Advances in Botanical Research, Volume 2

Advances in

BOTANICAL RESEARCH

This Page Intentionally Left Blank

Advances in

BOTANICAL RESEARCH Edited bu

H. D. PRESTON The Astbury Deparlnaml of Biophp-ics The University, k d s , England

VOLUME 2

ACADEMIC PRESS, INC. (tiarcourt Brace lovanovic h. 'Publishers)

London Orlando San Diego New York Toronto Montreal Sydney Tokyo

ACADEMIC PRESS INC. (LONDON) LTD.

BERKELEYSQUAREHOUSE BERKELEY SQUARE LONDON, W.l

US.Edition published by ACADEMIC PRESS, INC.

Orlando, Florida 32887

Copyright 0 1965 by Academic Preas Inc. (London) Ltd.

All Rights Reserved NO PART OF THIS BOOK MAY BE REPRODUCED IN ANY FORM, BY PHOTOSTAT. MICROFILM, OR ANY OTHER MEANS, WITHOUT WRITTEN PERMISSION FROM THE PUBLISHER

Library of Congress Catalog Card Number : 62-21144

PRINTED IN THE UNITED STATESOFAMERICA

85 86 87 88

98765432

CO-UTORS

TO VOLUME 2

M. B. DALE,Departmend of Botany, The Univeraity, Southampton, England (p. 35). DEREKT. A. LAMPORT, Now at the Michigan State University Atomic Energy Commi8eion Plant Research LuboraWy, RIAS, Baltimore, Maryland, U.S.A. (p. 151). JOHN LEVY,Botany Department, Imperial College, University of Londun, 8.W.7 (p. 323). I. MLCNTON, Botany Department, University of L e d , England (p. 1). P. MAHESHWARI,Department of Botany, Univereity of Ddhi, Delhi, India (p. 219). N. S. RILNOASWAMY, Department of Botany, Univerdy of Delhi, Delhi, India (p. 219). P. A. ROELOFSEN, Laboratory of General and Technical Biology, Technological Univeruity, Delft, Netherland8 (p. 69). W. T.W~LIAMB, Department of Botany. The UniuerSitgv.8wuthampton, England (p. 36).

This Page Intentionally Left Blank

In preparing the first volume of this series we were well aware that, whatever eke we were doing and however valuable and rewarding this new series might prove to be, we were inevitably adding to the burden of reading which all professional saientists, and not leaat among them, the botanists, are called upon to bear. In the prefaoe to that volume we were therefore rather careful to point out what we hoped would be found to be an approach somewhat different from that of most other texts reviewing piecemeal a whole field of study. Though we ouraelves had faith in this new venture we neverthelees waited with more than a little nervousness for the reaations of readera mainly, though not entirely, voiced by reviewers in the learned journale. In the event all the reviewers were favourable and we pmoeed to the second volume, not so much perhaps with more assurance, as with relief both that we were not too far in error in assessing the need for and the merits of the approach we outlined and that the publisher, the authors, and the editor had together made a book which fulfilled theee aims. In presenting this second volume we have in mind the comment of one reviewer that we may 6nd difficulty in maintaining the standard set in the first volume, We are sure that this second volume will set his fears at rest and we are grateful to the distinguished authora in this volume that they have spared no effort in ensuring that t h b ehould be ao. It is naturally not pomible in two volumes to cover the whole field of botany, and an editor is therefore faced a t the outset with the task of selecting topics in such a way as to achieve a fair balance in each volume between the interests of the wide variety of readers at which it is aimed. The inclusion in this volume of two articles dealing with cell walls therefore calls for comment, particularly since this happens to he our own particular interesf. An editor is inevitably guided by one overriding principle. The topics he chooses must be in course of rapid development, must be significant and must be dealt with by an author who is, and is known to be, distinguished. This l e d to a deliberate choice of author rather than of topic. The editor must, however, be ready to accept an opportun$y if one unexpectedly comes his way. In choosing the growing cell wall as one of the more important topi08 of the day the only possible author was Prof. Roelofaen and we am Vii

viii

PREFACE

more than grateful to him for the warmth with which he accepted an onerous task. This was intended to be the only article on cell walls, for although Dr. Levy’s article on the recently discovered importance of soft rot fungi is involved with cell walls this is not its only, or even its major, importance. When, however, we saw the opportunity of a “stop preas” article on the moat recent cell-wall development we hsd no hesitation in pressing Dr. Lamport to write what in our opinion is an article complementary to that of Prof. Roelofsen but is also in itself a striking example of the sudden break-through which can occur in an old problem when an enquiring mind seizes upon a new clue. One of the most remarkable developments of our time in plant science has been the way in which hitherto purely observational regions are progressively becoming experimental or even mathematical. One case of the former is included here, in the remarkable developments in embryology a t the hands of Prof. Maheshwari and his colleagues and students; and one case of the latter is given in the mathematical approach to taxonomy represented here in the work of Prof. Williams and his associates, which is now giving a new look to this somewhat bewildering field. We do not need to emphasize the impact of electron microscopy in all fields of botanical enquiry; it is represented here in its position with regard to classification in which, as in other aspects of electron microscopy, Prof. Manton has played such a prominent part. We are again grateful to the authors in this volume for the time and care with which they have prepared the artioles and have thereby smoothed our path, and our thanks are due particularly to the publishers whose quiet efficiency we deeply appreciate. Finally we are indebted in this volume as we were in Volume 1 for secretarial assistance to Mrs. E. D. White (nee Lister), this time with the able assistance of Mrs. C. J. Parker.

R. D. PBESTON

beds, 1906

CONTENTS

......................................... v .......................................................... vii

CONTRIBUTORS TO VOLUME2

PREFA~E

Some Phyletic Implicatione of Flagellar Structure in Planta I . MANTON .. Introduction ................................................. Distribution of Flagella of the 8 +2 Type ......................... .. The Heterokont versus Isokont Condition ........................ Other Aspects of Flagellar Numbers and Relative Length .......... V . F l a ~ ~ a r.aS h..............................................

I I1 I11 IV

.

........................................ ........................................... ............................................ ............................................ . ............................................... . ............................................ . .................................................. x. summery ................................................... References., ..................................................

V I Flagellar Appendages A Flagellar Spines B Flagellar Hairs C FlagellarScales VII Flagellar Beeea VIII Flageller “Roots” IX Conclueions

..

. . .

.

6 8 10 13

13 14

17 17 18 20 20 33

Fundamental Problem8 in Numerical Taxonomy w T WILLIAMS and M . B DALE

. .

.

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

I Introduction I1 The Nsture end Propertiea of Claseificetions A The Basic Axiom B Monothetic and Polythetic Claseifications C Maximization D Hierarchical and Non-hierarchicalClaaaificetiona E Probabilistic and Non-probabilistic Cleeeificationa 111 The Choice of Mathematic&%l Model A Introduction: Metrics B Metric Properties of Pair-functions C Intrinsicslly Non-metric Systems D Non-EuclidmnSyateme E Conclusions IV The Beeio Euclideen Model A . Duality: The R/Q F’roblem ................................. B. Adjustmenta to the Model C Heterogeneity v. StretegyofAnSlyais .......................................... A Simplification Methods ..................................... B. Pertition C Non-hierarchical Methods D Hierarchical Methods Achowledgementa References

.

1

3

. . .. .

.

. . . .

. . . .

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

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

.................................. ...................................... ........................................... ................................................... ix

36

37 37 37 38 42 43 48 48 49 51 62 63 64 64 66 56 69 59 61 61 62

67 07

X

CONTENTS

Ultrastructure of the Wall in Growing Celle and ita Relation to the Direction of the Growth P

. .

. A . ROELOFSEN

................................................. 09 ............................................. ................... 70 70 ........ 78 ........................ 82 ................. 85

I Introduction I1 Morphological Aspects of Constitution. Synthesis and Breakdown of theprimery WEll A Constitution and Morphology of Microfibrils B Constitution of the Amorphous Matrix in Primsry Wells C Some Aspecte of Microfibril Coherence D Site of Synthesis of Prim~ryWall Substances E Questionable Evidence of Breakdown in Primary Walle I11 Survey of the Microfibrillsr Arrangement in Different Typee of Growing Celle A Freely Growing more or less Isodiawetrio Cells B Freely Growing Tubular Cells or Perts of Cells C Tissue Cells with Isodiametric Growth D Tissue Cells with Predominant Growth in Length E Tissue Cells that Predominantly Widen., F Tips of Tissue Cells with Tip Growth I V Interreletion between Growth end Wall Ultrastructure A Effect of Growth on Wall Structure B Effect of Well Structure on the Direction of Growth C Theories on the Mechenism of Orientated Jnitiel Synthesis of Cellulose Microfibrils References

.

.. . . .

. . . .

. . . . . .

......... 89 ................................................ 91 91 ................ ................ 98 ......................... 104 .............. 106 ..................... 112 113 ......................... ............ 114 .......................... 114

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

128 139 145

The Protein Component of Primary Cell Walls DEREK T

. A . LAMPORT

. A. LJcopeandDefinitione ...................................... 161 151 B. HistoricalPerspective1888-1969 ............................ 152 I1. Experimental Methods end Materi~ls............................ 165 A. Cell Suepension Cultures....................................166 I Introduotion .................................................

............................................. 167 168 I11. The Hydroxyproline-rich Wall Protein: “Extensin” .............. 180 A . Intra-cellular Location of Hydroxyproline..................... 160 B. Chemical Cheracb&tion of 4-tmnu-hydroxy-~-pIine......... 167 C. The Amino Acid Composition of Primepy Cell Walle., .......... 168 D. Enzymic Degradation and Characterizefion of Wall Protein ..... 171 E. Disdphide B r i d e in Cell-Wdl P r ~ b i n....................... 172 F. Distribution of the Hydroxyproline-rich Wall Probin in the Plant Kingdom ................................................. 174 IV. The Biosynthesia of “Extensin” ................................ 177 . A . Uptake and Incorporation of W-Proline by Intact Cells........ 177 B. ProlineHydroxylation ..................................... 184 V . Veristion of Cell-wall Hydroxyproline Content ...................188 A . Walls Isolsted from Tissue Culturw .......................... 188 B. Wells Isolated from Plsnt Psrts ............................. 189 B. WholePlante

c. Ana~yticdTechniques......................................

CIONTENTS

xi

.................. 189 ...................................... 193 ...................................... 194 ........... 198

.

V I Degredation of the Sycamore P r b 8 r y Cell Well A ChemioelDegradation B Enzymic Degradation VII A Tentative Piature of “Extensin” in the Primsry Wall VIII The Contribution of “Extensin” to Wall Form end Teneile Strength I X EnzymiaW8llProtein A hoorbio Aoid Oxidaae B Hydrolyeeee C Other Wall-bound Enzymee D How does the Wall Rind Emymtw ? X The Role of “Extensin” Aoknowledgemente Referenoes

. . ... .. .. .

200 ......................................... . 204

...................................... 204 .............................................. 206 ................................. 206 ......................... 206 ....................................... 209 ........................................... 213 ................................................... 213

Embryology in Relation to Phyeiology and Genetice P MAHESHWARI and N 9 RANQA8WAMY

.

. .

..

................................................. 219 ....................................................... 221 . ........................................ 221 ...................................... 223 . ............................................... 227 ........................................ 281 . .................................... 232 . ..................................... 232 ....................... 234 ..................................................... 237

I Introduction 11 Pollen A Longevity of Pollen B. GerminationofPollen C Pollen Tube 111 Control of Fertilizetion A Treetmentoftheat igma B Treatment ofthe Spyle C . Intrmvarien and in vitro Fertilization IV Embryo A Growth of Vmbryo in Relation to Seed Development B Dependence of Embryo on Endoeperm C Bpeofioity in Nutrition of Embryo V Endoeperm . A Conetituent4Jof Endoeperm B Role of Endusperm in Seed Development C CultmofEndosperm VI Embryo culture A Cultural Conditions B GrowthMedia C Applications of Embryo Culture D Limitations of Embryo Culture M I. Cultureof0vulee M I. Culture of Ovsriea and Flowera M Perthenocsrpy X Polsembryony A Adventive Embryony B EmbryonslBudding XI P ~ h e n o g eeeis n XI1 . . .g.A n X I I I Antherculfure XIV Control of Sex Expression X V Conclueions Aoknowledgementa Refe~

.

. . . . . . . . . . .

..

. .

. .

. ..

..

........... 237 ....................... 239 ........................... 289 .................................................. 240 ................................. 240 ..................... 242 ...................................... 243 .............................................. 246 ........................................ .............................................

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

247 260 266 262 263 266 278 280 280 286 296 800 801 804 309 810 310

xii

CONTENTS

The Soft Rot Fungi: Their Mode of Action and Sicanee in the Degradation of Wood JOHN LEVY

. . .

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

I Introduction 323 I1 Histology of Soft Rot 329 I11 A Technique for Studying Soft Rot Fungi........................ 337 IV Mode of Action of Soft Rot Fungi 339 A Peseive Penetration and Decay Penetration 339 B . Effect of Species of Wood on the Mode of Attack by the same Fungue 340 C Effect of Species of Fungus on the Mode of Attack in the same 344 Wood D Soft Rot Fungi on Posts in Ground Contact 348 V List of Fungi known to came Soft Rot 348 VI Dieoussion 349 Acknowledgements 366 356 Referenoes AUTHOSINDEX 369 SWBJEOTINDEX .................................................... 371

. . .

. .

.

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

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

Some Phyletic Implications of Flagellar Structure in Plants I. MANTON Botany Department, Univereily of Lee&, Enghnrl 1. Introduction..

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

1

II. Distribution of FhgeUe of the 9 +Z Tspe. .................................. 3 5 III. The Hebmkont veraue Isokont condition.. ................................ Iv. Other depeote of Flegellsr Nnmbera and Relative Length.. .................. a

v.

BlEgdhrlSaape

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

M. W q & u Appendzigw ........... .+ ...................................... A. F h g e k Spinen ..................................................... B. FbgellmHeire

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

c. Flegellsr Scales .....................................................

10 13 13 14 17 I?

........................................................ ...................................................... ia .......................................................... 20 ..............................................................20

VII. FhgellarBBaee YIII. Flr4gellar “Roote” IX. ConolUBioM,. X. Summery

References.. ...........................................................

33

I. INTRODUCJTION Discovery of the essential similarity between cilia and flagella of both plant and animal kingdoms has exposed a linguistic contradiction which impinges too closely on taxonomy to be easily resolved. So long as protistologists must refer to major p u p s of organkms as Ciliates and Flagellates, recommendation to use the one word “cilium” for the filamentous appendages of both, leaving the other word “flagellum” for the very differently constructed filamentous appendages of bacteria, is a council of perfection that can scarcely be carried out. That inextricable confusion has not resulted from this linguistic impasse ie a tribute both to the resilience of biologists and to the clear-cut basic character of the organelles themselves. Illogical verbal usage at this level is indeed harmless in the sense that it rarely, in practice, leads to a fundamental misunderstanding of facts. I n contrast, the unforeeeen obstacles to mutual comprehension that have been raieed by reoent attempts to standardize nqmenclature a t a more intimate level ie a topic about which more will be said below. In the study of cilia anp flaplla of the non-bacterial sort by electron microscopy, three periods pan conveniently be distinguished. In the firat phaae, extending roughly from 1946 (Jakus and Hall) to 1964 (Fawcett and Porter), with the most intense period from 1980 to 1982 (Manton and Clarke), the ubiquity of the 9+2 fibre pattern waa B

2

I. MANTON

established. That each member of the peripheral ring is double and not a single strand was also established at the close of the same period (Manton and Clarke, 1962; Fawcett and Porter, 1964), though general agreement that the fibres of the central pair are not similarly constructed came later. A second phase of study, beginning with Af?elius (1969) and culminating in Gibbons and Grimetone’s classio paper of 1960 (see also Gibbons, 1961a), introduced better fixation, the use of new plastics and above all the successful applioation of new electron dense stains to thin sections, thereby opening up the study of additional structures which is still continuing. The new information on flagellar bases discussed below was made possible by thia work. Lastly a third phase, involving application of negative staining to dismembered flagella, is beginning to provide details of the fine structure within individual fibres (Andr6 and Thikry, 1963; Pease, 1963). It is not the intention to review here any of these phases as such, but only to discuss those aspects of flagellar structure which may have phyletic implications. Phase 3 so far has no phyletic implications and mention of it has been made only to bring the record up to date and to alert biologists who are not themselves electron microscopists to the existence of a very interesting new field of study. Phyletic implications, where they occur, are only rarely dependent on the micro-anatomical details associated with the 9+2. They are usually dependent on ancillary structures or on other details such as relative length and arrangement of flagella, some of which have long been known, in a general way, to light microscopists. The early history of knowledge of the existence of tile heterokont condition among algae (Thuret; 1861, Pringsheim, 1866; Thuret and Bornet, 1878; Luther, 1899) can be cited in illustration of this. Endorsement of the liiht microscopist’s view using the same type of evidence seen larger is not, however, the main function of the study of fine structure. The phyletic implications that I wish to discuss here are those that could not have been effectively worked out by light microscopy unaided and on wbich the electron microscope has supplied a considerable number of entirely new criteria and with them some new points of view. Phyletically significant new criteria have occasionally been sought by deliberate electron microscopical enquiry, but usually they have been encountered incidentally as unexpeatd additions to an investigation directed primarily towards other objectives. At the start of electron microscopy (phase 1mentioned above), when the main technical method was that of shadow-cast whole mounts, the great variety of external appendages borne by plant flagella in contrast to those of animah was an incidental h d h g of this kind. The reviews published by myself in 1962, 1964 and 1966 cover sufficiently this phase of the investigation

FLAUELLAR STRUOOTURE IN PLASTB

3

of plants. Subsequent developments have however been of two kinds. There is real progress when new methods have proved to be applioclble to new types of material, but there are also some unfortunate mistakes which, for one reason or another, have begun to enter the litemture. These, if accepted as facts by the inexperienced, which is all too easily done, can impede progress substantially. It is this last consideration as much as any other which has prompted selection of the present title at this time. Finality is not yet attainable in many matters on which knowledge is rapidly growing. A progress report based on &&-hand experience may nevertheless have positive value. There will necessarily be strong personal bias in the selection of topics for discussion, since an exhaustive treatment of literature alone is not the intention. Fortunately two recent publications can be used to supplement this report. There is Pitelka’aexcellent little book on the fine structure of protista both colourless and coloured (Pitelka, 1963) and a recent algological survey by Christensen (1962) which lends itaelf admirably to discussion of phylogeny. Since the text of the latter is in Danish (an English edition is promised but not yet available) it may perhaps be of value to Englieh readers if the giat of Christensen’sphyletic scheme is reproduced in the form of a list from his table of oontents (see p. a), supplemented by his diagram (Fig. l), translated where necessary, reproduced on p. 5. 11. DISTRIBUTION OF FLAGELLA OF TEE 9+2 TYPE An important biological concept not based on flagella but summing up much information on the micro-anatomy of protoplasmic structures of other kinds is that ofthe procaryotic versus the eucaryotic cell (expressed as Procaryota and Eucaryota in Christensen’s liat). Anyone unfamiliar with these terms can usefully consult the excellent short paper by Stanier and van Niel(1962). Procaryotic cells lack membranebounded internalorganelles though they do not lack internal membranes. If organs of locomotion exist, aa in bacteria, they are never of the 9+2 type. Eucaryotic cella (i.e. cells of all the main plant and animal groups other than bacteria and blue-green algae) possess membranebounded organelles of various kinds, e.g. nuclei, mitochondria, chloroplasts. Even when oell size is reduced to within the dimenaiom common in bacteria, as in the tiny green flagellate iKrmnOnae p u d h (Manton, 1969a), the eucaryotic nature of the main cell components remains sharply distinct. The introduction of the membrane-bounded space within which biochemical activities of some kinds can be carried on in relative isolation from biochemical aotivities of other (perhaps mutually incompatible)

4

I . MANTON

kinds is clearly an evolutionary advance of the greatest importance which must have preceded most of the phyletic diversity with which modern taxonomy is concerned. Membrane-bounded spaces in the sense of ccdouble-membranea”(for fiwther discussionof this terminology see Manton, 1961) occur conspicuously in cells of blue-green algrte, notably in their pigmented regions. It is membrane systema of this type, or still more elaborate variants of them (but not simple membranes), which delimit the organelles within eucaryotic cells. Such ceUe are also the only ones to possess oilia and flagella of the 9+2 type, not, however, in all eucaryotic groups. The absence of flagella of any kind from the red algae has long been known (of. Fritsch, 1936), though this fact has scarcely disturbed the prevailing view that, apart from blue-green algae, a flagellate anoeatry lies behind all the main algal groups. An early loss of flagella is of course not difficult to imagine and the prevailing tendency to & o w Rhodophyceae (Fritsch, 1936, or Rhodophyta in the terminology of Smith, 1951), late in the soheme of major algal groups has consciously

List of mujor algal groups according to ChriStemen, 1962 PROCARYOTA Cyanophyta Cyanophyeeae EUCARYOTA ACONTA Rhodophyta

RMphyme CONTOPHORA Chromophyta

Cryptophgceae Di?WphgWZ$ B h a p h i ~ ~ ~ Cb~OPkP= H*topkJCrw*h-

Bt7Cill&C?phyC4?4W Xadwphym

Chlorophyta

p*hY-

EWkVhYLxOPhYPrmhphyeeae Chlorqphyctw

FLAGELLAR S T R U C T U R E I N P L A N T S

6

or unconsciously led to a general feeling that they are late arrivals and intrinsically specialized. Christensen (1962) is most explicit another way : specialized they undoubtedly are in many ways (somatic structure, mode of growth, life histories), but this could be mainly sign of extreme antiquity within the eucaryotic scheme, an antiquity which could have resulted from an origin before the 9+2 flagellum had been evolved. The extreme simplicity of their plastid lamellations (cf. Bouck, 1962; a180 A. D. Greenwood, personal communication) is an important additional micro-anatomical character consistent with such a view. Christensen’s phyletic scheme (Fig. 1) thus places the Rhodophyta MI the loweet and most ancient eucaryotic group, preoeding in origin the

Fro. 1. Christensen’s Phyletic scheme of probable relationehip between the main group of algae and other main groups of living organisms. Algae are indieetad by thick lines and lettera, other organisms by thin lines and letters. After Christensen, 1982.

6

I. MANTON

simplest true flagellates and ante-dating the separation of the plant and animal kingdoms. The importance of the term Aconta as opposed to Contophora is thus greater, and the term itself more significant than the mere absence versus presence of flagella had previously suggested.

111. THE HETEROKONT VERSUS ISOKONT CONDITION There is now a good deal of evidence in favour of extreme antiquity for the heterokont condition. By this is meant the possession of flagella in pairs, the two members of each pair Wering in length, type of motion and presence or absence of external appendages, and wully with the basal bodies of a pair mutually attached at a wide angle. Among pigmented forms, at both the monad and algal levels, other features of cell structure accompanying the heterokont flagellation include absence of chlorophyll 6 , presence of a rather simple internal plastid structure devoid of grana (perhaps more correctly described as plastids with a different lamellar system from that of the higher green plants), a more intimate relation between chloroplasts and endoplasmic reticulum than in organisms possessing chlorophyll byand the possession of mitochondria with micro-tubules and not flattened cristae. It is significant that some of these characters, notably the tubular equivalents of cristae, are also encountered among Protozoa. It is therefore virtually certain that the latter antedate Metazoa phyletically and that the heterokont pigmented flagellates are likely to be ancestral not only to these but to the main groups of heterokont algae and fungi also. These views are adequately expressed in Christensen’s diagram (Fig. l), and are, on fhd whole not controversial. An important innovation in Christensen’s scheme is, however, the use of the term Chromophyta to cover all groups of Contophora lacking chlorophyll b. Such an arrangement may not last permanently, but in the present state of knowledge it brings clarity where confusion previously existed and is therefore to be welcomed. In addition to the basic heterokont groups (Chrysophyceae, Xanthophyceae, Yhaeophyceac) some new names are included on which comment will be added later (pp.12, I’letc.). Attention should however be directed here to the importance of the present position of Vauckeria. That t h h familiar classroom type is closely relevant to questions of origin of the water moulds is well known, but as long as it was treated a8 a member of the green algae, a total conflict of evidence was inevitable together with a tendency to over-estimate the antiquity and relative importance of Chlorophyta versus Chromophyta, which still adversely affects phyletic thinking. The heterokont condition of the Vuwkria spermatozoid

FLAGELLAR STRUUTURE I N PLANTS

7

waa however well known to Pringsheim (1866),though at that date the concept of Heterokontae (Luther, 1899) as a group did not exist. It took nearly a century before the chemistry of pigmentation (for litemture see Strain, 1961) plus supplementary details of flagellar struoture (Kooh, 1961) could counteract the seductive influenoe of its bright green colour and lead to the realization that its true afRnitiee are with the Chromophyta. Everything that has happened since (eee especially Greenwood et al., 1967;Greenwood, 1969)has confirmed thie conclueion and there can be no doubt that the tranafer of Vaw&& to Xanthophyceae (Perke, 1962)is correct. This, it may be said in passing, could in fad have been done half a century sooner had the value of flagellar characters as phyletic indicators been better understood. The Chlorophyta (containing chlorophyll b) include the familiar isokont group^ enumerated in Christensen’s liat under Chlorophyceae together with euglenoids and some additional new p u p a which are not iaokont in flagellation. The essential unity of all the major land plants (archegoniates,gymnosperms,angiosperms)with green algae is endorsed by rapidly accumulating information about salient details of fine structure of the cell as a whole which cannot be described here (see however Manton, 1964b,c,d). This unity neverthelees depends almost certainly on chlorophyll b (and its physiological and structural conaequences) rather thah on flagellation, and though the origin of chlorophyll b as such can scarcely be discussed here it ia important to notioe the conservative treatment of thia problem given by Chriabnaen. In placing Chromophyta and Chlorophyta on a level, he is in fact avoiding a question which he might have treated differently. There are many members of the Chromophyta containing chlorophyll a but no other chlorophyll, e.g. many Xanthophyceae and Chrgeophyoeas (Bogorad, 1962). A caae might therefore perhaps have been made out for greater antiquity of some groups of Chromophyta compared with Chlorophyta had Christensen wiahed to do 80. He has wisely avoided this question, but it ia perhaps uaeful to note that an argument in favour of the converaeposition,i.e. for greater antiquity of Chlorophyte, versus Chromophyte, could scarcely have been sustained at all. It ie therefore important that in the list (p. 4) the group to be mentioned last ie no longer the Rhodophym (Rhodophyta) aa in Fritsch (1936) and Smith (1961), but the Chlorophyta themselves. With this point of view I personally am in full agreement. The origin of iaokont flagellation aa such is not aa olosely linked with the origin of the Chlorophyta aa has sometimee been thought. It ia important here to realize that the tacit assumption eometimes made by algologiets that isokont flagella are in themaelves primitive in an aeeump tion baaed on indirect rather than positive evidence. We have in fsot

8

I. MANTON

no information regarding the nature of “the primitive flagellate’’ and it is probably important to realize that while many Chlorophyceae are isokont other members of the Chlorophyta (Euglenophyceae, Loxophyceae and Prasinophyceae on Christensen’s terminology) are not. Conversely some members of the Chromophyta, notably the group of genera sopnrated from the Chrysophyaeaeby Christensen under the title Haptophyceae, are isokont, e.g. leochryuiu (Parke, l949), Chy80chrmulinu (Parke et al., 1966,1962etc.), P q r n d u m (Manton and Leedale, 1963) and the Coccolithophoridae (Parke and Adams, 1900). The differences between these and the isokont Chlorophyceae are nevertheless profound and include general features of cell organization as well as details of flagellar bases, some of which will be discussed below. The possibility must be accepted that isokonty may have arisen more than once within the Contophora, though in all 0&888 from an essentially unknown type of ancestor, and its presence is thus in some ways less informative than is heterokonty when all the various ancillaries are fully developed.

IV. OTHERASPEUTB OF

B’LAGELLAR

NUMBEW AND RELATIVE LENGTH

The uniflagellate condition, encountered sporadically in many groups, is sometimes manifestly a secondary development by suppression of one member of a former (usually heterokont) pair. An extremely clear example of this is the brown alga Dictyota (Manton, 1969c)in which the uniflagellate spermatozoid retains a basal body for the hind member of a heterokont pair which fails to develop further. The single flagellum is clearly the of members of the Chrysophyceae such as Mdequivalent of the front flagellum of their heterokont relatives within the same group (Pihlka, 1949).It is possible that the single flagellum of fungi such as A l h y c m or OZpidium is aim&& the equivalent of t b hind flagellum of a former heterokont pair, but this is less certain. On the other hand, the nature of the single flagellum in green flagellah such as Pedinmnonas, Fig. 4, (Ettl and Manton, 1964), hficronumae (Manton and Parke, 1900),hfowmmtiz,Fig.6, etc.,is entirely uncertain. These organisms are small and apparently simple. They lack sexual processes and often lack true starch, though they posseas chlorophyll 6. The single flagellum they contain could be primitive. Whether this is in fact the case is however almost impossible fo prove. The special case of Euglena can be dismissed &om this category. While it is probable that all members of Euglenophyceae will eventually prove to be biflagellate (G. F. Leedale, personal communication), it is well established (Pdngsheim, 1966; Leedale, 1966) that all representatives, both aolourlessand coloured,previously described as hawing aaingle

FLAGELLAR STRUUTURE I N P L A N T S

9

flagellum bifurcate at the base, are biflagellate, though the moond flagellum if3 80 short that it fails to omerge from the oantll (formerly termed reservoir or gullet). Much elementary teaching could usefully be corrected on these points. Multiple flagella (beyond two) have certainly been achieved independently more than once and major phyletic conclusions based on this character have to be made with caution. The best known examples of multiple flagella are at present Oedoqonium (Hoffman and Manton, 1962, 1963) among algae and the gametes of many archegodate land plants (compareMsntonandClarke,1961a,withMmton, l952,1959b), all chlorophyll b-containing organism. The quadriflagellate condition is however a special case which should perhaps be mentioned first. It might have been expected that a group of four equal flagella would have been directly related (as a relatively simple precursor) to a large ring such a8 that of Oehqorjium (Fig. 10). This however is not 80. A recent study carridout in detail from this point of view (Manton, 1964a) has shown that at least in 8tigeoclonium and D r a v m U i a (members of the Chdophorales 8emu Christensen) the quafdrifiagellate condition is achieved by the grouping of two isokont flagellar pairs in a mirror image relation (Fig. 11A) and the same is almost certainly true for Ulothriz (Manton, 1962). Other quadriflagellate green cells, e.g. P y ~ a milnonas (Manton et at., 1963),Platymonas, 8 p ,may relate back to uniflagellate or biflagellate relatives in which flagellar number has increased in a different manner. Some information about these will be given below (pp. 9,11,17 etc.). It is enough to say here that the grouping of fourflagella seems to have occurred within the green algae many times and few, if any, are in a direct line towards either Oedogonium or a fern s*rmatozoid. While the origins of the special types of multifbgellate motile oelh characteristic of Oedogonium on the one hand and many land plants on the other are therefore still obscure, a positive finding relevant to the latter is perhaps important. The biflagellate condition of bryophyte spermatozoids is sometimes (e.g. 8phugnum, Manton and Clarke, 1962) undoubtedly the expression of two flagella of different lengths, attached at different points on the cell and therefore not a pair in either the isokont or ordinary heterokont mnse. Whether this oonditioa is primitive or specialized (i.e. due to reduction from a multiflagellate type) is unknown but it indioates that flagellar number is not in itself the moat significant character, but rather the overall symmetry of the cell. The spermatozoids of archegoniates are fundamentally aaymmetrical no matter whether they are bi- or multiflagellate. There is therefore no ease for attributing an isokont ancestry to them. The common textbook practice of treating C?damydummaa aa an unequivocally primitive

10

I. M A N T O N

type is thus valid, if at all, only for certain groups of green algae but not for the mainstream of phyletic change leading to the vegetation of the land. It is perhaps important to note that flagellar size as such is not directly connected with flagellar number, though minor phyleticallydetermined differences exist. In the large compound zoospores of Vaucheriu (Greenwood, et el., 1967) the average length of the shorter member of each flagellar pair was 9-8p and of the longer member 11.1 p. In Oedogonium, for which abundant numerioal data have ale0 been assembled (Hoffmann and Manton, 1962, 1963),an averago length of 174 p was encounterod both in epermatozoids with 30 flagella per cell and in Z O O E ~ O ~with ~ E over 120. The biflagehte gametes of bryophytea may give measurements of slightly greater length, e.g. 20p for both Marchantia (Manton, 1962) and 8phagnum (Manton and Clarke, 1962), but those of a fern (Dryoylleris) are only slightlylessthanthisat 18p (MantonandClarke, 196la;Manton, 1969b), though flagellar numbers in ferns may reach three figures. Uniflagellate cells are more varied, though lese so if the exceptional condition of the very reduced species Nicronumas puailb (see below and Fig. 6A) is excluded. P e a l i v rniw (Fig. 4A)with a cell only a few microns larger than N.puailb has a length for its single flagellum ranging from 10-17 p according to conditions. The extreme shortness of the hind flagellum in many heterokont members of the Chromophyta iR thus a sign of fundamental asymmetry of the cell as a whole, since it is in no way conditioned by flagellar number as such.

V. FLAGELLAR SHAPE The only feature of flagellar Rhape of importance in the present context is that of the tip. ThiR region, however, oannot be effectively studied by light microscopy uncontrolled by electron microscopy since flagellar tips are often the sites of structural breakdown either in vivo or after fixation, leading to artifacts of very peculiar kinds which need to be recognized for what they are before conclusions can be drawn. If distal breakdown ocour~without rupture of the flagellar membrane, osmotic swelling may lead to apparently spherica1 apices which have been desoribed aa such more than once by light microscopists (see for example Strasburger, 1880,for Vuucheria).Discussion of this artifact in relation to bryophyte epermatozoids will be found in Manton and Clarke (1962).On the other hand the flagellar membrane may beoome wholly or partially stripped offleading to an appearance aa of a long distal thin hair. This artifact is almost certainly in part the r w n for the special term “acronematic”, introduced by Deflandre (1934)for

FLAUELLAR S T R U C T U R E IN PLANTS

11

flagella lacking lateral hairs, since the early drawing8 of stained flagella (e.g. Fischer, 1894) show extreme developments of this kind whioh are mrtainly paztly artifacts in the light of subsequent eleotron mioroscopical investigations. The most extreme o w e of all, in which the whole flagellum is disrupted, displaying its fibrillar oompition, ie usually recognized for what it is (oompare for example Ballowitz, 1888, with Manton, 1960, and Manton and Clarke, 19/51), but even extreme disruption commonly starts at the flagellar tip and may not affect the base. Since the component fibrils of a disrupted flagellum are individually below the limit of effective resolution with the light microscope unless dried,suoh a flagellum in a stained preparation may appear to show a faint, splayed, distal extremity on a more densely staining narrower base. Another complictction that has to be remembered is that of occwional abnormalities which may simulate one or other of these conditions. An early example (in Fucw 8erratw) of an exceptional plant giving a long terminal “hair” as a genuinely structural addition to the front flagellum of its male gametes which other plants of the same species do not normally show, was illustrated in Manton and Clarke (1961~).When therefore a similar condition is described for the zoospores of Chordaria on the coast of Denmark (Petersen et al., 1968), such a description nee& to be followed up by a re-investigation of the same species in another locality before the suspicion of structural abnormality in the material can be excluded. Abnormalities and damage apart, there may be other difficulties in determining the true shape of flagellar apices in dried material examined optically or electron microscopically.The flagellum itself may be coated in some way and its tip not directly visible. The apparently te-1 tuft, of hairs described for M&mnnmua 8qu.amata (Manton and Parke, 1960), Pyrarnimmw, etc. (Manton et al., 1963), belongs to this category and further information about thew genera will be given below (see also Fig. 9). Knowledge of the internal structure is also probably more necessary to interpretation than has generally been recognized and this has rarely been provided for the flagellar apices in plants (see however Roth and Shigenaga, 1964, for some pertinent details for animals). The smoothly tapered cilia of ferns and some bryophytes seem likely to depend on the early termination of the two central strands followed by gradual petering out of the strands of the peripheral ring (evidencefrom dismembered speoimens, Manton and Clarke, 1961a, 1962). In other cases, notably AZicrowwncce pwilkc, Fig. 6A (Manton, 1969a), the central strancb are prolonged beyond those of the peripheral ring giving a long and slender but membranebounded terminal extremity, suggesting a hair-point when seen with

12

I. MANTON

the light microscope. One of the commonest types of flagellar ending8 is, however, shortly mucronate (Fig. 2B),but there is no information on the internal tip structure in any of the organisms with tips of this kind and it is therefore impossible to know whether a mucro L only a short “hair-point” or a genuinely separate morphological category. I personally suspect the latter, but morphology alone provides insufficient guidance in this’case. While therefore it is probable that phyletic information may ultimately be obtainable from apical shapes of more than one kind, there is only one context at present in which these seem likely to be important in the immediate futuk. Among small green flagellates in the group provisionally assigned to Loxophyceae by Christensen, the long “hairpoint” of Nicromoms pmilh has already been mentioned. It is roughly 2 p long on the tip of a flagellum only about 1p long. The hair-point is smoothly covered by the flagellar membrane and is unquestionably not an artifact (Fig. 6A). This peculiar little organism, the smallest known true flagellate, remained for several years in an anomalous taxonomic position (for furfher discussion, see Manton and Parke, 1960). Two other recently investigated genera of small green flagellates are nevertheless showing (Fig. 6) with four signs of bridging this isolation. In Sp-sir, flagella and Mononu;cstix with one flagellum (Fig. 6B), each flagellum is effectively bipartite ,possessing a long distal extremity conspicuously thinner than the almost equally long proximal part. This feature waa correctly recorded by light microscopy at leaat for Spemul-aie (Korschikoff, 1913) and ie undoubtedly a good morphological character. Since an affinity between these genera and M. pueilh can almoet certainly be justified on other grounds it is probable that these oharao. teristic flagellar apices may prove to be useful phyletio indicators within this group when more is known. However, it is precisely for this reason that the continued u88 of the word “acronematicJJfor any of these flagellar types, even those last discussed, is undesirable. The term suggests the presence of a terminal hair comparable to the lateral hairs which will be discussed in the next section, which is in fact never the caw. The word Peitechenge&8d (whiphh flagellum) is not inappropriate for the hind flagellum of a heterokont pair (Fig. 2A) a8 a contrast to F&mmrqek?e,! (hairy flagellum, Fig. 2B)for the front flagellum of the same pair. In all other cases ordinary botanical language is greatly to be preferred, e,g. mucronate, tapering or bipartite. For the most frequently needed distinction, that between hairless and hairy (with lateral hairs), latinized words such as glabrous and hirsute or their translated equivalents are available for use in languages such as French which lack a

PLAGELLAR STRUCTURE I N P L A N T S

13

simple vernacular word for “hairy”. My recommendation is therefore strongly that “acronematic” should be deleted from the vocabulary. Further comments on nomenclature will be found in the next motion.

VI. FLAGELLAR APPEXDAGES Information on flagellar appendages of the sort variously termed Flimmer or mastigonemes (see below) dates from the nineteenth century, more eapeoially Loeffler (1889) and Fisoher (1894). Qood reviewrt of the light microscopy will be found in Petersen (1929), Vlk (1931) and Deflandre (1934), with important personal observations by Petersen (1918) and Koch (1951). Apart from this brief list of excellent papers, the first fifty years after Loeffler were on the whole dominated by indifference if not frank disbelief by most botanists. The importance of electron microscopy in changing this picture has been reviewed several times, see especially Pitelka (1949, 1963) and also Manton (1952, 1954, 1956), and no attempt will be made to diacuas the whole field again. Consideration can nevertheless usefully be given t o topics not adequately covered by previous reviews and some of these are indeed to be found in relation to each of the more diverse categories of flagellar appendages encountered in plants, namely, spines, hairs of various kinds, and scales. A. FLAGELLAR SPINES

th el^^ spines are of special phyletic interest in relation fo certain more recent observations on the nature of the primitive Fucoid to be disoussedbelow (pp. 19-20). Thereareonlythreeknowncases,allmembeFs of the brown algae. There R i a Ringle spine on the front flagellum in Hinucntlralia (Manton el al., 1953) and another in a 8irnihr portition in Xiphqhura (Manton,1956), a Fucoid from the southern hemisphere. I n the third example, Diclyola (Manton et al., 1963), there GIa row of spines. A spine seem to be an excrescence caused by a lump of some unknown material borne on one of the peripheral fibm and therefore beneath the flagellar membrane. A phyletic interpretation is not available for Dictyota though the character waa of importance in determining the plane of “symmetry”* aa passing between but not through the two central strands (Manton, 1959). A phyletic interpretation

*

That strict bilateral symmetry does not exist in fbgella follows from the demonetration of spiral aqmmetry so clearly made for flagellar beses by Gibbone (1961). Gibbons and Grimetone (1960) and others. Nevertheleae the central strands per ae have 8 symmetrydoftheir own m d e fixity of position in relation to other attributes of e flagellum which it ie impo.tent to determine. Tbie ie therefore the sense in which the words “fhgeUar eymmetry” rn intended in the present context.

14

I. M A N T O N

supporting a common origin for X i p h p h m a and Himanthalia, on the other hand, is a matter of considerable btereet. Xip?uq~homdiffere from Himccnthalia in lacking the basal “button” and is therefore probably more primitive. It is otherwise not dissimilar since it mnsists essentially of dichotomous ribbon-like thalli with conoeptacles on both surfaces as in HimanthuZkz. An a w t y between these two genera in spite of their present wide geographical separation could be argued on general morphological grounds. The demonstration that they share a character as unusual as that of their flagehr spine nevertheless gives force to this comparison in a way which is perhaps important. B. FLAOELLARHAIRS

FZageZlar hair8 have recently become a serious source of confusion aa a result of three entirely different factors, namely, the nomenclature, the accidental introduction of errors of fact into the literature and genuine increases of knowledge. Nomenclature can at this point be relegated to a footnote” since my own strong recommendation, a8 already explained, is to discard for the time being all special terms except perhaps tho two oldest (Flirnmergei88d and Peit8chengei88d) limiting these however to the special case of the two differentiated members of a heterokont pair in the mnventional heterokont groups (Xanthophyceae, Chrysophyceae, Phaeophyceae and the water moulds). For the appendages themselves in these and other cases the word “hair”, at least to readers of English, is unambiguous and I propose to use it in the account which follows since it carries no overtones suggesting uniformity in all contexts which a single specially coined technical term such as “mastigoneme” inevitably does. The scepticism of biologists in accepting the concept of b i r a on flagella at all is connected with the obvious diffioulty in distinguishing fixation artifacts fiom real structures which are at or just below the threshold of visibility with the light microscope. A mistake of a Werent kind is exemplified by Thalaasomonae Butcher, a genus of small green flagellates, described (Butcher, 1969) as possessing full heterokont flagellation. However, this description has recently been shown (Parke and Rayns, 1964) to be due to confmion betwoen two *The more commonly encountered terms other than IrlhmrgeisseZ end Peitachengeiesel(Fiecher, 1894) were introduced by Deflandre (1934) end englicized by Pitelke (1049). who ale0 lists them in Pitelke (1963). On this terminology, the noun w t i g o n e m denotea 8 &gellar hair of any khd; the adjmtive pantonemtk denotes e f h g e h n which when Been in profile ehowa hsh on both sides; 8 t h h m ~ a C Aagellum with bira on one aide only; end amonematic 8 terminal hair (diecuseed on p. 12). Other terms, of which there are an increasing number, do not yet 8igniScSntly effect the general literetuw.

FLAQHILLAR STRUOTURID I N P L A N T S

15

different organism present together in the same culture, namely, a pigmented organiem identioal with Micromo7~cs8qtubmatQ (Menton end Parke, 1960) and a much larger colourless chrysophyoean predator. This naturally displayed hairs on its front flagellum of the normal heterokont type charaoteristic of Chrysophyceae, when examined electron microsoopioally by another observer whose miorographs wem reproduced. The genus Th.akceeomonas has thus to be rejected aa a nomen confusurn, synonymous otherwise with Xkwmunaa 8quu&, and all statements based on it, claiming the presence of true heterokont flagellation in a member of the green algae, should be deleted also. This does not, however, mean that hairy flagella as such are absent from members of the Chlorophyta and indeed there are hairs of more than one kind and in several different types of arrangement. For comparison Fig. 2 illustrates the type of hair present on a true Ff!irnwrgeissef! of a brown alga, in this case represented by the front flagellum of the zoospore of Scytoeiphon. The hairs are co&1ge,arranged in two rows (one on each side of the flagellum), but becoming sparaer towards the tip which is shortly mucronate. Each hair appears to be bipartite with a slender distal extremity emerging from a somewhat thicker proximal part. Whether these slender extremities are present in life or represent molecular threads ejected from the centres of tubes (the thicker bases) as a fixation artifact is unknown. F i p 4, on the other hand,illustrates the very different type of hair present on the single flagellum of the type-species of Pedinomrmc~e(Loxophyceae within the Chlorophyta8ewu Christensen).Thereare again two rowsof lateral hairs, but each hair is sothin aa to make the detachedbacterialflagellumpresent in the same field appear coarse in comparison. Moleculer threads are again suggested, though in this case the idea of a f h t b n a r t W i s m o r e difficultto introduce in view of their very regular arrangement. What effect, if any, their presence may have on the mechanics of swimming is impossible to determine, but a substantial mechanical effect seems excluded for threads as thin as this. In a somewhat similar caw, Chlorochitridim t u b e r c u b Vischer (Pedinomonacr tuberculata (Vischer) Cams), the hairs are more numerous and longer (Manton and Parke, 1960) but they are individually no thicker than in P, minor. Their exact arrangement in ChZorochitridion should be re-examined, since the “plume-like” appearance might in that o w have been caused by an all-over hairiness rather than by restriction of the hairs to two ranks. This possibility is excluded for P.m i w (rig. 4) and also for a true li”limmergei88d since the alignment of the point6 of origin of hairs, either singly or in tufts, w(t8 clearly demonstrated at an early date (Manton and Clarke, 1966;see also Manton et aZ., 1952). The special flagellar characteristicsof the euglenoids have often been

16

I. M A N T O N

illustrated (see especially Mainx, 1928; Manton, 1962; and may be taken as generallyknown. Thelimitation of the longest hairs to oneside of the flagellum (Fig. 3) was detected by Fischer (1894) and hae not subsequently been seriously disputed. Nevertheless to w e the name PZimmergeisseZ for this as well as for the front flagellum of a hetemkont alga is almost certainly ill-advised. Not only is the arrangement of hairs different but the individual hairs, not separately visible with the light microscope which can detect only hair tufts in a stained preparation, are very slender and, though longer and more copious than in the type species of Pdinommure, are definitely not comparable either in structure or in arrangement with the hairs of the heterokont algae and fungi. The presence of thick hairs of more than one kind has, however, also been recorded on unequivocal evidence within the Chlorophyta. Among members of the Volvocales the surface of the whole cell, both wall and flagellar membrane, is sometimes covered by a close tomentum, at least when seen in the iixed condition (Fig.7). The evidence of dividing ceUs (Fig. 7D) indicates that this apparent tomentum is almost certainly derived from material deposited upon the new daughter cells before liberation from the old wall of the mother cell. An extreme case (Fig. 8) is that of Haemtococcus pluvialis which has not yet been studied developmentally but which has a very conspicuous tomentum limited to the flagellttr surfaces only. The hairs in this c a m are very translucent when examined in whole mounts (Fig, 8A) though they were recorded as present as early as 1961 (Miihlpfordt and Peters). I n sections on the other hand they are extremely compicuouc(, Wme being apparently compound and curly. They cover the flagellar surface all over including the tip, but they are not an integral part of its structure since they are clearly related to a separate layer of material of unknown nature laid upon the surface. A different type of thick hair (Rg. 9) has been encountered among many genera of Christensen's new group Prasinophyceae (see List p. 4) and also in some colourless Cryptomonads, notably Chilomonae (Pitelkaand Schooley, 1966). Examples illustrated here are on the tip of the flagellum of illicrommras s q u a m t a (Fig. 9A), the surface of the flagellum in Heteromastix (Fig. 9B)and scattered fiee on the field near a group of flagella from which they have become detached in H&ephmra (Fig.9C).Such hairs are often slightly curved and tapered at both ends. They commonly show a pattern of transverse s t ~ t i o m when seen in section and they have recently been shown, in the special case of Heteromaetix (Manton et al., 1965), to be msnuftlcturod in vesicles within the body and to be depoeited thenos, together with scales, upon the surface.

FLAGELLAR STRUCTURE I N P L A N T S

17

C. FLAGELLAR SCALES

These last examples lead naturally to some of the most peculiar flagellar appendages so far discovered, namely, jlagellccr scales (Fig. 9). These were first encountered in Micromonas s q w w t a (Manton and Parke, 1960)as a single layer of fairly large scales covering the whole of the cell including the single flagellum (Fig. 9A). For some years this remained a unique occurrence, but now many other cases are known (Figs. 9B and 9C), with possible consequences to the taxonomy and nomenclature (includingthe naming of M . aqwmata itself) which have not yet been fully determined. Nephr08ehi8 gilva (Parke and R a p s , 1904) has a single layer of fairly large scales over body and flagella, though here the flagellar scales are not quite like those on the body, and the flagella themselves are two, somewhat unequal in length. Still more complex examples with scales in two layers (Fig. 11B) and with morphological differences affecting both layers according to location on the organism have been encountered in Pyramimonae, Hahaphaera, Pteromonas,Heteromastix,Prasimladw, Plutymonm to name only a few for which the facts have been published or are nearly ready for publication (Manton et aE., 1963; Manton et al., 1905; M . Parke and I. Manton, unpublished). It is possible that Christensen’snew group Prasinophyceae may eventually be used to oontain them all, though t o do so will involve bisecting the genus Micromonas since M . pwilkc, the t p species discussed on p. 12, has already been assigned to Loxophyceae. These problems need not, however, seriously affect a general botanist. The grouping of difficultgenera into larger units is a special taxonomic exercise which not many people will want to undertake. The important point in the present context is not the names but rather the undoubted fact that flagellar scales as well as flagellar hairs seem likely to provide some phyletic information of a new and interesting kind even though this has not yet been completely assimilated into the nomenclature of any existing system.

VII. FLAGELLAR BASES The use of flagellar bases as phyletic indicators is only just beginning. The absence of the central strands was the only significant observation before the introduction of0the improved methods mentioned under phase 2 (p. 2), but these, in the hands of Gibbons and UrimrJtone (1900), gave results of such dazzling clarity that they have remItined unchallenged. These authors showed among other things that, in the flagellar bases of certain colourless endoparasites, the peripheral strands became triplets instead of the doublets present in the 9+2 region. Each triplet group when seen in transverse section appeared tilted radially, C

18

I. MANTON

giving the type of spiral asymmetry referred to above, and within this region of tilted triplets or at least for a considerable part of it an elaborate central pattern, sometimes referred to as the “cartwheel pattern”, occupied the centre of the basal body. These observations have been substantially confirmed by other workers on several animals and plants and, until recently, major diversity was not bclieved to occur. Nevertheless there is diversity, at least among algae. That the flagellar bases of certain green algae are not identical in all respects with those of Gibbons and Grimstone’s material has been imperfectly shown in a general way several times. Some good transverse and longitudinal sections through the transition region in a species of Polytoma (colourless relative of Chlamydomonas) published by Lang ( 1 963), and two detailed studies by myself, carried out independently on several pigmented algae (Manton, 1964a,b), have clarified the factual position. There is no doubt that Gibbons and Grimstone’s triplet strands (Fig. 11G) and a form of the cartwheel pattern (Fig. 11H) are present in all these bases, the cartwheel pattern being characteristic of the extreme lower end of a flagellar base, but the transition region between base and flagellum proper shows a spectacular stellate pattern which was not observed by Gibbons and Grimstone. In the green algae, and indeed probably in most if not all chlorophyll b-containing organisms, the stellate pattern is so strongly developed as t.0 arrest the attention (Fig. l l D ) , but elsewhere it has to be looked for with care even when present and it is not yet known whether indeed it is always present outside the Chlorophyta and land plants. In a search for it among the Chromophyta it was found with difficulty (Fig. 11C) in Prymnesium parvum (formerly Chrysophyceae, now Haptophyceae sensu Christensen). The flagellar base of this organinm nevertheless proved to be so unlike those of both Gibbons and Brimstone’s material and the members of the green algae studied that a view of any one of three or four different levels would have differentiated it phyletically at a glance. Nevertheless even here the extreme base had triplet strands and a cartwheel pattern in no way different from that of almost any other plant or animal. The fixity of the 9+2 thus seems to be shared by some but not all the structural features of flagellar bases and it R i certain that these will repay closer study in a wider range of forms than have hitherto been examined.

VIII. FLAGELLAR “ROOTS” The phyletic signficance of the fibrous appendages internal to the cell, which radiate from the region of the flagellar barns, repreuents a new field which has only recently been effectively studied in plants.

FLAGELLAR STRUOTURE I N PLANTS

19

Only a limited number of examples are yet known in detail, but these are all in their various ways important. One of the simplest to describe though by no means the simplest t o explain is Oedogonium. The well-known crown of cilia in the zoospore and spermatozoids of this familiar green alga were studied in detail b y Hoffman and Manton (1962, 1963) who showed not only the fibrous substructure giving stability to the ring (Fig. 1OB) but the presence of radiating roots arising singly between each pair of ciliary bases and therefore equal in number to the cilia themselves (Fig. lOA). Each root proved to be bipartite (Fig. IOC) with a relatively short rod of crossbanded densematerial underlying a ribbon of three tubes or coated fibres running longitudinally beneath the plasmalemma for possibly the whole length of the body. No exactly similar case has yet been encountered elsewhere since in other green algae there are normally two kinds of roots arranged alternately no matter what the total number of cilia (flagella) may be, i.e. whether one (Pedinomonua), two anisokont ( H e t e r o w t i x ) , two isokont (Chaetomorpha) or four isokont (Stigeoclonium, Darprnaldia, Ulothrix). Roots of all these genera contain tubes or coated fibres in very definite numbers. The commonest fibre number ie two (Fig. 1lE), encountered in all these genera though in each case in only two of the four roots present. The other two roots, arranged alternately with these in tt cruciform manner, have been Rhown to possess five fibres (Fig. 1IF) in Stigeoclonium, Draparnaldia, Ulothrix (Manton, 1952; Manton et al., 1955; Manton, 1964s) and three fibres in Pedimmna.9 (Ettl and Manton,, 1964) and probably Chaelomorpha (Manton et al., 1955). This degree of numerical uniformity was unexpected and is difficult to explain except perhaps in terms of an ultimately monophyletic origin for very diverse chlorophyll 6-containing organisms now distributed in several different sections of Chlorophyceae, F’rasinophyoeae and Loxophyceae. This sqbject therefore deserves further attention with the expectation that the exact phyletic meaning may become better interpretable when more is known. The same may be said for flagellar roots in members of the Chromophyta, though here the pattern just outlined is not repeated. Instead we have some extremely peculiar but equally characteristic structures which have been studied most fully among brown algae. Thn need for doing so was determined by the existence of a special organ, the 00called proboscis attached to the flagellar apparatus of Fucuis (Manton and Clarke, 1951c), in a position apparently identical with that of an internal “root” detected later (Manton, 1959~)in the uniflagellate spermatozoid of Dictyota. This matter has recently been explored in greater detail (Manton, 1964c) with the discovery that the proboscis

20

I. M A N T O N

of Fucus is indeed homologous with a special type of internal root present generally in brown algal zoospores as well as in the male gametes of more primitive Fucoids, notably Halidrys and C‘ystoseira. The phyletic importance of this is considerable, since it indicntos among other things that Fueus itself is more specictlized thtin has sometimus been thought, while Cg~toseiruis Rubstantially more primitive.

IX. CONCLUSIONS

It is thus clear that flagellar characters have value as phyletic indicators in many more ways than could have been known even ten years ago. Emphasis can nevertheless usefully be laid here on two major sources of confusion which must at all costs be avoided if the full value of these somewhat unfamiliar criteria is to be utilized. One is the type of careless mistake leading to seriously erroneous records as discussed on p. 14 with respect to “Thahsomonae”, and the other is premature systematization of nomenclature which can obscure genuine and phyletically significant differences under the artificial unity of a system of Latin names, as discussed under flagellar apices (p. 12) and flagellar hairs (p. 14). The two most important attributes for work of this kind are meticulous care in recording and an open mind in interpreting. Given these, together of course with reasonable botanical knowledge and manipulative skill, accumulated information will almost automatically clarify interpretation provided that it is collated with other sources of evidence. It cannot be emphasized too strongly, however, that for phyletic interpretations there is no a priori guidance as to what characters will in themselves be of major and what of minor importance. Only when the work is finished will this become apparent. It is essential therefore to use information from fine-structure as additions to the pool of knowledge provided by general morphology, biochemistry, life histories, etc., so that phyletic conclusions when reached are based on all the evidence and conflict with none. Knowledge of very few plant groups is in this happy state and therefore it may assist a general reader without first-hand experience of the particular characters under discussion if a summary of what I personally regard as reasonably well authenticated conclusions in the present state of knowledge be appended.

X. SUMMARY (1) Absence of flagella in the case of Red Algae, combined with other facts of fine structure (notably the platid lamellations) in which members of this group seem primitive, is I believe correctly interpreted

FLAGELLAR STRUCTURE I N PLANTS

21

(Christensen, 1962) as indicating greater relative antiquity for this group than for any other type of pigmented algae after Blue-Green Algae. (2) On thesubject of isokont versus heterokont flagellation, the following points have been brought out. (a) Heterokonty in the strict sense is not found in Chlorophyta though it occurs in several groups of’ organisms lacking chlorophyll 6, collectively termed Chromophyta by Christensen (19f32),together with their colourless tlorivativen, notably the water moulds. The Chromophyta are tLt leant an d t l and could be oltlcr than the Chlorophyta. ( b ) lxokonty H i found in the Huptophyceae among Chromophyta and in the Chlorophyceao among Chlorophyta. There is no close affinity between these two groups. Anisokont members of the Chlorophyta occur in the Euglenophyceae, the Loxophyceae and the Prasinophyceae. The fundamental asymmetry of the motile cells of land plants points to an anisokont ancestry within the Chlorophyta, for which therefore Chlamydomonas is not on the direct line. ( c ) This means that the nature of the “primitive flagellate” within the Contophora is unknown, but it must not be assumed that it contained chlorophyll b and it need not have been isokont. (3) External flagellar characters of value as phyletic indicators include : (a) apical structure (especially in certain anisokont greens), ( b ) flagellar spines (brown algae), (c) flagellar hairs (different in Chlorophyta from Chromophyta but of value in different ways in both), ( d ) flagellar scales. (4) Internal flagellar chaructw of value an phyletic indimtore include : ( a ) flagellar bases (different in Chlorophyta and Chromophyta), ( 6 ) flagellar roots (different in Chlorophyta and Chromophyta). ( 5 ) A special case for which the sum of flagellar characters (including external morphology and the micro-anatomy of roots) has given important new insight is in the Fucales, where Cyatoseim now appears t o be more primitive and Fucua less so than has commonly been thought.

22

I. M A N T O N

LEGENDS OF FIQURES 2-11 FIG.2. Scytosiphon zoospore to show normal heterokont flagellation. A: showing part of the body with a romplete hind flagellum (Peitschengeissel) and part of the hairy front flagellum (Flimmergeissel); micrograph C5933, x 15,000. B: the tip of a front flagellum, C3706, x 20,000 (Bfter Manton 1964r). FIG.3. Euglena apirogyra flagellar tip x 20,000 (courtesy Dr. G. F. Leedalc).

FIG.4. Pedinomonae minor. A: complete cell showing the single flagellum, 5797 /21, ~3000. 13: flagellar tip showing very fine lateral hairs with an S-shaped detached bacterial flagellum lying awow the wpecimen; mirrograph C5006, ~ 3 0 , 0 0 0(after Ettl and Manton, 1964). B:

h a . 5. A : Micronionnr pueilh Hection of a complete re11 xE0,000 (after Manton, 1959). Yottomtinliz eczeca (courtesy 1)r. J. H. Belchor). Micrograph C5015, x 20.000,.

Pro. 6. l~lugellsrtips from R freshwatcr plankton sample (courteay 1)r. H. Ettl); two dagells from ~tL‘hlrrmydomonnn species below; four flagella with long “hair-pointa” from a Spermloroopin cell abovo. Micrograph C3028, X 15,000. PIG. 7. Chloropnium ronue (courtesy Dr. H. Ettl). A: transverse section of a flagellum ahowing the superficial tomentum; micrograph C3137, X 50,OOO. B: oblique longitudinal section of a flagellum near the surface of the subtending cell, both showing tomenturn; C3151, ~40,000.C: longitudinal section through a cell tip with emerging flagellum t o show distribution of tomentum and differing thicknesses of wall layers below; C3170, x40,OOO. I): part of two daughter cells within the mother cell wall to show probable origin of tomentum; C3156, X40,000. FIG.8. Haemutococcus pluvialis (courtesy Dr. H. Ettl). A: shadowcast flagellar tip showing very transparent tomentum (contrast with the detached spine of another organism top right), C7299, ~30,000.B: transverse section of a flagellum with tomentum attached to a separate layer of material outside the plasmalemma; C8208, ~60,000.C: oblique longitiidinal sertion of a flagellum with tomentum (N.B. the lower end of the profile is not a flagellar tip); C7439, x 40,000.

Fro. 9. Examples of flagella with hairH and wcalen. A: Micrornrmna eqmmulo, x30,CNN (after Parke & Rayna 1964). B: Hetrromu8lix rotunda, x30,WK) (after Manton el ul., 11965). C: Halosphnera sp. (courtesy Dr. M. Parkc), mic-rograph BW2r5, ~30,000. Fro. 10. Ocdogonium cardiaeum npermatozoicl (aftor Hoffman and Mariton, 1963). A: flagellar ring with bases and ‘ ‘ m O t H ” , X 16,000. B: section whowing the cartwheel p&ton, in flagellar bases and the various fibrow hands connecting them together, xf%,fw). (;: two adjacent ‘‘mots” cut transversely showing in each the nolid rod below and tho ribbon of three fibres or tubes above, Xm,oOo.

Frcr. 11. Flagellar bases and “roots” in variouH algae. A: Stigeoclonium Hertion of a tip of a settling zoospore ~30,000 (from Manton, 1964a). B: Heleromaslix dovclb wction o f a flagel. lum covered with two scale-layers (see also Pig. BB); micrograph C5552, x 100,CXw). C: the “stellate pattern” in Prymnenium parvum, X 100,OOO (after Manton. 1964h). L): the “st&ate pattern” in Stigeockmium, X 150,000 (after Manton. 1984b). E :Stigeoeloniurn, two-strandd root, ~ 5 0 , 0 0 0(after Manton, 1984a). F: Sligeoclonium, five-stranded root, x IOO,ooO (after Manton, 1964a). G: triplet strands from the middle region of&flagellar base of P r ~ w i ~ m p r v u m ; micrograph C4238, x 100,OOO. H: triplet strands and cartwheel pattern a t innemoet end of flagellar base of Prymnesium parvum; micrograph C4613, x 1OO.OOO.

Fro. 3. Euglena.

Fro. 3. Euglena.

FIG.5. Micrononae and Monomaetix.

FIG.6. 8permatozoopie and Chkzmydomnas.

Fio. 7. Chloroqniurn rome.

29

FXQ.8. Uaemaloeoccua.

FIG.10. Oedogonium.

FIG.11. FIagellar “roots” and flagellar bases.

B L A Q E L L A R STRUOTUR‘BI IN PLANTS

53

REPERENOES A d d , J. and Thi&y, J. P. (1963). J. M h ~ p i 2,s 71-80. Afzelius, B. (1969). J. Bwphya. Bioohem. Cytol. 5, 289-78. Bdlowitz, E. (1888).AT&. dk.A M . 82, 401-73. Bogorad, L. (1982). I n "Physiology and Bioohemistry of Algae” (R.A. h*, ed.). AccEdemio P m , New York. Bouok, (3. B. (1962).J . OSU BWl. lQ1,153-70. Butoher, R. W. (1969). An introduotory account of the smaller algae of Brithh O O M t d Wetere. Part I. Fkh. Invest. hd.8er. 4. Chriateneen, T. (1962). Botanik. Bind 11: Syetemtik Botamk, Nr. 2, A b r . 1-178. Munkegeerd, Copenhegen. Dei%uuh,0. (1934). C.R. A d . Rcd., Paria, 198, 497-8. Ettl, H. and Menton, I. (1964). Nowa Hedw@a. In the preae. Fawoett. D. W. end Pgrt-ar, K. R. (1964). J. Morph. 94,221-82. Fieoher, A. (1894). Pringa?&m’e Jahrb.f. +a. Bot. XirVI. 187-236. Frifech,F.E. (1936).“structureesldreproductionof the Algee.”Vol.I.Wbri+ Gibbons, I. R. (1961a). J. Biophya. Bwchern. Cytol. 11, 179-205. Gibbona, I. R. (1961b). N d w e , Lond. 19, 1128-9. 7, Gibbona, I. R. and Grimatone, A. V. (1960). J. Bhphye. Biochem. U@. 879-716. Greenwood, A. D. (1969).J. exp. Bot. 10, 56-88. Greenwood, A. D., Manton, I. and Clarke, B. (1967).J. exp. Bot. 8,71-86. H o ~L.,and Manton, I. (1962). J. EXP.Bot. 18, 443-9. Hoffmsn, L.and Menton, I. (1963).Amer. J. Bot. 60. Jakus, M. A. and Hall, C. E. (1948). BioZ. Bull. 91, 141-4. Koch, W. (1961).J . E k h a Mitohell eci. SOC.87, 123-31. Korechikoff, A. A. (1813).Ber. dtsch. bot. Qw. 81, 174. Lang, N. J. (1963).J. cell. Bwl. 19, 631-4. Leedale, G. F. (1964).Arch. Mikrobiol. In the preaa. Loeffler, F. (1889). Zbl. Bakt. 6, 209-24. Luther, A. (1899). K . Bvenska, Vetensk A M . L@ud. 24, 1-22. Maims, F. (1928). Arch. ProtLqtenk. IJL, p. 306. Manton, I. (1960).Demonstration of compound cilia in a fern aperm8tozoid by meene of the ultra-violet microscope. J. exp. Bot. I,89-70. Menton, 1.(1962).The fine structure of plant cilia. R p p . Roc. exp. Bwl. 8,300-19. Manton, I. (1954). Microsnstomy of cilia, flagella etc. 139. Recent work on the internal structure of plant cilia. Proc. Int. Confevmce on EZectron Micro-

-,

London. 694-9.

Manton. I. (1966).In ‘lCellulsr Mechanism in Differentiation mid Growth’’, @. Rudnick, ed.). Princeton, U.8.A. Manton, I. (19698). Electron microscopical observetione on 8 very amall fl8geht.e. The problem of C h r d i m gnceilla Butcher. J . Mar. biol. A M . U.K. 88. 319-33. Manton, I. (196Qb).Observations on the mi&oantatomy of the spermatozoid of the Bracken Fern.(Pteridiwnaqudinwn.)J . Bwphya. B i o c b . Cytol. 6,413-18. Manton, I. (19690).Obeervationa on.the internal structure of the spermatozoid of Ddclyola. J. exp. BOt. 10, 448-61. Manton, I. (1961). P l a t Cell 8tmcture, In “Contemporery Botanioal Thought” (MoLeod and Cobley, eds.) pp, 171-97. Oliver and Boyd, London. Manton, I. (19848). ~bsewationson the Ane structure of the zooapore and young gemling of Stigeoolonium.J . exp. Bot. 16, 399-411. Msnton, I. (1964b). The poeeible signi5usnce of aome detaile of flagellar baeee in plants. J . R.nric?.. 800. 111,82,279-80. Mmton, I. (1964~).A contribution towarda underatanding of “the primitive Fucoid”. New PWZ. 68, 244-64. Manton, I. (1964d). Obeervstions with the electron microeoope on the division cycle in the flagellate Prymneeium p a r u n . J. R. Mh.6oc. 88, 317-26.

34

I. MANTON

Manton, I. and Clarke, B. (1960). Electron miaroscopical observatione on the spermetozoid of FUGUE. Nature, Lo&. 160, 973. 8 fern Manton, I. and Clarke, B. (196la). Demonstration of compowld C i h eprmetozoid with the electron microscope. J. Wp. Bat. 2, 126-8. Manton, I. end Clarke, B. (1061b). Electron microscopical observations On the zoospores of Pylaielb and Lanrimrb. J. q. Bot. 2, 242-0. Manton, I. and Clarke, B. (10610). An eleatron miorowope etudy of the sp-tozoid of Fucw aewatw. Ann. Bot. N.B. 15, 461-71. Manton, I. end Clarke, B. (1962). An electron miommope study of the spennetozoid of19phgnum.J. exp. Bot. 8,206-75. Manton, I. and Parke, M. (1860). Further observetione on small green fl&lab with special reference to possible relatives of 0-m Mlla Butohm. J . MW. h i . AEE.U.H. a 9 , 2 7 ~ ~ Manton. I., Clarke, B. end Greenwood, A. D. (1953). Further observatiom With the electron miarwoope on spermatozoids in the Brown Algae. J. c ~ pBat. . 4, 310-29. b t o n , I., Clarke,B. and Greenwood, A. D. (1965). Obeervationswith the eleotron microscope on biciliate end quadriciliate zoospores in Green &m. J . q B O ~6,. 126-8. Manton, I., Oates, K. and Perke, M.(1963). Observations on the fine structure Of the Pyramimorure stege of Hahaphaera end preliminary observations on three species of Pyramimonae. J. Mar. bwl. Aae. U . K . 48, 225-38. Manton, I., Clarke, B., Greenwood, A. D. snd Flint, E. A. (1962). Further observations in the struoture of plant, cilie, by a combination of Visual and electron microscopy. J. exp. Bot. 3, 204-16. Manton, I., Raym, D. G., Ettl, H. and Parke, M. (1066). Further observations on green flagellates with scaly flagella: the genus Heterommtiz Korshikov. J. Mar. biol. Aas. U.K.45, 241-66. Muhlpfordt, H. and Peters, D. (1061). 2001.Anz. 16 (Suppl.) 163-0. Parke, M. (1949). J , Mar. Biol. Aaa. U.K. 28, 266-86. Parke, M. (1962). J. Mar. biol. Aaa. U.K.32, 407-620. Parke, M. and Adams, I. (1060). J. Mar. Biol. Aaa. U.K.89, 263-74. Parke, M. and R a p , D. a. (1964). Studies on Marine Flagellates. VII. Nephro. aebnie gilwa sp. nov. and some allied forma. J. Ma?.. biol. Am. U.K. In the press. Parke, M.,Lund, J. W. Q. end Manton, I. (1062). Arch. Mikrobiol. 42, 838-52. Perke, M., Manton, 1. end Clarke,B. (1066). J . Mar. Liiol. Rae. U.K. 84,579-608. Pea~e,D. C. (1063). J . cell. BWl. 18, 313-26. Petersen, B. (1918). k w k . Ndurh. For. 69, 346-7. Petereen, B. (1029). Bot. Ti&&. 40,373-89. Petersen. J. €3.. Caram, B. and Haneen, J. B. (1968). Bot. Tidssb. 64, 67-60. Pitelka, D. R. (1940). Uniw. C d q . Publ. 2002.68, 377-480. Piteh, D. R. (1963). “Electron-microscope etruature of Protom8.” Pergamon h, Oxford. Pitelka, D. R. end Schooley, C. N. (1966). Univ. Cdif. Publ. 2002.61, 79-128. Pringeheim, E. a. (1948). BWZ. Rev. 28, 46-61. Pringsheim, E. G. (1966). N o ~ Acts a Leop. CWOZ.18.1-188. m h e i m , N. (1866). M d . A M . Wku. Berlin. 133-66. Roth, L. E. and Shigenege, Y. (1064). J. cel2. Bwl. 20, 240-70. Bmith, 0.M. (1061). Menu81 of Phycology. N.8. 27, Ohmica Botunim 00. W d W , Maaa. Stanier, R. Y. and van Niel, C. B. (1062). Arch. Mikrobiol, 4a, 17-36. strain, H.H.(1061). In “Menual of Phyoology” ( a. 116. Bmith, ed.) p. 262. Cronioe Bot~nic& CO., W81tham, Maasaehueette. Strasburger, E. (1880). Znltbildzcngund 2-w. 3rd edition. Jena. Thuret, G. (1861). Recherches sur lea Zoospores dea Alguw. Ann. 8ei. W . (bot.) 38 10, 1-13. Thuret. a.md %ma+, Ti! ‘11S’711\ “F+.4- DL----l-A---- r t D-21

Fundamental Problems in Numerical Taxonomy

.

W. T. WILLIAMS and M B . DALE Department of Botany. The Univeraity. 8outhamptm. Bngland

.

............................................................. 86 .................................. 87 . ............................... ;..................... a7 B. Monothetio and Polythetic Clwi5~lltione.................................87 C. Maximizcltion ......................................................... 88 D. Hierarchical and Non-hierclrahical Clessiflmtione........................... 42 E. Probabilfstic and Non-probabllisttoClwifirntiom.......................... A3 I11. The Choice of Mathemstioel Model.......................................... 48 A . Introduction: Metrica .................................................. 48 B. Metric Properties of Pair-functions ...................................... 49 C. Intrinaidy Non-metrio Byatam ........................................61 D. Non-Euclidean Syntamn................................................. 52 E. Concluaionn ........................................................... 63 IV. The BM~C Euclidesn Model ................................................. 54 A . Duslity: The R/Q Problem .............................................. 64 B. Adjustments to the Model .............................................. 66 C. Heterogeneity......................................................... MI V. Strategyofhlysis ....................................................... 60 A. Simplification Methods ................................................. 69 B. Partition ............................................................. 61 C. Non-hierarchical Metbode ..............................................81 L). Hierarchical Methods .................................................. 82 Acknowledgements........................................................ 67 Referenw ..............................................................87 I

.

Introduction

I1 The Nature and Propertien of Clessiflcutione A TheBaeicAxiom

I. INTRODUCTION In any field of endeavour which transgrosoos the boundary between fundamental and applied disciplinefl there tend to be two alternative approaches: the user’s approach. “What do I wish to do. and how can it best be done?” and the more fundamental “What can most efficiently be done. and what can it be used for?” The approaches are more different than is commonly realized. and both are necessary. Them refleotions are prompted by the appearance of the firet major text-book devoted to numerical taxonomy. that due to Bokal and Sneath (1964). This will provide an admirable introduction for those botanists wishing to enter this rapidly developing field; and it is no denigration of this important work to suggest that the authors are less rigorous in their examination of the methods than they are in their Uee and htemreta-

W. T. WILLIAMS A N D M . B. D A L E

36

tion, for it is with these latter aspects that they are primarily concerned. The user’s interests in plant ecology are similarly met by Greig-Smith (1964) and, in a more limited context, by an article which to some extent complements our own (Lambert and Dale, 1964). Excellent bibliographies have been provided for taxonomy by Sokal and Sneath (1964) and for ecology by Goodall (1962) and Greig-Smith (1964). Our intention is different. The newcomer to this field is faced with a formidable diversity of methods, all apparently fullilling closely similar functions. It is nevertheless our contention that the number of fundamentally distinct methods is very small, and that criteria can be erected which will olarify the distinctions between them, and between their numerous variants. This is tho aim of this communication. We shall not be concerned with the problem of allocation to an existing classification, which is the province of discriminant analysis. Although all the methods we shall discuss are in principle applicable to botanical problems, few have yet been ao applied; our references will therefore of necessity be drawn from a wide variety of disciplines. Symbols used will be conventional; but in the 2 x 2 contingency table arising from the possession ( J , K )or lack (j,k)of two attributes J and K, two conventions now exist for the number of individuals in each class: the alternatives are set out below:

SJ

-J

SK

a

6

-K

C

d

+J

-J

(J)

Scheme (i)is older, and has long been used in elementary statistical texts; scheme (ii) is used by Sokal and Sneath. The latter is more informative, but is clumsy in algebraic expressionsand in our experience is easily misread. When such a table is at issue, we shall therefore adhere to the (a,b,c,d)convention.

FUNDAMENTAL PROBLEMS I N NUMERICAL TAXONOMS

37

TI. ‘I‘HIGNATUREAND PROPERTIES OF CLASSIFICATIONS A. THE BASIC AXIOMS

Most, ge~i(wtldixcurrsionn o t i c:lrcndicr~t h r two co~icc~iwd tto clc?fina. t’o (tistiiigiiirrli bctwcon, csistiirg typw of clrwnification; such , for examplc, are tlie discussions in Lawrence (IYGl), Beckner (196Y), Gilmour (1961)and Sokal and Srisatli (1964).It is important for our purposes, however, to establish the minimum requirements which all cliissifications must meet, and wc rcstnte tlie problem as follows. A population consist8 of elerncnts, each of which can be iridividually described by reference to a predetermined list of “relevant characteristics”. This population is subdivided into sets of elements. what requirements must be fulfilled by these sets for the subdivision to rank as a classification? We submit that the following axioms will suffice. iuid

(1) Within every many-membered set there must be, for every member of the set, at least one other membes with which it shares at least one relevant characteristic. (2) Membership of the set may not itself be a relevant Characteristic. (3) Every member of any one set must differ in at least one relevant characteristic from every member of every other set.

Axiom (1) introduces a concept of “likeness” and ensures that tin elenlent cannot be classified if nothing is known about it. Axiom (2) has two important ConsequenceH. First, division into groups defined solely as possessing a stated nuniber of membcrs (such a8 dividing a population into groups of ten, or dividing it equally into eight parts) is excluded ; secondly, all classifications mu& be open-ended-there may be no known members to add to a set, hut it must not be impowittlle by definition to add more. Axiom (3) not orily cnsurm that identicah cannot be distributed between different sctq, hit makes provision for the Mingle-membered set. Although these axioms will suffice to define a classification, they are not in general sufficient to define one which is useful. We therefore need to discover what additional constraints must be imposed to enable our classification to meet specific external requirements, and it is from this point of view that we now proceed to examine some of the basic problems in numerical taxonomy. B. MONOTHETIC AND POLSTHETIC CLASSIFICATIONS

These terms were introduced by Sneatll (1962) to replace Beckner’fi (1959)terms “monotypic” and “polytypic” (without changing Beckner’s definit.ions),since these terms have other meanings. The Nets in a mono-

38

W. T . WILLIAMS A N D M. B . DALB

thetic classification are completely defined by the presence or absence of specific characteristics. Since such classifications are always generated in practice by successive sub-division, it follows that there must always be at least one set all of whose members share at least one relevant charaoteristic. It is quite possible to oonstruct a population, classifiable by reference to the axioms, from which no such set can be extracted; in such a case monothetic classification is impossible. Monothetic classifications may nevertheless be useful. They have proved valuable in ecology, where the concept of “indicator species” has long been familiar; they may well be needed in criminology, in which a decision may have to be taken quickly and based on as few attributes as possible. They are normally unacceptable in taxonomy; in medical taxonomy, for instance, one does not wish a man to be treated for the wrong disease because he has one aberrant symptom. The occasional criticism that monothetic systems produce misclassification is, however, invalid, since the criticism automatically assumes that a polythetic system is desired, and the argument is circular. The real objection to manothetic classifications is that they assme a property of the population which it may not in fact possess. Polythetic classifications imply no properties beyond those invoIved in the basic axioms, and are therefore always possible. 0. MAXIMIZATION

1. principlee of maximization The basic axioms will serve to define a large number of alternative cla&kations, And a further constraint is needed to select from among these. The constraint universally required by users is that, in a sense yet to be defined, the members of any one set are to be as alike as possible and as unlike the members of other sets as possible. Differences within sets are to be minimized, differences between sets are to be maximized. Formal work in this field, usually loosely known a8 “maximization”, has been largely confined to discriminant situations, particularly in the field of pattern recognition (vide, e.g. Sebestye~, 1962); but the diverse methods of numerical taxonomy are simply variant methods of maximization. The method8 fall into two fundamentally distinot groups.

i. r%f-~tnLcturingmpdhodi~ (u)A function of the relevant characteristics ia defined between paim of elements. ( b ) An element may be either a member of a population or an entire set; if a set, then the set may be defined by one of its members, by all

BUNDAMENTAL PROBLIDMS I N NUMERIUAL TAXONOMY

39

of its members, or by an element constructed from all of its members. (c) Sets are to be construoted so that the funotion is minimum (or maximum) within them, maximum (or minimum) between them, or both.

ii. Derived-~tmturing?ne.t)wda (a) A function is defined between pairs of relevant oharaoteristios over a given set of members. (b) A characteristic, or a group of characteristics, is found for which the funotion, or a derivative of the function, is maximal. (c) Sets of members are defined in relation to the oharacteristic(e) so selected.

For certain purposes it is desirable that the analysis oan be “inverted”, in the sense that the elements and characteristias change plaaes. For this to be possible the data must fulfil certain oonditions which we explore later (Section IV A). The apparent diversity of methods in the literature largely concerns self-structuring methods, and in these the diversity is largely one of the function selected. Monothetic methods necessarily employ derived-structuring. 2. Intern1 and external ch&$cationa It is assumed in the foregoing paragraph that the members as defined by their relevant oharacteristics form a self-sufficient set within which maximization is to be effected; such systems, which comprise almost the wholo of existing literature in numerical taxonomy, we shall call “internal” classifications. It may nevertheless be desired to impose a restraint in the form of an external element or set of elements (selfstructuring) or an external characteristio or set of oharacterietics (derived-structuring). In such casm the maximization is entirely between the reference unit on the one hand and the internal eete on the other, the internql sets needing only to satbfy the besic olassifiwtory axioms. The process of maximization is, however, itself different from the all-internal case. The pkimary maximization is of the range of the selected function, in that the internal sets are to be as like or unlike as possible to the reference sef. The main use of these “external” classificatiom is likely to be predictive; if the population is heterogeneous in the sense we shall define in Section IV C, they will be more powerful than the classical regrmsions taken over the whole population. Their possible application t o problem in plant ecology is also under investigation. The only example known to us in the literature iu the derived-structure “predictive attribute ctnalysis” of Macnaughton-Smith (19SS), with whom we are currently collaborating in the development of more general S Y E ~ I W .

40

W. T. WILLIAMS A N D M , B. D A L E

3. Simultaneous alternative chsi$cations: clump Suppose the system be restricted by the requirements ( i ) that maximization shall extract only one sub-set from the population, and (ii) that this sub-set shall be subject to a specified constraint; the constraint normally imposed is that the subset must contain a specified element or group of elements which will act a8 its nucleus. Such a subset is normally termed a “clump” and the remainder of the population is without interest. Let this process be successivly repeated on the entire population by specifying a new constraint on each occasion ;the ultimate result is a set of clumps. This set is sometimes loosely termed an “overlapping olassification”, but such an extension of the term “classification” is not to be recommended; the clumps need not exhaust the population, and any one element can, and usually does, occur in more than one clump. Systems of this type are particularly associated with the work of Needham (vide,e.g. Needham, 1962; Needham and Jones, 1964) on linguistic data arising from problems in documentation and information retrieval ; but they have also found some application in anthropology and medicine (Bonner, 1964). They have been developed to meet circumRtances in which simplicity and speed of computation are more important than power, and they may well require re-examination before they can satisfy the more rigorous demands of plant taxonomy and ecology. A system of clumps can similarly be generated by the use of a changing external criterion as constraint. The groups delimited by the “deme” terminology (Gilmour and Heslop-Harrison, 1954) of plant taxonomy together form a system of precisely this nature, but it seems never to have been the subject of numerical study. We shall not be further concerned with clump systems in this article. 4. Weighting Sokal and Sneath (1964) accept the Adansonian postulate that “every character is of equal weight”. We need not so restrict ourselve8, and we shall first distinguish between a primi and a posteriori importance.

i. Importance a priori Classifications in, for example, modical or criminological corrtoxtn may be used as guides to action; in such cases particular characteristics may be of overriding importance. It might be regarded as undmirable to send epileptics to prison, no matter what their other characteristics suggested. Such cases do not disturb the systems we are considering, since they do not alter the classifications, but only the UBB that ie made of them. It has, however, frequently been suggested (wide, e.g.

FUNDAMENTAL PROBLEMS IN N U M E R I C A L T A X O N O M Y

41

Proctor and Kendrick, 1963)that characteristics should be assigned a differential importance from prior knowledge of the field ; as we have already pointed out (Williams et al., 1984) this destroys the objectivity which is the single most valuable feature of numerical taxonomy, and we cannot recommend it. ii. lmportunce a posteriori Derived-structuring methods maximize some function of the characteristics. After maximization, therefore, each characteristic will be associated with a numerioal value which reflects its contribution to the overall maximization, and which may therefore be regarded as tt measure of its importance. The application of this concept presents different problems in different systems, and the situation may best be explorcd by comideration of firstly, monothetio derived-struoture, and secondly, polythetio self-structure. (a) Mmthetic derived-etructure.If a population is such that it contains many shared charaoteristias and 80 can be defined as a set of final classes, a very large number of alternative monothetic classifications is possible. The characteristics used may be selected solely for external convenience, or even indiscriminately, and there is no internal maximization. Such are the “special claasifications” (into, e.g., food- or fibreplants) and the dichotomous keys in floras. These, which are in fact perfectly good external classifications, are commonly termed“artificia1”. It is therefore tempting to equate “artificial” with “absence of internal maximization”; but we defer to the views of Sneath (in Zitt.) to the effect that the terms “natural” and “artificial” have been so variously used that to provide them with new statistical definitions would confuse rather than clarify the situation. In contraat to these classifications, the method of Aseociation Analysis, whose properties are discussed in Section V D 2, ie a monothetio method whose defining CharacteriRtics have been obtained by a process of internal maximization. The characteristios now differ in a p8teriori importance, and this has by some workers been regarded an “weighting”. (b) Polythetic self-structure. Here again it is theoretically possible to effect classification without maximization, but since most real-life populations already themselves satisfy our Axiom (1) for a classification, the solution is usually trivial. A single maximization is therefore necessary in practice. All the “similarity” methods discussed in Sokal and Sneath (1964) are of this type : they use the least maximization which is in practice essential. However, the first step in such an analysb might be a derived-structure maximization, so that the characteristics were aa a first step provided with “importance” measures; a second

42

W.

T. WILLIAMS A N D X. B . DALE

maximization, using these weighted characteristics, would be necessary to complete the classification. The only such doubly-maximizedmethods known to us are those in whose development we have ourselves collaborated (Macnsughton-Smith et al., 1964; Williams et ul., 1964). 5. Unintentional weighting: “nui~ancecorrelations” The selection of attributes is not itself the concern of numerical taxonomy. Nevertheless, methods which employ derived-structure functions are prone to difficulties arising from so-called “nuisance correlations”-groups of attributes linked for reasons unconnected with the purpose of the analyeis. This problem does not arise in those ecological studies in whiah the attributes are plant species, since these are necessarily different things. The questionnaires of sociologiortl studies, however, normally contain much redundant information ; this is deliberate, since a question which may be avoided in one form may be answered readily in another. Some of the attributes are therefore logically linked, and these linked groups may dominate the subsequent analysis. It must be remembered that questionnaires have not normally been deaigned with modern numerical methods in mind, and the increasing use of these methods will doubtless in time iduence the design of questionnaires; but meanwhile the problem exists. The nature of the problem, however, has not always been clearly understood. The objection,to these links is simply that they can be known to be links without recourse to analysis ;if they could not be so known they would be of interest. It does not follow that they are in every caw easily recognized, and a preliminary numerical analysis may serve to este;blieh them. This is possible if the system is such that elements and chamteristics can change places, so that the characteristior, can be grouped into sets; if such a set inescapably suggests the hypothesis that the mombercl are linked for reasona-auch as intrinsic redundancy in a questionnaire -in which the investigator is not interested, the group can be replaced by one or more of its members or by a new attribute constructed from all of them. Despite statements to the contrary in tho literature, we submit that the objection to nuisance correlations does not lie in their logically necessary links; the sole criterion is the interest or otherwise of the wer. D. HIERARCHICAL AND NON-HIERILRCHICJAL CLASSIFICATIONS

Hierarchical classifications are of very real advantage to the taxonomiat, since they enable him to compare taxa at any desired level. This has probably contributed to the fact that the vast majority of existing numerical methods are hierarchical in nature. However, it

F U N D A M E N T A L P R O B L E M S IN N U M E R I O A L T A X O N O M Y

43

may also generate a requirement that each level in division is aesoaiated with some measure which shall fall as the hierarohy deswnds. It is not always realized that this places an additional oonstraint on the ohoioe of maximizing function ; some functions (notably Euclidean distanoee and information statiatios) possess this property, whereas others (most of the derived-structure ooefficients and the “statistioal distance” coefficients) do not. The term “reticulate classification” seems .to include two quite different concepts. The f h t ia the unmaximized external classification with an embarrassing choice of alternatives, such aa arises in classifying books; this need not concern us. Truly reticulate claasifications arise out of an interest in inter-set relationships after division into sets has been complgted. If only the inter-set functions are required, a completely non-hierarchical method could be used; but, as shdl later point out, the choice of such methods is extremely restricted. In most caaes, therefore, both the hierarchy and the inter-set functions are of interest, and the problem is to generate either from the other. We ahall later demonstrate that maximization may, or may not, be uniform over the entire mathematical model in me. If it is uniform, as with unweightd Euclidean distances or information statiatios, no difficulty arises : inter-set functions and hierarohical divisions are everywhere oompatible. In those methods with which we ourselves have been associated, the maximization is deliberately non-uniform over the model; in these cases, which are hierarchical, no compatible inter-set function has yet been defined (vide Sections I11 D (i)(ii)).It is not permissiblo to define a completely new function,laincethe original hierarchical maximization may then fail; this is the cauae of the “recombination of aets” difficulty which Goodall (1953a) experienced in his pioneer studies in divisive methods. E. PROBABILISTIC AND NON-PROBABILISTIC CLASSIB+ICATIONE

This particular dichotomy has generated more confusion-and probably more rancour-than any other. It underlines the commonlyexpressed doubts a~ to whether these methods can, or cannot, be classed as statistics, and so has caused Greig-Smith (1964) to use the term “quantitative” and Sokal and Sneath (1964) and ourselves to fall back on “numerioel”. It underlies, too, the misgivings that authors frequently express concerning the “significance” of their results. The difficulty has been exacerbated by the fact that modern statistios is almost entirely concerned with estimates of probability, so that if well-known statistical parameters-x* or the correlation coefficient, for example-are uaed for maximizing, it is aaeumed that these am

44

W . T . W I L L I A M S A N D M . B. D A L E

estimates which should be associated with measures of probability. I n fact, the methods of numerical taxonomy are not, or need not be, probabilistic systems at all, but hypothesis-generating systems. We shall outline the two alternative approaches.

I . The non-probabilistic approach From this point of view, the methods of numerical taxonomy may be regarded as stemming from a branch of statistics of respectable antiquity-that concerned with finding mathematical formulations which will serve as a concise and economical description of an otherwise intractably oumbersome mass of data. Though superficially so dissimilar, their logical relatives are to be found among such projects as the fitting of Pearson curves to actuarial data (Elderton, 1938); the search for a flexible growth-curve (Richards, 1969; Nelder, 1961); and the application of contagious Poisson distributions to distributions of plants in the field (Archibeld, 1948).The fitting of regression lines is itself a member of the mme family, extended by the probabilistic concept of the significance of the parameters which the fitting requires. Now, these concise mathematical descriptions can with perfect validity be used to generate hypotheses concerning the nature of the data, but only if two conditions are rigidly satisfied. First, aa always, the hypotheses must be capable of being tested; secondly, any test must depend on new observations, and cannot again use the data from which the hypothesis was generated. Generation of the hypothesis may not be used as its own evidence; we forbear to cite examples of this practice, contenting ourselves by remarking that they can be found in biological literature. The precise statistical context of these methods can most clearly be demonstrated by comparing a vegetation survey in ecology with an agronomic experiment in, say, mineral nutrition. In the agronomic context the hypothesis is set from previous experience, and this determine8 the details of an experiment, which issues in a quantity of data; etatistical methods are applied to thaw data in order to test the hypotheHis-ueually in the form of the probability of obtaining a given deviation from a null hypothe6iB by chance alone. Tn tho ecological context, although experience may have informed its collection, the data is the starting-point ; function6 are elected and appropriately maximized in order to reduce the data to eimplor form; thin simpler form is used to generate a hypothesis-often in the form of “there is a change of some sort in this region”; and the hypothesis is tested by new, direct observations in the field. Examples of this type of hypothesisvalidation may be found in the work on Association Analysis (Williams and Lambert, 1960).

B U N D A M E N T A L YBOBL10MS 1N N U M E R I U A L T A X O N O M Y

46

Nor is validation difficult in applied taxonomy. In medicine, for example, the individuals classified may be diseme-producing organisms, or symptoms; in criminology they are normally delinquents. I n these cases the hypothesis takes the form of a suggestion for treatment. It may be remarked in passing that the power of the mathematical methods used is all-important in these fields, for although falsification of a hypothesis might gratify a dispassionate experimenter, it is apt to be disastrous if a human individual is concerned. The problem is more difficult in “classical” taxonomy. Here it is tempting to enunciate a phylogenetic hypothesis, normally based on inter-set functions, but fossil records are such that hypotheses of this type are rarely testable (vide, e.g. Sneath and Sokal, 1962). The basic requirement of taxonomy aemu stricto is stability, both of the membership of sets and of the pattern of characteristics that their members display within them. I n the first case (the membership of sets), addition of new characteristics followed by re-maximization should not ohange the membership of the sets. In the eecond (the pattern of characteristics), let a new element be disooverod whom chctracteristios are imperfectly known; if from the known chasacteristics it can be allocated unequivooally to an existing set, the pattern of its remaining characteristics, when these are examined, should conform to the pattern for the set. On this approach, therefore, the methods of numerical taxonomy are hypothesh-generating systems; and a hypothesis-generating system is neither valid nor invalid. Probability enters only, if indeed it enters at all, in the testing of the hypotheses that are generated. This approach exposes a possible danger, which we do not believe taxonometric writing has always avoided. This is that computer classifications might be regarded aa in some sense absolute-as “objective” and therefore “better”. They are not objective, since they depend on the user’s personal choice of maximizing function; and they are only better if they can be shown to fulfil a stated requirement more efficiently.

2. The probabilistic approach It is, &B we shall show, easy to conceive of probahilietio clamificatione in theory; but we are here concerned to defend tho thesitc that auch classifications are usually both impracticable and unprofitable. Fimt, it should be noted that a probabiliRtic classification requiree a null hypothesis; this will normally take the form of stating that the pairfunctions available for maximization in a given population or set could have been generated by a random process. The null hypothesis cannot, in fact, be independent of the function selected for maximization,

46

W. T. WILLTAMS A N D M. B. DALHl

i. DifJicultiea inherent in null hypothea Since the null hypothesis depends on the maximizing function, it will be convenient to select two well-known cases for consideration. (a) Multivariate nomnal popubtiom. In this case the null hypothesis would state that the observed variation in characteristics could have been generated by a set of independent normal variates, usually the characteristics themselves. The function available for test would probably be the correlation matrix. Now, Bartlett’s (1950) test for the roots is not available if the matrix is singular, and experience suggests that it is sensitive to departures from normality. To demand that all the coefficients be individually significant is nomally regarded aa too stringent ; and Goodall (19G3a) has in effect suggested that the ooefficients be themselves treated as normal deviatos, so that the proportion of them which exceeds the individual signifioanoe level be regarded &B a test of significanoe of the whole matrix. There is, in fact, no simple, unequivocal and robmt test available. ( 6 ) Qualitative populations. The function used (though others are available) is often related to the Euclidean diRtance between elements (or set centroids) plotted in an n-dimensional space where the j t h co-ordinate for an element is 1 if it possesses the j t h attribute and 0 if it lacks it. The problem now is to state a null hy&hesis at all. Use of the binomial expansion would imply that possession of all characteristics was equally likely; and the solution obtained by Rohlf (1962) for even n makes assumptions as to the distribution of the frequencies. If we assume, however, that the hypothesis should not involve the frequencies, an obvious solution would be to retain the frequency totals and to construct from them the entirely dissociated class-frequencies; that is, the numbers of individuals that would be required in all possible sub-classes if, without change in the total numbers possessing each attribute, all pairs of attributes were to have zero association. It is straightforward, though tedious, so to calculate the probabilities (for 0,1, d2)in the two-characteristic case; but the resulting algebraio expressions are extremely cumbersome, and lend little hopo of extension. In any case, construction of the general null population may present formidable difficulties. If we write (A) for the number of individuals possessing attribute A, (AB) for the number posmsing both A and B, and so on, then in the oompletely dissociated population W C .

N

*.

.)

( 4 * ( B )((7... .

=x

“3

Unfortunately, for more than two characteriafics, this relationship is necessary but not s d c i e n t (vide, e.g. Yule and Kendall, 1960), and oannot therefore be UBePx M a generating function.

F UNDAM INT A L PROBLIMS I N N U Y I R I O A L TAXONOMY

47

It has been suggested to UB (Macnaughton-Smith, in W.) that information statistics might provide a solution of the qualitative problem, in view of their remarkable additive propertiea and their relationship to x*. Let a p u p of n individuals be specified by the possession or lack of p attributes, and let the number possessing the j t h attribute be a,; we have made preliminary ObservationA, using ecological data, on the behaviaur of the statistic: I = Iyn log It-

E [u, log a,+(n-aj)

log (n-tzj)l*

j-1

Ecological data not uncommonly contain groups of identical or nearidentical individuala, and these groups may vary greatly in size; the data will have the properties of a stratified, rather than of a random, sample. Unfortunately, we find that the stati8tio above is seneitive to this particular form of non-randomnew, and is therefom unduly senaitive to set size--sets tend to be fused if they contain comparable numbers of membera. This is incompatible with our second classificatory axiom, since it implies that a function of the set may determine the allocation of one of its members. This difficulty may be removed by normalizing for group size, though, in some forma of analysis, a t the expense of replacing it by the generation of an “ambiguity” problem related to that arising from unweighted Euclidean diatanwa (Section V D 3 (i)). Nevertheless,these statistics have many desirable properties and would repay further investigation. (c) GoodaZl’8 coe&ien$. Very recently Goodall (1964) has proposed a probabilistic similarity index. For every pair of individuals, the probability that the two are as similar as in fact they are is calculated for each attribute separately , and the attribute-probabilities then combined. The method is cumbersome for qualitative b t a , but it is the only method known to us which ia in principle applicable to mixed data i.e. data in which the attributes are 80 unlike that any common scaling would be unrealietic. No example of it8 use has yet been published. ii. AelicaliOn Of $WObabili8tic Ch8$C&iO?l% Suppose an appropriate criterion of significance, and therefore an appropriate null hypothesis, to be available; and s u p p a a popubtion to have been divided by maximization into two eete whose dididididididididididididididididkincfio fails to reach signihance. It still does not follow that the division should not be effected. For the population may be so intractably large that the best possible sub-diviaion, though non-significant,may be mom uaeful than none at all. However, although the overall characbristicpattern may not define a significantdifference, sub-seta of charmbh-

48

W . T . WILLIAMS A N D M. B . D A L E

tics may exhibit stability (this phenomenon may occur if the population exhibits n o h , which are briefly discussed in Section I V C 4). I n either case, it is the usefulness of the division which will be of importance; the division will therefore in any case be subjected by the user to a second, pragmatic, test which will override the fist, probabilistic, test. We are therefore not convinced that any useful purpose is served by the probabilistic test, quite apart from its inherent difficulties. 111, THECHOICEOF MATHEMATICAL MODEL A. INTRODUOTION : METRIOS

The ultimate test of a numerical method is whether the u8er fin& it useful. However, all mothods are of some we to the usor; and if he ie to bear the sole responsibility of deciding between them, he will be faced with an immense amount of empirical work,still with no assuranm that the method may not fail under extreme conditiona-as, we believe, some existing “similarity” methods have s h a d y failed. The literature contains many despondent remarks on the paucity of available information relating to comparison of methods. This is particularly true of the pair-functions themselves, often loosely classed as “similarity coefficients”. The best-known have been reviewed by Goodman and Kruskal (1954, 1959), Dagnelie (1960) and Sokal and Sneath (1904); but it k doubtful whether even these extensive collections are complete. The problems would be relatively unimportant if all such functions were jointly monotonic, in the sense that, if element-pairs are so ordered that one function forms a monotonic series (i.e. a series which either increases or decreases over the whole of its length), the remainder will also be monotonic, To take only three well-known functions, 2a/(%+b+c), (a+d)/(a+b+c+d), and the correlation coefficient, it ie easily shown that no one of these is jointly monotonic with either of the other$. A choice is therefore necesmy; and the testing diffioulty oan be overcome, at lea& in part, if the methods and function8 are required to fulfil appropriate mathematical conditions. We consider it essential that any meamre used for maximization should define a model, and, if possible, a model in Euclidean space. The advantages of such Rystems are threefold. First, many simple, robust and powerful methods are available in Euclidean systems that are not available outside them. Secondly, as mentioned in Section I1 D, they have hierarchical advantages. Thirdly, and perhaps most important, our daily experience gives us an intuitive perception of Euclidean systems, and thereby enables us to grasp their properties and to predict these properties in extreme cases. If the function is such that it is not known

FUNDAMENTAL PROBLEMS I N N U Y E R I U A L TAXONOMY

49

to be associated with any particular probabilistic or spatial model (models which me neither probabilistic nor spatial are possible, but we know of no published work on them) we propose that it must, aa a minimum requirement, be a metric; it will then necessarily define a spaoe whose properties oan be explored. We deal in this motion with the general problem of metrios, and it will be oonvenient Brat to etate the conventionaldefinitione. The subject is fully dieoueaedin geometrioal texts; the formulation we use ia substantially that of Kelley (1965). De$dtion. A numerical function d(z,y) of pairs of points of a eet E is said to be a metric for E if it satisfies these oonditions: (1) d(z,y) = 4Y,430 (symmetry) (2) d(z,z)ioiiwith i L soft rotting f ~ ~ n g ~ snry t II”! sy t II 1)tA )I ti s ( ) f t,tic: (I(!c:rLy (1 0 V(! lo]I(:( I . I,lo,ytl ( I !)(iO) slrgg(!st.(:(lt~tiiLt!ijlic iriovcrri(!rit ( 1 1 wiLi,(!l*i ~ t t f ’ ~ J I l ttic ~~1 soft,wootl tini her W;LH necessary c i t h r t,o r(!rnovc ti11 i i i t i i t)it,or. prcserit in the tiniI)cr or formed :LS a hy-product of the fungal iitt,tk(:l(,or even in order to introduce oxygen into t,he wood once the fiingus had penetrated the surface layers. She, anti later (lorbett (1963),described unsuccessful experiments to establish this ascertion and the latter decided that the only reliable means of determining the presence of soft rot attack in its early stages was t o cut sections of the timber and examine them under the microscope. It was soon observed that, there appeared to he differences in the early stages of attack by Chaetomium globosurn in beech (Fagus sylvatica) and in Scots pine (Pinus sylvestris), and a technique was worked out to attempt to examine these details more closely and to see what correlation, if any, there might he with the great difference in the rate of attack of these two species (as measured by loss in weight) when exposed to the fungus in a petri dish. The technique evolved ( b h ? t t ,1:JtX) was to use small c u t m of wood, hetween 0 . 7 cm and 1.0 c:rn in cr1gu clirtiension. ‘hw: tiloc:ks W(!IC w r c fully cut SO that t w o fikcos worc i r i thc: i,r;lrlsVCrHC l ) l i L r t c : , I,wo i t 1 tjll(! rJ1,fli;t,i longitudinal plane and ttio rciri;iirtirig !,WO i r i ttic! i,rirtgc:tti,iriI I ~ i t ~ ~ i f ~ i r ~ plane. Four faces of catch hloc:k w(!r(! swhd with “Ar:rl~lit,c:” (11 I W f J prietary formulation of a n epoxy rcsiri ;ulhesivo), HO t , h i ~ tttic c:rit.ry of the fungal hyphae was restricted to two ftices only, arid these were always opposite faces of the cube and therefore in the same plane. In this may it was possible to present a particular orientation of cells to the active growth of fungus on an agar medium, so that it might be possible to observe the way in which the fungal hyphae penetrated through the block. The size of the block was a convenient one for mounting directly on t o a sledge microtome so that sections could be cut in any plane without having to re-shape t.he block. Using this technique, experiments were set up with the sapwood of Scots pine (Pinus sylvestris) and beech (Fugue sylvuticu). Three blocks

I i ~ i d

T H E S O F T R O T E 'IrN 0 1

439

with a different pair of faces exposed in each case were placed on a mycelial mat of the fungus Chaeforniwm globosum growing on either a modified Abrams medium or on 1.5 per cent malt agar. With both species, considerable differences were observed in the rate in which the fungal hyphae penetrated through the block from the various faces and emerged on the upper unexposed face. With both species the transverse face provided the quicker entry and the hyphae emerged on the upper face sooner than with the tangential faces exposed and much sooner than the blocks with the radial faces exposed. In the case of the transverse faces, profuse sporulation occurred arid completely covered the upper face ; sparse sporulation occurred on the tangential face and little or no sporulation occurred on the radial face. With both species it appeared that the fungus was following the line of least resistance and W ;Lpvrt urc thtL1, rriiglit o w u r to I);LSS taking ;Ltlv;Lnt;qc of' iLny S ~ ) ~ Lor t t IP t I y 1)t i ;LC (1 11i(* k I y t I t roug t I t I I(! Id( k . 'I' I 10 1)cn(:trn t i o r 1 frc ) 111 tmr t s vcrsc f i ~ c cto trtLrlsvcwc! fiwe t ~ p p e ; ~ rto ( ~IrrLvc. l oc.c.irrrct1 prir11iLrily through the ~ I r i tho ( m e of' the tangentiul lririieii of trwlieids, vessels ~ L I Iti hex. faces the quickest route seemed to 1r;tve been d o n g tho length of' tlie medullary rays and the contents of' these cells may well have assisted in the nutrition of the fungus. With the radial faces, however, there appeared to be no straightforward path unless the fungal hyphae could penetrate readily from cell to cell through the bordered pits of the tracheids and simple pits in the fibres. Microscopic observation confirmed that this penetration did, in fact, occur.

Iv. MODE OF ACTIONOF

SOFT

ROT PUNGJ

A. PASSIVE PENETRATION A S D I I E C A Y I'PNIFTltAl'ION

In the case of Scots pine the rays were quickly colonized h y the fiirrgus, which appeared to penetrate fairly rodily throrigh thc: sirtilll(: [)it,H between parenchyma cells o f ttic r;Ly ; ~ r i ( I t h vcr1,ic:nl t,rii(*h(:iilsol' 1,tw wood. However, although h y p h e wcro s w r t f'rorn l,irri(? ~ J J 1 i r r i c : tlf) penetrate thrcmgh the bortlcred pits froni om trti(:ttttid t,(J t1h:rtr:xt,, this was by no means common in the early stiLgcs of infec:t,icm i ~ r r ( 1whiLt Corbett (1963) describes as the passive pcnetration of the furtgus apr)wm to proceed from ray t o tracheid and tracheid to ray more easily than directly from tracheid to tracheid. In the case of beech, the main path of passive penetration from cell t o cell was through the simple pits. These pits in beech are of small diameter and frequent occurrence and it seemed highly probable that one of the reasons for t,he rapid loss of weight of beech in the laboratory

310

JOHN LEVY

by these cultural methods in comparison with the niuch slower loss in weight, of Scots pine was due primarily to the speed with which the fungus could make a pasfiive penetration through the former species and so with a greater volunie of timber in contact with the actively growing fungal hypliae t’he suhsequent decay penetration into the tva,ll was likely t>oo(:cur o i l i L greater nrinilier of occasions. It, seems iiiiporttuit therefore to tlistinguish hetween (a)tlie gross penetration of ,the fungiis into itnd through the hlock, ; ~ n d( b ) the pllct,riLt,iotiof i n d i v i d ~ i i f1111jii~1 ~l hyl)t~aeinto the cell wall C ~ L I I S ~ breakJI~ tlowti o f t h t w d l . ( lorl)et,t,(1!163) uiitlcrlitit!n this tlistiric:tioii a i i t l suggests t IIO t,(:r.liis “l)ilsfiivc! I)(!ii(!tr;st,ioii”il11(1 “tl~c:;~yi)(!ti(!t”iLt,ioli”t,o c o v w ttliosc! t,w(J li)i-tiis of l)rogrwsioti of i tic I’itiigys. I’ii.ssivc! l~ciict,riit,iotii r r I i c d i Wi1.s i~iiic:lic!r tjlirrii iii Si:ot>s1)itic.

R. EFFECT OF SPECIES O F WOOD O N THE MODE OF ATTACK BY THE SA3TE FUNGIJS

Corbett and Levy (1963) described the initiation of decay penetration of Chaetomirim ylohoswvn into Scots pine. The hyphae enter the lumina of the tracheids via the simple pits between t,he tracheid and t,he ray parenchyma cells. The hyphae t,hen align themselves parallel to the long axis of the trnclieitl and usudly contiguous with its wall. Lateral branches develotackare characterized by the presence of darkly stained hyphae in the cell luniina and ia pits connecting adjacent fibres. Because of the thickness of the fibre wall and the small size of the lumen, observation in longitudinal sections of the initiation of attack presents much more difficulty tlinn wibli Scots pine and in most, of the material examined the most common evidence of t'he rot was mature cavities wit11 conical ends having darkly stained hyphal contents and generally lying almost p t d l e l to the long axis of the fibre. Transverse hyphar piwxiiig ;Ldjac:ent fihre \\riLIIs were rilrcly seen. I I I scveribl iiistiLiiws ;I firic? I i ~ y p h iW;LN ~ sttl:Il in tlie f i l ) r ~\ ~ i l l I (x)liI1
341

JOHN LEVY

occurred in the S, layer of the a d j w m t fibre traclieid wall. followed by ciivit,y forniiit,ion i t s found in Scots pine. I n the majorit,y of cases, however. i m l partioularly wlien transverse or tangential fares of the test blocks were exposed to the fungus. it' w a s found t,liat the Iiyplinc in t,he lumen proceeded to erode the wall IvIicrever it came into rolititct with t,he wall surfiice. Figure 8 s1~on.sthe characteristic appearance of the tiniber destroyed in this way. In some institiices i i lurninal hyplia formed a very short lateral t)rttnc:h which hegan to crode the wall at, ttio p i n t of i~ st>dnctl.'I'his iriitiiil erosion cwntw:t iLIi(\ w a s found to b o c o ~ densely af)l)ciLrl>(l iLs it stiiirI)-Sitlctl V-slii1~lte~l nick in the OCII wiilt wllen viewed in longitutli~li~1 scotion, c:onsitlcrtLttl,y I;trger a~icimore sl1iirl)l,y tlefiiied t.hm t h i k t o1)served 1j.y I'ro(:t,or (I!U1) witti t)ro\\.n t*ot.i i t l ( l 1 ' 1 ~ j 1 1 ~ 1 i t ~ l y cxtcn(1itig ILX fw iLs t1ho 1tri111iir,yWiLIl. Sonictinit:s i L liit,(?riil t1,y1)tIi11 t)r;m(:ti pssitig iiit,o,t,ll(: wit11 ;i,t, SII(:II it, t i i d ( t)(!(;ii,lll(! v(!ry litic! ii,ritI t8r;ivc!rsc(Ithis will1 i t r l ( l t,llc! w;LII of' t IIO i L ( l j i ~ ( ~ ( ~(:ell l l t ~ ill; right ;I.IIKICS to t,hcir Ioiig iwis ~ ~ n limiiccl tl i L l l o t l l C r V-SlliiItc(1i i i o k iit t,tle otlicr siirfii~x~. 'I'his 1)ror:i?ssof destruction let1 to nlirrketl scU1J)tUriIlgof the wit11 when viewed in longitudinal section. I n trimsverse sections, however, the cavities were almost always found t.o have rounded edges. Cavities within the thickness of the wall, usually forming a series of spiralling chains?of cavities, were found only at the extreme edges, beneath a sealed face, of t'hose blocks where transverse or tangential faces were exposed: Other parts of' these blocks showed only erosion of the wall from the lumen. In blocks where the radial longitudinal faces were exposed, however, decay penetration was restricted t o the cells nearer the exposed €ace and was mainly in the form of helically directed chains of cavit'ies in the S, layer and only rarely by erosion from the lumen. The initiation of decay penetration in the three tirnhers n1iL.y t h i i s tw seen to follow quite different paths. The reasons for these diflccrenr;es are not entirely obvious, but must he related to the fine structure and chemistry of the cell wall; they are tliscusscd later in this p a p r . F(*rthe present kt it suffice to say thtit the SiLnic fungus, ~~hudorrtiurn, yhhwrr~,, does not hchavc i n tho SiLnlC w;iy in e i ~ ( : Hhp C i C H o f woo(l. C. EFFECT O F SPECIES O F FI'S(;I.S

OX THE MODIC OF ATTACK

I?J THE SrlMlE WOOD

Corbett (1963) tested several species of fungus in culture against bloclrs of birch (BetitZa sp.) which had all faces unsealed iind one tar]gential face presented to the fungus. The fungi were L'mnim/hyriurn f i i c l i e l i i . Sphaeronema sp.. Stysmztis sfpmonilis, all of which had actively attacked the wood after three n,esks' exposure, Camarosporizrm amhiens

THE SOFT ROT FVNGI

315

T H E SOFT R O T F U N G I

347

adjacent cell; the erosion of the wall by hyphae in the lumen, forming V-shaped nicks; and cavities with conical ends within the S, layer, were all observed. Destruction of the wall was severe, even in cells with a low density of hyphal growth. Corbett (1963) figures four electron-micrographs taken by Preston and Levi (1963) using this material, one of which is reproduced as Fig. 9. This shows R hypha lying in a mature cavity within the wall with a portion of the finer lateral hyphn from which the cavity was initiated shown crossing the middle lamella and primary wall region. This is a very remarkable picture indeed. Sphaeronema sp. The exposed (tangential) face of the block was severely rotted after three weeks’ exposure. Hyphae were confined chiefly to the S, layer, and of relatively large diameter in comparison with wall thickness. Erosion from the lumen was rarely observed. Stysanw &emonnitis. The edges of the block and patches of tissue associated with vessels were severely rotted after three weeks’ exposure. Hyphae did not erode the wall from the lumen, but penetrated the wall and formed discrete vertical chains of helically arranged cavities within the S, layer. The hyphal walls stained quite strongly with picro-aniline blue compared with the hyphal contents which showed only a pale coloration. Rotted cells retained their form even though the hyphae ramified completely through the wall. I n longitudinal sections rotted tissue had a lacy appearance, because only the primary walls and middle lamellae remained and the hyphae filled the whole space between the undigested framework. Hyphal diameter increased with the development of the cavities. There was an extremely sharp differentiation between rotted and unrottecl zonen and it was more difficult to identify any early rstagea of penetration of the wall. Camarospwhm ambiem. Slight wulpturing of tho wall kJy liirnerid hyphae was observed in fibre-tracheidn on the e x p m d edge of tho block, but only when dense “ropes” of hyphac were present in the lumen. Cephubsporium sp. After three weeks the walls of some fibre-tracheids on the extreme edge of the exposed face were undergoing crouion hy luminal hyphae and there was also evidence of cavities within the wall. Cross-penetrartion hyphae were extremely common. The pattern of attack by all f i v ~fungi is similar in nature to that already observed for Chetomium globosum. Although differences in detail occur, the similarities are striking. It would be interesting to test these organisms on other timbers and see whether the effect of the dpecies of timber is similar in all species of fungus.

348

JOHN LEVY D. SOFT ROT FUNGI ON POSTS IN GROUND CONTACT

Corbett (1963) made some st,riliing observations on tlie occurrerice of soft rot in birch and Scots pine fence posts on three different sites. Soft rot cavities were observed to be present in birch posts a t the ground-line to a depth of 30-45 cells from the surface eleven months after the posts had been driven into the ground. After eighteen months’ exposure, soft rot cavities were also observed in the top of the birch posts. A similar pattern was observed in Scots pine, but the depth of penetration was less and showed considerable variation with the site conditions. Birch and Scots pine posts of a similar type t o those examined had previously been observed to fail a t the ground-line due to attack by Basidiomycete fungi, after three years’ exposure. This a t once lays open the possibility that the soft rot fungi may play a part in a succession of fungi colonizing these fence posts and could well be an important precursor of attack by Basidiomycetes. More work t o elucidate this point would be of very considerable interest and value.

V. LIST OF FUNGIKNOWN

TO CAUSE

SOFTROT

Corbett (1963) compiled a list of all the fungi which have been recorded as having the ability to induce soft rot in hardwoods or softwoods in pure culture. Trichoderma viride was not included in this list since there have been conflicting reports as to the ability of this species of fungus to produce soft rot. The list is by no means complete, since it omits marine fungi (0.g. Jones, 1963), and some other species, but is still a useful compilation, in spite of the difficulties with identification that often occur. Her list is as follows. ASCOMYCETES Chaetomium cochliodes C . elatum C . funicola C. globosum Ophiostoma coerulesccns 0 . piceue 0 . pini Xylariu sp.

FUNGIIMPERFECTI Acremoniella sppp. Acremonium Alternariu sp. A . humicola

Reference* 2, 4 4

2, 5 1, 2, 4, 7 3 6 3, fi 2

c

1

2 2 6

349

THE S O F T ROT FUNGI

Biapora effuaa B. pwilla Biaporomycea sp. Bobordeniellrt sp. Canta7osporiu7n ambiena Cephaloeporium sp. Ceratocyetis pilqera CWTOPaiS sp. Conwthyrium sp. C . fuckelii cyto8porelEa Rp. Dendryphium sp. Diplococcium sp. Diacula pinocolu v. mamnwaa Haplochulara sp. Hel&coaporiurnaurcum Hormkcium sp. Nematogenium sp. Orbicuh sp. 0 . porietinu Pestaloztia sp. Phialophora richardsiae P . faatigiata Phoma sp. Pullularia sp. Sclerotium sp. Sphaetonema sp. Sperocybe sp. Styannus sp. S . stemnitis Torula sp. Trichoapwium heteromorphuin Tridhurua terrophilua

* 1. Corbett, 1963. 2. Duncan, 1960. (Identifications stated t o be tentative.) 3. Krapivina, 1960.

2

5 2 2 1 1, 2

2 2 2, 4 1 2 2

2 3 2 2 2, 3 2 2

2 2 2 2 2

2

a 1 2 2, 4 1 2

3 2, 4

4. Savory, 1954. 5. Snvory, 19548.

6. Tumanyan, quoted by Krapivina, 1960. 7. Z)a Costa and Kerruixh, 1963.

VI. Drsccssioiv The differencesobserved between the initiation of the decay penetration of Chaetomitcm ylobosum into beech, birch and Scots pine can reasonably be explained only on a basis of differences in the habitat provided by the three species of wood. These differences, as Corbett (1963) and Courtois (1963) have pointed out, can be found in variations of the gross anatomy of the timbers; in variations of tho fine structure of the wall of the cells; and in chemical differences between the wall components.

350

JOHN LEVY

Scots pine. Passive penetration depends very largely on the existence of free space for the growing hyphae to develop. This may well involve the utilization of cell contents, but does not include enzymic breakdown of cell wall material nor the mechanical intrusion into the wall structure such as might be envisaged by the development of appressoria. In Scots pine there appears to be no difficulty in colonizing the rays, where proliferation of hyphae takes place, and the hyphae appear to pass freely through the large simple pits between ray parenchyma cells and vertical tracheids. Penetration of hyphae from tracheid lumen to tracheid lumen through the bordered pits is not as regular or as frequent as might be expected and would seem to suggest a barrier, probably associated with aspiration of the pits. The resistance to passi ve penetration in ICScots pine, therefore, must depend primarily on the ease with which growth occurs in the ray parenchyma cells with subsequent movement of the hyphae through the Bimple pits into the traclieici lumina. Corbett (1965) gives a photograph which show8 perithecial and mycelial development in Scots pine blocks, twelve days after exposure (Fig. lo), which suggests different rates of penetration. In reporting observations on such blocks exposed to Chaetmium globosum for only three days, she notes that the hyphae were observed, in those blocks with tangential faces exposed, preferentially to enter the ray parenchyma cells and to pass into the tracheids through the pits between the two cells. Little hyphal development was noted in the tracheids. In blocks with exposed radial faces, the hyphae were observed entering the tracheids and passing from tracheid to tracheid via bordered pits, a type of penetration which appears to occur to a greater estent in the spring wood than in the summer wood. Since the ohwrved rate of growth on the block was slower, the penetration through 1mrderc:d pits must therefore be a slower process and one can infer that cithcr the hordered pits form a barrier to progrcssion or else ti. build-iij~of flYpJILL1 material is necessary in the my prenchyrna to riU[)[JlY ttla tiritlgotwid and necessary logistics for an attack to develop. It is axiomatic so far as soft rot fungi are aoncernorl that unless tho hyphae reach the cells, the walls will not be decayed. It therefore follows that the rate of decay of a piece of timber must depend in the first instance on the speed with which passive penetration is effected. If this is slow, then subsequent decay will be slow and the chief reason for the difficulty experienced in using Scots pine as a laboratory test species could well be the slow passive penetration. Duncan's results with the leaching technique could well be due to the leaching action opening up the bordered pit apertures sufficiently to permit easy passage of growing fungal hyphae.

T H E SOFT ROT FUNGI

35 1

Decay penetration in Scots pine is less emy bo explain. The ptLtt,ern of development of first the lateral bmnch, then its penetration t'lwough the first wall and vertical branching developing in the S , layer of the adjacent wall is strange, to say the least. Why should vertical branching occur more frequently in the S2layer of one ccll and not in the other,

FIQ.10. Block8 of Pinus eyloestris 12 days after exposure to the fungus Chnefomirm globosum (Corbett, 1963) ( ~ 2 . )

when both have apparently equal chances of being attacked? Why does the lateral branch never enlarge in size or produce a cavity round i t ? The answers must lie in the physical and chemical nature of the cell wall. One question that may well be pertinent concerns plasmotlosmata. These exist in the primary wall between two young cell^, a8 cytoplasmic

352

JOHX LEVY

connectlions between the cells,.Do they persist i i s t,lw ~ c c ~ o ~ i t l nlayers iy of the wall nre laid c1o\vn? If t,liey do. their prexcn(:c ~\oul(lprovide n very fttcile esplaiint,ion of the observed fiwts. OIIOcould sugpcst, t,li;it t,hc initiathi of the developnient of the hteral brniicli f’rom Iiyphite in t.he lumen of the cell is in response to a stiniulus from possible cytoplasmic or protein residues remaining in the plasmadesma. The branch, through fine, may then produce sufficient enzymes to break down t.he protein residues when the growth of tthe hyplia would then follow the path of enzymic dissolution. ‘J’liose hyphac that penetrate complntely through t,he wall would be growing i n a plasmatlemia that reniains coruplete in spite of the disturbance caused hy the production of the secondary wa,ll. This, however, i m y not be true of all such plasmadesniata and where an unconformity occurs the hyphae can 1jenetrat.eno longer and extension growth is suppressed. Meanwhile, the fine hypha has heen producing ina all imounte of cellulolytic enzynies which, when extension growth ceases, accmnniulate near the tip of the hyphae. Roelofsen (1959) suggests that flow is likely t o be easier along the microfibrils than across t.liem clue to the encrustiitk)ns of lignin and so dissolution may be init,iiLt,cdl).v t,lic flow or tliflusion of’ cnxyincs a l ~ i i gthe easiest path, iw it, is il(!cll~~iiIliLt(!d ;it tlic I ~ y p l i t ~tip. l ‘J’liis would ;~c:c~oirnt for the ‘I’-xhapc?d vc!rt,ic:id hr;~iwIiingof’tliu liypl~~io. r 1 I his i s ~1)(:~:11l~ttiot1 t)ilsctl 0 1 1 i l l ] irriknowri fii(:tor, ~ t r i f lif it is ~ t r o ~ t l t l ( ~not, mist i n the sacoriclory wr~11, tlic WIIOIC of t h t , ~~li~sti~;~tlestiiiiti~ the foregoing pragtxph hccoiiics ~o ~iiuc:l~iionser~sc. H(JLV~\.W: it explains so many of t,he observed facts that further investigations ~ tt proportion of the ultrashould lie nlikde. Who kiiows? it inay I Jthat microscopic: cnvit,ies mentionetl by many ;uit,hors an! scctions of plasrnadenmat,tL which Iiuve heen l~rol<enup (luring the find strlgcs of‘ formation of the secondary \mil. Beech. Passive penetration is effected rapidly hecausc of’ tlie erne with which tlie fungal hyphae can penetrate from cell to cell via the simple pits. Tlie pits are relativeiy abundant in the fihre walls i111t1 so the fungus has inany spaces through tvhich i t rnay pass from cell to cell. The lumen of t,he fibres often appears to be filled by a liirgc hyjhri, from which branches pass through the pits, completely filling the pit apertures. There is thus a close contact between the hyphal material and the wall, which may amount to a higher percentage of the total hyphal surface area than with the other species of WCJOd cx~miirifxl.‘I‘hin may be a reason for the speed with which rleoiiy mcirrs i r i t w d i , clac:ay penetration 1)cginning as ;I t)r;iric:h frorn IL t~,yptiai r i ;L pit, t l p W t , l l ~ ( ? going straight into the w d . ‘I’hc siiiitll clir~rnctorof’ tlic: f i l m s nltulc observation difficult from longitudinal sections, h u t thu (:losoj)roxirnity of hyphae starting to form cavities in the S, layer with latcraf tryphac

:

THE SOFT ROT FUNGI

363

penetrating through the Pits seems to provide at least circumstantial evidence of this. Birch. Passive penetration is effected, as with the other two species, wherever there is space for the hyphae to grow, and there is considerable colonization of the rays. Decay penetration is effected in one of two ways, either lateral penetration and T-shaped branching in the wall, or by erosion of the wall from the lumen. This latter attack is effected by the formation of V-shaped nicks in the wall which may expand considerably but which invariably show a characteristically angled end to the erosion pattern (Fig. 8). This is similar to the angles observed and discussed by Roelofsen (1956) and Frey-Wyssling (1966) at the ends of cavities in the S, layer of the wall. The orosion of tho cell wiills from the lumen in birch is tt Htriking feature and must be related to the nature of the cell wall. Wardrop arid Dadswell (1967) suggest that the S, layer is absent in birch, slthough Meier (1965) shows illustrations in which it appears to be present. Courtoia (1963) quotes Meier (1957) in suggesting that the S, layer of birch fibres contains little lignin, which is not true of softwood tracheids. On the other hand he further suggests that differences in erosion may be due to differences in the mannan and xylan content of this layer and quotes Meier and Yllnev (1956), Liese (1963) and Sachs, Clark and Pew (1963) in support of this assertion. More work is obviously needed before this point can be elucidated, but it should be a relatively easy matter to discover how soft rot fungi react to mannan and xylan in culture and to observe the decay penetration into other timber species with a similar composition to the S, layer, or into reaction wood. From these observed differences in reaction to the three species of wood it does become apparent that the soft rot fungi may well prove a useful tool in the elucidation of the nature of the plant cell wall. Further work, with or without fungi, could well be designed to to& the validity of some of these epeculations. A t thc mme time, thwe in ti n:al need to follow up the work already done with other npcciw of tirntwr and with other species of fungue; not only soft rot fungi, but I m w n nJt, white rot and staining fungi as well. The same fungus, Chaetomium globoeum, can, in one spooies of wood, show afhities with the characteristic habit of staining fungi on the one hand and of Basidiomycete attack on the other. Cartwright and Findlay (1958) describe staining fungi as making a lateral penetration of two adjacent cell walls from a hypha in the lumen of one cell to a hypha in the lumen of the other. The hypha penetrating the wall R i very fine by comparison with the lurnenal hyphae. Chaetomiumglobosum can do this in both birch and Scots pine. This at once raises the question that, if the soft rot fungi pan behave like staining fungi, can the latter r

384

JOHN LEVY

type of fungus produce cavities in the wall like the former? Krapivina (1960) claims that this is so and illustrates typical early stages of cavity formation. More observatioii is required with stnining fungi nnd work is needed to throw light on the nature of the penetration ttlirough the wall in this particular case. Do plasmadesmata play a part here, or is the penetration purely mechanical, by means of appressoria? The V-shaped nicks observed by Corbett (1963) during erosion of the wall of fibres of birch by hyphae of Chuetomiumgloboaum are larger and more sharply defined, yet nevertheless show remarkable similarity to the V-shaped nicks observed by Proctor (1941) during brown rot attack. It would be interesting to discover the conditions which give rise to this effect. What factors determine the way in which the soft rot fungus will behave ? There is a very real need to combine morphological and histological studies, such as are described here, with biochemical studies. What do these observations mean in biochemical terms? How far are chemical differences in the wall materials suppressing or stimulating one enzyme or enzyme system at the expense of others? The amount of enzyme produced by the fungus must be an important factor in the way in which the wall is attacked. Those fungi which produce much extracellular enzyme are able to react with the wall materials at some distance from the hyphal tip, by the simple diffusion of the enzyme ahead of the fungus, such as occurs with the brown rot fungi. Does this mean that the soft rot fungi produce only small amounts of extracellular enzyme? Does the size or shape of the enzyme molecule prove a limiting factor to any extent? Why is the dissolution of the S, layer such a characteristic feature of this type of fungal attack? TH there a correlation between the T-shaped branching prior to cavity formation in the S, layer of the adjacent wall and the fact that the hypha has just penetrated the middle lamella and primary wall? Could this be the reason why no branching occurs in the first wall? If the reason for t h e decay of the S, layer depends on the high cellulose and low lignin content of that layer, how do these fungi compare with the fungi and bacteria known to attack cotton and other cellulose fahrics? Can bacteria cause soft rot, or do they act as a precursor of the soft rotting organisms? Wood preservatives provide another series of investigations that should be combined with these histological studies. Preston [1959) showed that waterborne mixtures of inorganic salts penetrated into the cell wall. How far is this true of other fungistatic and fungicidal materials used as wood preservatives? Is it possible to find a material that has a preference for the S, layer in softwoods and could therefore be used specifically against the soft rot fungi? Alternatively, i t might

T H E S O F T ROT FL‘NOL

355

well be that a material that penetrated only the S, layer could be more effective against soft rot attack in birch. Madhosingh (1961) has shown that a soil fungus, Fwrarium oxysporum, may well reduce the toxicity of certain wood preservatives to the entry and spread of a wood-rotting Basidiomycete. How far is this true of the soft rot fungi and their apparent tolerance of wood preservatives present in wood in conmntrations sufficient to inhibit the development of wood-rotting Basidiomycetes1 The main conclusion to be drawn from this is that much remains to be done to understand the behaviour of wood rotting fungi in wood. Work along these lines could well throw light on the nature of the cell wall and on new ways of arresting the decay penetration of fungal hyphae into the cell wall. The soft rot fungi &re an interesting group and rewarding to work with. They are of economic importance only where insufficient or excess moisture or low dosage of preservative treatment inhibits the growth of the more vigorous Basidiomycetes. They may also turn out to be a stage in the colonization of wood in ground contact and precursor of attack by wood rotting Basidiomyoeta fungi.

ACKNOWLEDGEMENTS The writer wishes to express his thanks to two research assistants, Mrs. Flora Deverall (nbe Lloyd) and Dr. Nanette Corbett, who carried out much of the work described in this article, and who listened patiently to much speculation on their observations. Thanks are also expressed to Prof. R. D. Preston, F.R.S., for his many kindnesses and discussions. Mr. J. G. Savory, earns specis1 thanks for making his list of references and reprints available and giving ungudgingly of hifi time to discuss the subject. Mr. A. Horne, who took most of the photomicrographs, is another to whom special thanks are due.

REFERENCES Abrams, E. (1948). C‘irc. US.Bur. Stand. 188. Armstrong, F. H. and Savory, J. G. (1069). Holzforschung 13 (3), 84-9. Baechler, R. H., Blew, J. 0. and Duncan, C. 0.(19fJl).Americnr~,So&i?~ ./ Mechanical Engineere, P8Per No. 61-PET-5. Bailey, I. W. and Vestal, M. R. (1937). J . Arnold Arbor. 18, 106-208. Barghoorn, E. S. and Linder, D. H. (1944). ParlowuA 1, 396-467. Barn, S. N. (1948). J . Tert. Id., 39, 232. Becker, GI. and Kohlmeyer, T. (1968). A T C ~fiir ~ V~ i 8 C h ? z k & ~ 8 ~ CQ h(1). f$ 2940.

Bryant, Sir A. (1942). “The Years of Endurance 1793-1802”, p. 332. Coiihe, London.

356

,JOHN L E V Y

Campbell, W. G. (1952).“Wood Chemistry” (L. E. Wise and E. C. Jahn), Vol. 2, pp. 1061-1116. Reinhold, New York. Cartmight, K. St. G. and Findlay, W. P. K. (1958).“Decay of Timber and its Prevention”. 2nd ed., H.M.S.O., London. Cartwright, K. St. G., Findlay, W. P. K., Chaplin, C. J. and Campbell, W. G. (1931).Bulletin. Forest Products Research, London, NO. 11. Corbett, N. H. (1963). “Anatomical, Ecological and Physiological Studies on Microfungi associated with Decaying Wood”. Ph.D. Thesis, University of London. Corbett, N. H. and Levy, J. F. (1963).Nature, Lond. 198 (4887). 1322-3. Courtois, H.(1963).Holzforschung und Holzverzuertung 15 (a),88-101. Da Costa, E.W. B. and Kerruish, R. M. (1963).Holzforschung 17, 12. Dippel, L. (1898).Dae Mickroskop Pt. 2, 116-19. Dost, W.A. (1959).Proceedings ofthe 20th Annual Water Conference, Pittsburgh. Pennsylvania. Duncan, C. G. (1960). U.S. Forest Products Laboratory, Madison, Report No. 2173. Findlay, W. P. K. and Savory, J. G. (1950).Proceedings of tho V I I International Botanical Congress, I. 3 15. Findlay, W. P. K. and Savory, d. (4. (1954).Holz Rub-u. Werkst. 12, 293-96. Frey-Wyssling, A. (1956).HoZz u Roh-u. Wwkst. 14 (6).210. Gandy, D. G. (1955).MOA Bull. 64, 551-2. Goff, J. It. and Exeell, J. S. (1961).American Society of Mochanical Engineers 61 (8). Hartig, It. (1878). “Die Zersetzungsersclicinungon des Holzcs der Nadelholzbliume und der Eiche”. Berlin. Johnson, T. W. (1956).M?jcoZogia 48, 841-76. Johnson, T. W., Fercliau, 11. A. and Gold, H. 8. (1959).Phyfon 12 (l),65-80. Jones, E.53. G. (19625).Trans. Brit.mycol. SOC.44, 93-114. Jones, E. B. G. (1962b). British Wood Preserving Association Convention Record, 41-3. Jones, E. 1%. G. (1963).J . Inst. Wood A%. No. 1 I , 14-23. Krapivinrt, I. G. (1960).Lesnoi Zhurnal 3 (l), 130-33. (C>.S.I.I.,81, 145 Brown, C. L., 250, 311 Brown, J . R., 173, 213 Brown, R., 165, 172, 213 Bryant, Sir A., 324, 355 Buchholz, J . T., 234, 311 Bukovac, M . J., 276,277,306, 311,321 Bulard, C., 250, 311 Biinning, E., 142, 143, 144, 145 Burgeff, H . 133, 145 Burk, D., 208, 215 Burkholder, P. R., 252, 318 Burstrom, H., 84, 145, 156, 213 Butcher, R. W., 14, 33 Butenko, R. G., 283, 311 Butt, V . S., 204, 213 Buvat, R., 86, 89, 118, 119, 14.5 Bystrom, B. G . , 97, 104, 148

C Calvin, J. R., 88, 116, 145 Ctmara, A,, 240, 311 Campbell, R. C., 276, 312 Campbell, W . A., 206, 217 Campbell, W . G., 324, 327, 956 Campos, F. F., 300, 311 C a p h , S. M., 241, 319 Cappelletti, C., 234, 311 Caram, B., 11, 34 Cartwright, K . St. G., 324, 326, 353, 356 Caruso, C., 225, 318

Castle, S., 103, 127, 136, 145 Catell, R. B., 50, 67 Chandler, C., 229, 311 Chang, c. w., 145, 251, 311 Cheng, L. O., 81, 148, 167, 217 Chapeville, F., 188, 213 Chaplin, C. J., 324, 356 Chase, S. S., 298, 301, 311 Chayen, J., 202, 203, 213 Cheadle, U . I., 89, 146 Chon, J. C. W., 125, 134, 140, 146 Chong, K. C., 299, 321 Chibnall, A. C., 158, 214 Ctiiiiwd, F. P., 159, 214 Ching, P.T., 277, 319 Ching, K . K., 221, 315 Ching, T . M., 221, 315 C'hopra, R. N., 265, 268, 281, 311 ('houclhury, I%.,259, 308, 311 C'Iiri&rtrmn, G. S.,152. 160, 214 L'liristonson, T., 3, 4, 5, 21, 33 Clarkc,U., 1,2, 7 , 9, 10, 11, 13, 15, 19, 33, 34 Clark, I., 353, 357 Clauson, R. E., 300, 312 Cleland, R., 87, 147, 154, 172, 181, 214, 216 Cochran, G., 65, 67 Coe, E. H., Jr., 298, 312 Coltrin, D., 81, 149 Colvin, J . R., 73, 14, 75, 76, 144, 145, 146, 147, 149 Compton, It. J., 226, 312 Condit, I . J., 277, 319 h n k l i r i , M. K., 241, 25(4 251, 2x0, J2I) Conxtth:l, P.,HI, 1.16

(,'tJOlJf'r, u. c., 257, ,311 ('oopar, F. P., 88, 116, f 4 5 Corbett, N. H., 324, 32!), 330, 331, 337, 338, 34(J, 341, 342, 343, 344, 345, 347, 348, 349, 350, 351, 354, .J56 Corc:oran, M. R., 25.5, 273, 512 Cormack, R. G. H., 96, 145 Coronado, A., 186, 214 Corrcns, C., 226, 312 Corrigan, J . J., 168. 214 Cottone, M . A., 157, 214 Courtois, H., 328, 329, 335, 342, 349, 353, 356 Cowan, S. T., 64, 68 Cox, L. G., 258, 317 Craig, W . R., 272, 273, 274, 320

Crane, J. C . , 231, 274, 276, 277, 311, 312, 31 7 Curtis, J. T., 60, 67, 263, 290, 292, 312 Cutter, V. M., 263, 312 Czaje, A. Th., 104, 112, 145

D Du, Costa, E. W. €3.. 349, 356 Dadswell, H . I., 6, 7, 10, 13, 15, 19,

33, 34 Gregory, W. C., 301, 313 Greig-Smith, P., 36, 43, 67 Grekoff, P. I., 240, 313 Greyson, R. I., 272,273, 274,320 Griggs, W. H., 223, 274, 313 Grumstone, A. V.. 2, 13, 17, 18, 33, 152, 214 Guha, S., 271, 294, 296, 313, 314, 318 Guignartl, L., 220, 313 Giinthcr, I., 72, 76, 97, 146 Custafuon, I?. G., 238. 273, 279, 313 Gut, M., 185, 214 Guzan, I. I., 226, 314

H Haagen-Smit, A. J., 241, 250, 262, 313, 320 Haberlandt, G., 280, 313 Haccius, B., 287, 288, 313 Hadley, H. H., 308, 319 Haeckel, A., 228, 313 Hagedorn, H., 94,146 Hagiaya, K., 223, 313 Hall, C. E., 1, 33 Hall, M. A., 165, 205, 214 Hall, 0. L., 240, 313 Hallaway, M., 204, 213 Halperin, W., 284, 321 Halsoy, I)., 277, 311) Hancock, V. J. I?., 204, 213 Hannig, E., 220, 240, 313 Hansen, J. B., 11, 34 Hamon, C. H., 222, 3 1 3 Harberd, D. .J., 52, 67 Harrington, W. P.,187, 203, 214, 215 Harrison, J . S., 158, 217 Harteck, P., 222, 311 Hartig, R., 324, 356 Hartloy, B. S., 173, 213 Hassid, W. Z., 13, 75, 81, 146, 147 Haurowitz, F., 158, 214 Haustein, E., 300, 313 Hayano, M., 186, 214 Hecht, A., 232, 233, 313 Hecht, K., 161, 214 Hedemann, E., 280, 313 Hcilmann, J., 156, 169, 214 Heinen, W., 231, 513, 516

Hendricks, R. H., 166,217 Hensz, R. A., 307, 308, 313 Hepler, P. K., 86, 142, 143, 146 Heslop-Harrison, J., 40, 67, 272, 306, 300, 308, 313

Healop-Hamison, Y., 308,313 Regs, B., 211, 214 Heme, C. O., 268, 313, 314 Hestrin, S., 72, 74, 147 Heyn, A. N. J., 164, 204, 214, 215 Hillyer, I. O., 306, 307, 321 Hindman, J. L., 272, 273, 274, 320 von Hippel, P. H., 187, 203,214 Hirarnatsu, A., 172, 215 Hiroe, M., 296, 313 Hoffman, L., 9, 10, 19, 22, 33 Hofmeister, W., 219, 313 Holionka, J . , 224, 314 Holsten, R. D., 204, 298, 319 H(ilze1, H., 150, 167, 100, 178, 186, 218 Honda, S. I., 204, 215 Honoyman, J., 71, 146 Honma, S.,238, 249, 268, 313 Hopkins, C. E., 66, 67 Hopp, R. J., 308, 316 Horne, R. W., 78, 79,147 Houwink, A. L., 77, 78, 100, 104, 123, 126, 126, 127,145,146,148 Hudson, C. S., 167, 215 Hughes, D. E., 167, 215 Humphreys, T. E., 211, 215 Huskins, C. L., 299, 313 Hussey, H., 160, 217 Hutterer, F., 169, 215

I Iizuka, M., 227. 314 Ingraham, L.L.,187, 216 Inoh, S., 296, 313 Irreverre, F., 167, 215 Isenberg, I., 211, 217 Islam, A. S., 261, 314 Ito, H., 263, 272, 314 Ito, I., 306, 314 Ito,M., 286, 314 Ivanov, M. A., 280, 314 Iwakiri, B. T., 223, 274, 313 Iwanami, Y.,226, 228, 229, 314 Iwanowskaya, E. V., 249, 266, 314 Iyer, R. D., 206, 318

J

Jackson, D. S., 186, 215

Jackson, G. A. D., 276, 314 Jackson, W. T., 166, 215 Jakus, M. A., 1, 33 Jancey, R., 62 Jang, R., 79, 87, 146, 164, 206, 215 Jansen, E. F., 79, 87,146, 164, 200, 215 Jansen, L. L., 260, 314 Jaworski, E., 70, 146 Jeffers, J. M. R., 61 Jenkins, J. A., 191, 215 Jennings, A., 266, 317 Jensen, E. V., 212, 216 Jenuen, W. A., 79, 114,146, 264,312 Johnson, T. W., 326, 326, 356, 357 Johri, J3.M. 229,230,271,292,294,314 Jones, E. 13. a,,320, 348, 356 J O ~ O SJ., K. N., 197, 21li Joncu, K. S., 40, 68 Jonas, L. I.

*7J

IAt~cuwctril~oo1~, van A., 280, 315 Lemay, P., 96,146 Leopold, A. C., 273, 315 Lesley, J. W., 258, 315 Letham, D. S., 81, 147 Lcvan, A., 298, 299, 315 Levi, M. P., 329,340,346,347,356,357 Levy, J. F., 324, 337, 340, 341, 356 Lewis, D., 229, 279, 315 Lcwis, M., 117, 148 Liese, W., 77, 147, 353, 356 Limayc, W. J., 100, I 4 7 Lintlcr, D. H., 328, 356 Lint-weaver, H., 206, 216 Liriskcns, H. F., 201, 215, 230, 231, 31.7, 315 Lipmann, F., 188, 213 Liverman, J. L., 305, 316 Livingston, G. K., 221, 315 Lloyd, F. J., 337, 338, 356 Loefflcr, F., 13, 33 Loewus, F. A., 224, 319 Lofland, H. B., Jr., 253, 315 Logan, M. A., 159, 167, 216 Lord, W.J., 276, 315 Love, A., 304, 315 Lvve, D., 304, 315 Luckwill, L. C., 238, 242, 273, 274, 315 Lund, H. A., 279, 315 Luther, A., 2, 7, 33 Lyndon, 1%.F., 81, 147

315

Lamport, D. T. A., 81, 89, 105, 147, 156, 160, 161, 162, 169, 179, 183, 184, 185, 188, 195, 197, 198, 205, 207, 215 Lance, G. N., 03, 68 Lang, A., 307, 308, 312 Lang, N. J., 18, 3 3 Larter, E. N., 269, 315 LaRue, C. D., 189, 201, 217, 244, 206, 315, 319

AllcLarw,S.R., 243, 31.5

AUTHOR INDEX

McLean, S. W., 261, 290, 315 MacMillan, J., 279, 315 Macmughton-Smith, P., 39, 41, 42, 47, 63, 66, 67, 63, 65, 66, 68

Madhosingh, C., 366, 356 Maekawa, F., 284, 315 Magoon, M. L., 296, 298, 315 Maheshwari, P.. 232, 236, 236, 237, 239, 264, 266, 209, 270, 273, 280, 280,290,292, 309,314, 315, 316 Mainx, F.. 16, 33 Majumder, S. K., 230, 311 Makita, M., 158, 217 Manner, G., 186, 188, 215 Manning, J . M., 185. 216 Manton. I., 1, 2, 3, 4, 7, 8, 9, 10, 11, 12, 13, 16, 10, 17, 18, 19, 22, 3 3 , d l Mapos, M. 0.. 281, 294, 295, 299, 316,

319 Marco, G., 76, 146 Mardones, E., 186, 214 Margerie, C., 87, 147 Marks, E., 231, 312 Mar& E., 211, 215, 217, 243, 316 Martens, P., 163, 215 Martin-Smith, C. A., 87, 145 Marx-Figini, M., 75, 147 hesart, J., 273, 316 Matchett, W . H., 79, 81. 87, 90, 147 Mathan, D. S., 191, 215 Mathon, C. C., 240, 316 Mathur, R. S., 278, 320 Matsubera, S., 251, 316 Mauney, J . R., 249, 253, 254, 316 Maxwell, E. A., 60, 68 Mears, K., 281, 319 Meeuse, A. D. J., 89, 147 Mehrota, N., 229, 317 Meier, H., 77, 89, 147, 327, 3211, 383, 356 Meister. A., 168, 186, 188, 214, 216, 217 Merry, J . 215, 316 Mertz, D., 204, 216 Meyers, S. P., 326, 356, 357 Miller, E. K., 165, 217 Miller, R. W., 168, 191, 218 MiUett, M. A., 168, 197, 217 Millman, B., 73, 75, 76, 144, 147 Mills, J. A., 168, 214 m e r , H . W., 167, 214 Mitchell, R. L., 168, 197, 217 Mitra, J., 299, 316

365

Mockett. L. G., 42, 83, 56, 66, 68 Mohan Ram, H. Y., 241, 285, 319, 321 Mohr, H . C., 307, 308, 313 Monty, T