, 61, 64-9, 80-1, 84 .... Etendeka, 19, 23, 36, 108, 116, 171,177, 221,223, 245, 247-8, 252, 263 Etendeka/Paran~i-Walvis Ridge, 45 Ethiopia, 19, 32-3, 96, 293, 301 Europe-Greenland separation, 305 Explora Wedge, 143-6, 158 Explora-Andenes Escarpment, 186 Exposure Hill Formation, 168-9 Faeroe--Iceland-Greenland Ridge, 19, 22, 315 Faeroe-Shetland Channel, 389 Faeroe-Shetland sills, 377 Faeroes, 327-8, 330 failed rifts, Antarctica, 157, 187 Falkland Islands, 141,152, 159, 172, 215 Falkland Plateau, 146, 209-10, 213, 216 Falla Formation, 167-9, 173 Farallon plate, 64, 69, 79 faulting Beardmore Glacier, 168-9 offshore Namibia, 269 Voring margin, 307 Yemen, 295-6, 301

397

Felicitt, 279 Fenris graben, 306 Ferrar, 35-6, 45-6, 54, 140--1,146, 152, 156--8, 186, 188, 205, 231 dates, 152 geochemistry, 169-71,177, 196--200, 197 petrography, 196 plume association, 54 Sr isotope ratios, 141,201 Sr ratios, 34 tholeiites, 165, 167 xenoliths, 35 Filchner lee Shelf, 188 fission track dates, 294 Yemen, 296-8 fission-track dates East Greenland, 360 Jameson Land, 371 Flat Tops, 78, 80 Fortune Bank, 274 fractionation, 33 fracture zones, 18 Fulla Ridge, 306-7 gabbros Antarctic Peninsula, 153 Dufek intrusion, 201 Karoo, 137 Seychelles, 279 West Greenland, 327 Galapagos plume, 45 Galicia Bank, 4 Gastre Fault System, 152, 155, 159, 209-10, 212-13, 215,217 geoid highs, 101,104-108, 111 geothermal profiles, 127, 129 Goban Spur, 1-3 Gondwana break-up, 33, 52, 91, 95, 105, 115-16, 137, 142, 156 flood basalts, 110 geochemical province, 166 plumes under, 54 reconstruction, 210 Gough Island, 116, 199-200, 249 grabens Namibia, 259 Patagonia, 213 Rio Grande rift, 64 Gramado basalts, 223, 225--6, 228-30, 236 granites Antarctica, 173 Ellsworth-Whitmore Mountains, 174 Jameson Land, 368 Karoo, 137 Pirrit Hills, 177 Seychelles, 279 Thurston Island, 155 Yemen, 296, 302 granitoids Antarctic Peninsula, 153, 155 Patagonian Batholith, 215 South America, 155 gravitational instability, 157-8 gravitational potential, 10, 13

398 gravity, offshore Namibia, 268-9 Great Basin, 82, 302 Greece, flood basalts, 117 Greenland, 305,314, 317, 321-2 basalts, 47-8, 50 dykes, 9 lavas, 45 low velocity anomalies, 115 rifting, 117 Tertiary lavas, 6 Greenland-Faeroes-lceland-Ridge,44, 345, 347, 358 Gulf of Aden, 109, 293 Gulf of California, 109 Haag Nunataks, 216 half-grabens, South Africa, 142 Hareoen Formation, 349, 351,353,355 Hatton Bank, 1-2, 312 Hatton-Rockall Basin, 5, 10, 317 Hawaii, 36-7, 101,108, 110-11,284, 286 plume, 5, 10-11, 45, 52, 79, 342 Hawaiian Arch, 284 Hawaiian-Emperor seamount chain, 24 hawaiites, Espanola Basin, 67 heat flow, Siberian Traps, 117 heat transfer mechanisms, 42 Hebridean--Greenland craton, 34-5 Hebrides, 343 Heimefront Range, 174, 187 Hel graben, 306 Hercynian orogeny, 117 high-Ti basalts, 196-7, 233,236 East Greenland, 361 Ferrar, 169, 177, 196 Paran~i, 51,223-4, 228 high-Ti CFB, Nuanetsi, 202 high-velocity anomalies, 105-6 HIMU, 246, 248-50, 252 Hoggar, 10, 93 Hold-with-Hope, 344 Horingbaai dolerites, 225 hot lines, 109 hot mantle sources, 24-5 hotcells, 110--I 1, 119-20 hotspot longevity, 25 hotspot sources, 23 hotspot trails, 41,244, 251 hotspots, 100, 111 Atlantic, 106 fixity, 25, 100, 114, 119-20, 119-21 and geoids, 104, 108 Iceland, 358 Indian Ocean, 103 initiation, 119 insulation, 234-5 and lithospheric thinning, 129, 133 in low velocity anomalies, 104, 109 Huab dolerites, 226 hyaloclastites East Greenland, 360 Greenland, 330 Jameson Land, 371 Kirkpatrick Basalt, 168 West Greenland, 322-4, 326, 359

INDEX hydrous phases, 51, 54, 79, 82, 233--4 hypabyssal intrusions, Karoo province, 174 hypersthene tholeiites, 196 Iceland, 46-8, 111, 117, 247, 341,345 basalt composition, 353-4 hotspot, 315,387, 390, 392 low velocity anomalies, 108 mantle sheet, 12 plume, 5--6, 10, 25, 36, 79, 316, 318, 342, 344, 346, 349-50, 358--61 position, 358 Iceland Ridge, 41 igneous chronology, Africa, 92 ignimbrites Marifil Group, 212 Yemen, 302 incompatible elements East Greenland, 361 Greenland, 345-6, 355, 358 Paran~i, 226 Rio Grande Rift, 74 Seychelles, 273,279 West Greenland, 345, 351 India, basalt flows, 7-8 Indosinia block, 106 intraplate rifting, 96 intrusive complexes, Seychelles, 271 Iran, flood basalts, 117 Ireland, 330--1,378 basalts, 327 Irminger Formation, 351,353 isostatic uplift, 127, 129, 145 Jabal al Nar, 301 Jabal Khariz, 301 Jameson Land, 365-83 basin structure, 370-2 Jan Mayen, 115 Jemez lineament, 8 l Jones Mountains, 173 Jornado Basin, 76 Jurassic-Cretaceous boundary, 246-7 K-Ar ages Deccan, 273 "Ferrar, 172 Jameson Land, 368 Paran~i, 226, 229 Seychelles, 274 South AtlantiC, 10 Theron Mountains, 188 K/Nb ratios, 47-8 Kaapvaal craton, 33, 35, 94 Kalahari craton, 94, 96 Kangaroo Island tholeiites, 170 Kangerdlugssuaq, 342-4, 349-51,353, 358, 360-2, 368, 373 Kangilia Formation, 324 Kara massif, I 17 Karoo, 5, 9, 45, 46, 142, 152, 231 Africa-Antarctica breakup, 19 associated plume, 117, 138-9, 141,158-9 asthenospheric source, 36

INDEX enrichment, 140 geochemistry, 177 geology, 137 lithosphere involvement, 33 lithospheric mantle, 32 picrites, 337,346 plume, 141 plume association, 45 tholeiites, 169 velocity anomalies, 108 xenoliths, 35 Karoo Basin, 259 Kenya rift, 92-3, 104 Kerguelen, 19, 22-3, 45, 46, 52, 103, 110, 242, 247-8, 250 crustal contamination, 22 hotspot, 115 ocean breakup, 23 Pb isotope ratios, 52 plume, 55 velocity anomalies, 108 very low velocity anomalies, 102 Kholan Formation, 295 kimberlites, 33, 93-4, 96, 114, 118, 121,246 Kirkpatrick Basalt, 141,168-9 Kirwan Escarpment, 171, 174, 177, 187 komatiites, 33, 50, 103, 112-13, 118 Kraul Mountains, 172, 174 Kudu Wells, 259, 263, 269 Labrador Sea, 6, 317, 329-30, 342-4, 346-7, 359 Labrador Shelf, 389, 392 Lake Baikal, 107, 109, 120 lamproites, 51, 114, 118, 121,233, 236--7 Karoo, 202 Nuanetsi, 231 Rio Grande rift, 62 lamprophyres, 327, 369 Languedoc, 33-4 Laramide orogeny, 64, 157 Large Igneous Provinces, 17-30, 24, 41-3 Larsen Harbour Complex, 155 Latady Formation, 175 lateral temperature gradients, 112 laterite, Yemen, 295 Lebombo, 9, 140-1,144-6, 173, 177-8, 337 Lebombo monocline, 137, 174 Lesotho, 138-40, 146, 172 leucitites, Eifel, 83 lherzolites, 33-4, 83,337--8, 340 Limpopo, 138-9, 142-3, 145 Line Islands, 6, 23 Lipetrrn Group, 212 Liquine-Ofqui fault zone, 210 lithosphere Archaean, 33 enriched, 35-6 layers, 50 lithospheric interactions, 49 lithospheric rifting, 9, 12-13, 41 lithospheric stretching, 62, 112, 117, 125, 133, 257 Antarctica, 175 Columbia River, 11 Davis Strait, 6

399

Europe, 83 time periods of, 3-4 lithospheric susceptibility, 18 lithospheric thickness, 33, 101 Colorado, 80--1 Eifel, 83 Great Plains, 80--1 Sardinia, 83 Siberia, 117 lithospheric thinning, Rio Grande rift, 64 Littlewood Nunataks, 177 loading stress, 127 Lomonosov Ridge, 360 Lonco Trapial Group, 215 Long Normal Polarity superchron, 246-7, 253 Louisville plume, 45 Louisville Ridge, 248 low velocity anomalies Antarctica, 101 Atlantic, 102, 108, 115 and hotspots, 118 Iceland, 108 New Zealand, 101 North Atlantic Tertiary Province, 115 Pacific, 101,102 low-Ti basalts, 196-7,233, 236 Ferrar, 169, 177, 196 Paran~i, 223-4, 226 low-Ti CFB, Gondwana, 204 Lower Lavas, East Greenland, 351,353, 355-6, 358,360 Luderitz Basin, 259, 263-4 Mackenzie dyke swarm, 8 Madagascar, 46, 47 basalts, 49-50 separation, 145,213,273 Madagascar Ridge, 23 Madagascar-Marion Island Ridge, 45 magma supply rates, 45--6 magma types, Paran~i, 223 magmatic evolution, Paran~i, 228, 236 magmatic incubation, 55 magnetic anomalies, Voring margin, 312 magnetic reversal frequencies, 247 Mahabaleshwar, 279 Mahr, 275, 279 major elements Greenland, 354 West Greenland, 341,353 Maligfit Formation, 322--4, 327, 351 Malpais lavas, 76 Malvinas Islands, 215 Malvinas Plateau, 209-10, 213, 216 Mangaia, 250 Manihiki Plateau, 46, 242, 246--7, 250 mantle, mineralogy, 233 mantle contamination, 114, 118 mantle convection, 24, 110, 119 mantle decompression, 1, 3, 13, 25, 41, 54, 62, 106, 114, 117, 155, 159, 231,235, 250, 315, 317-18, 341, 361 mantle melting, 337-9 mantle sheets, 5--6 mantle structure, 100

400

INDEX

mantle wedges, 269 Maputo, 139, 144 Maranhao, 246, 250 Marie Byrd Land, 174 Marifll Group, 174, 212-13, 215 Marion Island, 46 Marion plume, 55, 117 Marshall Mountains, 169 Martin Vas plume, 246, 250 Mascarene Plateau, 19, 273--4 Massif Central, 33-4 Mateke-Sabi monocline, 137 Maurice Ewing Bank, 213 Mawson Formation, 168-9 mechanical boundary layer, 50-1,234--6, 252, 343 melilitites, 94 melt flow, 1, 6-7 melt intrusion, 1, 3 melt migration, Voting margin, 314 melting experiments, 34 mesosphere, 102-8, 112-13 metasomatism, 6, 34-5, 37, 100, 104, 110, 113, 121,140 meteorite impact, 221 microplate assemby, 117 microplates Antarctic/Pacific, 187 Antarctica, 209, 216 Mid-Atlantic Ridge, 102-3, 108, 115-16, 241,246, 358 midplate stresses, 24 Mikis Formation, 351,353 minettes, 62 mobile belts, 343-4 Mogollon-Datil lavas, 75 Moho, 4, 54, 125, 127, 257, 269, 378 monoclinal folds, Marshall Mountains, 169 Moray Firth, 10 Moray/Ninian delta, 388 MORB, 32, 199-200 Atlantic, 76 Ferrar Magrnatic Province, 203 K/Nb ratios, 48 Karoo, 202 Lesotho, 141 Nuanetsi, 140 Patagonia, 215 source, 100 Taos, 78 Voting margin, 310 West Greenland, 351,353, 356, 360 More Basin, 5, 10 Morocco, flood basalts, 117 Mount Bumstead, 169 Mount Erebus, 101, 110 Mount Fazio tholeiites, 169-70 Mount Hill Formation, 174 Mount Poster Formation, 175 Mozambique, 137, 144-6, 152, 155 Mundwara, 279 Murihiku terrane, 173-4 Murud dykes, 279 mylonites, Gastre Fault, 212 Nagssugtoqidianmobile belt, 344 Nahuelbuta Mountains, 210

Namibe Basin, 259 Namibe Desert, 263-4 nanoplankton zones, Greenland, 324-5, 331,336 Naramada tiff, 275,283-4 Natal, 137 Naturaliste Plateau, 19 Nauj~guit Member, 322--6 Naujfit Member, 324 Nauru Basin, 23, 45--6 Nazca plate, 104 Nd isotope ratios Coats land, 200-201,203--4 East Greenland, 353--4 Eifel, 83 lithosphere, 34 Rio Grande Rift, 69, 76, 78-9 West Greenland, 353-4 xenoliths, 35 Nd/Pb ratios, South Atlantic, 249 necking, 125, 133 nephelinites African rift, 93 Eifel, 83 Espanola Basin, 67 Karoo, 137 New Zealand, 174 Jurassic magmatism, 173 low velocity anomalies, 101 subduction, 175 Newark Group, 103 Niaqussat Member, 327 Ninetyeast Ridge, 22, 248 non-volcanic margins, 1-4, 12 Nordffjord Member, 327 North America, pre-drift volcanism, 106 North American plate, 64 Noah Atlantic, 6 opening, 10, 213 spreading rates, 4 Noah Atlantic opening, 321 Noah Atlantic rift, 6 North Atlantic Tertiary Province, 315 North Atlantic Tertiary Province, 5--6, 19, 36, 108, 321 associated plume, 315 velocity anomalies, 108, 115 North Island (Seychelles), 183, 275, 279 North New Guinea Plate, 102 North Patagonian Massif, 155, 209-10, 213 North Sea, 7, 330, 387-90, 392 Norway, 305,314, 316, 389 Norwegian--GreenlandSea, 387, 392 Nossob Sandstone, 263 Nuanetsi, 137-40, 143, 173, 178, 202, 231,337 Nuussuaq, 322--6, 343-5, 349 Ocean basin flood basalts, 23 ocean plateaus, 19, 41, 48, 55, 104 oceanic crust, thickened, 41 OIB, 31, 33, 37, 46, 199-200 Cameroon Line, 84 Coats Land, 203 Espanola Basin, 64-9 K/Nb ratios, 48 northern hemisphere, 61, 76 Paramt, 228

INDEX Rio Grande rift, 64-5, 68-9, 78-9, 81 trace-elements, 248 Trans Pecos province, 83 West Greenland, 356 Yampa, 78 Omega dolerites, 188, 196 Ontong Java Plateau, 19, 23-4, 45--6, 55, 111,242, 2467, 250, 252 ophiolites South America, 155 South Georgia, 158 Orange Basin, 259, 263-4 Orange Free State, 139 Orange River, 266 Ordlingassoq Member, 322-4, 326-7 Orphan Basin, 389, 392 Oslo Graben, 117 Pacific plate, 23 palaeomagnetism Jameson Land, 368 West Greenland, 325-8 Palmer Land, 173-5 Pangaea, 102, 109 formation, 103 Pangaea breakup, 95, 111,117, 131,133, 135 Panhala Formation, 10 Paramt, 50, 142, 169, 171,177, 221,242-3, 247-8, 250, 252-3 At-At ages, 223, 229-30 associated plume, 110, 116-17, 245 asthenospheric source, 36 comparisons with Jameson Land, 377 Dupal character, 46 dykes, 8, 224-5 enrichment, 140 extrusion locus, 12 high-Ti basalts, 51 K-At ages, 229 magma types, 223 melt flow, 5 plume association, 45 sills, 9 South America-Africa break-up, 19 subsidence, 9 trace elements, 224 velocity anomalies, 108 volume, 7 Paran~i-Chacos Basin, 242-3 Paranapanema, 224, 228-9, 236 partial melting, hydrous peridotite, 204 passive rifting, 24-5, 257 passive stretching, Ferrar province, 178 Patagonia, 174-5, 178 Patagonian Batholith, 155, 173-5,212, 215-16, 209-210 Pb isotope ratios, 46, 51,248-9 Coats Land, 201 Ferrar, 197 Kerguelen, 52 Madagascar, 50 St. Helena, 246 Southwest Indian Ridge, 51 Pensacola Mountains, 157, 174, 186 peridotites, 33, 113

401

perisphere, 113-14, 116, 121 permissive magmatism, 91, 96 Phenai Mata, 279 Philippine Sea plate, 107 phlogopite, 51, 54, 233-4 Phoenix plate, 23 picrites, 9, 46, 53, 112-13, 118, 121 analyses, 337-42 Deccan, 337 East Greenland, 353, 356, 358 Greenland, 317, 330-1,335-6 Karoo, 137-8, 202, 337 West Greenland, 322-3,327, 329, 351,358-9 Pigafetta Basin, 23 pigeonite tholeiites, 196 pillow basalts, 359 Jameson Land, 371 West Antarctica, 168 Pirrit Hills granite, 177 Pitanga, 224, 228-9, 236 plate deformation, 54 plate motions, 18, 52, 95, 108, 110-11,117 plate-wide activity, 94-5 plume heads, 44, 45, 108, 111 plume models, Jameson Land, 383 plume temperatures, West Greenland, 342 plume-lithosphere interactions, 41-59, 51-3, 342 plumes contact times, 55 dimensions, 44, 236, 245, 251 evolution, 42-4, 54, 111-13 initiation, 9 lithospheric impact, 44 melting, 44 sources, 113 stability, 102 plutonic complexes, Karoo, 137 Poladpur, 279 Ponta Grossa Arch, 224--6, 228-9, 236 Porcupine Basin, 388-9. 392 Porcupine Seabight, 328 Poseidon Ocean, 8 Potrillo-Palomas lavas, 64, 68-9, 76, 79-80, 84 Praslin, 279 pre-existing magmatism, Rio Grande rift, 80 Prebble Formation, 167-9, 173 pure shear, 257 Rajasthan, 279 Rajmahal Traps, 19, 115 Rajmahal-Ninetyeast-Kerguelen Ridge, 45 R~s Sub-basin, 306 Rb-Sr ages Chon Aike formation, 155 Ferrar, 174 West Greenland, 336 Recovery Glacier, 188 Red Sea, 33, 293-4, 298, 300-2 Red Sea Hills, 94 REE ratios, Dronning Maud Land, 202-3 regional uplift, Rio Grande rift, 64 R~.union, 22, 44, 46-8, 110-11, 115, 221,231,273, 288, 358 plume, 55

402

INDEX

source, 23 Rhinegraben, 104, 109, 120 rhyodacites Parami, 223 Seychelles, 274 rhyolites, 9 Antarctic Peninsula, 153, 217 Karoo, 137 Karoo province, 173 Marifii Group, 212 Paran~i, 223 Patagonia, 215 Rio Grande rift, 64 Seychelles, 274 Ribeira basalts, 223 ridge migration, 103 rifting Africa, 241,243 Antarctica, 204 Atlantic, 316-18 duration, 11 East Greenland, 360 Gondwana, 177 Labrador Sea, 359 North Atlantic, 305 offshore Nan~ibia, 265 Parami, 241 Red Sea, 300-1 South Atlantic, 242, 245, 251,269 Rinkian mobile belt, 343-4 Rinks Dal Member, 323-7 Rio de Janeiro, 224-6, 228-9, 231,236 Rio Grande rift, 61-89, 109 initiation, 64 Rio Grande Rise, 19, 45, 116, 244-5, 248 Rocas Verdes, 155 Rockall Basin. 5 Rockall Plateau, 329 Rockali Trough, 10, 310, 389 Rodrigues Ridge, 6 Rooi Rand dyke swarm, 144 Ross Sea, 167, 176 Rungwe rift, 94 Sable P,iver Formation, 171-2 St Helena, 46, 48, 115, 117, 241-2, 249 plume, 5, 245, 249, 25 i seamounts, 244-5, 249 St Peter-Paul islets, 116 Salado Basin, 242-3 Samoa, 110 San Jorge basin, 216, 264 San Luis, 76, 78, 80-1 Santos, 224-6, 228-9, 23 i, 236 Sardinia, 83 Sarnu-Dandali, 279 Saudi Arabia, 294-6, 298-302 Saya de Malha Bank, 274 Scarab Peak tholeiites, 169-71 Schirmacher Oasis, 174 Scoresby Sund, 314, 351,353-4, 360-1,366, 368-9, 371, 373, 383 Scotian Shelf, 389 Scotland, 327, 330-1,377, 392

seafloor spreading, 129, 135 Antarctica, 152 Atlantic, 4, 141,241,369 Canada-Greenland, 350 Greenland, 329-30 lndian Ocean, 283 Jurassic, 143, 145 Labrador Sea, 317 Mozambique Basin, 155 South Atlantic, 10, 242-3, 263 Weddell Sea, 187 seamounts, 23,244, 248, 251 seaward dipping reflectors, 5, 13, 19, 41, 46, 138, 158, 307 Africa, 145 Greenland, 369 secondary convection, 24-5 secondary melts, 9 seismic profiles, Namibia, 260-1 seismic velocity variations, 3D, 100 Serviileta Plaza centre, 80 Seychelles, 271-5, 279, 283, 288 trace elements, 273 Seychelles Bank, 274, 283 Shackleton Range, 177, 186, 188, 196 Shannon Island, 314 Shatsky Rise, 111 shear, non-volcanic continental margins, 3 shear velocities, outer shell, 106 shear velocity, 105-7 Shetland, 329-30 Siberian Traps, 19, 25, 45, 54, 108, 117 associated plume, 117 duration, 45 velocity anomalies, 108 Sierra de la Ventana, 157 Sierra de las Uvas, 75-6, 80 Silhouette Island, 274-5, 279, 283 silicic volcanism Antarctica, 169 Beardmore Glacier, 174 Ferrar province, 177-8 Siljan, 377 sills Jameson Land, 366, 368, 373-7 Karoo, 144 as magma chambers, 377 Paran~l, 8-9 simple shear, 257, 269 Sinai, 294, 298, 300 slab accumulation, 102 Slave craton, 116 Snake River basalts, 36 Somali Basin, 142, 145, 152 Somuncura Batholith, 215 Sonoma orogeny, 117 SOPITA superswell, 250 source intensity, 18 South America, 153, 155, 157, 159 South Atlantic, 7, 157 opening, 10, 221,223, 237, 241,244-6, 251 South Georgia, 153, 155, 158, 174, 205 Southern Coastal Batholith, Chile, 215 Southwest Indian Ridge, 51

INDEX Spitzbergen, low velocity anomalies, 115 spreading centres, oceanic, 3 spreading rates, 4, 24 North Atlantic, 4 spreading ridge, India, 116 Sr isotope ratios Antarctic Peninsula, 153 CFB, 33 Coats Land, 200-201 Deccan, 50 Dronning Maud Land, 186, 197 East Greenland, 50, 353-4 Eifei, 83 Ferrar, 34, 141,171,177, 186 lithosphere, 34 Rio Grande Rift, 69, 76, 78 South America, 155 Thurston Island, 155 West Greenland, 353-4 steady-state plumes, 24-5 Straumsvola, 174 stress analysis, 125-36 structural controls, Africa, 92 sub-continental lithospheric mantle, 62, 100 Colorado, 78 Eifel, 83 Espanola Basin, 69 subducted sediment, 233,249, 252 subduction, 100, 104 Gondwana breakup, 140, 156 high-velocity anomalies, 102-3 Karoo, 146 New Zealand, 175 and plate tension, 131 proto-Pacific, 215, 217 Tethyan margin, 158 Transantarctic Mountains, 187 Urals, 117 Western USA, 62, 69, 82 subduction pull, 133, 135 subduction zone sources, 51 submarine ridges, 22 subsidence Atlantic, 387-90 East Greenland, 351,360 Jameson Land, 373,381 Paran~i, 9 super-plumes, 242, 246-7,250-1 surface flows, 7 Svartenhuk, 345,359, 362 Swaziland, 138 Sweden, 377 syenites Jameson Land, 368 Seychelles, 279 Straumsvola, 174 Tahiti, 284" Tanzania, 93 Taos lavas, 64, 68, 78 Tarim shield, 103 Tasman, 152 dates, 152 dolerites, 174

403

geochemistry, 170 Tasmania, 110 Tawilah Formation, 295, 301 tensional stresses, 149 terrane accretion, Pacific, 117 Tethyan margin, 157-8 Tethys, 177 Thaba Putsoa, 35 thermal boundary layer, 42, 50, 104, 108, 114, 116, 120 thermal bulge, East Greenland, 36! Theron Mountains, 165, 169-72, 174, 177, 186, 188, 196, 200-201 thinspots, 44 Hebrides, 344 Jameson Land, 377 West Greenland, 330 tholeiites, 9, 19, 31, 35-6, 45, 50, 54, 121 African rift, 91 Albuquerque Basin, 78 Antarctic Peninsula, 154 Deccan, 272, 279, 288 East Greenland, 353 Espanola Basin, 65, 67, 76 Ferrar, 158, 165, 167, 169 Greenland, 335, 351 Karoo, 137 Paran~i, 223 Ponta Grossa Arch, 226 Seychelles, 271,274 Voring margin, 310, 314 West Greenland, 322, 351 Thurston Island, 153, 155-6, 174-5 Tierra del Fuego, 216 Tithonian, 230, 236 Tobffera, 155, 174, 215 trace elements Antarctic Peninsula, 154 CFB, 231 Greenland, 355-6 Paran~i, 224, 226 Rio Grande Rift, 68 Seychelles, 273¢ 279 West Greenland, 35 !-3 trace-elements, OIB, 248 trachytes, Seychelles, 279 Trans Pecos province, 83 Transantarctic Mountains, 140-1,154, 157-8, 165-9, 175-6, 178, 185-8, 199, 201,204 Transbaikal, 54 transient volcanism, 23-4 Transvaal, 139 trench rollback, 157 Trinidad-Columbia seamount chain, 246 triple junctions, 104, 108, 140 Karoo, 138 South America, 245 Yemen, 293 Tristan da Cunha, 23, 45, 46, 110-11, 115, 117, 198-9, 221,228,233,236-7, 241-2, 244-5,249 Dupal basalts, 51 plume, 231,249, 251-2, 263,358 Trondelag Platform, 306 tufts Seychelles, 274

4O4 Voting margin, 307,310 Tularosa Basin, 76 turbidites Namibia, 264 Patagonia, 211 South Africa, 142 Tvora, 174 Ubekendt Ejland, 327, 336, 345 Uganda, 93 Umiussat Member, 324 underplating, 3, 5-6, 9, 11, 13, 33, 43, 343 Deccan, 284 Karoo, 139 Rio Grande rift, 80 Voting margin, 312 Yemen, 298 uplift East Greenland, 371 Jameson Land, 381,383-4 North Sea, 390 Yemen, 295-6, 298 uppermantle currents, 111 definitions, 100 upwelling, 108, 110, 114, 119-20 Ferrar province, 178 origins, 103 Voting margin, 314 Ural-Taimyr, 117 Urubici basalts, 224 Vaigat Formation, 322-3,327, 337, 343, 351 Vanfaldsdalen Formation, 351,353-4 vein complexes, 62 very large velocity anomalies, 103, 108, 111 very low velocity anomalies, 102, 104 Victoria Group, 168

INDEX. Victoria Land, 165-71,174, 176, 185 Viking Graben, 10, 391 volatiles, 107, 118 volcanic continental margins, 1, 5 volcanic passive margins, 19 Voting Basin, 5, 10, 306-7, 310, 314 Voting Escarpment, 306, 310, 314 Voring Margin, 305, 307 Voting Plateau, 19 wall-rock reactions, 35-7 Walvis Basin, 259, 264 Walvis Ridge, 23, 41, 116, 199-200, 228, 244-5, 248-50, 257, 259, 264, 266, 268, 269 Wankie Sandstone, 263 Weddell Sea, 144, 152, 155, 157, 159, 187,213 West Antarctic, crustal blocks, 152 West Greenland, 335-6, 349-50 West Greenland Basin, 322 West Greenland magmatism, duration, 327 Western Australia, 19 wet lithosphere, 34-5 Whichaway Nunataks, 170--2, 177, 186, 188, 196, 201 White Mountains, 106 Whitmore Mountains, 159 Wollaston Foreland, 314 Wrangellia, 117 Wyoming craton, 35, 79 xenoliths, 33--7, 79 Yampa lavas, 68-9, 80 Yellowstone plume, 78-9, 81 Yemen, 32-3,293-5,298-302, 361 zeolitization, Beardmore Glacier, 172 Zimbabwe, 139 ZrfY ratios, 204--5

Magmatism and the Causes of Continental Break-up edited by

B.C. Storey, T. Alabaster and R.J. Pankhurst "...represents a timely and welcome contribution to the field as it comprises semi-review articles dealing with magma generation and break-up processes, as well as syntheses from selected examples of CFB provinces. In addition there are several case studies (both geochemical and geophysical) which examine specific issues in detail. Even in isolation many of these articles are exceptionally useful, either as distillations of current thinking, or as new contributions of ideas and~or data. Together, however, they combine to create a comprehensive volume covering much of our present understanding.., of the causal relationships between lithospheric extension, rifting, thermal anomalies and magmatism. '

Janet Hergt in Chemical Geology, vol. 109, p356 '...should be within reach of all geoscientists seriously interested in continental flood volcanism and mechanisms of continental break-up. It provides a fascinating picture of the current state of knowledge of continental break-up on a global scale and highlights the complexity of the driving forces of break-up and the origins of associated basaltic rocks... In general, this is an important book which provides a solid foundation for developing understanding of modern continental rift tectonics and for the interpretation of continental rift geology and magmatism in the geological record. Every university geology library should buy it.'

G. Wheller in Australian Geologist No 88, p41 '...breadth of subject matter, the diversity of the authors and the speed of publication all conspire to make this an excellent 'research in progress' volume. This is not a collection of similar papers reporting a consensual view, but a far more useful picture of the state of the art. In particular, it covers both the well-known examples of continent break-up apparently related to 'plume' magmatism, as well as examples where break-up manifestly has nothing to do with hot upwellings... This book should be in every library of Earth Science.'

D. Pyle in Geological Magazine, vol. 131, p732 '...provides a wealth of information and stimulating ideas for geologists interested in breakup tectonics, magmatism and stratigraphy.., should be a required acquisition for university and research libraries.'

Warren Manspeizer in Earth Science Reviews, vol. 35, p327

• • • •

416 pages 166 illustrations including 10 in c o l o u r 24 papers index

Cover illustration: Jurassic Ferrar Sill intruding Beacon Supergroup, Transantarctic Mountains

ISBN

0-903317-83-4

80