Oceanography And Marine Biology

OCEANOGRAPHY AND MARINE BIOLOGY AN ANNUAL REVIEW Volume 25 OCEANOGRAPHY AND MARINE BIOLOGY AN ANNUAL REVIEW Volume 2...

Author: Harold Barnes

160 downloads 1475 Views 8MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

OCEANOGRAPHY AND MARINE BIOLOGY

AN ANNUAL REVIEW Volume 25

OCEANOGRAPHY AND MARINE BIOLOGY AN ANNUAL REVIEW Volume 25

HAROLD BARNES, Founder Editor MARGARET BARNES, Editor The Dunstaffnage Marine Research Laboratory Oban, Argyll, Scotland

ABERDEEN UNIVERSITY PRESS

FIRST PUBLISHED IN 1987 This edition published in the Taylor & Francis e-Library, 2005. “To purchase your own copy of this or any of Taylor & Francis or Routledge’s collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk.” This book is copyright under the Berne Convention. All rights reserved. Apart from any fair dealing for the purpose of private study, research, criticism or review, as permitted under the Copyright Act, 1956, no part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, electrical, chemical, mechanical, optical, photocopying, recording or otherwise, without the prior permission of the copyright owner. Enquiries should be addressed to the Publishers. © Aberdeen University Press 1987 British Library Cataloguing in Publication Data Oceanography and marine biology: an annual review.—Vol. 25 1. Oceanography—Periodicals 2. Marine biology—Periodicals 551.46′005 GC1 ISBN 0-203-40068-2 Master e-book ISBN

ISBN 0-203-70892-X (Adobe eReader Format) ISBN 0-08-035066-8 (Print Edition) ISSN 0075-3218

PREFACE

1987 sees the publication of the 25th volume of Oceanography and Marine Biology: An Annual Review. The series was started in 1963 by Dr Harold Barnes because he thought the time was appropriate to draw together the work being done in all the marine sciences. The aim was stated in the preface to volume 1: “… to consider annually basic aspects of the field, returning to each at appropriate intervals, to deal with subjects of especial and topical importance, and to add new ones as they arise.” As far as possible this aim has been fulfilled. In the 25 years 254 articles have been published involving over 300 authors, some of whom have contributed more than once. At no time has there been any controversy between the editor and authors; the willingness with which they have acceded to editorial requests has always been appreciated and has made the editor’s task a pleasure rather than a tedious trial. The Annual Review has also been well served by its publishers who have attended to its production with meticulous care and so far have always managed to maintain the annual schedule. Manuscripts continue to be submitted to this series; many experts are still willing and even anxious to accept invitations to contribute to it. The desire to publish in it must reflect its importance and value to marine scientists in general. During the years since the death of Harold Barnes in 1978 I have been fortunate in having the advice of many friends and colleagues including, in particular, Drs A.D.Ansell, R.N.Gibson, and T.H.Pearson. Their help has been, and still is, greatly appreciated. It is hoped that this series of Annual Reviews will continue for many more years to fulfil the aims of its founder editor.

CONTENTS

PREFACE

iv

Phytoplankton Dynamics in Marginal Ice Zones WALKER O.SMITH JR

1

Sampling and the Description of Spatial Pattern in Marine Ecology N.L.ANDREWAND B.D.MAPSTONE

26

Flumes: Theoretical and Experimental Considerations for Simulation of Benthic Environments ARTHUR R.M.NOWELL AND PETER A.JUMARS

70

Larval Settlement of Soft-sediment Invertebrates: the Spatial Scales of Pattern explained by Active Habitat Selection and the emerging Rôle of Hydrodynamical Processes CHERYL ANN BUTMAN

89

Aplysia: its Biology and Ecology THOMAS H.CAREFOOT

139

A Review of the Comparative Anatomy of the Males in Cirripedes WALTRAUD KLEPAL

250

The Benguela Ecosystem. Part IV. The Major Fish and Invertebrate Resources R.J.M.CRAWFORD , L.V.SHANNON AND D.E.POLLOCK

305

The Association between Gobiid Fishes and Burrowing Alpheid Shrimps ILAN KARPLUS

458

The Ecological Impact of Salmonid Farming in Coastal Waters: A Review R.J.GOWEN AND N.B.BRADBURY

508

AUTHOR INDEX

520

SYSTEMATIC INDEX

547

SUBJECT INDEX

559

PHYTOPLANKTON DYNAMICS IN MARGINAL ICE ZONES WALKER O.SMITH, JR Botany Department and Graduate Program in Ecology, University of Tennessee, Knoxville, Tennessee 37996, U.S.A.

INTRODUCTION The marginal ice zone is an oceanographic front in which a transition from dense (those waters completely covered with ice) pack ice to one completely free of ice occurs (Fig. 1). The marginal ice zone is dynamic, responding rapidly to physical forcing; hence, the transition from 10/10 ice cover to open water can be abrupt or occur over hundreds of kilometres. The position of the ice edge can vary widely, with mesoscale variations occurring over the time scale of days and large-scale changes occurring seasonally (see Fig. 2). Furthermore, significant interannual variations occur (Niebauer, 1980; Zwally et al., 1983) which are related to global variations in air-sea interactions. An understanding of the processes which create changes in the ice-edge position will allow successful modelling of the marginal ice zone (both its physical and biological processes) as well as prediction of spatial and temporal variations in its position and gradients of ice concentration within it. This in turn will allow human activities in the region (offshore petroleum exploration, transportation, fisheries development) to increase. Polar regions are characterized by environmental extremes. An obvious major seasonal change occurs in the quantity of incident light, with long periods of darkness followed by a rapid change to continuous irradiance. As a result of the low amounts of incident radiation received annually, the local heat budget of polar regions is negative (i.e. a net flux of heat occurs to the atmosphere), which results in very low seawater temperatures, often near the freezing point. Both light, by virtue of its direct and indirect effects on phytoplankton growth, and temperature, via its overall control of microbial growth (Eppley, 1972), strongly influence autotrophic processes in polar regions. Despite the constraints that these factors (and potentially others) place on phytoplankton growth, they in and of themselves do not produce an environment conceptually different from temperate regions. Recent investigations into the physical and biological characteristics of marginal ice zones have, however, clearly shown that the marginal ice zone possesses attributes which are unique and which generate features which distinguish the ice-edge region from either polar or temperate areas. It has been noted for many years that the ice edge is a locus for activity of all trophic levels. For example, in the Antarctic, krill (Euphausia superba) seem to be associated with the marginal ice zone (Marr, 1962), as are many species of whales which feed upon krill (Mackintosh, 1970). Many species of birds are highly concentrated in the marginal ice zone, such as penguins (Spheniscidae) and snow petrels (Pagodroma nivea) (Ainley & Jacobs, 1981). In the Arctic polar bears frequent the ice edge, feeding on seals which surface in the leads during summer. The causes of the elevated higher trophic level biomass and activity in the marginal ice zone can be related to food abundance, using ice as a refugia from predation, and using ice as a nesting and breeding site (Ainley & Jacobs, 1981). Because the food web at the ice edge is ultimately

2

WALKER O.SMITH, JR

Fig. 1.—An aerial image of the marginal ice zone in the Fram Strait: the image was constructed by photography from repeated airplane overpasses; each ‘strip’ is approximately 10 km wide; note the irregular features along the ice edge, including a prominent eddy, ice bands, and meanders; the centre of the eddy is located at approximately 78°48′ N: 2°37′ W, and the mosaic is orientated in a north-south direction; composite provided by Dr R.Shuchman, Environmental Research Institute of Michigan, Ann Arbor, Michigan.

PHYTOPLANKTON IN MARGINAL ICE ZONES

3

dependent on the primary producers, such elevated standing stocks and activities of higher trophic levels must be supported by enhanced primary production within the marginal ice zone. Carbon fixation is conducted by algae attached to the ice and by phytoplankton. Ice algae contribute biogenic material early in the growing season and can reach large concentrations in a spatially confined zone, but the annual primary production associated with the ice when compared with that of the water column plankton is low (Whitaker, 1977; Garrison, Sullivan & Ackley, 1986). Ice-edge phytoplankton blooms (defined as standing stocks greater than those present in the absence of ice) have clearly been shown to be consistent features of marginal ice zones (Hart, 1934; Marshall, 1957; El-Sayed, 1971; McRoy & Goering, 1974; Alexander & Niebauer, 1981; El-Sayed & Taguchi, 1981; Schandelmeier & Alexander, 1981; Smith & Nelson, 1985a, b; Smith, Smith, Codispoti & Wilson, 1985). It is the purpose of this paper to discuss the features of marginal ice zones which make them unique among polar systems and how the oceanographic processes that occur within them influence phytoplankton growth and accumulation. PHYSICAL-BIOLOGICAL INTERACTIONS WITHIN THE MARGINAL ICE ZONE The major feature of the marginal ice zone is the physical presence of ice. The effects of ice are numerous. For example, during its seasonal advance and retreat, the freeze-thaw cycle changes the surface layer’s salinity and density, thereby changing the stability of the upper water column. Melting ice has been shown to create large density gradients in the Arctic (e.g. Marshall, 1957; McRoy & Goering, 1974; Alexander & Niebauer, 1981; Smith et al., 1985) and the Antarctic (e.g. Jacobs, Gordon & Ardai, 1979; Jacobs, Huppert, Holdsworth & Drewry, 1981; Smith & Nelson, 1985a); in contrast, freezing ice and subsequent brine rejection leads to instability and deep vertical mixing (Foster, 1968; Matthews, 1981). Changes in the stability of the upper layers of the ocean can have a rapid impact on plankton growth. The effect of stability on phytoplankton growth was first described by Riley (1942) and Sverdrup (1953), who mathematically derived the relationship between critical depth and the depth of vertical mixing to predict the onset of a temperate spring phytoplankton bloom. At a receding ice edge, low salinity (hence low density) melt-water is released, and the vertical stratification is greatly increased (Marshall, 1957; Alexander & Niebauer, 1981; Smith & Nelson, 1985a). This maintains the phytoplankton in the upper portion of the water column and provides sufficient light for growth to occur. In contrast, increased vertical mixing caused by brine rejection should decrease phytoplankton growth; such an effect has never been directly observed. The effect of melting ice on vertical stability appears to be the major factor in the initiation of a phytoplankton bloom within the marginal ice zone. Ice also modifies the quantity of surface irradiance which reaches the water column. Ice within the marginal ice zone can be either annual (produced within the past winter season) or multi-year ice (present for more than one winter). The optical characteristics of the two types are substantially different, but both greatly attenuate light. In general, pack ice (including any snow cover layer) will absorb from 80–99–95% of surface irradiance (Maykut & Grenfell, 1975; Sullivan, Palmisano & SooHoo, 1984). Any microbial community on or within the ice will further contribute to light attenuation. Despite the fact that the compensation light intensity of phytoplankton photosynthesis in polar waters is low (less than 1 µE·m−2·s−1; El-Sayed, Biggs & Holm-Hansen, 1983; Palmisano et al., 1985), phytoplankton photosynthesis in areas of heavy ice cover is severely restricted by the attenuation of light by sea ice per se. Ice can also influence the properties of the water column by modifying the air-sea interaction. For example, ice shields the water column from the wind’s energy, thereby reducing vertical mixing in areas with extensive ice cover. In areas with less ice cover, the ice moves more rapidly in response to wind than does the ocean’s upper layer, since the frictional drag on the rougher surface is greater than that on the

4

WALKER O.SMITH, JR

ocean. Another effect of ice is the generation of ice-edge upwelling. Upwelling has been observed within the marginal ice zone in the Bering Sea (Alexander & Niebauer, 1981) and in the East Greenland Sea (Buckley et al., 1979; Johannessen et al., 1983), although never in the Southern Ocean. The proximate cause of iceedge upwelling appears to be wind-driven off-ice Ekman transport (Johannessen et al., 1983; Niebauer & Alexander, 1985), and analytical (Gammelsrød, Mork & Røed, 1975) and numerical (Niebauer, 1982) models confirm this. Røed & O’Brien (1983), however, developed a model with a movable ice sheet which predicted surface convergence (rather than divergence) under the same conditions as the models of Gammelsrød et al. and Niebauer. This discrepancy may be the result of the sensitivity of all the models to various physical coefficients describing air-sea-ice interactions (Røed, 1983). Regardless of the conditions which give rise to upwelling, it is clear that such an event would have a significant impact on phytoplankton growth in the marginal ice zone of the Arctic by injecting nutrients into the euphotic zone, particularly during the summer months after nutrients had been depleted. Any occurrence of ice-edge upwelling in the marginal ice zone of the Southern Ocean, however, might be counter-productive, in that nutrients are rarely limiting, even in the densest blooms (Nelson & Smith, 1986). Upwelling in the Antarctic would reduce vertical stability caused by ice-melt, increase vertical mixing, and thereby result in a reduced light environment for phytoplankton. Therefore, if ice-edge upwelling were to occur in the Antarctic, a decrease in phytoplankton productivity and biomass might result. The ice edge also plays a rôle in the formation of eddies. They have been frequently observed in the marginal ice zone of the Fram Strait region (Johannessen et al., 1987; Shuchman et al., 1987; Manley et al., 1987), and there have been a number of mechanisms proposed to explain their genesis. For example, Hakkinen (1986) modelled the generation of eddies via a variable ice cover along an ice edge, where eddies are generated by a differential wind-induced circulation of water and ice. Topographical anomalies also have been shown to induce eddy formation (Smith, Morison, Johannessen & Untersteiner, 1984), and because the bottom topography in the vicinity of the ice edge of the Fram Strait is complex, eddies are frequently shed. Eddies are important biologically to the marginal ice zone of the Fram Strait for a number of reasons. First, they transfer heat into and out of the ice, increasing the rate of ice ablation and altering the stability of the water column. Secondly, they can move parcels of water under the ice (Manley et al., 1987), thereby reducing the amount of light available in the water column and light-dependent phytoplankton productivity. Finally, eddies induce vertical motion within their boundaries. The location within the eddy of the vertical motion is dependent on the direction of horizontal flow (cyclonic vs. anticyclonic) and whether the eddy is accelerating or decelerating, but in a manner similar to ice-edge upwelling, eddies will inject nutrients into the euphotic zone, and result in a significant stimulation of phytoplankton growth and accumulation during certain periods of the year. FACTORS INFLUENCING PHYTOPLANKTON GROWTH IN THE MARGINAL ICE ZONE Because marginal ice zones occur in different oceanic basins, each with unique physical, chemical, and biological features, it is difficult to generalize among the various ice-edge systems. Therefore, a comparative analysis of different marginal ice zones may be instructive. Four regions with well studied ice edges are included in this analysis: the Bering and Chuckchi Seas, the Fram Strait and the Barents Sea, the Weddell Sea, and the Ross Sea.


5

MARGINAL ICE ZONES OF THE ARCTIC Bering and Chuckchi Seas The Bering Sea is a shallow continental shelf which is normally completely covered during the winter, but completely ice free by early June (Fig. 2). In some years the ice cover retreats over 1000 km (Niebauer, 1982), with the mean distance of melt-back along 170° W being 920 km (Konishi & Saito, 1974). The retreat occurs over a period of three months, so that the mean rate of ice retreat is 10 km per day. There are also variations in rate of retreat during the three months; mean distances of retreat are 135 km in March, 460 in April, and 325 in May (Konishi & Saito, 1974). It has been shown that a large phytoplankton bloom occurs within the Bering Sea marginal ice zone (Alexander & Niebauer, 1981; Schandelmeier & Alexander, 1981), and that this bloom is related to the creation of a vertically stable region in the vicinity of the receding ice. Extremely high concentrations of phytoplankton biomass occur in the bloom (chlorophyll a concentrations of more than 25 µg·l−1) and completely deplete the nitrate within the euphotic zone (Alexander & Niebauer, 1981). The bloom appears to be restricted to the zone of increased stability, but the distance of this stratified region is variable, ranging from 25 km (Alexander & Niebauer, 1981) to 100 km from the ice edge (Niebauer, Alexander & Cooney, 1981). The difference in horizontal extent is a function of the rate of meltwater input, the wind direction and speed, and the local current patterns. Given that there is nutrient depletion within the bloom, further new production (sensu Dugdale & Goering, 1967; Eppley & Peterson, 1979) requires introduction of nitrate into the surface waters. This apparently is accomplished by two mechanisms in the Bering Sea. The first is ice-edge upwelling (Alexander & Niebauer, 1981; Niebauer, 1983; Niebauer & Alexander, 1985), and the second is deep vertical mixing, often enhanced at the 50 m isobath by tidal forces (Niebauer & Alexander, 1985). Without the introduction of nitrate into the euphotic zone after the ice-edge bloom, significant primary production would not be expected to be sustained. That is, the ice-edge bloom would change the strength and timing of primary productivity, but would not increase the total annual carbon fixation. It is clear that mixing and upwelling do occur; their frequency is, however, unknown, and hence an estimate of their impact on phytoplankton blooms in the marginal ice zone is not yet possible. McRoy & Goering (1976) estimated the importance of primary production within the ice-edge system, and concluded that it accounts for approximately 40% of the annual production on the continental shelf. More recently, Alexander (unpubl.) has measured primary production within the marginal ice zone (Table I); her data show the intensity of the bloom during ice retreat and also the expected seasonal cycle in the open waters. When the ice-edge and open-water productivity is integrated through the entire year, approximately 50% of the total production is contributed by the ice-edge system (Table I). Recent estimates of primary productivity in the western portion of the Bering Sea by Sambrotto, Goering & McRoy (1984) indicated that production in that portion of the continental shelf had been significantly under-estimated because the stimulation of growth by nutrient input via cross-shelf flow had not been taken into account. Regardless of the absolute level of open-water production, it is clear that ice-edge phytoplankton blooms in the Bering Sea provide a highly significant contribution to the annual rates of carbon fixation in the region. It is also possible to estimate the impact of marginal ice zone-related blooms at one location by combining data on the extent of the bloom, the duration, the area covered, and the primary productivity within the bloom (Smith & Nelson, 1986). Because the magnitude of the production is variable through time (Table I), the localized impact is also variable. At a location on the outer shelf, primary production is limited to only the early melt-back period (March), and hence production is small (about 0·32 g C·m−2·day −1).

6

WALKER O.SMITH, JR

Fig. 2.—The average maximum and minimum extent of pack ice in the Northern and Southern Hemispheres: data for Southern Hemisphere ice limits from Zwally et al. (1983); data for Northern Hemisphere after Dietrich, Kalle, Krauss & Siedler (1980).


7

TABLE I Primary productivity of open water locations and within the ice-edge phytoplankton bloom: values in brackets are linear extrapolations; data are means of 14C-productivity measurements taken on the continental shelf south of 60° N; data by courtesy of V.Alexander, University of Alaska Primary productivity Ice edge Daily C·m−2·day−1)

Month

(g

January February March April May June July August September October November December Total (g C·m−2·yr−1)

– – 0·32 3·16 6·60 – – – – – – –

Open ocean Monthly (g

C·m−2)

– – 4·03 94·80 204·60 – – – – – – – 309·32 (50·1%)

Daily (g

C·m−2·day−1)

[0·10] [0·11] 0·13 2·05 4·00 1·50 [1·10] 0·69 [0·48] [0·27] 0·06 [0·07]

Monthly (g C·m−2) [2·85] [3·15] 4·03 61·50 124·00 61·50 [33·95] 21·39 [14·40] [8·37] 1·80 [2·40] 308·44 (49·9%)

Because ice retreat occurs at a rate of 10 km·day−1 and the maximum extent of the bloom is 100 km (Niebauer et al., 1981), a bloom would be maintained at one location at most for ten days, and primary productivity contributed by the bloom would be 7·35 g C·m−2. This calculation assumes that the major factor in initiating and sustaining ice-edge blooms is the introduction of melt-water from the retreating ice edge (Alexander & Niebauer, 1981; Smith & Nelson, 1986). Similarly, for a mid-shelf location influenced by the ice-edge bloom in April, the primary productivity at one point would be 20·6 g C·m−2, and for a shallow inner-shelf area which supports a bloom in May, the primary productivity would equal 63·0 g C·m −2. Given that total production on the continental shelf is estimated to be at least 300 g C·m−2·yr−1 (Sambrotto et al., 1984), ice-related production is a modest source of organic matter on the outer shelf. Clearly, as the bloom intensifies it becomes more significant to the yearly production for a single location, and the major impact occurs in areas uncovered during late spring. Significant interannual variations in the extent and concentration of ice cover in the Bering Sea have been reported by Niebauer (1980). These are apparently related to basin-wide air-sea interaction anomalies, in that during years when significant ENSO (El Niño-Southern Oscillation) events occur (Barber & Smith, 1981), ice cover is much less extensive than in ‘normal’ years. During ‘warm’ years the Aleutian low pressure system shifts southward, bringing warmer air temperatures over the Bering Sea which to a large degree restrict the development of ice. A one- or two-year time lag occurs between an ENSO event and the Bering Sea temperature anomaly, indicating that such large scale interactions are of substantial magnitude (Niebauer, 1984). Because the ice-edge blooms in the Bering Sea are related to vertical stability induced by melt-water, it is clear that during warm years the effect of these blooms on the annual productivity cycle is lessened; the magnitude of the interannual productivity fluctuations is, however, unknown.

8

WALKER O.SMITH, JR

The ice-edge system of the Chuckchi Sea has been investigated to a lesser extent than that of the Bering Sea. Significant ice-melt does not begin until June, and large portions of the region remain ice-covered the entire year. Those areas that are uncovered during ice-melt exhibit strong stratification (Fleming & Heggarty, 1966; Hameedi, 1978), and nutrient levels are extremely low during the growing season. For example, nitrate was virtually undetectable within the euphotic zone during the summer at stations near the ice edge (Hameedi, 1978), and surface chlorophyll a levels were moderate (ranging from 0·3–0·8 µg·l−1). An extreme chlorophyll maximum was observed at nearly all stations in the marginal ice zone, being 1–2 orders of magnitude greater than surface levels. This maximum was associated with the base of the pycnocline; in view of the surface layer’s depletion of inorganic nitrogen, the maximum probably formed from surface biomass production and subsequent passive sinking of intact, metabolically active cells. The mean integrated euphotic zone chlorophyll a value was 147·1 mg·m−2, which represents nearly the entire water column in this shallow continental shelf region. Primary productivity ranged from 150 to 6857 mg C·m−2·day−1, with a mean of 1914 mg C·m−2·day−1 (all values reported in Hameedi, 1978, were doubled, since the original data were given on a half-day basis). These values are among the largest ever reported for a marginal ice zone and may be partially reflective of the elevated temperatures at which incubations were conducted. For example, one station had water column temperatures ranging from 0·9 to –1·6°C, but productivity incubations were conducted at 10·2°C. In view of the susceptibility of polar phytoplankton to significant influences of temperature (Neori & Holm-Hansen, 1982), the productivity data of Hameedi (1978) should be viewed with caution until substantiated. From Strait and Barents Sea Although geographically close, the marginal ice zones of the Fram Strait and the Barents Sea are in many ways dissimilar. For example, the Barents Sea is quite shallow (less than 100 m), and its southern portion is permanently ice-free due to the advection into the area of warm (greater than 3°C) North Atlantic water (Sverdrup, Johnson & Fleming, 1943). It also is influenced by freshwater input from rivers. The Fram Strait region, however, is isolated from continental influences and is at least 2000 m deep in most places. The bottom topography is complex and apparently generates a quasi-permanent eddy centred around the Malloy Deep (Wadhams & Squire, 1983; Smith et al., 1984). The area is influenced not only by the northerly flowing North Atlantic water but also by the southerly outflow from the Arctic Basin, the East Greenland Current (Paquette, Bourke, Newton & Perdue, 1985). The boundary between these waters, the Polar Front, is a region of rapidly changing temperature and also separates distinct biological communities (Smith et al., 1985). The marginal ice zone of each region is also different. The ice edge in the Barents Sea begins breakup in May and retreats approximately 475 kilometres by early August (Rey & Loeng, 1985); ice retreat is controlled by localized solar heating. The Fram Strait marginal ice zone is relatively invariant in space, being controlled by the position of the Polar Front (Paquette et al., 1985). When ice is blown or advected over the warmer North Atlantic water, melting occurs independent of the time of year; similarly, if the ice retreats so that it covers only waters originating from the East Greenland Current, little ablation will occur. The distance between the mean minimum and maximum extent of ice in the Fram Strait at 80° N is 120 km (Vinje, 1977). During September-October primary production within the marginal ice zone in the Barents Sea averaged 23·9 mg C·m−2·h−1 (equivalent to 287 mg C·m−2·day−1 if a 12-h day-length is assumed; Heimdal, 1983). During this study the surface nitrate values were generally low, ranging from 0·1–1·6 µM; phosphate and silicate were also low. All stations exhibited marked vertical stability, with the pycnocline generally located near 20 m, although there was some variation in the depth of the mixed layer (from about 12–25 m).


9

Chlorophyll a concentrations were always less than 2·0 µg·l−1. Because pre-bloom concentrations of nitrate in the region are approximately 10 µM (Anderson & Dyrssen, 1981), it is obvious that substantial production and export had occurred prior to the study. No discernible increases in nutrient concentrations (via ice-edge upwelling), phytoplankton biomass or primary production were noted within the marginal ice zone, although the upward movement of isopycnals was observed in at least one section from which biological data were taken (Johannessen et al., 1983). Surface productivity and chlorophyll values were noted to range from 0·8–57·8 mg C·m−3·day−1 and 0.2–1.6 µg·l−1, respectively, in the western Barents Sea (Vedernikov & Solov’yeva, 1972); however, the stations occupied were very close to the coast and may have been influenced by continental processes. Rey & Loeng (1985) also studied the production within the marginal ice zone of the Barents Sea. They found that primary productivity was greatest during the spring (1465 mg C·m−2·day−1) but decreased to much lower rates in summer (239–495 mg C·m−2·day−1) as a result of the restriction of nutrient input due to strong vertical stability. No biological manifestations of upwelling were noted, probably because the sampling period encompassed only eight days in the period May through October. A phytoplankton bloom associated with the ice edge was observed and found to follow the retreat of the ice through the entire summer; the quantity and vertical distribution of phytoplankton biomass was, however, much different in the summer than in the spring, in that the summer distribution was characterized by marked subsurface maxima. This study was one of the few that was able to follow the temporal progression of a bloom associated with the marginal ice zone, and confirmed for this region the overriding importance of vertical stability not only to the timing of bloom initiation but on the limitation of primary production by restriction of nutrient influx. The Fram Strait marginal ice zone is, as stated, different in many respects from that of the Barents Sea. It is, however, similar in that initial nutrient concentrations prior to a spring bloom are alike (Anderson & Dyrssen, 1981), chlorophyll a concentrations within the bloom are somewhat similar (Smith et al., 1985; Smith, Baumann, Wilson & Aletsee, 1987), and the degree of vertical stability is similar in both. Smith et al. (1985) studied the distribution of nutrients and phytoplankton in relation to physical processes and found that nitrate concentrations were reduced in the euphotic zone to below 0·3 µM when adequate light was available for phytoplankton growth. Chlorophyll a concentrations reached a maximum of 11 µg·l−1. Based on the observed chlorophyll-nitrate relationship, a number of stations appeared anomalous (i.e. ‘excess’ chlorophyll a was present relative to the nitrate levels found). This suggested the introduction of nitrate into surface waters by some mechanism which subsequently stimulated phytoplankton growth and/or accumulation. In a study at the same location approximately one year later, Smith et al. (unpubl.) found slightly elevated levels of chlorophyll a in the same area, and that these were associated with lowered concentrations of nitrate (Fig. 3). That area is the site of a quasi-permanent eddy described by Wadhams & Squire (1983) centred in the vicinity of the Malloy Deep (a topographic depression). Because eddies can impart significant vertical motion within their structure, it was concluded that the increased chlorophyll was a result of growth enhanced by eddy-induced vertical flux of nutrients. Furthermore, eddies are a common feature of the marginal ice zone in the Fram Strait (Shuchman et al., 1987), apparently being generated not only via conservation of vorticity (i.e. induced by topographic anomalies; Smith et al., 1984) but by winds (Hakkinen, 1986), and variations in ice concentration (Hakkinen, 1986). The duration of the transient eddies in the vicinity of the ice edge appears to be at least 20 days (Shuchman et al., 1987), which is definitely long enough for an increase in phytoplankton growth and biomass to occur. The spatial extent of the eddies is variable. The Malloy Deep eddy appears to be approximately 100 km in diameter at the surface (although its shape is by no means symmetrical; Fig. 3), and its horizontal motion extends to the bottom (Johannessen et al., 1987). The transient eddies, those associated with the ice edge or formed by instabilities generated by

10

WALKER O.SMITH, JR

the Malloy Deep flow, are smaller (about 30 km in diameter) and may have their vertical and horizontal motions confined to the upper 200 m (Manley et al., 1987). Determinations of primary production in the Fram Strait region are scarce and the seasonal cycle of production is poorly known. Smith et al. (1987) found large variations in productivity across the ice edge, with the largest values (1718 mg C·m−2·day−1) occurring within 10 km of the ice edge. The mean productivity was 426 mg C·m−2·day−1, with minimum values of 29 mg C·m−2·day−1 at a station 60 km from the ice edge in waters totally covered with ice and 178 mg C·m−2·day−1 at a station 50 km from the ice edge in open water. Mesoscale variations in phytoplankton biomass were also noted (Fig. 4) and appeared to be in part related to the movement of the ice edge through time. The data do, however, clearly show that the ice edge consistently is a locus of phytoplankton biomass in the region. Interannual variations in phytoplankton productivity in the Fram Strait marginal ice zone (as well as all ice-edge systems) are basically unknown, despite the well-known yearly variations in ice extent and concentration (e.g. Zwally et al., 1983). Two studies completed in 1983 and 1984 (Smith et al., 1985; Smith et al., 1987), however, suggest the potential for significant differences between years. The 1983 study took place in July-August, whereas the 1984 study was conducted in June-July, so that some seasonal bias in the comparison of data is unavoidable. Chlorophyll a concentrations within the Malloy Deep eddy in 1983 were greater than 300 mg·m−2; in 1984 integrated chlorophyll a values never exceeded 100 mg·m−2 in the same region (both data sets integrated through 100 m). Nutrient values were slightly higher in 1984, but still were less than 1 µM in surface waters during both years. The major difference noted was a difference in temperature; temperatures in 1983 within the eddy were greater than 4°C, whereas in 1984 temperatures were less than 2°C. Vertical density gradients were large in both years. It is unclear whether this difference was related to differences between the two years or whether it was a manifestation of a seasonal effect. Fig. 3 cont. The Fram Strait region, like the Bering Sea, is interesting in that ice-melt results in active phytoplankton growth which ultimately becomes nutrient-limited, and that further new production is dependent on introduction of nitrate into the euphotic zone. Therefore, new productivity is dependent on physical processes such as upwelling or eddy formation, and the annual primary productivity budget, particularly during the summer (and presumably nutrient-limited) months, is dependent on the frequency of these physical events. Because the ice completely disappears by June in the Bering Sea, other processes not related to the presence of ice such as tidal mixing and storm events must replenish nutrients. In the Fram Strait ice-related processes, however, occur throughout the entire growing season. A complete understanding of the phytoplankton dynamics of the Fram Strait will require knowledge of the duration, extent, and magnitude of vertical fluxes generated by the important physical processes (eddies, ice-edge upwelling) found in the marginal ice zone of this region. MARGINAL ICE ZONES OF THE SOUTHERN OCEAN Weddell Sea The Weddell Sea is completely ice-covered for much of the year, with the pack ice extending to near the Antarctic Convergence. The circulation within the Weddell Sea is essentially a large cyclonic gyre, and significant mixing with other water masses (Drake Passage water, Scotia Sea water) occurs at the gyre’s northern extension. Ice retreat begins in October, with ice-melt and compaction occurring at the northern edge and in some years in the centre of the gyre (Zwally et al., 1983). At the minimum extent of the ice


11

Fig. 3.—The distribution of temperature (a), σt (b), nitrate (c), and chlorophyll a (d) in a section taken across the Fram Strait: the influence of the eddy centred over the Malloy Deep can be seen in the density, nutrient and chlorophyll distributions (from Smith et al., unpubl.).

during February, ice remains along the coast of the Antarctic Peninsula (see Fig. 2), but the rest of the region is generally ice-free. The biology of the Weddell Sea marginal ice zone has been investigated since the early DISCOVERY expeditions (Hart, 1934, 1942). Hart (1942) separated various phytoplankton taxa into different groups, one of which included ice-edge forms. He also noted that this group included species whose relative contribution to the seasonal phytoplankton maximum was greatest. Thus, even early descriptive works indicated the biological importance of marginal ice zones. In 1968 an extensive bloom within the pack ice was observed by El-Sayed (1971). Chlorophyll a concentrations of 190 µg·l−1 were measured at the surface, and the bloom was overwhelmingly comprised of the diatom Thallasiosira tumida. The area covered by the

12

WALKER O.SMITH, JR

bloom was estimated to be 15500 km2. Unfortunately, only two stations were occupied which had chlorophyll a values at the surface greater than 2·0 µg·l−1, and water column characteristics were reported for only one of those. It is clear that the accumulation was not due to in situ growth, since nitrate concentrations even if quantitatively converted into phytoplankton biomass could not result in such large chlorophyll a levels. There must have been some concentration of cells from the ice and/or water column during the study. It is most probable that the massive accumulations of diatoms were the result of freezing ice which concentrated the algae from within the top 10 m of the water column by the mechanism proposed by Garrison, Ackley & Buck (1983). Even if ice crystals concentrated all the cells from the upper 10 m to the top few cm of the water column, the surface layer still must have had a chlorophyll a concentration of approximately 15–20 µg·l−1. It should also be noted that El-Sayed (1971) reported large concentrations of pancake ice in the area studied, indicating that water was actively freezing. Therefore the bloom was indeed unusual in its mode of formation, but its importance to seasonal production was probably small.


13

Fig. 4.—The distribution of integrated (from 0–100 m) chlorophyll a in the marginal ice zone of the Fram Strait, June, 1984: the ice-edge position indicated by a dashed line; the marginal ice zone is the major site of growth and accumulation of phytoplankton biomass in this region.

The first detailed study of the marginal ice zone in the Southern Ocean was conducted in 1977 by ElSayed & Taguchi (1981). The cruise investigated the variations in phytoplankton biomass and productivity along the ice edge in relation to local physical and chemical factors. Large differences in autotrophic biomass and productivity were noted between the northern (north of 72° S) and southern sectors; specifically, mean integrated chlorophyll a concentrations were seven times greater in the southern sector (31·6 compared with 4·4 mg·m−2), and primary productivities were four times as great (0·41 compared with 0·10 mg C·m−2·day−1). No difference between the degree of ice cover between the two sectors was observed, and the vertical stability profiles showed no obvious systematic difference. The large-scale biological differences were attributed to water column stability, grazing, and proximity to land masses. During the same study, an analysis of the quantity, distribution, and taxonomic composition of ice-algae was also completed (Ackley, Buck & Taguchi, 1979). In general, low amounts of algal biomass within ice were found; maximum chlorophyll a concentrations were 4·3 µg·l−1 (that value represents the integrated pigment concentration for an entire ice core in which subsamples had been melted and quantified). The potential contribution of ice-algae to overall autotrophic biomass appeared to be larger in the northern portion of the Weddell Sea, not as a result of increased growth and production of ice-algae, but rather as a result of lower levels of phytoplankton biomass observed within the water column. El-Sayed & Taguchi (1981) did, however, describe the potential for biological interactions between the communities in the ice and those in the water column. The phytoplankton dynamics of the Weddell Sea marginal ice zone has recently been studied as part of AMBRIEZ (Antarctic Marine Ecosystem Research at the Ice-Edge Zone). The purpose of this project is to describe the seasonal patterns at various trophic levels within one marginal ice zone. In November– December, 1983, the biomass and productivity of phytoplankton was determined in transects normal to the ice edge to determine the level of activity at the onset of ice retreat. In March, 1986, similar transects were occupied to compare the standing stocks and growth of an ice edge which is advancing with one which had been retreating (Fig. 5). The first cruise started approximately six weeks prior to the maximum solar angle (most direct radiation), and the second approximately nine weeks after the equinox. Vertical density differences observed during the 1983 cruise were not as great as those in 1986 (Nelson et al., 1987; Muench

14

WALKER O.SMITH, JR

& Husby, unpubl.). The mean depth of the mixed layer in spring was 55 m; in autumn it averaged 35 m. In fact, the vertical stratification observed in the autumn was nearly constant and independent of ice cover (Fig. 5), whereas the mixed layer depths in spring were highly variable, ranging from 15 to 130 m, and spatially correlated with the retreating ice edge. These conditions persisted despite (or as a result of) the surface freezing and melting that occurred during the autumn and spring, respectively. Chlorophyll a concentrations were much greater in spring than in autumn, with mean euphotic zone values in the spring of 3·1 µg·l−1 (maximum value 12·9 µg·l−1), and in the autumn the euphotic zone average being 0·18 µg·l−1 (maximum observed value 1·80 µg·l−1). Maximum standing stocks in the spring occurred in the vicinity of the ice edge, but no clear spatial correlation between the vertical stability and chlorophyll distributions in autumn was found. Primary productivity values paralleled the biomass distribution (Fig. 5), with integrated (through the depth to which 0·1% light penetrated) means for spring and autumn being 571 and 200 mg C·m −2·day−1, respectively. Thus, the conceptual model that an ice-edge phytoplankton bloom is initiated by melt-water induced stratification and is dissipated by the breakdown of this stability appears only partially correct for the Weddell Sea ice edge, in that strong stability was still observed in autumn but elevated phytoplankton standing stocks were absent. The water column characteristics were qualitatively similar to those found in the Scotia Sea (Rönner, Sörensson & Holm-Hansen, 1983; Glibert, Biggs & McCarthy, 1982), where mixed layer depths of approximately 50 m occurred in the absence of ice. The low biomass and primary productivity encountered in autumn, 1986 may have resulted from low ambient light conditions, substantial grazing pressure or the dependence on ammonium as a nitrogen source for growth; regardless of the cause(s), it is clear that an ice-edge bloom in this region did not persist into the austral autumn. Ross Sea The Ross Sea is hydrographically less complex than the Weddell Sea in that it is somewhat more isolated from the large-scale circulation patterns of the Fig. 5 cont. Fig. 5 cont. entire Southern Ocean. Water movement within the Ross Sea can be described as a cyclonic gyre (Ainley & Jacobs, 1981), with water flowing along the coast of Victoria Land towards the equator and along the Ross Ice Shelf towards Victoria Land. The entire sea is ice-covered until late October, when open-water areas begin to appear along the edge of the Ross Ice Shelf at about 175° E (Zwally et al., 1983). The open areas continue to expand equatorward and towards Victoria Land until the entire region is ice-free east of 160° E. Ainley & Jacobs (1981) reported that in December areas with little or no ice exhibited the colouration and light attenuation coefficients characteristic of water with large accumulations of phytoplankton (no chlorophyll a measurements were made), and found that these bloom areas were confined to shelf waters which had been recently uncovered via ice-melt. Holm-Hansen, El-Sayed, Franceschini & Cuhel (1977) occupied a number of stations in the Ross Sea, but none of them can be considered to be within the marginal ice zone. Their data do provide, however, a good estimate of the background (i.e. areas not recently influenced by ice) biomass and productivity in the Antarctic region. Mean integrated (through 200 m) chlorophyll a and carbon assimilation values were found to be 10·4 mg·m−2 (range 0·73–30·8 mg·m−2; n=16) and 140 mg C·m−2·day−1 (range 40–294 mg C·m−2·day −1; n=23), respectively. El-Sayed, Biggs & Holm-Hansen (1983) investigated the nutrient and phytoplankton distributions in the vicinity of the Ross Ice Shelf during January but, again, these locations cannot be considered to be part of the marginal ice zone at the time of the study because any effects of the pack ice had been dissipated. The mean integrated (through the euphotic zone) chlorophyll a value was 19·1 mg·m−2,


15

Fig. 5.—The distribution in 1983 (a, c, e) and 1986 (b, d, f) of density (as sigma-t), chlorophyll a, and primary productivity from sections normal to the ice edge in the Weddell Sea: the 1983 section was conducted in the austral spring (November) and was orientated along a north-south axis at about 61° S; the 1986 section was completed in autumn (March) and was orientated in an east-west direction at 64° S: data for a and c from Nelson et al. (1987); data in b courtesy of Drs R.Muench and D. Husby; data in d, e and f from W.Smith (unpubl.).

16

WALKER O.SMITH, JR

slightly higher than that reported by Holm-Hansen et al. (1977). El-Sayed & Turner (1977) reported a mean of 20·2 mg·m−2 for chlorophyll a concentrations within the euphotic zone in the Ross Sea.


17

The distribution of phytoplankton biomass, productivity, and nutrient uptake at the ice edge off Victoria Land in January–February, 1983 was investigated by Smith & Nelson (1985a, b). Transects occupied

18

WALKER O.SMITH, JR

normal to the ice edge found that a large freshwater lens was present at the surface which created a large density gradient within the water column, reducing the mixed layer depth to less than 20 m. The spatial distribution of phytoplankton biomass (as evidenced by chlorophyll a, particulate carbon, particulate nitrogen, biogenic silica, and cell numbers) was highly correlated with the spatial extent of the zone of reduced salinity (and density). At locations removed from the ice-melt influence, vertical stability and phytoplankton biomass decreased markedly. The bloom was confined to an area about 250 km from the ice edge. Some variations in biomass along the ice edge were noted (Smith & Nelson, 1985b) and were attributed to fast-ice effects on the source water. The stratification induced by ice-melt was observed in all transects occupied. Because a number of estimates of biomass were measured concurrently in this study, it provided insights into what may be unusual adaptations of some species of Antarctic phytoplankton. Chlorophyll concentrations in the mixed layer of the bloom averaged 3·7 µg·l−1 (Smith & Nelson, 1986). Particulate carbon levels for the same samples averaged 39·5 µmol·l−1; therefore, the carbon/chlorophyll (w/w) ratio was 118. Such a value is extremely high when compared with temperate and tropical systems, and may reflect the low light conditions within this bloom (due to self-shading). Other studies, however, have also observed elevated C/Chl. ratios (Li, 1980; Sakshaug & Holm-Hansen, 1984), so it is possible that this is an adaptation to the low temperatures encountered in polar waters. The biogenic silica concentrations (a measure of the diatom, silicoflagellate, and radiolarian opaline material) were the highest ever measured in any ocean, averaging 24·4 µmol·l−1 within the bloom’s mixed layer (Smith & Nelson, 1986); in fact, the concentrations observed were higher by a factor of two than the previous maxima (Smith & Nelson, 1985a). The biogenic material was composed almost entirely of diatoms (Wilson, Smith & Nelson, 1986). The silica/carbon molar ratio was 0·60; normal oceanic diatoms have ratios of approximately 0·12 (Brzezinski, 1985). Therefore, the bloom contained only moderate levels of chlorophyll, very high levels of particulate carbon, and massive amounts of biogenic silica. If observations had consisted of only chlorophyll, an erroneous impression of the magnitude of the bloom would have been obtained. Microscopic examination of whole water samples indicated that extremely low levels of recognizable detrital material were present; nearly all of the observable particles were intact diatoms. Primary productivity within the bloom averaged 960 mg C·m−2·day−1 (Wilson et al., 1986), and outside of the high stability region it averaged 617 mg C·m−2·day−1. Integrated (through the euphotic zone) chlorophyll a concentrations for the bloom stations averaged 63·7 mg·m−2 (Smith & Nelson, 1985b). Species composition within and outside the bloom was similar (Smith & Nelson, 1985a), and growth rates within and outside the bloom were also similar. This was interpreted as evidence that at the edge of the bloom, the depth of vertical mixing increased, distributing the biomass through the entire water column, and hence over short time periods no difference in species composition would be noted. Furthermore, growth rates were similar because both areas were low-light environments; within the bloom the high standing stocks of phytoplankton caused self-shading, and outside the bloom deeper vertical mixing (the average mixed layer depth outside the bloom was 40 m) reduced the average light environment encountered by the phytoplankton. Few studies have investigated the form of inorganic nitrogen removed by phytoplankton within marginal ice zones of the Southern Ocean. Outside the marginal ice zone it has been found that nitrate provides anywhere from 5 to 50% of the total nitrogen required for growth (Rönner et al., 1983). It is this percentage of primary production that can be converted to higher trophic level biomass and is available for export (Dugdale & Goering, 1967; Eppley & Peterson, 1979). In view of the massive amounts of nitrate that are almost always present in surface waters in the Antarctic, such extensive use of recycled nitrogen (ammonium) is somewhat surprising, although it has been suggested on the basis of nutrient utilization


19

patterns (Kamykowski & Zentara, 1985). Nelson & Smith (1986) measured ammonium and nitrate uptake within the marginal ice zone of the Ross Sea and found that on average 65% (range 32–95%) of the primary production was supported by nitrate. Such extensive amounts of new production (i.e. exportable to higher trophic levels or as particulate flux out of the euphotic zone) are among the highest measured values to date in the ocean, and may help explain why marginal ice zones are often the site of large accumulations of herbivores (krill, copepods, baleen whales, salps), birds (penguins, fulmars, petrels), fishes (myctophids), and marine mammals (seals, whales). IMPACT OF ICE-ALGAL COMMUNITY ON ICE-EDGE PHYTOPLANKTON BLOOMS Numerous species of algae grow on and within the pack ice of both polar regions (Horner, 1976; Horner & Schrader, 1982; Palmisano & Sullivan, 1983; Garrison, Sullivan & Ackley, 1986). It is now clearly established that the ice-algae begin active growth prior to any phytoplankton bloom by virtue of their relative position in the light field (Matheke & Horner, 1974), and upon the disintegration of the ice, the algae are released into the water column. It has been suggested that the ice-algae, by virtue of their extreme biomass (chlorophyll a concentrations within a small section of ice often are greater than 100 µg·1−1; Bunt & Lee, 1970; Whitaker, 1977; Palmisano & Sullivan, 1983), can give rise to bloom conditions within the marginal ice zone (Meguro, Ito & Fukushima, 1967). Recent observations have indicated that ice-algae sink very rapidly (Horner & Schrader, 1982) and that the residence time for most species within the water column is short. The fact that most ice-algae do not remain suspended for long time periods does not, however, preclude selected species from actively growing in the highly stratified upper layers of marginal ice zones. Wilson et al. (1986) found that Nitzschia curta, a pennate diatom usually found as a member of the ice-algal community, dominated the bloom in the western Ross Sea. N. curta also occurred in the ice, but it was not as common in the ice as it was in the water. Furthermore, micro-autoradiographic studies confirmed that it was actively photosynthesizing in the water column. Wilson et al. (1986) suggested that N. curta was released during ice-melt, selectively seeded the stratified surface waters, and accumulated to produce the large biomass observed in the study. Garrison & Buck (1985) used a statistical approach to determine the similarity in taxonomic composition of ice-algal communities and water column phytoplankton. They found that there was a significant overlap in taxa between ice samples and plankton taken from water directly below the ice, and suggested that the under-ice pelagic populations are derived from those in the ice. In a similar fashion, phytoplankton are concentrated from the water column during the formation of frazil ice (Garrison et al., 1983). Ice in the Weddell Sea consists of 50–70% frazil ice (Clarke & Ackley, 1984) which forms as the surface loses heat to the atmosphere i.e. where polynias and leads remain open due to wind action. Therefore, substantial amounts of frazil ice can form in energetic environments. As ice crystals form, they rise to the surface (their formation is limited to approximately the upper 10 m), scavenging and concentrating algae as they rise. In such conditions a strong species similarity would be expected. During conditions of ice retreat, a bloom forms however, in the wake of the ice; that is, there should be a spatial separation of the ice and the maximum concentration of bloom biomass. The magnitude of the spatial separation should depend on the rate of ice retreat and water-column seeding. In the Bering Sea, where rates of retreat are rapid, ice-algal communities and bloom communities are strikingly dissimilar (Schandelmeier & Alexander, 1981), whereas in the Ross Sea there seemed to be more overlap (Wilson et al., 1986). The degree to which icealgae serve as innocula for the surface waters undoubtedly depends not only on the input rate of algal cells, but on the. physical characteristics of the ice and the surrounding oceanographic environment.

20

WALKER O.SMITH, JR

PARTITIONING OF PRIMARY PRODUCTIVITY WITHIN MARGINAL ICE ZONES Ice-edge phytoplankton blooms are conspicuous mesoscale features of marginal ice zones, but an assessment of their importance to food-web dynamics and biogeochemical cycles has just begun. Just as numerous physical, chemical, and biological differences exist among the marginal ice zones studied to date, there is no reason to expect that the fate of ice-edge primary productivity will be the same among these various systems. The Bering Sea ice-edge bloom appears to contribute to a demersal food web. Little phytoplankton biomass appears to be utilized by herbivores within the water column during the life of the bloom (Alexander, 1980; Alexander & Niebauer, 1981), but rather sinks to the benthos where it is ingested, remineralized and/or resuspended during storm events. Walsh et al. (1985) suggest that a significant portion of primary production of the Bering Sea is advected and deposited in the sediments of the Chuckchi Sea, but the proportion of the ice-edge production being deposited is unknown. In the Fram Strait region, total annual marginal ice zone primary production appears to be less than in other systems, and zooplankton (large calanoids such as Calanus hyperboreus and C. finmarchicus) appear to remove 80% of the daily primary productivity despite their relatively low biomass (S.Smith, pers. comm.). The region also supports an active pelagic microbial food web (nanoplankton, bacterioplankton, and heterotrophic flagellates). Some biogenic material is lost from the euphotic zone via sinking of particles. Wefer & Honjo (1985) found that at 2100 m the maximum vertical flux occurred in September to October and was not temporally correlated with the maximum period of surface productivity. They also found that the flux of organic matter in autumn reached 7 mg·m−2·day−1, which is approximately 2·4% of the surface production. During the spring and summer seasons, a much smaller percentage of organic matter reaches the deep sea, indicating that surface production is being efficiently cycled at most times of the year. The Ross Sea marginal ice zone may be quite unusual among ice-edge systems in its manner of partitioning of biogenic material. There are large deposits of biogenic particulates in the region, and it has been assumed that these deposits reflect a large surface productivity (Noriki, Harada & Tsunogai, 1985). These deposits of siliceous material are large enough to make the area one of the largest locations for silica removal in the global silica budget (DeMaster, 1981). Carbon productivity for much of the growing season is low, with the exception of the ice-edge bloom which presumably occurs when- and wherever vertical stability results from melting ice. When the annual bloom-related carbon production is calculated from the data of Wilson, Smith & Nelson (1986) and compared with the total yearly production calculated by HolmHansen et al. (1977), the bloom supplies about 67% of the annual (bloom+non-bloom) productivity. When compared with carbon contents and deposition rates of the sediments below (Dunbar, Anderson & Domack, 1985; Ledford-Hoffman, DeMaster & Nittrouer, 1986), it is clear that most (about 90%) of the surface organic production is remineralized within the water column or at the surface of the sediments. When a similar calculation is made for biogenic silica, nearly all of the silica produced at the surface is being deposited in the benthos. This implies that opaline sediment accumulation in the Ross Sea occurs not because of extraordinarily large surface production but by a decoupling of organic and siliceous cycles either within the water column during particle flux or at the sediment surface, and that the marginal ice zone has an extremely important rôle in the global biogeochemical cycle of silica. Micro-paleontological data also indicate the importance of the ice edge to the benthos in the Ross Sea. Nitzschia curta, the overwhelmingly dominant species in the ice-edge bloom (Wilson et al., 1986), is also the dominant form found in the sedimentary record (Truesdale & Kellogg, 1979). Furthermore, it has been deposited consistently throughout the past 18000 years, indicating that the surface bloom is indeed a predictable feature. The contribution of N. curta increases in the biogenic sediments closer to the coast of Victoria Land (Truesdale & Kellogg, 1979), which is consistent with the idea that an ice-edge bloom is


21

initially seeded by ice-algae but the taxonomic composition of the bloom will change in space and time as a result of selective growth of released species. Concentrations of herbivores, particularly euphausids, and other pelagic higher trophic levels may be somewhat less in the Ross Sea than in the Weddell Sea (Marr, 1962; Everson, 1977). It is possible that a large proportion of the biogenic material is utilized in the benthos rather than the water column. Sedimenttrap collections indicate that much of the material involved in vertical flux is whole phytoplankton cells rather than faecal material (Dunbar et al., 1985); therefore, the Ross Sea may function like the Bering Sea and support an active demersal food web. Little quantitative data exist (other than the geochemical evidence of siliceous deposition) to substantiate this hypothesis. The dependence of the Weddell Sea food web on the marginal ice zone production also has not been quantified. Much greater standing stocks of krill (Euphausia superba) have, however, been observed in the South Atlantic sector of the Southern Ocean than any other (Marr, 1962), including massive swarms in the vicinity of Elephant Island (Shulenberger, Wormuth & Loeb, 1984). The numbers of pelagic birds, penguins, and marine mammals also seems to be greater in the Weddell Sea than in other regions (Everson, 1977; Ainley, O’Connor & Boekelheide, 1984; Fraser & Ainley, 1986), but the degree of dependence on production associated with the marginal ice zone is unknown. Extremely high rates of biogenic silica accumulation have been measured in the South Atlantic sector near the northern limit of the Antarctic Convergence (DeMaster, 1981); the rate of organic matter deposition is unknown. Studies of the food web of the Weddell Sea ice-edge system are at present in progress, so that the uncertainties involved in the trophic relationships of this area should become resolved. CONCLUSIONS Marginal ice zones are unusual regions within polar seas; they not only have a distinctive physical setting, being covered by a moving ice pack during portions of the year, but have a characteristic biological system associated with them. The ice edge is a physically dynamic region, with eddies, fronts, jets, and other mesoscale features often present. They are also the sites of certain oceanographic processes such as ice melting or freezing, and upwelling, all of which have immediate impacts on biological communities. Communities in physically dynamic habitats are often coupled to the energy provided by the physical processes in the environment (Margalef, 1978); those in marginal ice zones, and phytoplankton assemblages in particular, are no exception. Recently developed techniques for the study of phytoplankton distributions need to be applied to marginal ice zones. For example, the Coastal Zone Color Scanner (CZCS) has been used in temperate and tropical regions to document the meso- and large-scale distributions of chlorophyll in the ocean. Although problems remain in its application within regions with a high solar angle, preliminary studies of ice-edge systems have been made (Maynard, 1986). Further use of airborne and satellite-based imaging techniques will greatly stimulate traditional biological oceanographic studies, as they have in other regions. Continuous monitoring of biological properties (e.g. optical transmission, fluorescence, bioluminescence) should also be utilized and will enable researchers to have a greater appreciation of the spatial and temporal variations of these properties in marginal ice zones, as well as the interactions between physical and biological processes in these systems. The phytoplankton assemblages of ice-edge regions are often much more productive than those removed from the ice edge or those within the heavy ice pack, and hence a potential exists for the marginal ice zone to be a locus of activity for the entire food web and for the biogeochemical cycles of the region. In many ice-edge systems this has been shown to be true, but the physical and biological processes differ among ice-

22

WALKER O.SMITH, JR

edge systems. Much remains to be learnt about the trophic structure within these ecologically important regions. For example, the annual progression of primary production in a pelagic environment has not been well documented, so that the quantitative impact of marginal ice zones on carbon cycles cannot be accurately estimated. The relationship of the water column plankton to the ice-algal community of the pack ice remains elusive, as does the quantitative significance of nanoplankton and microbial grazing. Finally, the relationship of ice cover (and the biology of the marginal ice zones) to large-scale variations in global circulation is unknown, despite evidence that such teleconnections exist. Given that so much about marginal ice zones remains to be learnt, future research efforts directed at understanding these regions will undoubtedly contribute much to our knowledge of the structure and function of polar ocean systems. ACKNOWLEDGEMENTS Financial support for research in marginal ice zones has been provided by the Office of Naval Research and the Division of Polar Programs, National Science Foundation. Many colleagues and students have collaborated on various aspects of this research and have contributed generously to the synthesis of concepts presented in this manuscript. I would especially like to acknowledge Drs D.Nelson, H.J.Niebauer, S.Smith, and L.Codispoti. REFERENCES Ackley, S.F., Buck, K.R. & Taguchi, S., 1979. Deep-Sea Res., 26, 269–282. Ainley, D.G. & Jacobs, S.S., 1981. Deep-Sea Res., 28, 1173–1185. Ainley, D.G., O’Connor, E.F. & Boekelheide, E.F., 1984. Ornithol. Monogr. 32, Amer. Ornithol. Union, Washington, D.C., 97 pp. Alexander, V., 1980. Cold Reg. Science Tech., 2, 157–178. Alexander, V. & Niebauer, H.J., 1981. Limnol. Oceanogr., 26, 1111–1125. Anderson, L. & Dyrssen, D., 1981. Oceanol. Acta, 4, 305–311. Barber, R.T. & Smith, R.L., 1981. In, Analysis of Marine Ecosystems, edited by A.R.Longhurst, Academic Press, New York, pp. 31–68. Brzezinski, M.A., 1985. J. Phycol., 21, 347–357. Buckley, J.R., Gammelsrød, R., Johannessen, J.A., Johannessen, O.M. & Røed, L.P., 1979. Science, 203, 165–167. Bunt, J.S. & Lee, C.C., 1970. J. mar. Res., 28, 304–320. Clarke, D.B. & Ackley, S.F., 1984. J. geophys. Res., 89, 2087–2095. DeMaster, D.J., 1981. Geochim. Cosmochim. Acta, 45, 1715–1732. Dietrich, G., Kalle, K., Krauss, W. & Siedler, G., 1980. General Oceanography. J. Wiley & Sons, New York, 626 pp. Dugdale, R.C. & Goering, J.J., 1967. Limnol. Oceanogr., 12, 196–206. Dunbar, R.B., Anderson, J.B. & Domack, E.W., 1985. In, Oceanology of the Antarctic Continental Shelf, edited by S.S.Jacobs, American Geophysical Union, Washington, D.C., pp. 291–312. El-Sayed, S.Z., 1971. In, Biology of the Antarctic Seas, edited by G.Llano & I. Wallen, American Geophysical Union, Washington, D.C., pp. 301–312. El-Sayed, S.Z., Biggs, D.C. & Holm-Hansen, O., 1983. Deep-Sea Res., 30, 871–886. El-Sayed, S.Z. & Taguchi, S., 1981. Deep-Sea Res., 28, 1017–1032. El-Sayed, S.Z. & Turner, J.T., 1977. In, Polar Oceans, edited by M.Dunbar, Arctic Institute of North America, Montreal, pp. 463–503. Eppley, R.W. 1972. Fish. Bull. NOAA, 70, 1063–1085. Eppley, R.W. & Peterson, B.J., 1979. Nature, Land., 282, 677–680. Everson, I., 1977. The Living Resources of the Southern Ocean. FAO, Rome, 156 pp.


23

Fleming, R.H. & Heggarty, D., 1966. In, Environment of the Cape Thompson Region, Alaska, edited by N.J.Wilimovsky & J.N.Wolfe, U.S. Atomic Energy Commission, Oak Ridge, TN, pp. 697–754. Foster, T.D., 1968. J. geophys. Res., 73, 1933–1938. Fraser, W.R. & Ainley, D.G., 1986. BioScience, 36, 258–263. Gammelsrød, T., Mork, M. & Røed, L.P., 1975. Mar. Science Comm., 1, 115–145. Garrison, D.L., Ackley, S.F. & Buck, K.R., 1983. Nature, Land., 306, 363–365. Garrison, D.L. & Buck, K.R., 1985. In, Marine Biology of Polar Regions and Effects of Stress on Marine Organisms, Proc. 18th Europ. Mar. Biol. Symp., edited by J.S.Gray & M.E.Christiansen, J.Wiley & Sons, Chichester, pp. 103–122. Garrison, D.L., Sullivan, C.W. & Ackley, S.F., 1986. BioScience, 36, 243–250. Glibert, P.M., Biggs, D.C. & McCarthy, J.J., 1982. Deep-Sea Res., 29, 837–850. Hakkinen, S., 1986. J. geophys. Res., 91, 819–832. Hameedi, M.J., 1978, Mar. Biol., 48, 37–46. Hart, T.J., 1934. ‘Discovery’ Rep., 8, 1–268. Hart, T.J., 1942. ‘Discovery’ Rep., 21, 261–356. Heimdal, B.R., 1983. J. Plankton Res., 5, 901–918. Holm-Hansen, O., El-Sayed, S.Z., Franceschini, G.A. & Cuhel, R.L., 1977. In, Adaptations within Antarctic Ecosystems, edited by G.Llano, Gulf Publ. Co., Houston, pp. 11–50. Horner, R.A., 1976. Oceanogr. Mar. Biol. Ann. Rev., 14, 167–182. Horner, R.A. & Schrader, G.C., 1982. Arctic, 35, 485–502. Jacobs, S.S., Gordon, A.L. & Ardai, J.L., 1979. Science, 203, 439–444. Jacobs, S.S., Huppert, H.E., Holdsworth, G. & Drewry, D.J., 1981. J. geophys. Res., 86, 6547–6555. Johannessen, O.M., Johannessen, J.A., Morison, J., Farrelly, B.A. & Svendsen, E.A.S., 1983. J.geophys. Res., 88, 2755–2769. Johannessen, O.M., Johannessen, J.A., Svendsen, E.A.S., Shuchman, R.A., Campbell, W.J. & Josberger, E., 1987. Science, in press. Kamykowski, D. & Zentara, S.-J., 1985. Mar. Ecol. Prog. Ser., 26, 47–59. Konishi, R. & Saito, M., 1974. In, Oceanography of the Bering Sea with Emphasis on Renewable Resources, edited by D.Hood & E.J.Kelley, University of Alaska, Fairbanks, pp. 425–450. Ledford-Hoffman, P.A., DeMaster, D.J. & Nittrouer, C.A., 1986. Geochim. Cosmochim. Acta, 50, 2099–2110. Li, W.K.W., 1980. In, Primary Productivity in the Sea, edited by P.Falkowski, Plenum Press, New York, pp. 259–279. Mackintosh, N.A., 1970. In, Antarctic Ecology, edited by M.W.Holdgate, Academic Press, London, pp. 195–212. Manley, T.O., Villanueva, J.Z., Gascard, J.C., Jeannin, P.F., Hunkins, K.L. & van Leer, J., 1987. Science, in press. Margalef, R., 1978. Oceanol. Acta, 1, 493–509. Marr, J.W.S., 1962. ‘Discovery’ Rep., 32, 33–464. Marshall, P.T., 1957. J. Cons. perm. int. Explor. Mer, 23, 173–177. Matheke, G.E.M. & Horner, R., 1974. J.Fish. Res. Bd Can., 31, 1779–1786. Matthews, J.B., 1981. J. geophys. Res., 86, 6653–6660. Maykut, G.A. & Grenfell, T.C., 1975. Limnol. Oceanogr., 20, 554–563. Maynard, N.G., 1986. Mar. Tech. Soc. J., 20, 14–27. McRoy, C.P. & Goering, J.J., 1974. In, Oceanography of the Bering Sea with Emphasis on Renewable Resources, edited by D.Hood & E.J.Kelley, University of Alaska, Fairbanks, pp. 403–421. McRoy, C.P. & Goering, J.J., 1976. Mar. Science Comm., 2, 255–267. Meguro, H., Ito, K. & Fukushima, H., 1967. Arctic, 20, 114–133. Nelson, D.M. & Smith, Jr, W.O., 1986. Deep-Sea Res., 33, 1389–1412. Nelson, D.M., Smith, Jr, W.O., Gordon, L.I. & Huber, B., 1987. J. geophys. Res., in press. Neori, A. & Holm-Hansen, O., 1982. Polar Biol., 1, 33–38. Niebauer, H.J., 1980. J. geophys. Res., 85, 7507–7515. Niebauer, H.J., 1982. Cont. Shelf Res., 1, 49–98.

24

WALKER O.SMITH, JR

Niebauer, H.J., 1983. J. geophys. Res., 88, 2733–2742. Niebauer, H.J., 1984. Bull. Am. meteor. Soc., 65, 472–473. Niebauer, H.J. & Alexander, V., 1985. Cont. Shelf Res., 4, 367–388. Niebauer, H.J., Alexander, V. & Cooney, R.T., 1981. In, Eastern Bering Sea Shelf: Oceanography and Resources, Vol. II, edited by D.W.Hood & J.A.Calder, U.S. Dept Commerce, Washington, D.C., pp. 763–772. Noriki, S., Harada, K. & Tsunogai, S., 1985. In, Marine and Estuarine Geochemistry, edited by A.C.Sigleo & A.Hattori, Lewis Publ., Chelsea, MI, pp. 161–170. Palmisano, A.C., SooHoo, J.B., White, D.C., Smith, G.A., Stanton, G.R. & Burkle, L.H., 1985. J. Phycol., 21, 664–667. Palmisano, A.C. & Sullivan, C.W., 1983. Polar Biol., 2, 171–177. Paquette, R.G., Bourke, R.H., Newton, J.F. & Perdue, W.F., 1985. J. geophys. Res., 90, 4866–4882. Rey, F. & Loeng, H., 1985. In, Marine Biology of Polar Regions and Effects of Stress on Marine Organisms, Proc. 18th Europ. Mar. Biol. Symp., edited by J.S.Gray & M.E.Christiansen, J.Wiley & Sons, Chichester, pp. 49–64. Riley, G.A., 1942. J. mar. Res., 5, 67–87. Røed, L.P., 1983. J. geophys. Res., 88, 6039–6042. Røed, L.P. & O’Brien, J.J., 1983. J. geophys. Res., 88, 2863–2872. Rönner, U., Sörensson, F. & Holm-Hansen, O., 1983. Polar Biol., 2, 137–147. Sakshaug, E. & Holm-Hansen, O., 1984. In, Marine Phytoplankton and Productivity, edited by O.Holm-Hansen et al., Springer-Verlag, Berlin, pp. 1–18. Sambrotto, R.N., Goering, J.J. & McRoy, C.P., 1984. Science, 225, 1147–1150. Schandelmeier, L. & Alexander, V., 1981. Limnol. Oceanogr., 26, 935–943. Shuchman, R.A., Burns, B.A., Johannessen, O.M., Josberger, E.G., Campbell, W.J., Manley, T. & Lannelongue, N., 1987. Science, in press. Shulenberger, E., Wormuth, J.H. & Loeb, V.J., 1984. J. crust. Biol., 4, 75–95. Smith, D.C., Morison, J.H., Johannessen, J.A. & Untersteiner, N., 1984. J. geophys. Res., 89, 8205–8208. Smith, S.L., Smith, Jr, W.O., Codispoti, L.A. & Wilson, D.L., 1985. J. mar. Res., 43, 693–717. Smith, Jr, W.O., Baumann, M., Wilson, D.L. & Aletsee, L., 1987. J. geophys. Res., in press. Smith, Jr, W.O. & Nelson, D.M., 1985a. Science, 227, 163–166. Smith, Jr, W.O. & Nelson, D.M., 1985b. In, Antarctic Nutrient Cycles and Food Webs, edited by W.R.Siegfried et al., Springer-Verlag, Berlin, pp. 70–77. Smith, Jr, W.O. & Nelson, D.M., 1986. BioScience, 36, 251–257. Sullivan, C.W., Palmisano, A.C. & SooHoo, J.B., 1984. In, Ocean Optics VII, edited by M.A.Blizard, SPIE, Bellingham, WA, pp. 159–165. Sverdrup, H.U., 1953. J. Cons. perm. int. Explor. Mer, 18, 287–295. Sverdrup, H.U., Johnson, M.W. & Fleming, R.H., 1942. The Oceans. Prentice-Hall, Englewood Cliffs, New Jersey, 1087 pp. Truesdale, R.S. & Kellogg, T.B., 1979. Mar. Micropaleon., 4, 13–31. Vedernikov, V.I. & Solov’yeva, A.A., 1972. Oceanology, 12, 559–565. Vinje, T.E., 1977. Norsk Polarinst. Arbok 1975, 163–174. Wadhams, P. & Squire, V.A., 1983. J. geophys. Res., 88, 2770–2780. Walsh, J.J., Premuzic, E.T., Gaffney, J.J., Rowe, G.T., Harbottle, G., Stoenner, R., Balsam, W.L., Betzer, P.R. & Macko, S.A., 1985. Deep-Sea Res., 32, 853– 884. Wefer, G. & Honjo, S., 1985. EOS, 66, 1271 only. Whitaker, T.M., 1977. In, Adaptations within Antarctic Ecosystems, edited by G.A. Llano, Gulf Publ. Co., Houston, pp. 75–83. Wilson, D.L., Smith, Jr., W.O. & Nelson, D.M., 1986. Deep-Sea Res., 33, 1375– 1387.


25

Zwally, H.J., Comiso, J.C., Parkinson, C.L., Campbell, W.J., Carsey, F.D. & Gloersen, P., 1983. Antarctic Sea Ice, 1973–1976: Satellite Passive-Microwave Observations, NASA Sp. Publ. 459, Washington, D.C., 206 pp.

Oceanogr. Mar. Biol. Ann. Rev., 1987, 25, 39–90 Margaret Barnes, Ed. Aberdeen University Press

SAMPLING AND THE DESCRIPTION OF SPATIAL PATTERN IN MARINE ECOLOGY N.L.ANDREW and B.D.MAPSTONE* Institute of Marine Ecology, Zoology Building, School of Biological Sciences, University of Sydney, Sydney, N.S.W. 2006, Australia

INTRODUCTION The description of pattern is of fundamental importance in ecology. Irrespective of the field of study, all marine ecologists are faced with the problem of establishing and quantifying patterns in nature. Observed patterns are the building blocks of the models from which we generate hypotheses, both about the patterns themselves and about processes that may govern them. Predictive hypotheses about processes suggested to explain observed patterns are tested by experiments and thus experiments are dependent on the patterns we perceive. More subtly, our perceptions of patterns often colour the sorts of questions we ask, thereby canalizing the design of experiments and the sorts of answers we get from them. Furthermore, observed patterns frequently provide the context within which the results of experiments are interpreted. Information on the distribution and abundance of organisms is often the sole basis for ecological and management decisions. The accurate and precise description of pattern is, therefore, essential to most aspects of ecology. Our reading of the literature suggests that the adequacy of sampling methods and designs has not often been demonstrated. Most studies apparently proceed more by custom and tradition than by careful consideration of potential biases and problems inherent in sampling different organisms. Certain procedures or methods are so popular in some fields that they have become “standards”; a good example is the use of the 0·25 m2 quadrat in intertidal studies. In other fields enormously divergent procedures have been used, apparently without justification. For example, in studies of tropical fish, a great variety of methods, ranging m transects, have been used to quantify from counts per unit time of observation to counts per abundances of fish. Many authors and/or editors of journals may consider demonstrations of the adequacy of sampling methods to be too preliminary or mundane to be incorporated into papers. Hence, this information may lie unpublished in filing cabinets or in inaccessible theses or reports. Alternatively, the appropriateness of the design may be determined completely informally, with intuition and experience guiding the design in an unstructured manner. Uncritical acceptance of standard techniques is, however, apparently widespread. Popular methods were often used despite overwhelming evidence of great mprecision and consequent low reliability and repeatability (see Downing, 1979; Resh, 1979). There are obviously many different methods of gathering data on the abundance of organisms. These can vary from the use of microscopy to count bacteria, to the use of cores to collect meiofauna, to aerial surveys of cetaceans. Each method has a special set of conditions and problems. Some methods are definitely more

*Order of authorship determined alphabetically.

SAMPLING AND SPATIAL PATTERN

27

limiting than others: it is much easier to estimate the abundance of algae on a shore than it is to count deepsea fish. Despite the many and varied methods used, the data so collected have two features in common: (1) all are subject to problems of inaccuracy and imprecision arising from the application of the sampling methods; and (2) all ecologists are constrained by time and funding, which restrict the placement and number of samples taken. Although many problems of sampling are specific to the particular methods used, there are considerations in design that have broad generality. Our aim in this review is to assess how marine ecologists have addressed the description of patterns in the arrangement and abundance of organisms and to discuss principles that have general applicability across all disciplines. We shall discuss two aspects of the description of pattern. The term “spatial pattern” will be used to refer to the arrangement of individuals within an area and “patterns in abundance” will refer to the abundance of organisms within and among areas. We have not used the more common term dispersion in our discussion of the arrangement of organisms in space to avoid confusion with the mathematical usage of the term. The term dispersion is reserved to describe the spread, or variability, of estimates about their mean (Sokal & Rohlf, 1981). We shall discuss the design of sampling programmes involving univariate data gathered to answer specific questions about spatial pattern or patterns in abundance of organisms. For example, how many bivalves are there in an estuary and are they non-randomly arranged? Are there more in one estuary than in others along the coast? We offer nothing more novel than the synthesis and interpretation of a large and scattered literature. We hope that this synthesis will promote the careful design of sampling programmes that are tailored to the demands of particular studies. While the review focuses on how marine ecologists have approached the description of pattern, relevant literature about terrestrial or freshwater organisms will be discussed where necessary; the principles are, after all, the same. For convenience, we shall concentrate most of our discussion on the estimation of abundance, but the discussion may be applied to most other variables, e.g. estimates of biomass or behavioural data. Two different aims may be identified when discussing the determination of abundance. One is to estimate the number of organisms in one area, and the second is to examine differences among areas. The design of sampling programmes to fulfil these different aims follows different guidelines. The procedures that best estimate the total number of organisms in an area may require uneven replication within sub-areas, adjusted to the exact requirements of the area (stratified sampling). On the other hand, when comparing estimates among areas, the requirements of the analysis are often best met by having equal numbers of replicates in each area (balanced multi-stage sampling). We discuss the best ways to analyse data collected for these different purposes in the relevant sections. Because we are primarily interested in univariate data sets gathered to answer the sorts of questions posed above, we have not considered multivariate techniques such as classification and ordination. Such techniques are more exploratory or hypothesis-generating in nature and are concerned with answering fundamentally different questions from those that will be discussed here (Green, 1979). Studies concerned with the description of whole communities or species diversity, and measures of association between species are also outside the scope of this review. We introduce sampling programmes designed to assess the abundance of organisms via a discussion of some general considerations such as the formulation of objectives and the interaction between statistical adequacy and biological realism. A discussion of the concepts of accuracy and precision leads on to more specific problems such as the determination of appropriate sizes of sampling units, replication, optimization procedures and the design of single-stage programmes to estimate the number of organisms in one area. This is then extended to more complex designs such as stratified random sampling and multi-stage

28

N.L.ANDREW AND B.D.MAPSTONE

sampling. Power analysis is reviewed as a tool for designing sampling programmes and for analysing the sensitivity of tests. In most marine studies, the description of spatial pattern entails the demonstration of non-random pattern. The various indices used to detect non-random pattern are reviewed along with their application to marine studies. The techniques used to describe non-random pattern in more complex situations, and in more detail, are then briefly outlined. In reviewing the literature we were primarily interested in those papers that concentrated on the description of spatial pattern or patterns of abundance, and especially those that were concerned with procedural aspects, i.e., studies that introduced new tests or justified the use of methods. Journals were searched from 1965 (or their inception) to June 1986; a list of the journals searched is provided in Table I. We have made selective forays into other journals and the earlier literature where necessary, usually as a result of citations from papers found in the above search. We have deliberately avoided citing unpublished papers and those in the “grey” literature, such as local or internal reports, because these are generally not available to most readers. We apologize to anyone who feels a paper of his/hers has been unjustly omitted. TERMS AND CONCEPTS In general we shall keep to a minimum the use of mathematics in this review. There are, however, a number of mathematical terms used that should be explicitly defined to avoid confusion. As far as possible we shall maintain consistency with terminology already used in the ecological literature. The term “population” will generally be used in a biological sense to refer to the local assemblage of organisms being studied (Caughley, 1977; Elliott, 1977). Where “population” is used in a statistical sense, clear qualifiers will be used to avoid ambiguity (e.g. population of values from which sample data are drawn). TABLE I List of journals searched American Naturalist Annual Review of Ecology and Systematics Canadian Journal of Fisheries and Aquatic Sciences (formerly Journal of the Fisheries Research Board of Canada) Coral Reefs Ecology Ecological Monographs Estuarine and Coastal Shelf Science (formerly Estuarine and Coastal Marine Science) Journal of Animal Ecology Journal of Applied Ecology Journal of Ecology Journal of Experimental Marine Biology and Ecology Journal of the Marine Biological Association of United Kingdom Journal of Wildlife Management Limnology and Oceanography Marine Biology


29

Marine Ecology Progress Series Oceanography and Marine Biology: an Annual Review Oecologia (Berlin) Oikos

It is unusual in biological, especially ecological, research to be able to count or measure the entire population being studied (Cochran & Cox, 1957; Cochran, 1963; Southwood, 1966; Elliott, 1977; Seber, 1982). Consequently, data are collected from a subset, or “sample”, of the population and inferences about the population are made based on these sample data. The size of the sample is usually very small relative to the size of the population (Caughley, 1977; Elliott, 1977; Green, 1979). In those cases where a relatively large proportion (>5–10%: Cochran, 1963; Elliott, 1977; Green, 1979; Seber, 1982) of the population is being measured, corrections for finite sampling may be necessary, details of which can be found in one of the many texts that treat sampling design (Cochran, 1963; Scheaffer, Mendenhall & Ott, 1979; Snedecor & Cochran, 1980). We use the term “sampling method” to describe the procedures used to obtain samples (e.g. coring, visual census, vertically hauled plankton nets). The actual device, implement or operation used to obtain mm core, m transect, 250each reading in a sample will be called the “sampling unit” (e.g. mm diameter plankton net). The way in which sampling units are allocated in space and/or time will determine the sampling design. To illustrate: a population of meiofauna might be sampled by taking cores (the sampling method) using a simple random sampling design, with ten replicate cores of 20 mm diameter mm depth (the sample unit) being collected. In this example, each core might be subsampled to economize counting time, in which case a similar statement could be made about the method, design, unit size, and replication of the sub-sampling procedure. “Variables” are the characteristics of the organisms or population being measured. Examples would include standard length, test diameter, density or TABLE II A list of symbols and their meaning as used throughout this review Symbol

Meaning

Description

n x µ

Sample size Variate Population mean Sample mean Population variance Sample variance Population standard deviation Sample standard deviation Standard error

Number of replicate units taken Single measure of variable, e.g. number of individuals in quadrat Mean of the population Mean of sample: estimator of µ Variance of the population Variance of sample: estimator of σ2 Standard deviation of population Standard deviation of sample: estimator of σ Standard deviation of

σ2 s2 σ s SE

distance from nearest neighbour. “Parameters” are mathematical or numerical values used to describe some characteristics of the entire population of the variable being measured (Winer, 1971; Scheaffer et al., 1979; Snedecor & Cochran, 1980). In most situations, the values of parameters are never known since they are the exact or “true” measures of an entire population of values. Normally we only have a sample of those values.

30


For example, the true mean and variance of the weights of a population of snails are parameters of the population. Estimates of parameters are represented by the corresponding values obtained from a sample of the population; e.g. the arithmetic values of the mean and variance of the weights of a sample of snails taken from the above population. Winer (1971) and Elliott (1977) refer to estimates obtained from samples as “statistics”. We shall use “statistics” in two specific senses: sample or descriptive statistics being synonymous with the sample estimates, as in the above usage; and “test statistic” meaning the derived numeric value (e.g. F-value, t-value) used in a statistical test of a hypothesis about the data. Clearly, when an entire population is measured, the sample will be the population, and the parameters and their estimates will be identical. As noted above, this is unlikely in ecological contexts. The symbols and their meanings used throughout this review are given in Table II. Note that we have listed only the arithmetic mean as a measure of location and standard deviation, variance and standard error as the associated measures of dispersion. These are, by far, the most often used parameters or estimators in the ecological literature. ESTIMATING ABUNDANCE Virtually all ecologists are faced at some stage with the task of estimating abundances of organisms in the field. In recent years there have appeared a few publications specifically concerned with the design of sampling programmes and/or experimental studies (Holme & McIntyre, 1971; Caughley, 1977; Green, 1979; Underwood, 1981; Hurlbert, 1984; Underwood & Denley, 1984; Millard, Yearsley & Lettenmaier, 1985; Millard & Lettenmaier, 1986). These recent publications differ from excellent works such as Cochran & Cox (1957), Cochran (1963), Winer (1971), Scheaffer et al. (1979), Snedecor & Cochran (1980) and Seber (1982) in that the former were more concerned with ecological problems than the latter texts, which tended to concentrate on statistical procedures, agricultural experiments or sample survey design for human populations. Our purpose in this section is to outline the procedures that have been, or might be, followed by researchers intending to sample marine organisms. This will mostly involve reiterating the points made in the above publications. Green (1979), in particular, has thoroughly outlined the proper development of sampling programmes for environmental impact studies. Our discussion will largely parallel his “ten principles”. QUESTIONS, SCALE AND GENERAL CONSIDERATIONS As Green (1979) has pointed out, the interpretation of the results of a sampling exercise will be only as clear as the statement of the objectives of the study. It cannot be emphasized too strongly that before a sampling programme can be designed efficiently, all the questions to be addressed must be stated clearly and explicitly. Ambiguity in the intentions can only result in corresponding weaknesses in the design of the programme and the results that come from it. For example, if the objective of a study is to estimate the total number of holothurians on a particular coral reef, then a stratified sampling design should be used. If, however, the objective is to compare abundances among reefs, then a multi-stage sampling design should be used. It is most important to identify the sources of variation that might influence the data and, hence, the perception of pattern. Sources of variation in sampling programmes are often associated with phenomena operating at various scales, or can be estimated at several scales. For example, Caffey (1985) studied the recruitment of barnacles at several sites on each of several rocky shores at several locations along the New South Wales coast and thus accounted for variation at four spatial scales. The spatial scale at which a study


31

is done can have major consequences for both the results obtained and their subsequent interpretation (recent references include Harris, 1980; Denman & Gargett, 1983; Dayton & Tegner, 1984; Irish & Clarke, 1984; Millard et al., 1985; Wiens, 1986; Wiens, Addicott, Case & Diamond, 1986). Perceptions of “scale” are entrenched in our perception of problems, and subsequent generation of hypotheses. As Resh (1979) and Harris (1980) pointed out, however, we must deal not only with the convenient scales of our own perceptions and methods, but also with the functional scales at which the organisms are likely to respond to their environment. Clearly, it is logistically impossible to deal with all scales and/or sources of variation and careful choices have to be made. Results cannot be interpreted at scales, or for sources of variation, that the sampling programme did not address. Careful consideration has to be given also to the nature of the factors being investigated: are the questions being posited for only specific situations (e.g. on a particular coral reef) or is more generality required (the whole Great Barrier Reef)? Resolution of this question influences not only the formulation of testable hypotheses and the interpretation of results, but also the selection of study conditions (such as locations): studies of a single coral reef in one region cannot be used to infer patterns for the whole Great Barrier Reef. Resolution will come partially from a greater attention to the framing of questions and partly from a recognition that many ecological questions need answering at more than one spatial scale. An explicit statement of the questions to be addressed by a sampling programme will not always be straightforward. The questions must be expressed sufficiently clearly that specific null hypotheses can be stated. The hypotheses to be tested will incorporate the original objectives of the study and decisions about the scales of interest. They will strictly determine the subsequent design of the sampling programme. Once a hypothesis is stated, the researcher has implicitly defined the ‘right’ and ‘wrong’ scales at which sampling should be done, so statement must be clear and explicit. Decisions about the appropriate spatial scales and sources of variation have to be made from such intangibles as intuition and experience, in addition to existing natural history and distributional data. These decisions will be easier if the objectives are clearly stated. Incorrect decisions are likely to result in a design that is inappropriate to the objectives of the study and inaccurate perceptions of real patterns. At this stage of the design process, the researcher should have a clear concept of the general structure of the sampling programme. It is imperative to anticipate the analysis of the data and to check that the proposed programme conforms to the assumptions of those analyses. For example, if a programme is designed with the expectation that a three-way Analysis of Variance will be used to analyse the results, the form of the Analysis of Variance—the F ratios and their degrees of freedom—should be examined before the data are collected. If, for example, a fully factorial analysis is anticipated in which two of the factors are considered random factors, then there will be no valid F test for the main effect of the fixed factor, unless at least one of the first order interactions containing that factor is clearly non-significant (P>0·25, by convention, Winer, 1971) so that a pooled denominator mean square can be calculated (see Winer, 1971; Snedecor & Cochran, 1980; Underwood, 1981). Changing a priori the designation of one of the random factors to ‘fixed’ may correct the problem of analysis but will also fundamentally alter the original hypotheses, the way in which sampling is done and the interpretation of the results (Underwood, 1981). Problems of this kind must be sorted out a priori rather than a posteriori. As Green (1979) has commented, the time for statistical advice is when the programme is being designed, not after the data are in hand and one is wondering what can be done with them. The investigator must next give careful consideration to the way in which samples will be collected. In many situations, the identity and spacing of the basic unit of replication is relatively clear e.g. discrete units such as patch reefs, tide-pools or islands. This is less clear, however, in situations where the sample unit is arbitrary (e.g. a quadrat) and the area being studied cannot be divided, a priori, into natural units. For

32


example, if a large and relatively uniform area has to be arbitrarily divided, then how far apart should the basic units be: 1, 10, 100 or 1000 metres? The determination of the spatial scales at which a study will be done involves an explicit judgement about the relative importance of variation at different scales. Variability among these scales, and its importance, will vary with the species and physical environment studied but can be estimated before the main study by doing multi-stage pilot studies. Replication at all levels to be considered is essential (Green, 1979; Hurlbert, 1984). Furthermore, Green (1979), Bernstein & Zalinski (1983), Hurlbert (1984), and Underwood & Denley (1984), among others, have all discussed several aspects of confounding in experimental or sampling designs. All have emphasized, that data must measure variability within as well as among the levels of a sampling programme and that the sources of variation about which hypotheses are posed must be unambiguously identifiable in the data. Hurlbert (1984) presented an extensive discussion of the ways in which designs have often been confounded. In general, random collection of sample units within the lowest level (smallest source of variation considered) of a design is recommended to ensure independence of the data and to avoid systematic error. True randomization is often difficult to implement but great care should be exercised in the substitution of ‘haphazard’ sampling for random sampling: ‘haphazard’ sampling may introduce subtle, unanticipated biases. Millard et al. (1985) have clearly demonstrated the effects of non-independence of sample data. It is also important to consider the proportion of an area or population that will be sampled by some chosen number of replicates, particularly where random samples are to be taken from an area or set of areas on a number of occasions. If the proportion of the local population or area sampled each time (the sampling fraction) is greater than 5–10%, then the estimate of abundance may need to be adjusted for finite sampling (Cochran, 1963; Elliott, 1977; Seber, 1982). Where the sampling fraction is larger, there may be a high probability of samples substantially overlapping and introducing unplanned confounding in the data. Replicates will not be independent and variability will be under-estimated. When comparisons of abundances are of interest, it is also desirable that the sampling programme remains balanced—that is, that replication at each level of the programme is consistent across all treatments. Comparative analyses of unbalanced data are possible for many situations, but in general the results of unbalanced analyses are less reliable than those of a corresponding balanced analysis (Bradley, 1968). When an initially balanced sampling programme becomes unbalanced it may be preferable to rebalance the data set by omitting some values prior to analysis. When the objective of a sampling programme is to estimate the total number of organisms in an area (e.g. the population of cockles in a bay), however, the best design may be an unbalanced stratified sampling design. Where sampling is designed to measure an effect of some perturbation or treatment, it is essential that control situations are also sampled so that the effects of treatments can be distinguished from stochastic processes or other, unspecified events that affect only some sampled areas (Green, 1979; Millard & Lettenmaier, 1986). Bernstein & Zalinski (1983) have elaborated on Green’s statements and have emphasized the need to consider various scales of change in both control and treatment conditions. Finally, after the sampling programme is completed, the data should be examined to ensure that they conform to the assumptions of the proposed statistical analyses and they should be transformed if necessary. This subject has been covered repeatedly, both in the statistical literature (Winer, 1971; Snedecor & Cochran, 1980; Sokal & Rohlf, 1981) and in the ecological literature (Downing, 1979, 1980, 1981; Green, 1979; L.R.Taylor, 1980; W. D.Taylor, 1980; Underwood, 1981; Morin, 1985) and will not be laboured here.


33

ACCURACY AND PRECISION In most standard English dictionaries, accuracy and precision are defined in terms of each other and are often noted as synonyms (Gove, 1969; Sykes, 1982). In scientific and engineering fields, however, the terms have come to represent quite different concepts (Cochran & Cox, 1957; Cochran, 1963; Lafedes, 1978; Sokal & Rohlf, 1981; Lincoln, Boxhall & Clark, 1982). The difference between common and technical usage has apparently led to inconsistent use of the two terms in much of the ecological literature. Clear distinction between the terms seems, however, to be both useful and desirable. Accuracy is defined as the closeness of a measurement or estimate to the true value of the variable being measured or parameter being estimated (Cochran & Cox, 1957: Cochran, 1963; Sokal & Rohlf, 1981; Lincoln et al., 1982). Thus, accuracy refers to the location of an estimate relative to the location of the true value. For example, if the true mean standard length (SL) of a population of fish is 43·7 mm, then a sample mean of 43·6 mm SL is a more accurate estimate of the population mean than a sample mean of 40·5 mm SL. A method that gives estimates that are repeatedly and predictably inaccurate is said to be biased. Bias, then, is the systematic deviation of an estimate from the true value and is caused by artefacts of the method used to obtain the estimate. Precision refers to the degree of concordance among a number of measurements or estimates for the same population (Cochran & Cox, 1957; Cochran, 1963; Sokal & Rohlf, 1981; Lincoln et al., 1982). Precision is reflected by the variability of an estimate. In the above example, if a number of samples of the population of fish returned a set of estimates of mean SL that ranged from 40·3 mm to 40·7 mm, then those estimates would have greater precision than a set with range 46·5 mm SL-52·0 mm SL. Note that accuracy cannot generally be inferred from precision. The preceding two sets of samples were both inaccurate (because the true mean was 43·7 mm) but the first was far more precise than the second. Clearly, then, estimates can be inaccurate but precise, both accurate and precise and so on. Precision and accuracy are truly synonymous only when no methodological biases exist (Cochran & Cox, 1957). The terms have been used in this sense (e.g. Winer, 1971; Underwood, 1981; Seber, 1982) but it is unlikely that bias will be absent in biological research and we will assume that some bias is always, at least potentially, present. We thus use the terms strictly and separately. Two other terms that have often been used with accuracy and precision are reliability and repeatability. These terms are apparently less clearly defined than accuracy and precision but have sometimes been used as synonyms for accuracy or precision, respectively (Cochran & Cox, 1957; Cochran, 1963). We shall use the terms reliability and repeatability to describe something of both accuracy and precision. Repeatability will be used to refer to the degree to which a sampling design or method can be expected to give consistently estimates with the same accuracy and precision. Reliability refers to the confidence that we can have in a given method or procedure: a reliable estimate would be both accurate and precise and an unreliable estimate would be neither. Accuracy and methods Inaccuracy in estimates can be attributed to two main sources: (1) inappropriate design of the sampling programme; and (2) biases inherent in the sampling methods. The first of these sources arises because the design of a sampling programme is inappropriate to the question(s) being investigated (see pp. 44–47). The second is a source of error that is systematically implicated in all sampling programmes. All estimators are dependent on the methods used to obtain the sample data and will thus incorporate the biases of those methods. Bias may arise from many sources, including errors made by observers, loss of organisms from sample units, over- or under-estimation of the areas sampled, disturbance caused by sampling, avoidance of

34


the sampling device by organisms, and others. Our main interest in this section is not the cause of methodological biases, but the ways in which researchers have investigated the accuracy of their methods. Although many authors have apparently ignored questions of accuracy, several have emphasized the importance of methodological bias in determining the results of sampling programmes (Wiebe & Holland, 1968; McIntyre, 1971; Wiebe, 1971; Russell, Talbot, Anderson & Goldman, 1978; Resh, 1979; Southward & Barry, 1980; Sale, 1980; Sale & Douglas, 1981; Weinberg, 1981; Omori & Hamner, 1982; Fairweather & Underwood, 1983; Sale & Sharp, 1983; Kennelly & Underwood, 1984, 1985; Pot, Noakes, Ferguson & Coker, 1984). Practical tests of accuracy will be specific to each methodology, and will usually rely on attempts to assess absolute abundance by an independent method. This independent assessment must be accurate, otherwise one is only comparing two sampling methods, both with unknown biases (Kinzie & Snider, 1978). There have been a number of reports where accuracy has been thoroughly tested, usually by knowing a priori the absolute abundance of organisms in the sampled population or enumerating completely a population after sampling (Haury, 1973; Weinberg, 1981; Pihl & Rosenberg, 1982; Bell, Craik, Pollard & Russell, 1985; Andrew & Stocker, 1986). In many systems, such as plankton or bacterial communities, the accuracy of sampling and/or processing methods is extremely difficult, if not impossible, to assess because direct, independent enumeration of populations is impossible. In these, and many less problematic instances, the relative accuracy of different methods has been assessed by sampling the same populations by different methods and then comparing the values of the estimates obtained. Where estimates differ one method can only be said to be more accurate than another on the basis of intuition, auxiliary knowledge, and argument rather than on unequivocal evidence that it gave a truer account of the population. When methodologies have been compared, it has often been argued that the greatest estimate of abundance is the most accurate (Leatherwood, Gilbert & Chapman, 1978; Sale & Douglas, 1981; Sale & Sharp, 1983; Stretch, 1985; Gray & Bell, 1986). This argument is based on the assumption that you do not count what is not present, so over-estimation is extremely unlikely (Caughley, 1977). Exclusion or loss of organisms from samples is far more likely than inclusion. Some authors, however, have pointed out that boundary effects, where the inclusion of individuals in a sampling unit such as a quadrat is somewhat subjective, often lead to over-estimation (references in Downing & Anderson, 1985; Downing & Cyr, 1985). For sessile or slow-moving organisms, such boundary effects can be avoided by procedures such as including all ‘borderline’ individuals on two sides of a quadrat and excluding them along the other two sides. The problem is more difficult to circumvent when the organisms being counted move rapidly across sample unit boundaries and the observer has to make immediate decisions about whether an organism was inside or outside the unit at the instant when it was first sighted. Ideally, inclusions and exclusions of such marginal individuals should be equi-probable but inclusion seems more likely than exclusion (references in Downing & Anderson, 1985). For example, when visually censusing fish along belt transects, inclusion of individuals that are seen to swim into the transect is more likely than exclusion of fish that are observed leaving the transect. A number of recent studies have attempted either to quantify absolute accuracy or quantitatively to compare the accuracy of alternative methods. Jones (1974) showed that direct counts of bacteria using epifluorescence techniques varied dramatically when the bacteria were treated with different stains. Youngbluth (1982) came to a similar conclusion about the effect of the design of emergence traps on estimates of abundance of demersal zooplankton. Mackay, Cooling & Berrie (1984) found considerable disparities among five methods of estimating primary production in estuarine angiosperms. Gray & Bell (1986) reported that catches of vagile macrofauna from sea-grass meadows varied greatly with the method used (poison compared with beam trawl). There have been several comparisons of methods for assessing the


35

abundance of fish, either destructively or non-destructively, and all report major differences in the estimates returned by different methods (Russell et al., 1978; Brock, 1982; DeMartini & Roberts, 1982; Pot et al., 1984; Kimmell, 1985). By contrast, Bouchon (1981) found that quadrat and line transect methods gave similar estimates of coral cover and diversity. Nie & Vijverberg (1985) found no differences between catches of plankton from a Friedinger sampler and those from a Schindler sampler. Leatherwood et al. (1978) obtained similar, although very imprecise, estimates of numbers of dolphins using different methods of aerial survey. They noted that estimates varied with observers for all methods. Kennelly & Underwood (1984) found no difference between estimates of abundance of epilithic micro-organisms obtained in situ using an underwater microscope and from chips of substratum examined with superior microscopes in the laboratory. The underwater microscope, however, detected only about 25% of the micro-organisms that could be collected with a small suction sampler (Kennelly & Underwood, 1985). In a few instances, computer simulations have been used to model the sampling characteristics of different methods and consequently determine their biases and accuracy. Kinzie & Snider (1978) found that four visual survey techniques commonly used in field studies of corals, consistently gave inaccurate results when simulated with coral populations with varying characteristics. Wiebe & Holland (1968) and Wiebe (1971) found in simulations that estimates of the abundance of plankton were affected significantly by both the diameter of the plankton nets used and the distance over which the nets were towed. Furthermore, several of the more inaccurate estimates were more precise than most accurate estimates, illustrating that precision cannot be used to infer accuracy (e.g. Keast & Harker, 1977). Unless such simulations are constructed on a sound knowledge about the distribution and behaviour of the organisms in natural conditions and of the properties of the sampling device, extrapolations from computer to nature should be treated cautiously. Wiebe & Holland (1968) found good agreement between the precision of their simulated samples and that of several field studies and Wiebe (1972) tested the predictions of his earlier simulation study in the field. Despite the problems involved a number of researchers have attempted to assess absolute accuracy. Van Vleet & Williams (1980) tested the accuracy of a number of methods of sampling organic films on the sea surface by sampling, in the laboratory, surfaces that had been contaminated only with measured quantities of known compounds. Haury (1973) experimentally seeded a Longhurst-Hardy plankton recorder during operation with known quantities of real and model organisms in order to quantify biases arising from the variable residence time of plankton in the catch net leading the recorder. Weinberg (1981) tested the accuracy of seven methods of surveying coral communities by estimating abundances of corals by each method in an area that had previously been mapped in detail. The methods differed greatly in accuracy. An example of careful assessment of bias and accuracy of a sampling method has been presented by Pihl & Rosenberg (1982). By sampling known populations of vagile decapods and flatfish within fenced areas of a shallow sediment flat, Pihl & Rosenberg were able to compare the estimates of abundance obtained from the use of their drop trap with the known true values. They assumed that sampling fenced organisms was equivalent to sampling unfenced organisms. They attempted to address this problem by observing the responses of many individual organisms to the movements of the trap carriers and the operation of the trap. In most cases, the true assessment of accuracy will be very problematic and almost always extremely expensive in time and effort. In many studies, it may be unwarranted because knowledge of absolute abundance is not of primary interest (Caughley, 1977). Where comparisons are the aim of a study, estimates of relative abundance are often sufficient, although such comparisons assume equal sampling bias for all sites or conditions being compared. Unknown variations in the bias of a method with its use in different circumstances (e.g. habitats) is particularly problematic and difficult to assess. For example, artefacts of

36


visual censuses of fish may change with changes in the complexity of the habitat surveyed. Similarly, Kennelly & Underwood (1985) found their suction sampler to be reliable when sampling clean, hard substrata of low complexity but to be of little use on more complex substrata such as silted surfaces or articulated coralline algae. Although estimates that are known to be inaccurate may be useful when accurate ones are not obtainable, some knowledge of the accuracy of the sampling methods used is always desirable. PRECISION The usefulness of an estimate is dependent on its precision as well as its accuracy (Cochran, 1963; Caughley, 1977; Green, 1979). Unlike accuracy, precision can be assessed relatively easily from characteristics of the sample data. Consequently, the well-established statistical tools for assessing precision are not specific to methodologies and can be considered as general principles. Precision is a function of the variance of the sample estimate: precision increases as the variance of the estimate decreases (Cochran & Cox, 1957; Cochran, 1963; Elliott, 1977; Eberhardt, 1978a, b). Precision is related to the confidence we have in the estimate, such as mean abundance. An important distinction must be drawn here between variation among sample data and the variation of the sample estimate, by which precision is measured. Variation in the sample data is an estimate of the variation in the arrangement of organisms in the real world. The spatial arrangement of individuals is inherently variable in space and time. In most populations, organisms are not regularly or randomly distributed but aggregated (Southwood, 1966; Pielou, 1969; Elliott, 1977; Downing, 1979; Resh, 1979). As we will discuss later (see p. 70), the variation among sample data depends on the size of the sampling unit relative to the scale of aggregation. Variations in the numbers of organisms in sampling units represent real phenomena and, if the methods used are reasonably reliable, sample variance should be relatively invariate with changes in sample size (=number of replicates). Variation in the sample estimate is, by contrast, implicitly dependent on the size of the sample. Consider the earlier example of estimating the mean size of fish. If a large number of fish (say 200) were measured for each repeated sample, then the sample means are more likely to be similar than if each of the samples included only 20 fish. Precision is, therefore, a characteristic of the sampling procedure rather than a reflection of some characteristic of the population being sampled. The Standard Error (SE) of a sample mean is the estimated standard deviation of a population of sample means from samples of that size. It is estimated from the standard deviation (s) of the sample data by:

where n is the sample size. Thus, for any s, which will be determined by the spatial arrangement of organisms and the size of units used to sample them, SE will decrease as replication increases. It follows that precision is a function of the SE of a sample. Precision has, however, been measured in terms of several sample statistics. The variance/mean ratio has sometimes been used in this context in addition to its use as a measure of the spatial arrangement of organisms (see p. 71). The ‘coefficient of of the data is also an expression often used in discussions of precision (Wiebe & variation’ Holland, 1968; Wiebe, 1971; Elliott, 1977; Resh, 1979; Pringle, 1984). These measures are not indicators of precision but of the relative variability of the sample data, standardized for the magnitude of the sample mean (Snedecor & Cochran, 1980). In other studies, precision is appropriately measured by comparing the 95% confidence intervals of the sample mean to that mean (Resh, 1979) or, more commonly, by the ratio (Southwood, 1966; Downing, 1979; Pihl & Rosenberg, 1982; Pringle, 1984; Downing & Anderson,


37

1985; Downing & Cyr, 1985; Morin, 1985). Unfortunately, the ratio has also been referred to as the ‘coefficient of variation’ (Irish & Clarke, 1983; Pringle, 1984) and in several papers it was not clear or was used to calculate precision. whether Precision is inversely related to the values of the ratios used to measure it. Precision is great when the SE is small relative to the mean and precision decreases as the ratio increases. The precision required for an estimate should be set a priori by the researcher and will be largely determined by the questions being to be 0·5 whilst another may demand greater precision, for asked. One researcher may require The important point to note here, is that by setting desired precision, the researcher example can then design a sampling programme to achieve that precision as economically as possible. OPTIMIZATION AND PILOT STUDIES It is usually desirable to describe patterns with the maximal precision and resolution possible with the available resources. Almost all studies are constrained by logistic and economic considerations (Cochran, 1963; Holme & McIntyre, 1971; Lewis, 1976; Saila, Pikanowski & Vaughan, 1976; Green, 1979; Resh, 1979; Morin, 1985; Millard & Lettenmaier, 1986). Optimization of the design of sampling programmes is achieved by determining the most efficient allocation of resources—i.e., minimizing decreases in precision and/or resolution imposed by cost or by logistical constraints (Cochran, 1963; Saila et al., 1976; Scheaffer, Mendenhall & Ott, 1979; Irish & Clarke, 1984; Downing & Anderson, 1985; Downing & Cyr, 1985; Morin, 1985). The notion of optimizing the design of sampling methods is certainly not new (see references in Cochran, 1963; Cochran & Cox, 1957; Snedecor & Cochran, 1980). Statistical methods for choosing appropriate replication and cost-efficient allocation of resources have been available for several decades. In this section we shall review the use of optimization procedures for designing sampling programmes in marine studies. Apart from considerations, such as ensuring that sampling programmes are appropriate to the question(s) being asked, there are two main procedural questions that need to be addressed empirically if a sampling programme is to be efficiently designed: (1) how big should each sample unit be, and (2) how many replicates are needed? For all sampling designs, optimization procedures require estimates of variances and/ or means before the main programme has commenced. These estimates can be obtained from three sources: (1) pilot studies; (2) previous studies; and (3) published data. If pilot studies are not possible, or if the same organisms have been studied in the same system then one of the other two methods may be used to obtain estimates of mean abundance, spatial pattern and variance. Clearly, recent prior studies done in the same system are far more likely to give useful estimates than studies done in a different system. Numerous authors have emphasized that pilot studies are preferable to the other two sources of estimates (Cochran, 1963; Gray, 1971; Holme & McIntyre, 1971, Hulings & Gray, 1971; Downing, 1979; Green, 1979; Resh, 1979; Underwood, 1981; Downing & Anderson, 1985; Downing & Cyr, 1985). Green (1979, p. 31) made the point emphatically: “Those who skip this step because they do not have enough time, usually end up losing time.” We shall discuss the implementation of pilot studies and draw attention to some of the problems likely to be encountered. In the first instance (see below) we shall assume that sample units are placed randomly over an area to obtain a single estimate for a population—Simple Random Sampling. We shall then discuss the more complex Stratified and Multi-stage designs.

38


The size and number of sampling units The two aspects of sampling that most affect the precision of sample estimates are the size of the sampling units and the number of replicates collected (Cochran, 1963; Greig-Smith, 1964; Southwood, 1966; Snedecor & Cochran, 1980; Elliott, 1977; Green, 1979). The size and shape of the sampling units used have been repeatedly shown, both theoretically and empirically, to have a great impact on the precision of an estimate (Wiebe & Holland, 1968; Stoddart, 1969; Gray, 1971; McIntyre, 1971; Wiebe, 1971; Elliott, 1977; Resh, 1979; Sale & Douglas, 1981; Sale & Sharp, 1983; Pringle, 1984; Downing & Anderson, 1985 and references therein; Downing & Cyr, 1985; Morin, 1985). The choice of the size of the sample unit is fundamentally related to characteristics of the population being sampled such as spatial arrangement. This is particularly true when organisms are aggregated. Units that are smaller than or equal to the scale of aggregation will often give more variable estimates of density than those that are large with respect to the scale of aggregation of the organisms (Smith, 1938; Pechanec & Stewart, 1940; Wiebe & Holland, 1968; Wiebe, 1971; Elliott, 1977; Watling, Kinner & Maurer, 1978; Green, 1979; Helshe & Ritchey, 1984). Estimates of abundance will be most variable when the size of the sample unit is approximately equal to the average distance between aggregations (Southwood, 1966; Elliott, 1977) (see p. 71). This occurs because these and smaller units are likely either to miss aggregations and so contain few or no organisms or to include an aggregation, and contain many organisms. Larger units will be likely to include part of at least one aggregation and so very small or zero counts are unlikely to occur. Thus, estimates of average abundance obtained from large sampling units will be less affected by the patterns in the spatial arrangement of the organisms. Consequently, for a given sample size, the precision of a sample estimate is likely to increase with increasing size of sampling units (Elliott, 1977; Downing, 1979; Resh, 1979). The rate of increase in precision with increasing size of sample unit will initially be great but will rapidly decline as the size of the unit exceeds the average distance between aggregations in the population. Resh (1979), however, cautioned that this trend is dependent on sample units being located wholly within habitats or other physical clines. If sampling units correspond to the scale of such factors as natural micro-habitat units, great variability may result from factors not simply related to the spatial pattern of the organisms. The shapes of sampling units may also affect the precision of estimates of abundance. Where boundary effects are important, the amount of boundary relative to the area or volume of the sample unit should be minimized (references in Downing & Anderson, 1985). Boundary effects are a function of the shape as well as the size of the sampling unit and may be particularly important where the shape and size of sampling units correspond to those of aggregations of organisms or topographic features to which organisms respond (Resh, 1979). Furthermore, the minimum linear dimension of a unit may be more critical than the other linear dimensions, area or volume. For example, Caughley (1977) and Sale & Sharp (1983) have demonstrated the importance of the widths of transects when using transect survey techniques for visual censuses. Hargrave & Burns (1979 and references therein) found that both diameter and aspect (diameter, depth of sampler) of sediment traps significantly affected their performance and that the effects varied with the turbulence of the water in which they were used. Four characteristics of the size and shape of sample units can be considered: length, area, volume, and time. The relative importance of each of these measurements to the precision and accuracy of the units will depend on the particular application. In general it would seem expedient to compare sampling units of a number of sizes at the beginning of a study (Southwood, 1966). Where there is some prior knowledge of the spatial arrangement of individuals in the population (see p. 70), the smallest unit considered should, if feasible, be larger than the average spacing among aggregations. Where the spatial arrangement of the organisms is unknown or unimportant, then the smallest unit should be at least one order of magnitude larger than the size of the largest organism being


39

counted; Green (1979) suggested at least 20 times larger. The largest unit tested should be as large as logistic and cost constraints will allow, with the provisos that it is not so large that it approximates or exceeds the physico-environmental scales of interest or causes overlap among replicates when sampling a limited area. It is difficult to suggest how many sizes of sampling unit should be compared, but three should be the minimum and more than six or seven is likely to be too costly for a pilot study. In the simplest case (simple random sampling), an equal number of (at least three) replicate sampling units of each size should be collected randomly within the area to be studied. The randomization of all units within one area is intended to minimize the chance that systematic biases caused by variations within the area will affect only the replicates of a unit of particular size. For example, if all units of one size are placed in one area and all units of another size are placed in another area, then the effects of unit size will be confounded with pre-existing differences between the areas (Hurlbert, 1984). The analysis of pilot data should consider two aspects of the sample units tested—their relative accuracy and their relative precision. Relative accuracy is assessed by comparing the estimates of mean density (number of organisms/size of unit) obtained from the units of various sizes. Significant differences among the standardized means indicate differences in the relative accuracies of at least some of the unit sizes tested, although it will not necessarily be clear which unit size is the most accurate unless the true density of the population is known. Differences among means do not necessarily indicate differences in absolute accuracy; it is possible that two significantly different means are equally inaccurate, one being larger than the true population mean and the other being smaller by about the same amount. Decisions about which size of sample unit is the most accurate can only be made in the same way as decisions about the accuracy of different sampling methods (see pp. 48–50). Sale & Sharp (1983) studied the effects of changing the width of transects used in visual censuses of coral-reef fish. They found significant linear relationships between width of transect and estimated abundances of a number of species and advocated the use of such relationships to predict the expected true density of fish. Using a separate method (which they argued was more accurate than censusing along transects) they independently estimated the abundance of one of the species and compared this estimate with the ‘true’ value predicted from their regression of abundance on width of transect. There was no significant difference between the two values, but nor was there a difference between the independent estimate and the estimate obtained from the narrowest transect used for that species. can be calculated A number of avenues can be followed to compare precision. In the simplest case, for units of each size and the values compared. The unit with the smallest value will give the most precise estimates. When desired precision (p) is set,

By re-arrangement of terms, necessary replication (n) can be calculated from:

In their detailed computer simulations, Wiebe & Holland (1968) and Wiebe (1971) considered the implications of sampling patchily distributed plankton populations with nets of various sizes towed over a number of distances. They concluded that, in general, nets of larger diameter gave estimates of abundance that were both more accurate and more precise than did smaller nets. They also found that increasing the length of the tow significantly increased both accuracy and precision for all net sizes. In both papers,

40


however, it was emphasized that both precision and accuracy increased with the size of net and length of tow at a decreasing rate. Thus, the most efficient sample units are not always the largest. For example, precision increased by about 26% with an increase in net diameter from 25 to 100 cm, but a further fourfold increase in the diameter of nets caused only an additional 8·2% improvement in precision. Pihl & Rosenberg (1982) took increasing After arbitrarily setting their desired precision at numbers of samples with a sampling unit of set size until they attained the required precision for estimates of abundance for the four species they studied (two fish and two crustaceans). Although they found that only 10, 15, 20, and 25 samples would be required for each of the four species, respectively, they conservatively chose to use 30 replicates in their main study. This study represents an example of the empirical rather than mathematical determination of appropriate sample size. With more complex but more potentially informative designs, a number of estimates of precision can be obtained for units of all sizes and the mean precisions compared using analysis of variance with the replicate measures of precision as data. This is possible where the pilot study is repeated at a number of similar sites or over a short time or when sufficient replicates of each sized unit are collected to allow subdivision of the data into ‘b’ subsets of n/b replicates for each size of unit (Scheffe, 1959). Measures of precision are calculated for each data set for each size of unit and the values analysed. Significant differences among the means of estimates of precision indicate better average precision for one or more unit sizes than for others. If the costs of using different sized units are similar (e.g. it might be only marginally more time consuming to count urchins in a 1 m2 quadrat than in a 0·25 m2 quadrat), then the analysis could (greatest precision) is the best choice if costs are stop here. The size of unit with the smallest value for not considered because fewer units of that size than of the other sizes will be required for any chosen level of precision. When differences in the costs among units are negligible, the cost of sampling is proportional only to sample size. In such a case, the choice of the size of the unit to be used is greatly simplified and the unit with the smallest CV should be used since that unit will yield greatest precision for any sample size. Green (1979) suggests that when precision is the same for units of all sizes then larger rather than smaller units should be favoured. It is unlikely, however, that all units will cost the same to use and hence efficiency (cost of attaining required precision) will be likely to favour the choice of one unit over others. For any size of sampling unit, precision will increase with sample size because the standard error and confidence intervals decrease with increasing replication. Increased precision is offset by the increased cost (time, effort, money) of obtaining and processing large samples. As with increases in the size of sampling units, the rate of increase in precision with increasing sample size is initially great but declines as sample size becomes large (Cochran & Cox, 1957; Cochran, 1963; Green, 1979; Scheaffer, Mendenhall & Ott, 1979). The declining rate of increase in precision suggests a situation of diminishing return for effort. It is usually economically desirable to use the smallest possible size of unit, but the cost of obtaining sufficient of them to obtain a required precision must also be considered. The total cost of sampling will be the cost per unit multiplied by the number of units required, plus any overhead costs, such as getting to the study site. To , fewer larger units will usually be required than smaller obtain a given precision, expressed as, say, units. In studies from several fields, it has been found that it is generally more economical to use many smaller sample units than a few larger ones (Gerard & Berthet, 1971; Elliott, 1977; Downing, 1979; Pringle, 1984; Downing & Anderson, 1985; Downing & Cyr, 1985; Morin, 1985). This result is not, however, universal and several authors have recommended the use of larger rather than smaller sampling units (Dennison & Hay, 1967; Wiebe & Holland, 1968; Gray, 1971; Wiebe, 1971; Kenchington, 1978). The relative economy of using sample units of various sizes should not be the sole arbiter of optimality. Particular biological, behavioural, physical or other factors will also be important in choosing units of appropriate, optimum size. For example, observer fatigue may increase bias and decrease precision when


41

large numbers of organisms are being counted (Caughley, 1977; Green, 1979; Kirchman, Sigda, Kapuscinski & Mitchell, 1982). It is also likely that small quadrats will be searched more thoroughly than large quadrats. Wiebe (1971) discusses, but does not analyse, the possible effects of costs and logistics, such as ship time and sorting time, on the choice of optimum size sampling units for studies of plankton. He also re-emphasizes suggestions by other authors that, in such studies, the choice of larger rather than smaller nets is likely to be an important means of decreasing avoidance of nets by real zooplankton in the field, an important source of inaccuracy in estimates of abundance of plankton. Such considerations will be specific to different fields. Where both cost and precision vary with unit size it has been repeatedly shown that efficiency is not always simply a function of the size of sampling units (Downing, 1979; Pringle, 1984; Downing & Anderson, 1985; Downing & Cyr, 1985; Morin, 1985). The estimate with the smallest variation will often be from the largest and probably the most costly unit. The efficiency (cost of obtaining a desired precision) of units of various sizes can be compared by cost-benefit analyses (Cochran, 1963; Saila et al., 1976; Snedecor & Cochran, 1980; Underwood, 1981; Irish & Clarke, 1984). In this case the cost-benefit analysis is the determination of the cost of using sufficient units of each size to estimate abundance with the desired precision. From the equation on page 55, number, n, of units of size u necessary for desired precision, p, is:

The toptal costs, Ct, of sampling with units of size u will be the sum of the overhead costs, co, and nu times the cost per unit, cu:

Evaluation of this expression for units of each size allows a direct comparison of their efficiency and, hence, allows the choice of optimum size of unit for the population being sampled. If overhead costs are constant and independent of size of unit, then only the last term in the above equation need be calculated. Note also that if a general cost function is desired, cost per unit can be expressed as a function of one or or size of unit (Downing & Anderson, 1985; Downing & Cyr, 1985; Morin, more variables, such as 1985). In most fields of ecology the relationships among the size of sampling units, replication, and efficiency have been examined only infrequently. Pringle (1984) compared the sampling efficiency (=cost per unit precision) of six quadrat sizes (0·25 m2−4·0 m2) used for estimating the biomass of a seaweed and found that although replication required for given precision varied linearly with size of sample units (quadrats), the relationship between total sampling time and size of quadrat was curvilinear. The relationship between cost and benefit (in terms of precision) was not a simple linear function of the size and number of quadrats used. Intermediate sized quadrats (1·0 m2 and 1.56 m2) were least efficient whilst the smallest quadrat tested was the most efficient. This means that it is more economical to take a larger number of the smallest units than a smaller number of larger units to estimate biomass with any given level of precision. Several Canadian scientists have recently reported similar investigations of optimal numbers and sizes of sampling units for the study of benthic organisms in freshwater lakes and streams. Downing (1979) and Morin (1985) analysed published data for a number of benthic animals from lakes and large rivers, and from streams, respectively. Downing & Anderson (1985) analysed original and published data for estimated

42


biomass of macrophytes; Downing & Cyr (1985) did likewise for epiphytic invertebrates. These authors derived empirical relationships between variance, mean abundance and the size of sample units based on the principles of Taylor’s power law (see p. 75) and the expected change in variance with changes in unit size. The relationships were developed by multiple regression of the logarithm of variance on the logarithms of mean density and sample unit size. In all cases the relationships were highly significant and accounted for between 88 and 95% of the variation in the large sets of data fitted. These empirically derived relationships and the terms re-arranged to give an were then substituted into the formula for precision, measured as expression for n (sample size) in terms of mean abundance, size of sample unit and desired precision. The conclusions from these syntheses of large bodies of published data were relatively consistent across all of the studies. In all cases, the sample size required to achieve a given level of precision was a tight function of population density and the size of the sample units. The sample size needed to achieve the desired precision declined as the density of the population being sampled increased and as larger sample units were used. In both cases, the rate of change in replication necessary for desired precision also declined. W.D. Taylor (1980) has criticized the approach of Downing (1979) arguing that conclusions taken from the synthesis of such a variety of species, systems, and methods are of doubtful use for studies of single species. For further comment see also Downing (1980, 1981) and L.R.Taylor (1980). When functions describing the costs of obtaining and processing samples were applied to the precision formulae in the above four studies the results diverged. Downing (1979) and Downing & Anderson (1985) concluded that it was always more efficient to use many small units than fewer large ones. In the first study, the cost was estimated by the total area of sediment that would have to be collected and sorted for the required replication of units of a given size. In the second study, costs per sample were measured when data were being collected for comparisons of performance of different sized sample units (quadrats). Downing & Cyr (1985) measured the costs of collecting and processing samples using quadrats of five different sizes and concluded that a quadrat of intermediate size (500 cm2) was the most efficient. Morin (1985) concluded that when population density was great, small sample units were most efficient but that larger units were more cost effective for sampling populations of low density. These studies illustrate the importance of replication and the size of sample units to sampling efficiency and precision. The relationships identified by Downing (1979), Downing & Anderson (1985), Downing & Cyr (1985), and Morin (1985) are useful indicators for the calculation of optimum sample size but encompass data from a wide range of taxa, methods, and locations. As these authors stress, pilot studies are the best means of fine tuning methods, size of sample units, and sample size for a particular study. Stratified sampling In the preceding section, we discussed optimization procedures on the assumption that one sample was to be collected to give a single estimate for a population. Estimates from simple random sampling are most likely to be reliable when the characteristics (e.g. aggregation, density) of the population are relatively consistent throughout the area sampled. Often the population or area being sampled is obviously not homogeneous and such characteristics as density and patterns of aggregation vary from place to place. In such cases, overall abundance or average population density can be estimated more precisely if a number of subsections of the population are sampled separately. In other words, various strata of the environment or population are identified and the organisms sampled in a way that is appropriate for each stratum. This is referred to as stratified sampling (Cochran, 1963; Scheaffer et al., 1979; Snedecor & Cochran, 1980). As with singlestage sampling, the objective is the estimation of the total number of organisms in the whole area (e.g. the


43

number of scallops in a bay). Precise estimates of parameters are obtained from the weighted combination of the estimates from all strata. There are several ways to design a stratified sampling programme. In the simplest case (Stratified Simple Random Sampling) little is known about the population and strata are identified on the basis of some auxiliary variable(s) (e.g. depth, type of sediment or habitat). Equal numbers of sample units are then collected randomly from each stratum. Alternatively, the total number of replicates feasible for the programme is allocated to the strata in proportion to their areas (Proportional Stratified Sampling). In both procedures, the researcher need know nothing about the population prior to the main programme, but requires a map of the area to be sampled. In practice, optimal sampling procedures may differ among strata. The most precise estimates of the total population will be obtained when sampling is optimal in all strata. To obtain approximately similar precision for all strata, sampling effort is weighted among strata in proportion to the variance of the data obtained from each stratum. Strata with very variable population densities are more intensively sampled than those with relatively homogeneous data. Here, the allocation of samples is proportional to characteristics of the subpopulations, rather than the areas of the strata. Sample allocation can be optimized according to three sets of criteria: (1) minimize the variance of the estimate for fixed total cost; (2) minimize the total cost of obtaining an estimate with a given variance, where costs per unit are equal for all strata; and (3) minimize the total cost of obtaining an estimate with a given variance, where costs per unit vary among strata. Formulae and full discussions of these may be found in Cochran (1963), Scheaffer et al. (1979), and Snedecor & Cochran (1980). To obtain estimates of identical precision from each stratum, each stratum should be treated as a single sample and the replication needed to give desired precision calculated as discussed on pages 52–58. In this case (Stratified Sampling with Optimal Allocation), the total number of replicates is the sum of the numbers of replicates from each stratum. If this total effort is beyond the limits of the programme, then one of the proportional allocation procedures must be used or the level of precision relaxed. It may also be desirable to optimize sampling methods and/or the size of sampling units for each stratum separately, provided that this does not introduce biases that differ among strata. For optimal allocation or proportional allocation based on variance, estimates of variation are required prior to the main study. These are best obtained from pilot studies in which a small sample is recorded from each stratum. Cochran (1963) has pointed out that stratification with sample allocation based only on auxiliary variables often fails to give estimates that are more precise than those from a Simple Random Sample design. Stratified Simple Random Sampling is the stratified design most commonly used in marine ecological studies. It is common, for example, to stratify sampling by habitat. In most cases, however, this design is used to obtain data for comparison of abundances in different habitats rather than to estimate the total number of organisms in all habitats sampled. Other stratified sampling designs, better suited to estimating total numbers, have apparently not been common in marine studies. Cuff & Coleman (1979) used a Proportional Stratified Sampling design to estimate the abundances of molluscs, polychaetes and crustaceans in Western Port Bay, Australia. They allocated samples according to the areas of the strata and used the data to calculate the expected variance of the estimated populations of the three taxa for three

44


designs of stratified sampling and for Simple Random Sampling. They found that the precision of the estimates was not improved by stratifying the sampling and concluded that stratification was generally no better than Simple Random Sampling. Green (1980) critically assessed Cuff & Coleman’s study and argued that they were not justified in extending their results, from only one study in one bay, to general recommendations for the design of future sampling programmes elsewhere. The failure of stratification to increase precision may well have arisen from the way in which Cuff & Coleman divided the bay into strata. More informed allocation may well have produced a more precise estimate than Simple Random Sampling (Cochran, 1963). Furthermore, Cuff & Coleman pooled counts of species and analysed the data by phyla. Green (1980) noted that this level of taxonomic resolution may be an inappropriate basis for stratification because it is unlikely that all species in a phylum will respond to environmental variables in the same way and have similar patterns of abundance. Heisig & Hoenig (1986) described a method of optimizing sampling design for estimating secondary productivity from size-frequency data. The allocation of effort among different times of the year (=strata) was optimized according to the expected abundance of organisms at each time. Multi-stage sampling Many studies are more concerned with examining patterns in abundance than with estimating the total number of organisms in an area. It may be important to assess whether a population can be considered homogeneous at various spatial scales, or whether differences in population parameters occur among, for example, different habitats. These situations have been referred to as mensurative experiments (Hurlbert, 1984). The simplest design for such comparisons is a stratified Simple Random Sampling design. In most instances, stratified sampling with proportional or optimal allocation of replicates would be more efficient for obtaining a single estimate for a population. Because of the unequal sample sizes, however, such designs are not as well suited to statistical comparisons among strata as Stratified Simple Random Sampling. Stratified designs can involve two or more hierarchically arranged levels of replication. Cuff & Coleman (1979), for example, collected replicate grab samples at each of several stations within each stratum of a bay. When replication is balanced among strata and at each level within strata, these hierarchical, multi-stage designs are perhaps the most powerful designs for comparing estimates. In general this class of designs is concerned with the estimation of differences among means for a single variable. The analysis of this type of data is often best accomplished through techniques of analysis of variance. No other analyses can formally test for both hierarchical and interactive effects within the same design. Formal introductions to these techniques of analysis may be gained from standard texts such as Winer (1971), Snedecor & Cochran (1980), and Sokal & Rohlf (1981). Underwood (1981) gave an introduction to techniques of analysis of variance and reviewed their use in marine ecology. Multi-stage designs allow the researcher to estimate variability at different scales within strata (such as habitats) and provide a stronger basis for generalization. For example, in the investigation of patchiness in the abundance of plankton, replicate samples should be taken in close proximity at sites separated by several kilometres, with several sites being sampled at stations separated by hundreds of kilometres. Another example might be the comparison of numbers of fish in kelp forests with the numbers in sea-grass beds. In order to draw general conclusions it would be necessary to sample each habitat at more than one location (separated by perhaps several kilometres) and to sample more than one site within each habitat at each location. The problem with complex designs such as these, is how to allocate effort efficiently among the various levels of the programme—how many locations, sites, and replicates give optimum precision to the estimates of abundances of fish in the two habitats.


45

Methods are readily available for such multi-stage cost-benefit analyses (see Cochran, 1963; Snedecor & Cochran, 1980; Sokal & Rohlf, 1981). Saila et al. (1976) and Underwood (1981) discussed the procedures and advantages of multi-stage cost-benefit analysis. Based on the hierarchical (or nested) analysis of variance, this form of cost-benefit analysis uses estimates of the proportion of variation explained by each level of sampling (such as locations, sites, and replicates), and the expected costs of replication at each level, to allocate sampling effort among levels so that overall variation of estimates of means is minimized. Bernstein & Zalinski (1983) have recommended the use of these procedures for planning environmental monitoring studies. Perhaps the most frequent application of multi-stage sampling is in studies of micro-organisms where hierarchical subsampling is employed to decrease the effort and expense of counting the large numbers of organisms collected in each field sample. Subsampling is routinely used in studies of plankton, benthic infauna, and bacteria. Well-developed and tested methods of subsampling and accurately splitting the samples into subunits exist, but there has been little attention to optimization. Subsampling has been done mostly by traditional rather than tested practices (Alden, Dahiya & Young, 1982; Kirchman et al., 1982; Montagna, 1982; Irish & Clarke, 1984). Kirchman et al. (1982) used cost-benefit analyses to determine the optimum number of subsamples, aliquots (=filters) and fields of view to be taken when using hierarchical Subsampling to estimate bacterial abundance in bodies of water (note that they refer to the field samples as subsamples of the water body). Irish & Clarke (1984) used cost-benefit procedures to determine optimal designs for sampling, and Subsampling, a tarn to estimate chlorophyll a concentrations and abundance of phytoplankton. Kirchman et al. (1982) found that by optimizing the design of the Subsampling procedures, “accuracy” (actually precision) of the estimates of abundance for the water bodies sampled increased by 20–50% compared with the traditionally used designs with only marginal increases in costs. They emphasize that most customary sampling designs fail to estimate variation at all levels (e.g. among fields or among filters) of the Subsampling procedure and, in the worst cases, cannot provide any statement of precision for the estimate of the number of bacteria in the sample. Montagna (1982) also found that the optimum design for Subsampling sediment samples to count bacteria involved replication of subsamples, filters, and fields of view. He also pointed out that the customary practice of counting 15–20 fields of view on only one filter per sample was inadequate. Venrick (1971) has also stressed that it is important to estimate the amount of variation at each level of Subsampling procedures. She gave a full explanation of the components of variation at each level and used a computer simulation to examine the relative effects of different subsample fractions and replication at each level on the estimation of the variation in the estimate of abundance at the highest level. She did not, however, deal with cost-benefit analyses. In other fields, spatial and/or temporal variation have only infrequently been investigated usefully by explicitly incorporating spatial scale into the design of sampling programmes. Replicates within levels (representing different scales) of hierarchical sampling programmes have often been combined with the result that variation at those scales could not be estimated (but see Lee & McAlice, 1979; Schwiegert & Sibert, 1983; Hankin, 1984; Kennelly & Underwood, 1984, 1985; Schwiegert, Haegele & Stocker, 1985; Andrew, 1986). Schwiegert & Sibert (1983) examined optimum designs for sampling Pacific herring from commercial catches to estimate the age structure and abundance of the fished population. The optimal designs varied among age classes of fish and among the different fishing ports sampled. In some instances, the analyses indicated that up to 100 fish should have been taken from several subsamples of each load that entered a port. In other cases, replication of subsamples was unnecessary and the optimal design was a two-stage

46


design (a number of fish taken from several replicate loads). For further discussion, see Smith (1984) and Schwiegert & Sibert (1984). Schwiegert et al. (1985) compared the precision of estimates of the abundance of herring spawn derived from several two-stage designs and a simple random sampling design. They found that carefully designed two-stage sampling gave more precise estimates than Simple Random Sampling but that other two-stage designs were no better than Simple Random Sampling. Hankin (1984) also compared a number of methods of allocating samples among levels of a two-stage sampling design implemented to estimate the abundance of fish in freshwater streams. He found that the most efficient design was one in which primary units (sections of stream) were variable in size and corresponded to discrete habitat units, such as riffles or pools. Random secondary units were then taken within each of these primary units. Hankin’s study essentially represented an empirical demonstration of the advantages of carefully designed stratified sampling programmes for estimating the total abundance of organisms in a heterogeneous habitat. Saila, Pikanowski & Vaughan (1976) used data from a prior survey in single and multi-stage cost-benefit analyses to estimate optimal designs for sampling several species of benthic invertebrates in the New York Bight. Kennelly & Underwood (1984, 1985) used pilot studies and cost-benefit analyses to derive the optimal replication at three spatial scales for sampling micro-organisms in a temperate sub-littoral kelp forest. All of the above studies have used the minimization of sample variance, and maximization of precision of estimated means, as the criterion of ‘benefit’ in cost-benefit analyses. Millard & Lettenmaier (1986) have discussed the use of multi-stage cost-benefit analyses based on balancing statistical power against cost rather than simply considering cost-precision relationships. They argue that in many situations optimal design based on the precision of means is inadequate since no account is taken of the likelihood that a hypothesized effect will be detected. Several authors have emphasized that patterns and processes can be better understood if estimates of variability at various scales are explicitly included in the design of sampling programmes. Caffey (1985) is a good example of such a study. The shortage of studies that have done this perhaps reflects attitudes of research workers to variability and its manifestation at various scales of investigation. As with more simple designs, pilot studies are required for optimization procedures for stratified and multi-stage sampling. Ideally, different sized units and methods should be examined in a number of the habitats, sites, etc., involved in these designs, although it may not always be possible to do this at all levels of the intended programme. Tests done under the extremes of density, habitat complexity or other independent variables may suffice. In pilot studies for multi-stage sampling it is important that each level of replication in the proposed programme be addressed by replication at corresponding levels in the pilot study. Note that cost-benefit analyses in this context can only be applied to hierarchically arranged factors. Replication at higher level orthogonal factors, when not fixed implicitly by the objectives of the study, can only be designated after estimates of replication at all lower levels have been calculated (Underwood, 1981). For example, in the earlier example of estimating the abundance of fish in kelp forests and in sea-grass beds, cost-benefit analyses would allow the determination of optimal numbers of replicate transects and sites in each habitat at each location. The number of locations to be sampled can then be calculated by dividing the total resources (time, money, etc.) available by the amount of resources required to sample the optimum number of sites and transects in each habitat. When the number of treatments in orthogonal or higher level factors is predetermined by the objectives of the programme, the total affordable cost is divided equally among the orthogonal treatments. The replication of nested factors within those treatments is then determined by cost-benefit analysis based on the sub-divided cost limits (Schwiegert et al., 1985).


47

POWER ANALYSIS Most sampling programmes are suboptimal because of logistic and/or cost constraints. Procedures such as cost-benefit analysis will ensure that the best use is made of limited resources, but the question remains whether the design is good enough to do the job asked of it. Recognition of this problem makes it desirable to know, a priori, how large a difference among means could be detected or, a posteriori, how sensitive a test was. Power analysis makes such statements possible. Statistical power is defined as the probability that a test will lead to the correct rejection of the null hypothesis; that is, that the null will be rejected when the alternative is true (Sokal & Rohlf, 1981). Power is, the probability of falsely retaining conceptually and probabilistically, the complement of Type II error (Yamane, 1967; Winer, 1971; Cohen, 1977; Sokal & a null hypothesis, and accordingly is expressed as Rohlf, 1981). A clear and extensive treatment of power analysis is given by Cohen (1977) who has provided tests, F tests, analysis of tables for the determination of power for correlation coefficients, t tests, variance for fixed effects and their interactions (for a limited number of designs) and analysis of covariance. A more statistically demanding treatment may be found in Winer (1971), who also considered the power of tests involving random factors in analysis of variance. Recent discussions in the ecological literature include Underwood (1981), Bernstein & Zalinski (1983), Toft & Shea (1983), and Rotenberry & Wiens (1985). Cohen (1977) has identified four parameters of statistical inference useful in power analysis : (1) power, (3) sample size (n), and (4) effect size. Knowledge of any three will allow the (2) significance level determination of the fourth (Cohen, 1977). The relationship between these parameters provides the research worker with the opportunity to exercise greater control over the design and analysis of sampling programmes and allows better and more careful interpretation of results (Bernstein & Zalinski, 1983; Toft & Shea, 1983). Power analysis is the analysis of the sensitivity of tests of null hypotheses against specified, quantified alternatives. It may be used either in the design phase of a programme or as a post hoc test of the adequacy of a design (Winer, 1971; Cohen, 1977; Sokal & Rohlf, 1981; Underwood, 1981; Bernstein & Zalinski, 1983; Toft & Shea, 1983; Rotenberry & Wiens, 1985; Millard & Lettenmaier, 1986). For example, the researcher may wish to know what sample size is required to detect an effect of given size subject to chosen probabilities of both Type I and Type II error. Alternatively, where a test failed to reject the null hypothesis, the sensitivity of the test can be examined: was the design too weak to detect anything but very large differences (Toft & Shea, 1983)? Despite this great potential, power analysis has remained a much under-utilized technique in ecological research (Toft & Shea, 1983). Several reasons for this lack of use have been suggested. Toft & Shea (1983) have argued that it is a reflection of our pre-occupation with Type I error and also a lack of awareness of Type II error and its importance. The reluctance (or inability) of ecologists to specify exact alternative hypotheses has been cited by Underwood (1981) and Rotenberry & Wiens (1985) as another possible reason for the lack of use of power analysis. By convention, we accept a 5% chance of committing Type I error without any real idea of the probability of Type II error. It is usually argued that this is acceptable because it is better to know the probability that an effect will be incorrectly claimed than it is to know the probability of not demonstrating a real effect. This is not necessarily so, particularly when strong assertions are made from the lack of a statistically significant result—when a null hypothesis is retained (Toft & Shea, 1983). These authors have argued that where a positive conclusion is drawn from a ‘negative’ (non-significant) statistical result, it should be exposed to the same restrictive standards as a positive conclusion drawn from a ‘positive’ (significant) statistical result (see also Rotenberry & Wiens, 1985).

48


In many situations, Type II error may have ramifications at least as important as Type I error, yet Type II error is rarely considered. As an ecological example, consider the situation in which research workers wish to assess the effects of commercial fishing on the numbers of adult fish. Areas that are regularly fished are sampled and mean abundance of the fished species in these areas is compared with that in areas that have been protected from fishing. A Type I error (asserting that fishing has depleted stocks when, in fact, it has not) is to be guarded against because such a decision would subject the fishermen to the imposition of strict quotas and the under-utilization of the resource. Further consideration, however, reveals that a Type II error is potentially at least as dangerous. The incorrect assertion that the stock is being fished at or below its sustainable yield (that is, existing fishing has no significant biological effect on stock size) could result in the total collapse of the fishery with consequences more severe than the imposition of quotas. Knowledge of the power of tests is particularly important in the description of pattern. Experiments test hypotheses concerning processes that might explain patterns. The pattern, or observation is taken for granted. In contrast, when hypotheses about patterns are tested, the observation itself is being questioned. The test is not concerned with why there are more gastropods here than over there, but rather whether it is true that there are more here than there. Incorrectly claiming no difference in abundance between the two areas will mean that the ecological processes influencing the density of gastropods will be assumed to be the same in both places. This could seriously delay understanding the processes that influence the number of gastropods in the two areas. of falsely In such cases, knowledge of is perhaps more important than knowing the probability rejecting the null hypothesis. Certainly, serious consideration should be given to the relative weighting of Type I and Type II errors. Setting power at a given level makes an explicit statement about the relative costs of Type I and Type II error (Winer, 1971; Cohen, 1977; Underwood, 1981; Toft & Shea, 1983). Winer (1971) has noted that the conventions of using 0·05 and 0·01 as acceptable probabilities of Type I error have little scientific or logical basis. Greater consideration of the relative costs of Type I and II error may lead to the de-sanctification of 0·05 and 0·01 as inviolate standards of Type I error. is not a The relationship between power and the more familiar probability of making a Type I error power will increase with simple one. If the size of an effect is held constant then, for a fixed value of increasing sample size. Conversely, if power and effect are fixed, then α decreases as sample size increases. That is, Type I error is less likely with larger samples than with smaller samples when power is set to a desired level. If sample size is fixed, as is often the case because of cost constraints, α and power will be directly related. More stringent requirements for power will be accompanied by decreased probability of power and effect size Type II error but increased probability of Type I error. The relations among for a two sample case are shown in Figure 1. Power analysis for planning If power analysis is to be used in the design stage of a sampling programme, then several alternative and n. The approaches may be taken. One is to determine the power of a test given effect size, are familiar and directly determination of appropriate sample size and significance level for Type I error under the research worker’s control. The determination of an appropriate effect size is much less clear, and great caution should be exercised when deciding the size of effect that is considered important (Cohen, 1977; Underwood, 1981; Rotenberry & Wiens, 1985). Effect size is defined as the standard deviation of means divided by a common standard deviation (Cohen, 1977). Where only two means are compared, effect size is the difference between the two means, expressed as a proportion of one of them. Where more than two means are being compared, effect size has no immediately intuitive meaning because a significant


49

result can be the product of differences among one or several of the means. Furthermore, the way in which similar means may be grouped and the number of groups may affect the interpretation of effect size and the power of the test. Winer (1971) and Cohen (1977) have provided various procedures by which an appropriate effect size can be calculated with reference to existing information. An appropriate effect size may be determined from previous studies in the system. Pilot studies should provide an estimate of error variance which can be used to calculate the effect size that a future sampling programme would be expected to detect. In such a case, after setting power, the only unknown becomes the numerator in the effect size ratio. Sweatman (1985) used the magnitude of differences among means that were significantly different in a similar study as the effect size criterion for calculating the power of non-significant results. In addition, Cohen has provided operational definitions of small, medium, and large effect sizes. He stressed, however, that these are relative measures, the specific value of which will be unique to the investigation. What constitutes a small effect in one system might be of overwhelming biological importance in another (Cohen, 1977; Underwood, 1981). It should be remembered that power analysis is a statistical procedure. Even if a sampling programme has great power, say 0·95 for all tests, it will be of little use if it is set within an unrealistic biological context. Millard & Lettenmaier (1986) have discussed the use of power analysis in designing environmental monitoring programmes. They advocate the a priori statement of an effect size considered to be important, possibly by legislation, and the use of cost-benefit analyses to design programmes that optimize the power of detecting an effect of that size. Unfortunately, they only consider the power of detecting main effects in complex designs. It is often the case that the interaction of these main effects (e.g. Event status×Time) are of greater interest in detecting environmental change than the main effects alone (Green, 1979; Bernstein & Zalinski, 1983, but see Hurlbert, 1984). Rotenberry & Wiens (1985) have argued that the suitability of a design be determined, not from the calculation of power for a given effect size, and n, but from the calculation of detectable effect size from and n. Decisions about the adequacy of the design would then be made after consideration of known a priori, the suitability of that effect size to the biological hypothesis being tested. By assigning subjectivity is shifted from the critical, a priori estimation of an unknown effect size, to the evaluation of suitability of a known one. That is, rather than stating an explicit, numeric alternative hypothesis, the relative importance of Type I and Type II error are stated and the worker decides upon the suitability of a calculated effect size that is expected to be detectable. Post hoc power analysis The second major use of power analysis is as a post hoc procedure to determine the sensitivity of an already completed test. The analysis here is essentially dependent on decisions already made by the research worker when the sampling programme was designed. Consequently, post hoc power analysis can only indicate the probability of erroneous retention of the null hypothesis; i.e. it can only be meaningfully applied to nonsignificant results. The only three published instances of post-hoc power analysis that we encountered (Sweatman, 1985; Andrew, 1986; Doherty & Sale, 1986) all sought to explain non-significant results. In both Andrew (1986) and Doherty & Sale (1986) the analyses revealed important deficiencies in design. A non-significant F ratio had 70–77% chance of being in error. Power analysis should not be used to justify the post hoc elevation of non-significant trends in the data to ‘pseudo-significance’, rather, it should point towards better design for future studies. It seems, therefore, that while power analysis opens the door to greater control over sampling design, the research worker should tread carefully. The greater control brings with it a greater responsibility in terms of

50


Fig. 1.—The relations among Power Effect Size (E), Sample Size (n), Type I and Type II error for a hypothetical two-sample situation. In each diagram, the two curves represent the distribution of the means of samples of size n taken from two populations. and are the grand means of the two sets of estimates and the best estimates of the population means. If the null-hypothesis is true, sample means greater than will result in the incorrect rejection of a true null-hypothesis (Type I error). Thus as a percentage of all values under curve i, is represented by the area under curve i to the right of in each diagram. Conversely, if the null-hypothesis is false, values of less than will result in the incorrect retention of the null-hypothesis (Type II error). In this case, the area under curve ii to the left of in each diagram will represent (as percentage of all results described by curve ii). The area under the remainder of curve ii (to the right of ) represents the correct rejection of the (false) null-hypothesis, i.e. Power It can be seen from diagram B, that increasing a with unchanged n and effect size, results in decrease in and increased power. Diagrams C-E show the effects of increasing sample size (n) and so increasing the precision of the estimates In C, is arbitrarily fixed at the same level as in A, with the result that decreases and power increases, and power are both fixed in D as in A and, clearly, declines. Finally, in diagram E, and power are set at the same level as in A and the detectable effect size is seen to diminish considerably.


51

the realistic setting of ß and effect size. The more specifically the hypotheses are stated, the greater the usefulness of the technique. It should be noted that tests of power are inherently one-tailed; tests have no power if any effect occurred that was opposite to that proposed (Cohen, 1977). Clearly in two-tailed tests of significance, Type II error is possible at either tail of the null distribution, but cannot occur at both. Consequently, the power of a test is meaningful only when it is discussed with respect to a single alternative hypothesis. This point has been treated more fully by Sokal & Rohlf (1981). CONCLUDING REMARKS It is important to know the limitations of any sampling programme. The concept of a pilot study is one of doing a smaller, cheaper version of the anticipated main study so that the main study can be optimized. It would, therefore, be pointless expending a large portion of the total budget, time, and resources on a pilot study and subsequently severely constraining the main programme. The amount of time and resources spent on the pilot study will be specific to each project and generalizations cannot be made. The cost of a pilot study, however, will generally be outweighed in the long run by the improved efficiency, accuracy, and precision in the main sampling programme (Green, 1979). Pilot studies along the lines of those suggested on pp. 44–52 could also be designed to compare methods of sampling, with different methods being substituted for the different sized units. More elaborately, various dimensions of units (e.g. core diameter and depth, transect length and width) can be varied orthogonally in pilot studies to establish not only the optimum size (volume or area) but also the optimum dimensions of sampling units. Different methods can also be compared across a range of unit sizes by testing unit size and method orthogonally in the one pilot study. Although orthogonally testing a suite of variables is more powerful and informative, to do so may prove prohibitively expensive. Many possibilities exist, but cost will limit the extent of a pilot study. It will generally be cheaper and often sufficient to evaluate methods, unit size and replication at various levels in a set of sequential, separate pilot studies than in a single big one. Some arguments against pilot studies have been made. Smith (1984) argued that estimates of variance obtained in pilot studies are frequently biased by small sample sizes (Cochran, 1963) and that the apparent benefits of pilot studies may be misleading when a population is sampled at a number of times: a design that is optimal at one time may be very poor at another time because of changes in the characteristics of the population being sampled. Schwiegert & Sibert (1984) replied that some indication of possible improvements in design is better than consistently sampling in ignorance. If temporal variation in population characteristics is likely to cause changes in the optimum design of a programme, then small pilot studies may be required at the beginning of each sampling occasion (Kennelly & Underwood, 1985). With respect to the problem of biased estimates caused by small sample sizes, Schwiegert & Sibert (1984) pointed out that Cochran (1963) detailed methods of estimating confidence limits for sample sizes chosen on the basis of pilot studies. In general, precision remains relatively stable for sample sizes within these confidence limits. In the absence of careful design procedures, there is the risk that methods and designs for sampling will be chosen simply by reference to previous studies. In some cases, such as successive studies of a system, this may be justified. In others, however, it leads down the potentially blind alley of standardization by convention. If a large number of independent studies come to similar conclusions and widespread standardization of sampling procedures in some systems is a result, then the use of standard procedures is to be encouraged. Standardization for its own sake, against evidence of variation in the usefulness of methods and designs when used under differing conditions, is, however, to be discouraged (Gray, 1971; Loya, 1978; Omori & Hamner, 1982).

52


From our discussion of power analysis, it should be clear that much is to be gained from the use of power analysis to examine the likely usefulness of the planned sampling programme and analyses. With data from the pilot studies, the proposed analyses can be thoroughly tested for their sensitivity with the expected nature of the data. It is at this stage that careful decisions have to be made about the relative importance of Type I and Type II errors. It may be that the desired levels of these errors are unattainable simultaneously with the design and replication proposed for the main sampling programme. In such a case, critical reexamination of the sampling design, of the desired levels for Type I and Type II errors, of costs of increasing replication (to increase power), and even of the basic questions being asked will be necessary if the programme is going to fulfil its intended purpose. The adjustments that have to be made will be entirely specific to each study. SPATIAL PATTERN The arrangement of organisms in space may range from aggregated (contagious, over-dispersed, clumped) through a random pattern to being, more rarely, regular (under-dispersed, uniform). In this context it is worth repeating Cassie’s (1963) explanation of the terms over- and under-dispersion. The terms refer to the distribution of density estimates about the mean rather than the spatial arrangement of individuals. In an aggregated population there will be a greater occurrence of small and large densities, hence the frequency distribution would be over-dispersed. In contrast, if the biological population were uniformly arranged in space, then density estimates would be less variable and their frequency distribution under-dispersed. Knowledge of the spatial patterns of organisms is a necessary prerequisite in establishing what processes might underlie the observed arrangement of organisms. If, for example, individuals were regularly spaced throughout an area then competition might be invoked and tested for experimentally. If they were aggregated, then a range of processes leading to that arrangement might be tested. The extent to which the arrangement of organisms deviates from a random pattern is also implicitly considered in the design of sampling programmes used to estimate their abundance. Clark & Evans (1954, p. 446) have provided a definition of spatial randomness: “In a random distribution of a set of points on a given area, it is assumed that any point has had the same chance of occurring on any sub-area as any other point, that any sub-area has had the same chance of receiving a point as any other subarea of that size, and that the placement of each point has not been influenced by that of any other point.” In this review, we are concerned only with the description of pattern and do not wish to infer processes to account for any identified patterns. The first step in the description of the spatial pattern of a population is to test the null hypothesis of spatial randomness. A null hypothesis of spatial randomness is advantageous because tests of this hypothesis are two-tailed: rejection unambiguously suggests one of the two possible alternatives, either clumped or regular. If the test fails to reject the null, then little more needs to be said about the arrangement of those organisms: random is random. The need for more detailed description, or explanation, only arises if there is evidence of non-random pattern. Furthermore, although the great majority of populations in nature are clumped, random and regular populations have both been described, (e.g. Holme, 1950; Barnes & Marshall, 1951; Stimson, 1974; Jumars, 1975a; Seapy & Kitting, 1978; Ebert & McMaster, 1981). The null hypothesis that the population under study may be random is realistic. DETECTING NON-RANDOM PATTERN Indices of spatial pattern fall into two general categories: those derived from estimates gained from some chosen sample unit, such as quadrats or cores; and those based on measurements of distances between


53

organisms. We shall consider several of the most commonly used indices from each category and discuss are based on estimates the limitations of their use. The coefficient of dispersion and Morisita’s (1959) derived from unit sampling and Clark & Evans’ (1954, 1979) R, Pielou’s (1959) and Mountford’s (1961) and Johnson & Zimmer’s (1985) I are based on estimates of distances between organisms. Measures from sample units Variance/mean ratio (Coefficient of dispersion). The variance/mean ratio is the simplest index of spatial pattern and perhaps the one most commonly used in the marine ecological literature. It is based on the characteristic that, for a population of individuals that are randomly arranged in space, the variance of a sample will equal the mean density. Because equality of the variance and mean is a defining characteristic of the Poisson distribution, deviation from this expectation provides the basis for a test of spatial randomness (Pielou, 1969). Under the null hypothesis of random spatial pattern therefore, the estimated variance (s2) and obtained from a sample of randomly placed units should be equal. mean The variance/mean ratio may be tested for significant departure from the expectation (under a Poisson process) of unity by calculation of its confidence intervals (Greig-Smith, 1964). Examples of this approach include Kosler’s (1968) investigation of the patterns of spatial pattern of meiofauna in the Baltic Sea and Dayton’s (1973) analysis of the rocky intertidal alga, Postelsia palmaeformis. More usually, departures from unity are tested for statistical significance after conversion to the statistic:

which is compared with a distribution with degrees of freedom (Southwood, 1966; Pielou, 1969; Fisher, 1970). Tests of significant departure of the variance/mean ratio from unity have been used in many studies of species ranging from polychaetes and molluscs (Rosenberg, 1974) to asteroids (Scheibling, 1980). In several papers reviewed it was unclear whether the variance/mean ratio was incorrectly compared distribution or was first converted to its test statistic I. Partial explanation for this may lie directly with a in semantic confusion, as the terms “variance/mean ratio”, “coefficient of dispersion” and “index of dispersion” appear to have slipped into synonymy (e.g. Southwood, 1966; Elliott, 1977; Diggle, 1983). The test statistic as used above (Fisher, term “index of dispersion (I)” has traditionally meant the 1970). The “coefficient of dispersion”, as originally proposed by Blackman (1942, p. 352), referred to the variance/mean ratio: “This estimate of dispersion, which might be termed a ‘reduced index of dispersion’, since it is the index of dispersion divided by the degrees of freedom, has been called by Clapham (1936) ‘relative variance’.” Blackman, therefore, clearly distinguished the “coefficient of dispersion” from the “index of dispersion”, as have many authors since then (e.g. Cassie, 1963; Greig-Smith, 1964; Pielou, 1969; Jumars, 1975a). The distinction between these terms should be maintained. Several authors have investigated the power of the “coefficient of dispersion” to detect non-random pattern (Bateman, 1950; Kathirgamatamby, 1953; Diggle, 1979; Perry & Mead, 1979; Helshe & Ritchey, 1984). In general, this statistic was a powerful test for non-randomness, even when the organisms were only mildly aggregated. The “coefficient of dispersion” was not, however, as powerful in detecting regular patterns. Bateman (1950) has shown that the “coefficient of dispersion” provided a powerful test of randomness only when sample size was greater than five. Bateman (1950) considered that if the mean density of organisms per sample unit was less than one, then the results of the test were invalid. Means of

54


between one and five organisms per unit produced conservative results, there being only a small probability of rejecting the null hypothesis of random spatial pattern (see also Cassie, 1963). This caution seems to be supported by results of studies in which “coefficients of dispersion” based on mean densities of less than five were tested for significant departure from unity (e.g. Holme, 1950; Barnes & Marshall, 1951; Kosler, 1968; Gage & Geekie, 1973a). In these studies there was a strong trend towards detection of significant departure from randomness with increasing density, and in many samples with a mean density of less than five there was no significant evidence for non-random pattern. The variance/mean ratio is a useful descriptive statistic. It is perhaps best regarded as an exploratory descriptive device, capable of detecting non-random pattern, but subject to a number of limitations. Its mean, and n make it of little use as a comparative test statistic when strong dependence on unit size, these variables differ among samples (Elliott, 1977). Goodness-of-fit tests. A test for non-random pattern may also be made by directly comparing the observed frequency distribution of estimates of density to the Poisson distribution with the same mean. The test, the more recently recommended comparison is made by a goodness-of-fit test such as the traditional G test, or their nonparametric equivalent, the Kolmogorov-Smirnov test (Sokal & Rohlf, 1981). Gage & Geekie (1973a) analysed a number of data sets from an investigation of benthic fauna with both the test and tests rejected the null hypothesis in a the Kolmogorov-Smirnov test. They found that, although the two similar percentage of tests the two tests produced conflicting results for some data sets. Fifteen of 43 tests test were not significant by the Kolmogorov-Smirnov test. Gage & judged to be significant by the Geekie argued that it is generally assumed that discrepancies arise in such cases from the different power of test, size classes with low expected frequency have to be combined, thus decreasing the two tests. In the the degrees of freedom of the test. Because such pooling is not required for the Kolmogorov-Smirnov test, it would not be similarly affected at the tails of the distribution. This index is based on an analysis of the proportion of the total number of organisms Morisita’s index found in each replicate sample unit. The index is calculated from:

where Σx is the sum of individuals found in all replicates (Morisita, 1959, 1962, 1971; Southwood, 1966; for maximum regularity and n when all Elliott, 1977). Its value can range between individuals are in the same replicate (maximum aggregation). It has a value of one for a random pattern, is, therefore, strongly dependent on sample size (Green, 1966; Elliott, 1977). The significance of may be tested against the F dis-tribution by calculating

(Southwood, 1966). Alternatively, the test statistic

is distributed with degrees of freedom (Elliott, 1977). Morisita (1959), Hairston, Hill & Ritte (1971), and Elliott (1977) have investigated the behaviour of under differing spatial patterns and found it to be strongly influenced by the size of sampling unit. For an aggregated pattern with individuals randomly gives a large and constant value with increasing unit size until the size of the arranged within clumps,


55

sample unit approximates the size of the clumps. Beyond this it declines towards one. If individuals are will increase with increasing unit size before uniformly arranged within clumps then the value of declining as above. The index gave unpredictable results when applied to a pattern that was known to be has been shown to be relatively invariant under conditions of uniform (see Fig. 18 in Elliott, 1977). different abundances (Morisita, 1959; Gage & Geekie, 1973b; Elliott, 1977). is relatively invariant with Gage & Geekie (1973b) and Colby & Fonesca (1984) used the fact that differing densities in their comparisons of spatial pattern between samples of infaunal molluscs and sand crabs, respectively. Interestingly, Gage & Geekie (1973b) found considerable concordance between the “index of dispersion” (I) and in the detection of significant departures from randomness in a number of was found to be slightly more conservative than I, but both tests gave more significant their samples, goodness-of-fit test or the Kolmogorov-Smirnov test. results than tests of a Poisson model using either a Other indices based on sample units. Other indices derived from measures of variance of density estimated with sample units have been discussed in several texts (Greig-Smith, 1964; Southwood, 1966; Pielou, 1969; Elliott, 1977). These indices have received little use in the marine ecological literature and will not be discussed here. Most are variants of the variance/mean ratio and are subject to the same limitations in interpretation (Elliott, 1977). One variant that deserves some mention is the index proposed by Green (1966). The index is given by:

and is valued at for maximum regularity, 0 for random pattern, and 1 for maximum and and may, therefore, be used as a aggregation (Green, 1966). Cx is independent of variations in n, comparative index even when these values vary between samples (Green, 1966). The usefulness of Cx as a comparative statistic is limited because there is at present no test of significance for departures from randomness. Hogue (1978) and Findlay (1981) have used Cx as a descriptive statistic in studies of pattern in benthic meiofauna. Jumars (1975a) has extended the use of the “index of dispersion” to consider the spatial pattern of more than one species. By comparing the summed “indices of dispersion” with the “index of dispersion” calculated from the summed data, differences in the spatial pattern of the species could be detected. If all species are similarly arranged then the two will be similar. If species tend to be segregated in sample units then they will differ. Two studies of deep-sea meiofauna have found evidence for patchiness in the arrangement of species using this test (Jumars, 1975b; Bernstein & Meador, 1979). Cautions for the use of indices based on sample units. The widespread use of these indices probably stems from the fact that data obtained to estimate abundance can be used to test for non-random pattern. Although this simplicity may be attractive, several points should be kept in mind when interpreting results. The first and most important point is that, for aggregated populations, the results gained may be an unpredictable consequence of the size and shape of the sampling units used (Skellam, 1952; Southwood, 1966; Payandeh, 1970; Perry & Mead, 1979; Diggle, 1983; Helshe & Ritchey, 1984). The observed pattern may vary from random, to aggregated and finally regular depending on the scale at which the population was sampled relative to the scale at which it was aggregated. This sampling artefact may arise from at least two sources. The relationship between variance and mean may change due to variations in abundance alone or through variations in the variance without attendant shifts in abundance, or both. Perry & Mead (1979) have shown that the capacity of significance tests to detect significant departures from random pattern for the variance/mean ratio increases with increasing density, and suggested that this is more the result of an increase in variance with increasing density than it is due to increases in density per se.

56


For aggregated populations of many different types of organisms, there is strong empirical evidence that the variance of a sample increases disproportionately with increases in its mean (Clapham, 1936; Clark & Milne, 1955; Taylor, 1961, 1971; Kosler, 1968; Gage & Geekie, 1973a; Gagnon & Lacroix, 1982). This relationship is known as Taylor’s power law, and is essentially a statement that the spatial behaviour of organisms is density-dependent (Taylor, Woiwod & Perry, 1978). Under these conditions, as the size of the sampling unit changes, aggregation, as estimated by a unit-based index may change as a function of mean density, even if the variance remains constant. Alternatively, apparent constancy of the variance/mean ratio may arise from changes in both elements of the ratio acting to cancel out each other (Downing, 1979). Indices based on distances between organisms From the above discussion, it will be clear that in the description of the arrangement of organisms in space, there are constraints imposed by the use of artificially defined sampling units. The use of quadrats or other sampling units imposes an order on the community being sampled that may mask true spatial relations among organisms. Consideration of these problems led to the development, by plant ecologists in the 1950s, of indices based on distances between organisms rather than on estimates of abundance gained from unit-based sampling. The data collected are, therefore, free from artefacts introduced by the sampling units. The general technique has been termed plotless sampling, or nearest-neighbour (N-N) analysis. Clark and Evans’ R. The earliest index based on distances between organisms to gain common usage was that proposed by Clark & Evans (1954). The measure is based on the fact that under the null hypothesis of random spatial pattern the expected mean distance between nearest neighbours (rE) can be calculated from knowledge of the density of organisms. where is an independent estimate of the density of organisms. The index R is the ratio of the observed mean distance (r0) and the expected number. R is tested for significance by the standard normal variate Z:

where

N being the total number of individuals in area A. R has been used in several studies of spatial pattern in marine organisms. An early example is Connell’s (1963) laboratory study of the spatial arrangement of colonizing Erichthonius braziliensis, a tube-dwelling amphipod. Immediately after settlement no non-random pattern was discernible, but then a regular pattern became established. Other examples include Stimson’s (1974) analysis of the spatial arrangement of corals and Wilson’s (1976) study of the bivalve Tellina tennis. Clark & Evans (1979) have extended their original index to include considerations of the spatial pattern of organisms in k dimensions. Practically, this allows the analysis of patterns along a line (one dimension) and in three dimensions and removes the restrictive assumption that organisms be positioned in two dimensions only. The extension of an analysis of spatial pattern to three dimensions might be useful in the study of infauna, or organisms inhabiting branching corals. Rohlf & Archie (1978) briefly discussed practical means of plotting the positions of organisms in three dimensions. The three dimensional form of R has apparently not yet been used in marine ecology.


57

Pielou’s and Mountford’s α Pielou (1959) has developed a different measure derived from the distribution of distances between randomly chosen points and the nearest individual rather than N-N distances. If individuals in a population are randomly arranged, then the distance between random points and nearest neighbours will be the same as N-N distances. The calculation of Pielou’s index requires a set of such distances and the density of the population. The index is given by:

where d is the density of organisms and is the mean, squared distance between random points and values greater than one indicating nearest-neighbours. Random pattern will be indicated by clumpedness, and values less than one indicating regularity. may be tested for significant departure from 1 by the test statistic

where n is the number of distances measured. A calculated value greater than the tabulated at a significance level of (usually) 0·05 indicates significant aggregation, while a value less than the (usually) 0·95 tabulated value indicates a regular pattern. Mountford (1961) has demonstrated that when density is estimated rather than known absolutely (by enumeration of the entire population) then may result in excessive numbers of erroneous departures from randomness, compared with that expected for a randomly arranged population. He provided a corrected version of the test statistic for that takes into account the estimation (rather than absolute knowledge) of density. has been used to detect non-random spatial pattern in studies of intertidal gastropods and subtidal sea urchins and gastropods (Underwood, 1976; Andrew & Stocker, 1986; Choat & Andrew, 1986). Mountford’s corrected test statistic was not used in these studies because the whole population used in the calculation of was counted. Density was therefore known, not estimated. Johnson and Zimmer’s I. Johnson & Zimmer (1985) have recently proposed a new index of spatial pattern (I). This index seems to offer several advantages over previous indices because all that is needed for its calculation is a set of distances between random points and the nearest individual. The calculation of I does not require an independent estimate of density. It can be used to consider the arrangement of organisms in three dimensions, as can R. When applied to sets of real data Johnson & Zimmer’s index compared favourably with results given by Fisher’s “index of dispersion” and Pielou’s and Mountford’s (Johnson & Zimmer, 1985). When tested with simulated data, it proved to be more powerful than in detecting regular patterns and an aggregated pattern that followed the Negative Binomial distribution. Although the behaviour of I under different spatial patterns and sampling properties has yet to be fully explored, it would seem to have great potential. No accounts of its use have so far appeared in the marine ecological literature. Cautions for the use of indices based on distances. The three indices considered are based on the assumption that an infinite population is being sampled, i.e. there are no boundaries to the area. Edge or boundary effects will arise if the individual (or random point) from which distances are measured is close to the edge of the area considered because there is a higher probability that its nearest neighbour will be outside the area of study. This is less likely to be true if the base individual were in the centre of the area. Selection of the closest individual within the area may therefore introduce bias by over-estimating the distance to nearest-neighbour.

58


Sinclair (1985) has shown that boundary effects may have a large impact on R in the above form, biasing results in favour of regularity by over-estimating rE. He went on to discuss a correction factor proposed by Donnelly (1978, in Sinclair, 1985) that counters this bias. Sinclair demonstrated that when the corrected test indicated a significantly regular pattern, with a probability of Type I error of 0·05, the probability of Type I error in the uncorrected test was actually 0·40. This enormous discrepancy arose from a population bounded within a square, which is the most common shape of quadrat. The magnitude of this bias will depend on the shape of the area considered with longer boundaries relative to the area enclosed meaning a greater bias. An alternative method of avoiding biases introduced by edge effects is to create a buffer, or border, around the edges of the area. Individuals in this area may be included as neighbours but may not be used as base individuals. Although this tactic was suggested by Clark & Evans (1954) in their original paper it has rarely been used (Sinclair, 1985). An appropriate size for such a buffer may be calculated as that distance within which a large proportion of N-N distances fall. Anderson & Kendziorek (1982) set that proportion to be 0·9 and found the required distance (d) by solving the equation

where is the estimated density within the area. The formula may be used for any proportion by appropriate substitution. Similarly, Harvey, Ryland & Hayward (1976, their p. 101) set the size of the buffer around a circular area “…such that individuals lying at the perimeter in a randomly distributed population would have a neighbour nearer than the edge of the disk on 95% of occasions…”. They arrived at that distance by solving the formula

for the square of the required width of the buffer, is the density. They noted that for aggregated populations, this distance would be an over-estimate. Several studies, e.g. Underwood (1976) and Andrew & Stocker (1985) have used buffers to avoid biases introduced by edge effects. Kinzie & Snider (1978) and Simberloff (1979) have considered the effect that the assumption of point processes has on the results gained from analyses of spatial pattern based on N-N distances. Both papers have demonstrated that the assumption that individuals were points in space, i.e. they had no area (or volume), can cause misleading results. The magnitude of the errors introduced will increase with the size of the individuals relative to the distances between them. Simberloff (1979) provided solutions to this problem for Clark & Evans’ R. He provided an approximate algorithm to correct this bias for situations in which the average diameter of organisms (assuming they are circular) is less than half of the expected mean N-N distance. Simberloff suggested the use of simulation procedures to estimate the magnitude of the bias for organisms with diameters greater than half the expected mean N-N distance. Simberloff’s consideration of these biases was extended to the three dimensional form of R. Simberloff re-analysed Connell’s (1963) data, taking into account the size of the amphipods, and although Connell’s interpretations were confirmed, the probability that the significant regularity was an error was increased. A preliminary re-analysis of Stimson’s (1974) and Wilson’s (1976) data suggested that results from both studies may have been similarly affected by the assumption that the individuals under study had no area. The expected mean N-N distance in the two studies was dependent on the density of the organisms, and ranged between 2·9 and 6·7 cm and between 16·7 and 28·8 cm, respectively. The magnitude of bias introduced would depend on the size of the organisms; the larger the organisms relative to these N-N


59

distances the greater the bias. Although the sizes of the corals and bivalves were not given, it would appear likely that, at least at the higher densities considered, the expected mean N-N distances would have been under-estimated, therefore favouring regularity. Stimson found that tests of R consistently indicated significant regularity. Interestingly this result was in contrast to a random pattern as indicated by the “index of dispersion” (but see p. 73 for limitations of that index). Wilson did not reject the null hypothesis of random pattern for all densities measured in the field. In the laboratory there was some evidence for aggregation but only in the lower two-thirds of the density range considered. In the absence of size estimates, the extent to which the statistical significance of these results, and interpretation, are in error, is a matter of conjecture. Anderson & Kendziorek (1982) have made use of the modified form of R in their detailed study of the spacing patterns of tube-dwelling polychaetes. They found significant evidence for regularity in the spacing of individuals. Pielou (1959) has demonstrated the importance of randomly selecting the base individual when using Clark & Evans’ R. If individuals are chosen by some short-cut method, such as choice of random areas followed by random selection of individuals within those areas, then the results of the test will be biased in favour of regularity. Assuming an aggregated population, randomly choosing small areas, and then individuals within them, would decrease the probability of selecting individuals within aggregations. True randomization would require the identification of all the organisms in the population and the selection of a sample using random numbers—a task that would be enormous if not impossible in many instances. Clark & Evans (1979) noted that this was one of the major practical limitations of their technique. Pielou (1969) has also pointed out an important distinction between those indices based on N-N distances, such as R, and those derived from estimation of distances between random points and nearest-neighbours. Pielou has identified two aspects of spatial pattern: intensity and grain. The intensity of a pattern may be thought of as the variability in density from place to place (Pielou, 1969). The grain of spatial pattern may be thought of as the way that variability in density is arranged in space. For example, a coarse-grained pattern may take the form of aggregations that are widely spaced with large areas of low density between aggregations. Alternatively, if there are large fluctuations in density over short distances, then the pattern may be considered to be fine-grained. The grain of a pattern may be independent of its intensity (Pielou, 1969). Indices of spatial pattern based on N-N distances measure only the intensity of pattern (Pielou, 1969). This is because they are derived from distances between organisms and in an aggregated population the majority of distances measured will be within clumps. The contribution from individuals outside clumps will be less than that from isolated individuals if random point to nearest-neighbour distances were measured. The latter indices are therefore influenced by the grain of the pattern to a greater degree than the N-N indices (Pielou, 1969). Advantages of the various indices It is not easy to compare the relative merits of tests for non-random spatial pattern based on numbers of individuals in sampling units with those based on measurements of distances. Both have advantages, whose relative worth will depend on the situation considered. Where distances between individuals can be measured, the distance techniques would seem to be preferable because they are less subject to artefacts introduced by the sampling units. For many research workers, however, the ability to measure distances among individuals may be an unobtainable luxury, e.g. those working on plankton ecology or on deep-sea fishes. Measures based on numbers of individuals per sample unit are the only option open under such conditions. Interpretations should always be constrained by the limits of the technique of sampling and

60


analysis. If indices based on data from sampling units are used, then explicit statements should be made about the scale at which any deviation from random pattern was observed. Detecting spatial pattern in heterogeneous areas The discussion so far has assumed that samples were taken from homogeneous areas, i.e. the probability of an individual occurring at any point was constant throughout the area. An ecological problem that has generated some recent work concerns the determination of boundaries of occurrence of species along gradients. The assumption of homogeneity clearly does not apply, so different probability rules are necessary. Pielou & Routledge (1976) applied Pielou’s (1975, as quoted in Pielou & Routledge, 1976) test for the independent assortment of boundaries to the occurrence of plant species along a gradient in a salt marsh community. The test is based on the determination of the probability of a boundary (i.e. presence and absence of a species in adjacent quadrats) occurring along a continuous transect of quadrats running along the gradient. Gardiner & Haedrich (1978) and Underwood (1978) have independently shown that the probability theory used by Pielou to establish the expected occurrence of boundaries within quadrats biased the test in favour of regularity (but see Pielou, 1979). A corrected version of the test based on probability rules that distinguish among species was provided in both papers. Gardiner & Haedrich applied this modified test to an analysis of boundaries of the occurrence of deep-sea fauna along a single transect. Underwood (1978) considered the problem of species’ boundaries in replicated transects on a rocky intertidal shore. Chaloupka & Hall (1984) also examined the occurrence of species’ boundaries with increasing height up intertidal shores. These authors have extended the Gardiner & Haedrich and Underwood test to consider situations in which the number of species boundaries per quadrat is restricted. Abel, Williams, Sammarco & Bunt (1983) have applied the same probability rules as Pielou & Routledge for the determination of the probability of a species boundary occurring in any given quadrat, but have constrained the number of boundaries that can occur within a single quadrat. They applied the test to a study of the distribution of corals in the Caribbean Sea. DESCRIBING NON-RANDOM PATTERN The indices so far described are sufficient only to detect deviations from spatial randomness, and provide no information on the details of those non-random patterns. For example, they cannot detect any underlying periodicity or constancy to the pattern or whether it takes the form of a mosaic of patches of high and low density; Jumars, Thistle & Jones (1977) provide an illustration of this point. Furthermore, the indices give no indication of scales at which pattern might be found other than the scale at which the data used to calculate the index are gathered (Hill, 1973). In Pielou’s (1969) terminology, the indices tell us only about the intensity (variability in density from place to place) of pattern and, although some are influenced by grain (Pielou, 1969), they say nothing explicit about the grain of a non-random pattern. Additional levels of structure may overlay pattern at the lowest level, i.e. aggregations of aggregations or large scale uniformity in the data. Pattern may be most evident at the finest scale (among-individual), and the knowledge that non-random pattern exists at that scale may often be all that is required (see earlier references). In some instances, however, this is not sufficient, and the investigator may wish to describe pattern more comprehensively. The description of pattern and the detection of higher orders of pattern are more difficult than demonstrations of non-randomness. The data required are more exhaustive and the analyses used more


61

complex (and consequently more open to mis-interpretation). Two techniques that have been widely used in the ecological literature are Spectral Analysis and Pattern Analysis. Both require intensive, systematic sampling, but only techniques based on this sort of data will provide extensive information on the nature of pattern in a community. The point estimates introduced by Jumars et al. (1977) provide some information based on a limited amount of systematic sampling, but if detailed descriptions of non-random pattern are required then extensive systematic sampling is required. Of course the optimal, but rarely possible, solution is to map the positions of all organisms in the entire area. Although detailed discussions of Spectral Analysis and Pattern Analysis will not be attempted here, a very brief introduction to the literature is appropriate. Spectral Analysis describes observed patterns by fitting a linear combination of wave forms to variability in data collected as a series of counts along a transect (Ripley, 1978). The use of Spectral Analysis in ecology has been well reviewed by Platt & Denham (1975), Fasham (1978), and Ripley (1978). Despite the enthusiasm Platt & Denham (1975) had for the future of the technique as a source of a new theoretical framework for ecological research, the only area in which it has been used extensively in the marine ecological literature has been in the study of patchiness of plankton. Fasham (1978) has reviewed this literature and so we shall not discuss its use in this field here. Ripley (1978) cautioned that some experience is required if Spectral Analysis is to be interpreted correctly. Pattern Analysis, or “contiguous quadrat analysis”, has been used extensively by ecologists studying terrestrial plants (Greig-Smith, 1952; Kershaw, 1957; see Greig-Smith, 1979, for review), yet has been virtually ignored by marine ecologists. The following brief discussion is intended more as an introduction to the literature than a review of usage. As first outlined by Greig-Smith (1952), Pattern Analysis involved the comparison of counts from contiguous quadrats of increasing size. The area under study was divided into a grid of contiguous quadrats and density estimated within each. Numbers from adjacent quadrats were then summed, producing estimates from quadrats double the original size, which were then doubled, and so on. Thus the size of quadrats considered increased as powers of two. Variances of the mean counts for each size of quadrat were then plotted against quadrat size. Peaks in the plots occur at scales at which there is great variability in estimates of density. These peaks are interpreted as being the scales at which non-random pattern occurred. Extensive descriptions of the technique may be found in Greig-Smith (1964, 1979). This original formulation has been shown to have several limitations (Thompson, 1958; Pielou, 1969; Kershaw, 1970; Errington, 1973; Mead, 1974; Usher, 1975; Ludwig & Goodall, 1978; Upton, 1984). The most serious among these arose from the non-independence of estimates of variance among sizes of quadrat. This reduced the usefulness of the technique because the F ratios used to test for the significance of peaks were based on non-independent mean square estimates, and therefore were invalid. Other problems lay in the critical dependence of the starting point of the transect on the estimation of the scale at which the pattern (s) occurred. This was especially a problem when there was a periodic pattern of variability in density along a transect. If the transect started in an area of low density, then the estimated distance between peaks in variance would differ from that given if the start coincided with a peak. Another problem was that the length of the transect increased by powers of two and, therefore, there were often relatively few data points to analyse (Pielou, 1969; Goodall, 1974). There have been a number of derivations of the original technique of Pattern Analysis proposed to circumvent the problems posed above, but none has been entirely successful. Mead (1974) modified the test test for anomalous groupings (i.e. groupings of high to compare quadruplet groupings of quadrats by a or low density). Upton (1984) has subsequently demonstrated that Mead’s test is subject to the same problem of critical dependence on starting point as the original formulation.

62


The only study we came across that had used classical Pattern Analysis in a marine ecological study was Grassle’s (1973) study of pattern in a coral community. Our failure to find more studies using this technique seems likely to be more a reflection of our limited search than a true picture of the popularity of the technique. It seems improbable that Grassle’s paper is the only instance where some form of Pattern Analysis has been applied to a marine problem, particularly in the phycological literature. Fitting contagious distributions Where significant departures from random spatial pattern have been demonstrated, many different statistical distributions have been fitted to summarize data. In particular, the negative binomial distribution, has proved popular as a model for contagious populations, e.g. Gärdefors & Orrhage (1968), Oviatt & Nixon (1973), Fasham, Angel & Roe (1974), and Todd (1978). See Cassie (1963) for a review. The negative binomial distribution is described by two parameters, mean density and the exponent k (Elliott, 1977; Taylor, Woiwod & Perry, 1979, and references therein). The reciprocal of k has been used as a measure of clumping of individuals: as the population tends to a random (Poisson) distribution 1/k tends to zero. Taylor et al. (1978, 1979) have investigated the utility of the negative binomial distribution as a model for the spatial pattern of organisms. In order for k to be a useful descriptor of spatial arrangement, it must behave in a predictable way when density and the degree of aggregation changes. It fails to do this. Taylor et al.’s (1979) analysis suggests that k has a rather dim future as a measure of dispersion. They found that even assuming that the negative binomial distribution adequately fitted the data, it had a number of severe limitations in its application to ecological problems. Numerous authors have cautioned against drawing ecological conclusions from the apparent fit of statistical models such as the negative binomial (Pielou, 1969; Bliss, 1971; Sokal, p. 374 in Taylor, 1971; Taylor et al., 1978, 1979; Todd, 1978). Although the data may fit a given distribution, it cannot then be assumed that the mathematical processes that determined that particular function are adequate descriptors of processes that determined the spatial arrangement of the organisms. Goodall (1974) suggested that the fitting of such contagious models has not been helpful in generating interesting and testable hypotheses about the arrangement of organisms in space. Taylor et al. (1979, their p. 301) concluded that the study of aggregation might best proceed if “based solidly on real data unprejudiced by preconceptions from models having little correspondence with reality”. CONCLUDING REMARKS If statements about spatial patterns are based on counts from a sample of randomly placed units, then only limited statements can be made about the intensity of the non-random pattern (Pielou, 1969). Non-random pattern can be claimed with more confidence from sets of distances between organisms, but unless the data contain detailed information about the spatial relations among individuals, i.e. the positions of organisms are mapped, then little more than a demonstration of non-random pattern is possible. Rapid methods for plotting the positions of organisms in the field may be found in Underwood (1977), Rohlf & Archie (1978), and Weinberg (1981). Ripley (1978) and Diggle (1983) have provided detailed discussions of the types of analyses that may be applied to mapped patterns. A review of the literature suggests that data of this completeness have only rarely been collected in marine systems. The overwhelming majority of studies concerning the spatial patterns of organisms seem content with testing for non-random pattern. As models attempting to explain the organization of communities become more complex, information more detailed


63

than demonstrations of non-random pattern will be required to test hypotheses related to the spatial relations of organisms. GENERAL CONCLUSIONS It will be evident that there are no easy prescriptive solutions to the problems of designing a sampling programme. To paraphrase Williams (1978), slightly out of context, there are only vague answers to such precise questions as: where?, how big? and how many? The best way to sample the abundance of an organism has to be determined for each individual study. We have reviewed the application of sampling procedures in the marine ecological literature and found considerable evidence of standardization of methodologies and of adherence to customary ways of doing things. Following a discussion of well-established criteria used to design better, more efficient sampling programmes, we have joined others (e.g. Underwood, 1981; Hurlbert, 1984) in seeking to attract attention to the advantages to be gained from good design. The reappraisal of methodologies continues to be a topic of much research and discussion in other fields. For example, in the literature on wildlife management there are many papers dealing with the types of issues we have discussed here. The great logistic difficulties facing many studies in marine ecology should prompt similar discussion in this field (see also Omori & Hamner, 1982). Our recommendations should not be misconstrued as a plea for the complete breakdown of established techniques. Hopefully, popular techniques will prove to have a sound basis justifying their common usage. The important point is that their use should depend upon their demonstrated suitability to each study, rather than simply on convention. It may be argued that the use of different sampling units will make comparative work difficult because the estimates will be expressed in different units. The ability to standardize estimates of abundance among samples from units of different sizes, circumvents this objection. The standardization is, however, subject to several assumptions and limitations. Assuming that different estimates of mean abundance are equally accurate and, therefore, that sampling units of different size will yield proportional results, data may be scaled to a different size of unit. Variance estimates can be scaled as the square of the change in unit size, i.e. if the unit has to be doubled to the standard unit then the variance of the mean should be increased fourfold. The description of spatial pattern from estimates based on units cannot be standardized so simply. As discussed previously, if the population is aggregated, the relationship between the mean and variance of a sample will vary unpredictably with the size of the sample unit. Unless the relationship between the variance and mean is known for the density range considered, the degree to which variance will change with a given change in mean cannot be calculated. Measures of spatial pattern derived from adjusted estimates of abundance will, therefore, generally be invalid. We suggest that the advantages to be gained from designing programmes so as to gain accurate and precise estimates, with minimal cost, far outweigh any disadvantages. It should be borne in mind that, given the evidence we have presented, the advantages of using a standardized method for comparative work may be illusory because estimates of abundance may be poor and, hence, of little comparative value in any event. If no statements about the accuracy and precision of estimates are made in studies that are to be compared then it can only be assumed that the estimates are equally reliable and therefore truly comparable. We do not suggest that extensive discussions of sampling methodologies be published in every paper. All that is required is a brief justification of why the organisms were sampled the way they were. Such brief statements will also be an aid to those who are designing similar programmes by providing a range of

64


possible methods to be considered. Toft & Shea (1983) and Underwood (1981) have made similar pleas for the publication of power of tests, or at least the provision of sufficient details for others to calculate power. Details of replication and sources of variation are now generally required in describing experimental studies. There is no lesser need in the description of sampling programmes. Ecologists have long recognized the importance of justifying their sampling methods. The German naturalist Victor Hensen, one of the first ecologists to estimate the abundance of plankton, was concerned with establishing the validity and reproducibility of his results (Hensen, 1911, 1912, as quoted in Lussenhop, 1974). The development of more sophisticated methodologies for the collection of data and the increasingly large array of statistical procedures for their analysis make the need to evaluate the accuracy and precision of the sampling even greater today. Fortunately, the tools to do that evaluation have also been developed considerably since Hensen’s day, so it is also easier. Optimization techniques, such as cost-benefit analysis and power analysis, put the modern marine ecologist in a good position to design sampling programmes that provide accurate and precise descriptions of natural systems. Our impression is that the description of pattern has become the poor cousin of experimental work in ecological studies. We do not take issue with the need for manipulative experiments to investigate processes in ecology. Experiments, however, will be difficult to design and interpret without extensive and reliable natural history and distributional information. We hope our discussion will prompt a greater attention to how marine ecologists describe spatial pattern. ACKNOWLEDGEMENTS We thank A.J.Underwood for his advice and extensive criticisms of the manuscript. We are grateful to J.H.Choat, P.G.Fairweather, K.McGuiness, P.F.Sale and H.P.A.Sweatman for discussion and constructive comments on the manuscript. We hope they will accept the path we chose through their sometimes conflicting criticisms. We are especially indebted to L.J.Stocker for her many contributions, in formulating ideas, giving us advice and encouragement, and in the preparation of the manuscript. Funding for this project was provided by a Commonwealth Scholarship and Fellowship Plan Award (to N.L.A.), a Commonwealth Postgraduate Research Award, and grants from the Great Barrier Reef Marine Park Authority and from the Australian Coral Reef Society (to B.D.M.). We are grateful for the support provided by the School of Biological Sciences, University of Sydney. REFERENCES Abel, D.J., Williams, W.T., Sammarco, P.W. & Bunt, J.S., 1983. Mar. Ecol. Prog. Ser., 12, 257–265. Alden, R.W., Dahiya, R.C. & Young, R.J., 1982. J. exp. mar. Biol. Ecol., 59, 185– 206. Anderson, D.J. & Kendziorek, M., 1982. J. exp. mar. Biol. Ecol., 58, 193–205. Andrew, N.L., 1986. J. exp. mar. Biol. Ecol., 97, 63–79. Andrew, N.L. & Stocker, L.J., 1986. J. exp. mar. Biol. Ecol., 100, 11–23. Barnes, H. & Marshall, S.M., 1951. J. mar. biol. Ass. U.K., 30, 233–263. Bateman, G.I., 1950. Biometrika, 37, 59–63. Bell, J.D., Craik, G.J.S., Pollard, D.A. & Russell, B.C., 1985. Coral Reefs, 4, 41– 44. Bernstein, B.B. & Meador, J.P., 1979. Mar. Biol., 51, 179–183. Bernstein, B.B. & Zalinski, J., 1983. J. Environ. Mgmt, 16, 35–43. Blackman, G.E., 1942. Ann. Bot. N.S., 6, 351–370. Bliss, C.I., 1971. In, Statistical Ecology, Vol 1, Spatial Patterns and Statistical Distributions, edited by G.P.Patil et al., Pennsylvania State University Press, Pennsylvania, pp. 311–335.


65

Bouchon, C., 1981. Mar. Ecol. Prog. Ser., 4, 273–288. Bradley, J.V., 1968. Distribution-free Statistical Tests. Prentice Hall, London, 388 pp. Brock, R.E., 1982. Bull. mar. Sci., 32, 269–276. Caffey, H.M., 1985. Ecol. Monogr., 55, 313–332. Cassie, R.M., 1963. Oceanogr. Mar. Biol. Ann. Rev., 1, 223–252. Caughley, G., 1977. Analysis of Vertebrate Populations. John Wiley & Sons, New York, 234 pp. Chaloupka, M.Y. & Hall, D.N., 1984. J. exp. mar. Biol. Ecol., 84, 133–143. Choat, J.H. & Andrew, N.L., 1986. Oecologia (Berl.), 68, 387–394. Clapham, A.R., 1936. J. Ecol., 24, 232–251. Clark, P.J. & Evans, F.C., 1954. Ecology, 35, 445–453. Clark, P.J. & Evans, F.C., 1979. Ecology, 60, 316–317. Clark, R.B. & Milne, A., 1955. J. mar. biol. Ass. U.K., 34, 161–180. Cochran, W.G., 1963. Sampling Techniques. John Wiley & Sons, New York, 2nd edition, 413 pp. Cochran, W.G. & Cox, G.M., 1957. Experimental Designs. John Wiley & Sons, New York, 2nd edition, 611 pp. Cohen, J., 1977. Statistical Power Analysis for the Behavioural Sciences. Academic Press, New York, revised edition, 474 pp. Colby, D.R. & Fonseca, M.S., 1984. Mar. Ecol. Prog. Ser., 16, 269–279. Connell, J.H., 1963. Res. Popul. Ecol., 5, 87–101. Cuff, W. & Coleman, N., 1979. J. Fish. Res. Bd Can., 36, 351–361. Dayton, P.K., 1973. Ecology, 54, 433–438. Dayton, P.K. & Tegner, M., 1984. In, A New Ecology: Novel Approaches to Interactive Systems, edited by P.W.Price et al., John Wiley & Sons, New York, pp. 457–481. DeMartini, E.E. & Roberts, D., 1982. Mar. Biol., 70, 129–134. Denman, K.L. & Gargett, A.E., 1983. Limnol. Oceanogr., 28, 801–815. Dennison, J.M. & Hay, W.W., 1967. J. Paleont., 41, 706–708. Diggle, P.J., 1979. Biometrics, 35, 87–101. Diggle, P.J., 1983. Statistical Analysis of Spatial Point Patterns. Academic Press, Inc., London, 148 pp. Doherty, P.J. & Sale, P.F., 1986. Coral Reefs, 4, 225–234. Downing, J.A., 1979. J. Fish. Res. Bd Can., 36, 1454–1463. Downing, J.A., 1980. Can. J. Fish. Aquat. Sci., 37, 1329–1330. Downing, J.A., 1981. Can. J. Fish. Aquat. Sci., 38, 127–129. Downing, J.A. & Anderson, M.R., 1985. Can. J. Fish. Aquat. Sci., 42, 1860–1869. Downing, J.A. & Cyr, H., 1985. Can. J. Fish. Aquat. Sci., 42, 1570–1579. Eberhardt, L.L., 1978a. J. Wildl. Mgmt, 42, 1–31. Eberhardt, L.L., 1978b. J. Wildl. Mgmt, 42, 207–238. Ebert, T.A. & McMaster, G.S., 1981. J. Ecol., 69, 559–564. Elliott, J.M., 1977. Methods for the Analysis of Samples of Benthic Invertebrates. Freshwater Biological Association, Scient. Publ. No. 25, Ferry House, U.K., 160 pp. Errington, J.C., 1973. J. Ecol., 61, 99–105. Fairweather, P.G. & Underwood, A.J., 1983. Oecologia (Berl.), 56, 169–179. Fasham, M.J.R., 1978. Oceanogr. Mar. Biol. Ann. Rev., 16, 43–79. Fasham, M.J.R., Angel, M.V. & Roe, H.S.J., 1974. J. exp. mar. Biol. Ecol., 16, 93–112. Findlay, S.E.G., 1981. Estuar. cstl shelf Sci., 12, 471–484. Fisher, R.A., 1970. Statistical Methods for Research Workers. Oliver & Boyd, Edinburgh, 14th edition, 362 pp. Gage, J. & Geekie, A.D., 1973a. Mar. Biol., 19, 41–53. Gage, J. & Geekie, A.D., 1973b. Mar. Biol., 20, 89–100. Gagnon, M. & Lacroix, G., 1982. J. exp. mar. Biol. Ecol., 56, 9–22. Gärdefors, D. & Orrhage, L., 1968. Oikos, 19, 311–321. Gardiner, F.P. & Haedrich, R.L., 1978. Oecologia (Berl.), 31, 311–317.

66


Gerard, G. & Berthet, P., 1971. In, International Symposium on Statistical Ecology, Vol. 1. Spatial Patterns and Statistical Distributions, edited by G.P.Patil et al., Pennsylvania State University Press, Pennsylvania, pp. 59–67. Goodall, D.W., 1974. Vegetatio, 29, 135–146. Gove, P.B., 1969. Editor. Websters Third International Dictionary of the English Language. Merriam, Massachusetts, 2662 pp. Grassle, J.F., 1973. In, Biology and Geology of Coral Reefs, edited by R.Endean & O.A.Jones, Academic Press, New York, pp. 247–270. Gray, C.A. & Bell, J.D., 1986. Mar. Ecol. Prog. Ser., 28, 43–48. Gray, J.S., 1971. In, Proc. 1st Int. Conf. Meiofauna, Smithson. Contr. Zool. No. 76, edited by N.C.Huling, Smithsonian Institute Press, Washington, pp. 191–197. Green, R.H., 1966. Res. Popul. Ecol., 8, 1–7. Green, R.H., 1979. Sampling Design and Statistical Methods for Environmental Biologists. John Wiley & Sons, New York, 257 pp. Green, R.H., 1980. Can. J. Fish. Aquat. Sci., 37, 296 only. Greig-Smith, P., 1952. Ann. Bot. N.S., 16, 293–316. Greig-Smith, P., 1964. Quantitative Plant Ecology. Butterworths, London, 2nd edition, 256 pp. Greig-Smith, P., 1979. J. Ecol., 67, 755–779. Hairston, N.G., Hill, R.W. & Ritte, U., 1971. In, International Symposium on Statistical Ecology Vol. 1. Spatial Patterns and Statistical Distributions, edited by G.P.Patil et al., Pennsylvania State University Press, Pennsylvania, pp. 337–356. Hankin, D.G., 1984, Can. J. Fish. Aquat. Sci., 41, 1575–1591. Hargrave, B.T. & Burns, N.M., 1979. Limnol. Oceanogr., 24, 1124–1136. Harris, G.P., 1980. Can. J. Fish. Aquat. Sci., 37, 877–900. Harvey, P.H., Ryland, J.S. & Hayward, P.J., 1976. J. exp. mar. Biol. Ecol., 21, 99– 108. Haury, L.R., 1973. Limnol. Oceanogr., 18, 500–506. Heisig, D.M. & Hoenig, J.M., 1986. Limnol. Oceanogr., 31, 211–215. Helshe, J.F. & Ritchey, T.A., 1984. Biometrics, 40, 877–885. Hill, O., 1973. J. Ecol., 61, 225–235. Hogue, E.W., 1978. Mar. Biol., 49, 211–222. Holme, N.A., 1950. J. mar. biol. Ass. U.K., 29, 267–280. Holme, N.A. & McIntyre, A.D., 1971. Methods for the Study of Marine Benthos, IBP Handbook No. 16, Blackwell, Oxford, 334 pp. Hulings, N.C. & Gray, J.S., 1971. In, A Manual for the Study of Meiofauna, Smithson. Contr. Zool., No. 78, edited by N.C.Hulings & J.S.Gray, Smithsonian Institute Press, Washington, pp. 16–24. Hurlbert, S.H., 1984. Ecol. Monogr., 54, 187–211. Irish, A.E. & Clarke, R.T., 1984. Brit. Phycol. J., 19, 57–66. Johnson, R.B. & Zimmer, W.J., 1985. Ecology, 66, 1669–1675. Jones, J.G., 1974. Limnol. Oceanogr., 19, 540–543. Jumars, P.A., 1975a. Mar. Biol., 30, 253–266. Jumars, P.A., 1975b. Mar. Biol., 30, 245–252. Jumars, P.A., Thistle, D. & Jones, M.L., 1977. Oecologia (Berl), 28, 109–123. Kathirgamatamby, N., 1953. Biometrika, 40, 225–228. Keast, A. & Harker, J., 1977. Environ. Biol. Fish., 1, 181–188. Kenchington, R.A., 1978. In, Coral Reefs: Research Methods, Monogr. Oceanogr. Methodol. No. 5, edited by D.R.Stoddart & R.G.Johannes, UNESCO, pp. 149–161 . Kennelly, S.J. & Underwood, A.J., 1984. J. exp. mar. Biol. Ecol., 76, 67–78. Kennelly, S.J. & Underwood, A.J., 1985. J. exp. mar. Biol. Ecol., 89, 55–67. Kershaw, K.A., 1957. Ecology, 38, 291–299. Kershaw, K.A., 1970. Ecology, 51, 729–734.


67

Kimmel, J.J., 1985. Environ. Biol. Fish., 12, 23–32. Kinzie, R.A. & Snider, R.H., 1978. In, Coral Reefs: Research Methods, Monogr. Oceanogr. Methodol. No. 5, edited by D.R.Stoddart & R.G.Johannes, UNESCO, pp. 231–250. Kirchman, D., Sigda, J., Kapuscinski, R. & Mitchell, R., 1982. Appl. Env. Microbiol., 44, 376–382. Kosler, A., 1968. Mar. Biol., 1, 266–268. Lafedes, R.N., 1978. Editor, McGraw-Hill Dictionary of Scientific and Technical Terms. McGraw-Hill, New York, 1829 pp. Leatherwood, S., Gilbert, J.R. & Chapman, D.G., 1978. J. Wildl. Mgmt, 42, 239– 250. Lee, W.Y. & McAlice, B.J., 1979. Estuar. cstl mar. Sci., 8, 565–582. Lewis, J.R., 1976. Oceanogr. Mar. Biol. Ann. Rev., 14, 371–390. Lincoln, R.J., Boxhall, G.A. & Clark, P.F., 1982. A Dictionary of Ecology, Evolution and Systematics. Cambridge University Press, Cambridge, 298 pp. Loya, Y., 1978. In, Coral Reefs: Research Methods, Monogr. Oceanogr. Methodol. No. 5, edited by D.R.Stoddart & R.G.Johannes, UNESCO, pp. 197–217. Ludwig, J.A. & Goodall, D., 1978. Vegetatio, 38, 49–59. Lussenhop, J., 1974. J. Hist. Biol., 7, 319–337. Mackay, A.P., Cooling, D.A. & Berrie, A.D., 1984. J. appl. Ecol., 21, 515–534. McIntyre, A.D., 1971. In, Proc. 1st int. Conf. Meiofauna, Smithson. Contr. Zool., No. 76, edited by N.C.Huling, Smithsonian Institute Press, Washington, pp. 149–154. Mead, R., 1974. Biometrics, 30, 295–307. Millard, S.P. & Lettenmaier, D.P., 1986. Estuar. cstl shelf Sci., 22, 637–656. Millard, S.P., Yearsley, J.R. & Lettenmaier, D.P., 1985. Can. J. Fish. Aquat. Sci., 42, 1391–1400. Montagna, P.A., 1982. Appl. Environ. Microbiol., 43, 1366–1372. Morin, A., 1985. Can. J. Fish. Aquat. Sci., 42, 1530–1534. Morisita, M., 1959. Mem. Fac. Sci., Kyushu Univ., Ser. E (Biol.), 2, 215–235. Morisita, M., 1962. Res. Popul. Ecol. Kyoto Univ., 4, 1–7. Morisita, M., 1971. Res. Popul. Ecol. Kyoto Univ., 13, 1–27. Mountford, M.D., 1961. J. Ecol., 49, 271–275. Nie, H.W. de & Vijverberg, J., 1985. Hydrobiologia, 124, 3–11. Omori, M. & Hamner, W.M., 1982. Mar. Biol., 72, 193–200. Oviatt, C.A. & Nixon, S.W., 1973. Estuar. cstl mar. Sci., 1, 361–378. Payandeh, B., 1970. For. Sci., 16, 312–317. Pechanec, J.F. & Stewart, G., 1940. J. Am. Soc. Agron., 32, 669–682. Perry, J.N. & Mead, R., 1979. Biometrics, 35, 613–622. Pielou, E.C., 1959. J. Ecol., 47, 607–613. Pielou, E.C., 1969. An Introduction to Mathematical Ecology. Wiley, New York, 286 pp. Pielou, E.C., 1979. Oecologia (Berl.), 44, 143–144. Pielou, E.C. & Routledge, R.D., 1976. Oecologia (Berl.), 24, 311–321. Pihl, L. & Rosenberg, R., 1982. J. exp. mar. Biol. Ecol., 57, 273–301. Platt, T. & Denman, K.L., 1975. Ann. Rev. Ecol. Syst., 6, 189–210. Pot, W., Noakes, D.L.G., Ferguson, M.M. & Coker, G., 1984. Hydrobiologia, 114, 249–254. Pringle, J.D., 1984. Can. J. Fish. Aquat. Sci., 41, 1485–1489. Resh, V.H., 1979. J. Fish. Res. Bd Can., 36, 290–311. Ripley, B.D., 1978. J. Ecol., 66, 965–981. Rohlf, F.J. & Archie, J.W., 1978. Ecology, 59, 126–132. Rosenberg, R., 1974. J. exp. mar. Biol. Ecol., 15, 69–80. Rotenberry, J.T. & Wiens, J.A., 1985. Am. Nat., 125, 164–168. Russell, B.C., Talbot, F.H., Anderson, G.R.V. & Goldman, B., 1978. In, Coral Reefs: Research Methods, Monogr. Oceanogr. Methodol., No. 5, edited by D.R.Stoddart & R.G.Johannes, UNESCO, pp. 329–345.

68


Saila, S.B., Pikanowski, R.A. & Vaughan, D.S., 1976. Estuar. cstl mar. Sci., 4, 119– 128. Sale, P.F., 1980. Oceanogr. Mar. Biol. Ann. Rev., 18, 367–421. Sale, P.F. & Douglas, W.A., 1981. Environ. Biol. Fish., 6, 333–339. Sale, P.F. & Sharp, B.J., 1983. Coral Reefs, 2, 37–42. Scheaffer, R.L., Mendenhall, W. & Ott, L., 1979. Elementary Survey Sampling. Wadsworth Publishing Company, Belmont, California, 2nd edition, 278 pp. Scheffe, H., 1959. The Analysis of Variance. Wiley, New York, 477 pp. Scheibling, R.E., 1980. Mar. Ecol. Prog. Ser., 2, 321–327. Schweigert, J.F. & Sibert, J.R., 1983. Can. J. Fish. Aquat. Sci., 40, 588–597. Schweigert, J.F. & Sibert, J.R., 1984. Can. J. Fish. Aquat. Sci., 41, 827–828. Schweigert, J.F., Haegele, C.W. & Stocker, M., 1985. Can. J. Fish. Aquat. Sci., 42, 1806–1814. Seapy, R.R. & Kitting, C.L., 1978. Mar. Biol., 46, 137–145. Seber, G.A.F., 1982. The Estimation of Animal Abundance. Griffin, London, 654 pp. Simberloff, D., 1979. Ecology, 60, 679–685. Sinclair, D.F., 1985. Ecology, 66, 1084–1085. Skellam, J.G., 1952. Biometrika, 39, 346–362. Smith, H.F., 1938. J. agric. Sci., 28, 1–23. Smith, S.J., 1984. Can. J. Fish. Aquat. Sci., 41, 826–827. Snedecor, G.W. & Cochran, W.G., 1980. Statistical Methods. The Iowa State University Press, Iowa, 7th edition, 507 pp. Sokal, R.R. & Rohlf, F.J., 1981. Biometry. Freeman Co., New York, 2nd edition, 859 pp. Southward, A.J. & Bary, B.M., 1980. J. mar. biol. Ass. U.K., 60, 295–311. Southwood, T.R.E., 1966. Ecological Methods. Chapman & Hall, London, 391 pp. Stimson, J., 1974. Ecology, 55, 445–449. Stoddart, D.R., 1969. Biol. Rev., 44, 433–498. Stretch, J.J., 1985. J. exp. mar. Biol. Ecol., 91, 125–136. Sweatman, H.P.A., 1985. Ecol. Monogr., 55, 469–485. Sykes, J.B., 1982. Editor. The Shorter Oxford English Dictionary. Oxford University Press, Oxford, 7th edition, 1368 pp. Taylor, L.R., 1961. Nature, Land., 189, 732–735. Taylor, L.R., 1971. In, International Symposium on Statistical Ecology. Vol. 1. Spatial Patterns and Statistical Distributions, edited by G.P.Patil et al., Pennsylvania State University Press , Pennsylvania, pp. 357–377. Taylor, L.R., 1980. Can. J. Fish. Aquat. Sci., 37, 1330–1332. Taylor, L.R., Woiwod, I.P. & Perry, J.N., 1978. J. anim. Ecol., 47, 383–406. Taylor, L.R., Woiwod, I.P. & Perry, J.N., 1979. J. anim. Ecol., 48, 289–304. Taylor, W.D., 1980. Can. J. Fish. Aquat. Sci., 37, 1328–1329. Thompson, H.R., 1958. Aust. J. Bot., 6, 322–342. Todd, C.D., 1978. J. anim. Ecol., 47, 189–203. Toft, C.A. & Shea, P.J., 1983. Am. Nat., 122, 618–625. Underwood, A.J., 1976. Oecologia (Berl.), 26, 257–266. Underwood, A.J., 1977. J. exp. mar. Biol. Ecol., 26, 191–201. Underwood, A.J., 1978. Oecologia (Berl.), 36, 317–326. Underwood, A.J., 1981. Oceanogr. Mar. Biol. Ann. Rev., 19, 513–605. Underwood, A.J. & Denley, E.J., 1984. In, Ecological Communities: Conceptual Issues and the Evidence, edited by D.R.Strong et al., Princeton University Press, Princeton, pp. 151–180. Upton, G.J.G., 1984. Biometrics, 40, 760–766. Usher, M.B., 1975. J. Ecol., 63, 569–586. Van Vleet, E.S. & Williams, P.M., 1980. Limnol. Oceanogr., 25, 764–770. Venrick, E.L., 1971. Limnol. Oceanogr., 16, 811–818.


69

Watling, L., Kinner, P.C. & Maurer, D., 1978. J. exp. mar. Biol. Ecol., 35, 109–118. Weinberg, S., 1981. Bijdragen tot de Dierkunde, 51, 199–218. Wiebe, P.H., 1971. Limnol. Oceanogr., 16, 29–38. Wiebe, P.H., 1972. J. Cons. perm. int. Explor. Mer, 34, 268–275. Wiebe, P.H. & Holland, W.R., 1968. Limnol. Oceanogr., 13, 315–321. Wiens, J.A., 1986. In, Community Ecology, edited by J.Diamond & T.J.Case, Harper & Row, New York, pp. 154–172. Wiens, J.A., Addicott, J.F., Case, T.J. & Diamond, J., 1986. In, Community Ecology, edited by J.Diamond & T.J.Case, Harper & Row, New York, pp. 145–153. Williams, B., 1978. A Sampler on Sampling. John Wiley & Sons, New York, 254 pp. Wilson, J.G., 1976. Mar. Biol., 37, 371–376. Winer, B.J., 1971. Statistical Principles in Experimental Design. McGraw-Hill Kogakusha, Tokyo, 2nd edition, 907 pp. Yamane, T., 1967. Statistics: an Introductory Analysis. Harper & Row, New York and John Weatherhill Inc., Tokyo, 919 pp. Youngbluth, M.J., 1982. J. exp. mar. Biol. Ecol, 61, 111–124.


FLUMES: THEORETICAL AND EXPERIMENTAL CONSIDERATIONS FOR SIMULATION OF BENTHIC ENVIRONMENTS* ARTHUR R.M.NOWELL and PETER A.JUMARS School of Oceanography, University of Washington, Seattle, WA 98195, U.S.A.

INTRODUCTION Flumes have been used to transport water since Roman times. The word flume is derived from Latin “fluere… to flow” and today means “an inclined channel for conveying water from a distance”. Flumes have been extensively used in sedimentary geology and civil engineering for approximately a century to examine the modes and rates of sediment transport. Their use in benthic biology resulted from the recognition that a variety of processes of biological interest were strongly influenced by water motions at the boundary with solid surfaces and that these processes in the bottom boundary layer can be modelled in flumes. The benthic boundary layer in the transport of scalar quantities such as nutrients plays an important rôle not only to benthic biologists and geochemists, but also to sedimentary geologists and boundary-layer fluid mechanicians. Our primary objective will be to review the characteristics of such flows and what constraints must then be placed on the design and operation of laboratory models of benthic boundary layers irrespective of the discipline to which the flume experiments are applied. As the application of flume techniques to biological problems in particular is not straightforward, our goal in this review is to present the fluid dynamic issues and compromises that govern the applicability and operation of laboratory flumes. Previous articles on flumes have focused on engineering considerations and construction techniques (Williams, 1971), on the differing types of flumes that are routinely used in sedimentary geology (Middleton & Southard, 1984), and on descriptions of specific flumes (for example, Vogel & LaBarbara, 1978; Nowell, Jumars & Eckman, 1981). Use of open-channel flumes in sediment transport studies ranges from early work by Osborne Reynolds in the 1880s, to the classical work by Gilbert (1914) and Shields (1935) to more recent studies by Guy et al. (1966), Grass (1971), and Sumer & Deigaard (1981). All these studies were concerned with the complex case of boundary-layer flow over a deformable (soft-sediment) bed, and the response of the bed to fluid forcing. All used fresh water and paid no attention to biota. Apart from the effort involved in filling the flumes, there were few limits on the sizes of the devices. Thus flumes used by these authors ranged up to 80 m in length. Once one incorporates a concern for the biota, especially for either controlling microbial abundance or for implanting a community of macrofauna and meiofauna, a severe pragmatic limitation is imposed on the size of the device. We take it as given that the purpose of a flume is not only to simulate realistic field conditions near the sea bottom, but also to simplify them so that the flow characteristics can be summarized in a small number

*University of Washington Contribution No. 1649.

FLUMES: SIMULATION OF BENTHIC ENVIRONMENTS

71

of parameters (i.e. those presented by Nowell & Jumars, 1984). Only then can the results in terms of effects of changes in those parameters be both replicated and generalized. There is no benthic flow environment simple enough to be characterized by a single variable, such as flow velocity (volumetric discharge divided by cross-sectional area or streamwise velocity at a given depth) along a flume or in nature, so there is no point in reviewing biological observations made where this was the only variable measured or used in matching field and laboratory flow regimes. The epitome of irreproducibility in this context is the taking of small-scale physical (e.g. Statzner & Holm, 1982) or chemical (e.g. Jørgensen & Revsbech, 1985) measurements about organisms without measuring or controlling the larger-scale fluid motions that determine these small-scale patterns. Nowell & Jumars (1984) review the important parameters, but do not show how to control them in a laboratory setting. Rules of thumb in flume design have been given previously in the marine ecological literature (Jumars & Nowell, 1984; Muschenheim, Grant & Mills, 1986) but their derivations have not. Our primary aim in this review is to expose the principles underlying the rules of thumb utilized in flume design, and to do so highly selectively. The problem of flume design can be broken into four parts, i.e. producing specific entrance conditions, providing certain exit conditions, driving the flow (including, in some cases, return of fluid with or without its particle load to the flume entrance), and lastly but most important tailoring the flow in the test section. Flows in the entrance, exit and drive sections are complex, defying simple and general parameterization, but conditions there are for the most part irrelevant to biological measurements of interest. Furthermore, the miniaturization often required of flumes for biological purposes usually makes no special demands here; the rules of thumb are reasonably independent of physical scale. For these reasons, we give brief treatment of these design considerations which are often satisfied with simple, empirically derived guidelines. Control of fluid dynamics in the test section is another matter. Here is where the serious and inevitable compromises of flume scale must be made. It is impossible to make them intelligently without understanding the underlying principles. We review those principles (of conservation of mass and momentum) by recourse to the basic Navier-Stokes equations, to which most oceanographers have at one or several times been exposed. Very readable introductions to such material, which focus on phenomenological understanding rather than mathematics, are presented by Bradshaw (1971), Francis (1975), van Dyke (1982), and Allen (1985). Finally, we exercise those principles by applying them to specific biological problems, including ones where it becomes obvious that laboratory flume work is impossible or impractical. We limit our scope explicitly to flumes for simulation of bottom boundary layers. Specifically we are excluding from consideration flow devices where the flow goes into solid-body rotation, such models generally being used to simulate geostrophic flows. We also exclude enclosed water tunnels for, although such devices are useful for measuring drag on bodies (e.g. fish fins or model submarine hulls) suspended far from a boundary, they are not useful without extensive corrections in simulating bottom boundary layers or in making drag measurements on objects attached to the bed or sidewalls. Thus, our major emphasis will be on flumes which permit us to study small-scale, viscous, and turbulent flows, most often flows with a single dominant velocity gradient. We consider in this review flows with a free surface, and in which stratification and rotation play negligible rôles. While the oceans are clearly stratified, rotating shear flows, the flow close to a body, and flows on the scales of importance to a single organism, or a local segment of a community, may be viewed as dominated by the relative balance of inertial, viscous, and body forces alone. As Shakespeare points out about life, and as more emphatically noted in “Trobriand Cricket”, by our entrances and exits shall we be remembered. The entrance conditions as well as the exit conditions can affect the nature of flow in the test section of a flume. What specific characteristics of fluid motion we need are specified by the question that we are trying to answer. How well we can answer the question will be

72

ARTHUR R.M.NOWELL AND PETER A.JUMARS

determined by how well we can tailor the boundary-layer flow in our flume to match the field situation. Our ability to mimic a field situation in the laboratory depends on selection of an appropriate flume design. ENTRANCE CONDITIONS The entrance to the flume is crucial for the development of simple flows. The problem can be viewed as getting the flow to ‘forget’ its recent history. Because the water usually is delivered by a pipeline which most likely has bends in it, the flow entering the flume often will have strong circulation from going round a curve. In addition the flow may have to diverge as it enters the channel. As an example of these problems consider the flow entering a flume from a 20 cm (diam.) pipe having just turned through 90° and diverging rapidly into a channel that is 30 cm wide. Clearly the flow from the pipe will form a centre jet which will take many pipe diameters to decay (based on many observations downstream of obstacles the flow will ‘remember’ for about 20 times the critical length scale which in this case would be pipe diameter). We could, therefore, not expect to develop a simple boundary layer for over 4 m. A smooth, gradual divergence at the entrance is required to avoid separation of the flow. Experience from the design of ship hulls shows that expansion angles exceeding 7° often show flow separation (Chang, 1970), so this angle should not be exceeded without detailed examination of the consequences. Rapid expansions (e.g. O’Brien, Tay & Zwart, 1986) lead to free jets and complex flows that are virtually impossible to describe in detail, again diminishing the value of bringing the problem into the laboratory. One common solution to the problem of removing the effects of the pipe flow and upstream delivery of the fluid is to force it through a diffuser aimed at the far upstream wall of the flume. The motion is broken into small scales and the flow distributed across the full width of the flume. Honeycomb grids downstream of the diffuser will break up remaining scales of motion that are slightly bigger than the grid scale, but such grids cannot markedly affect flow non-uniformity on the scale of the whole channel diameter. The use of grids for generating small-scale turbulence has a long and well-recorded history (Laws & Livesy, 1978) and it has become almost a shibboleth of flume building to include a screen at the entrance. Laws & Livesy (1978) detail the calculations required to estimate the effects of screens on the flow, but we note that in essence screens only help to tailor an already well-cut flow; they cannot generate smooth boundary layer velocity profiles out of non-uniform jets. Moreover in free-surface flows such as flumes, the more dense the screen, the bigger the pressure drop across it; consequently, a very dense or a very long section of mesh will result in a downstream hydraulic jump that propagates surface waves downstream, causing non-uniformity in the boundary layer. As a general rule the tubes comprising the mesh should be about 20 times longer than the mesh diameter in order to apply a sufficiently strong strain to the fluid to alter the incoming turbulence. Elegant mesh designs have been developed for wind tunnels in order to pre-shape the flow into a boundarylayer profile, but such designs require considerable care and many hours of computing. Furthermore, they are well suited to but a single flow rate. If a fully rough-turbulent boundary layer is desired, its development can be accelerated or ‘tripped’ by placing small-scale roughness elements across the flume bed on the downstream side of the honeycomb grid. EXIT CONDITIONS The exit problem can be viewed as not letting the flow know what is coming, i.e. of making a smooth exit without breaking cadence. At the exit of the test section the flow can either fall freely, be retained by a weir, or be directed through tail gates. In the case of a weir, especially if the flow is supercritical (with respect to Froude number defined below), upstream effects are clearly visible. The flow is suddenly deepened having


73

to go over the gate, and as a consequence a hydraulic jump may occur just upstream of the gate, or a backwater slope is developed (Henderson, 1966). But upstream effects also exist in subcritical situations and the weir causes the flow to diverge from the bed upstream of the gate. This effect clearly imposes extra length on our flume as it reduces the length of the useful test section. Similarly, a free overfall has an upstream effect at subcritical Froude numbers as the flow close to the bed is accelerated on approaching the overfall. As a compromise to the two extremes of a weir and a free-overfall, louvred gates, much like venetian blinds (set either vertically or horizontally) provide a minimum of upstream interference. A good picture of such gates is provided by Yalin (1977). DRIVING THE FLOW Flumes are used because we wish to control the mass or momentum flux or the boundary stress. Such control is achieved by manipulating discharge (volume per cross-sectional area per time) through the flume. Discharge is regulated either by using a pump and returning the fluid to the entrance, or by using a constanthead tank. For the types of velocities that are most commonly needed in oceanographic studies, a constanthead tank provides an inexpensive and effective method of providing a regulated, constant discharge. Pumps are often necessary for high discharges but require considerable attention to avoid problems of unsteadiness at low velocities (for example picking up the blade frequency from the impellor) and varying discharge at high rates when small voltage changes may cause large changes in effective discharge rate. Most flows of concern have boundary layer Reynolds numbers (free stream velocity times boundary layer thickness divided by kinematic viscosity) greater than 105, and because the fluid is well mixed through the return pipes and pumps we rarely have to worry about fluid stratification. At low-flow Reynolds numbers, however, fluid convection can cause numerous problems in the laboratory. Low Reynolds numbers of settling particles, such as larvae, provide a challenging design problem. Obtaining a totally still body of water is more complex than usually imagined and many settling columns are plagued with problems of convection. Although the velocities of convection are low, so are the velocities of larval settling. Temperature changes can also occur in flumes due to heating of the water as it passes through the pump; because of the high thermal capacity of water this only becomes a problem in very small flumes when short recirculation times may occur. Temperature variations should always be monitored for small changes in temperature can often markedly affect velocity sensors using heat transfer rate to infer velocity. TEST SECTIONS At last the stage is set. Well characterized, reproducible flow in the test section is the goal for the design of any flume. The test section should not be considered as the distance left over after we have moved far enough away from the entrance to avoid non-uniformity or advective effects and far enough upstream to miss the results of the flow leaving the flume. Rather the test section should be long enough to allow simulation of the flow field of interest. Unfortunately, unless one has carried out the calculations as to one’s needs before one either builds the flume or tries to use an already existing device, one may end up with no useful test section at all. MASS AND MOMENTUM BALANCE To obtain a valid simulation of any flow there is a hierarchy of similarity that needs to be considered. At the simplest level we wish to maintain geometric similarity; that is, the shape of the boundary (the bed in most

74


cases) should be the same from field to laboratory. Because the perimeter of a body in a flow where the noslip condition (zero velocity immediately at the body’s surface) applies is also a streamline, we will maintain kinematic similarity if we keep the velocity the same between laboratory and field. We usually wish, however, not only to maintain geometric and kinematic similarity but also to model the fluxes of material and momentum, in other words we wish to maintain similarity of forces, or dynamic similarity. To do so reliably we shall treat the equations that express the balance of forces within the flow; we shall use Newton’s second law. The governing equations are the continuity equation for an incompressible fluid (1) and the conservation of momentum equation (2) which may be written out in its scalar form as

The velocity is represented by ui, which has three components u, v and w, t is time, p pressure, g gravity, is density and and v are the dynamic and kinematic viscosity (and ) The terms on the left of the Navier-Stokes Equation 2 represent, respectively, the local and advective rate of change of momentum. They are balanced on the right hand side of the equation, respectively, by the pressure gradient, the viscous forces and the body force (which we take to be gravity). Equation 2 is just the familiar force balance F=ma written for a fluid parcel. Another way to conceptualize each term is as a force (MLT−2) per unit volume (L −3) of the water parcel, again yielding the proper units for each term (ML−2T−2). General solutions of these equations are not available, and simplification of them is achieved only if we can apply certain restrictions. To understand the physics represented by these equations we want to transform each of the terms into a nondimensional form. Let us define


75

where L is an independent length scale, V is an independent velocity scale and is an independent time for dimensional uniformity. We substitute these terms into the scale and we divide the pressure by Equation 2:

and multiplying through by L/V2

Each of the terms is now dimensionless, whereas in Equation 2 it had the dimension of mass/unit volume through multiplying the various times acceleration. There are four dimensionless groups, labelled terms. Term one, a frequency parameter, is called the Strouhal number, and is used as a measure of the unsteadiness of the flow. Most commonly it is used to relate periodic vortex shedding from cylinders to the velocity incident upon and diameter of the cylinder (Tritton, 1977), but its general interpretation is as a measure of unsteadiness. When the Strouhal number is zero the flow is steady and values rarely exceed unity for even the most unsteady cases. For example, in a tidal flow in an estuary where the tidal frequency seconds, the maximum velocity of order 50 cm·s-1 and the flow depth say 10 m, the Strouhal is thus modelling such a boundary-layer flow in the laboratory could be done using a number is quasi-steady flow especially if we were interested in the response of the bed, which is nearly instantaneous to changes in stress. As another example, consider shallow-water wind waves in the intertidal at a depth of 1 m where the mean flow velocity may be of order 20 cm·s−1, and wind waves may have a period of 5 s. Such parameters will result in a Strouhal number of 1·0. Clearly it would be unwise to attempt to model in the laboratory diffusional processes in such regions without including the unsteady term. By keeping the Strouhal number approximately the same between laboratory and field we ensure that we are correctly reproducing the kinematics of interest from the field. Term two is the ratio of gravitational to inertial forces, and is the inverse of the Froude number. It is important whenever we must consider free-surface effects, i.e. when the boundary layer extends through the complete flow depth, or when surface waves enter the problem. The Froude number, which as Lu (1977) notes is named after William Froude who ‘retired’ at age 36 to do research on ship rolling and resistance, also represents the ratio of the mean flow velocity to the phase speed of a shallow-water wave. As such it distinguishes between two regimes. Subcritical or tranquil flow has a Froude number less than unity, and the flow is deep and slow; information and boundary layer effects from downstream may be transmitted upstream by surface waves (viz. Lu, 1977). When the Froude number is greater than one, supercritical or shooting flow exists and information is transmitted only downstream. The Froude number is thus crucial when we attempt to model flows in estuarine and intertidal areas. Flows over most tidal flats are subcritical most of the time, although occasionally in run-off channels the flow is critical or supercritical, and standing waves may be observed at the transition.

76


The third term is the familiar Reynolds number (actually its inverse) which is the ratio of inertial to viscous forces. The ratio of terms two and four in Equation 2 namely the inertial to viscous terms may be written as

which dimensionally is

We see that changing the Reynolds number then represents changing the balance of force from two terms in the Navier Stokes equation. In order to model flows dynamically we must maintain similarity of forces; hence all the dimensionless groups must be kept numerically similar from field to laboratory. The fourth dimensionless group, called the Euler number, is a measure of the pressure gradient, and is of consequence mainly in flows where gravity plays only a small rôle and the flow is being driven by a pressure drop, such as in flow through a horizontal pipe. We thus have to consider three dimensionless groups when we attempt to model flow in the laboratory. If we consider only steady flow (no changes with mean flow parameters with time, however), then we can dispense with consideration of the Strouhal number. Flumes are traditionally very long relative to their widths, and customarily are operated with flow depths which are shallow relative to their widths. The reasons for this design can be seen if we take the Navier Stokes equations and simplify them. Let us consider a steady flow with no gradients in the cross-stream direction, much like the flow in a wide, straight river or the flow over a wide sand flat. With no gradients in the y-direction the steady-flow equations reduce to conservation equations for mass (3) and, dividing through by and ignoring gravity, for momentum (4) (3)

(4)

If we consider a flow over a flat bed and apply the no-slip condition at the wall (at z=0, U=0) then the only geometric length scale is the distance x from the leading edge. Observation of such flows shows that the is very small, so that If we let the x component of velocity be thickness of the boundary layer of order u, and d/dx be of order 1/x then du/dx is of order u/x and from the continuity equation dw/dz must be of the same order. Now as w is smaller than u and since d/dz is larger than d/dx then the order of dw/dz may


be met by considering w to be of order order

and d/dz of order

77

Then using the symbol 0( ) to mean of

we consider the order of the terms in the steady two-dimensional equations as (5)

(6) Equations 5 and 6 are termed the thin shear layer equations (Cebeci & Bradshaw, 1973) and can be solved to give the growth rates of boundary layers. They can be integrated over the boundary layer thickness to give the flux of momentum to the bed, that is the momentum integral yields the flux of momentum to the bed, or the stress acting on the bed. Now the two acceleration terms are of the same order in each equation, but the first viscous term in Equation 5 is very much smaller than the second viscous term so can be neglected. Since fluid parcels may be accelerated in boundary layers, and since strong viscous effects exist (having imposed the no-slip condition at the wall) the dominant viscous term is assumed to be of the same order of magnitude as the inertial terms, i.e.

This relation can be rearranged so that we see

Thus from a purely order-of-magnitude analysis we deduce that a viscous boundary layer thickness increases All the terms in Equation 6 are smaller than in Equation 5 and so to order may be ignored. as The shear-layer equations then become (7)

(8) The exact growth rate of a boundary layer can be predicted by using Equations 7 and 8 and solving for any imposed velocity profile, as done in many textbooks (cf. Shames, 1967) exactly for laminar flow and in an

78


approximate manner for turbulent flows. The resulting growth laws show that boundary layers are very thin relative to their lengths of development, and thus we need flumes that are long in order to develop boundary layers of reasonable thickness. Table I gives some typical thickness for viscous and turbulent boundary layers for commonly used velocities. Because our flume must have sides, the numerical artifice of saying that we have purely two-dimensional flow is not empirically obtainable. Boundary layers grow just as quickly on the sides of the flume as on the bed, as long as the roughness is the same. Because in the field the assumption of two-dimensional flow may often be fulfilled, we must attempt to make our flumes as wide as feasible. Increasing the width of the flume puts immediate demands on the pumping capacity, because the boundary layer must be fairly thick in TABLE I Upper part gives boundary layer thickness (in cm) for various lengths of channel, lower part gives corresponding boundary layer Reynolds numbers : the viscous cases, in which Re 90

1000

Sampled once

250

Intertidal

1000

4 months

20

0–117

1000

Sampled once

710

18–>60

800

Sampled once

175

5–11 500, 1000 20–60 1000 Intertidal 500 Intertidal 500, 1000 3–9 1000

5 months 310 Sampled once 100 3 months 90 1 month 4000 3 months 280

5–20

1000

Sampled once

55

Intertidal 0–9 22–131 13–70

250 1000 500 400

1 month Sampled once 1 month 2 months

65 NG 6390 500

38–168

300

3 months

500

97

scales (tens of metres to tens of kilometres) separating distinct assemblages and sediment types were, in part, dictated by the manoeuvrability of the sampling vessel and the accuracy of shipboard-operated navigational equipment; minimum distances between subtidal stations ranged from 50 m to 64 km, while intertidal communities could be sampled at closer intervals of 5 m to 800 m (Table III). In addition, once a relationship between species and sediment composition was observed, delimiting species distributions in relation to sediment type became the primary purpose of most benthic surveys, so relatively large distances between stations were desirable, since significant differences in bulk properties of sediments (e.g. grain size) could be easily detected at these spatial scales. As the topic of organism-sediment relations was experimentally dissected through the years, nearly all of the field and a good portion of the laboratory research was on the favourability of particular habitats to adults and on the interactions between different trophic and mobility types (Gray, 1974; Rhoads, 1974). Detailed studies of the feeding and mobility types of the infauna revealed that functional groups of organisms occurred in distinct types of sediment. Most authors did not speculate on larval settlement mechanisms which could have produced these patterns of distribution (in fact, larval settlement is not discussed at all in Rhoads’, 1974, review), but only discussed the favourability of these particular habitats to adults. Initially, the most popular explanation for these assemblages concerned the availability of food resources. For example, Sanders (1958) hypothesized that deposit-feeders dominate clays because these sediments are also rich in organics and microbes, while filter-feeders occur in sandier environments because the higher near-bottom flows deliver suspended particulates to the organisms at faster rates. Later experimental manipulations showed that interactions between functional groups are also important. Rhoads & Young’s (1970) classic “trophic group amensalism” hypothesis, for example, states that activities

98

CHERYL ANN BUTMAN

of deposit-feeding organisms interfere with the establishment and maintenance of populations of suspensionfeeders and that such amensalistic interactions are intimately related to the nature of the sedimentary environment (e.g. the degree of substratum motion). Recently, more complex interactions between the feeding and locomotive activities of benthic organisms and the structure of the bottom boundary-layer flow and sediment-transport regimes have been identified, stipulating a re-evaluation of the effects of functional groups on sediments and sediment transport (Jumars, Nowell & Self, 1981; Jumars & Nowell, 1984a) and of fluid- and sediment-dynamic effects on community structure (reviewed by Jumars & Nowell, 1984b). Few of these studies considered how the functional groups of organisms are initially established. The studies usually did not ask if the distinct assemblages resulted from differential larval settlement or differential post-settlement mortality, nor did they consider how the mechanisms controlling larval settlement (e.g. active habitat selection or passive deposition) would affect the establishment and maintenance of the assemblages (see also Dayton & Oliver, 1980). The licence to focus primarily on adults may have resulted because, concurrent with the early survey studies, a relatively small core of biologists (e.g. see studies cited in Table V, see pp. 128–9) conducted meticulous laboratory experiments on infaunal larvae and meiofauna, demonstrating that the organisms can actively choose between microhabitats. Thus, the rôle of larval settlement in creating the observed organism-sediment relations was generally assumed to be through active habitat selection (e.g. Thorson, 1957; Wilson, 1958; Meadows & Campbell, 1972a; Gray, 1974), even though scant direct evidence from the field was available to support this tenet (see later discussion, pp. 139–141). In fact, passive deposition of larvae also could have produced the observed patterns of organism distribution if, (1) larvae were deposited over broad areas, but differentially survived only in hospitable adult habitats (corresponding to particular sedimentary environments), or (2) speciesspecific larval fall velocities corresponded with particular sediment fall velocities so that hydrodynamically similar particles and larvae were deposited in the same environment. Interpreting the importance of amensalism or other interactions between established infauna and the flow or sediment environment to benthic community structure requires knowledge of the rôle of larval settlement processes (see also Jumars & Nowell, 1984b). For example, the trophic group amensalism hypothesis requires that initial distributions of larvae on the sea bed result from differential larval settlement, due to active habitat selection, or to differential post-settlement survival. If differential larval settlement results from passive deposition (i.e. settlement patterns depend on larval fall velocities and on the near-bottom flow regime), then it may not be necessary to evoke complex amensalistic interactions to explain the distributions of the adults. Thus, for example, suspension-feeders may not co-occur with deposit-feeders simply because the two functional groups have larvae with different fall velocities that are passively deposited in different fluid-dynamic environments. In both the survey and the process-orientated studies of soft-bottom community structure, the importance of larval ecology cannot be assessed a posteriori because larvae were rarely quantitatively collected in samples (Tables III and IV). Two methodological problems have especially prohibited an adequate consideration of the larval stages; Dayton & Oliver (1980), Santos & Simon (1980a), and Williams (1980) have discussed these problems. (1) Field sampling was usually too infrequent (monthly or even biweekly) to record initial settlement prior to post-settlement interactions. (2) The sieve screen size (500 µm) commonly used in recent benthic studies is too large to retain newly settled larvae of most invertebrate species. Even though the sieve screen size used in faunal surveys has decreased over time (note that Thorson, 1966, defined macrofauna as those organisms retained on a 2-mm sieve and meiofauna was originally defined by Mare, 1942, as organisms with body lengths between 0·2 and 2·0 mm), so that 300-µm screens are used in some contemporary survey studies (e.g. Grassle et al., 1985; Thistle, Yingst & Fauchald, 1985; Maciolek &

LARVAL SETTLEMENT OF SOFT-SEDIMENT INVERTEBRATES

99

Grassle, in press), 60- or 100-µm screens often are required to retain newly settled larval (or postlarval) stages (Eckman, 1979; Gallagher, Jumars & Trueblood, 1983; Hannan, 1984a). THE RÔLE OF LARVAL SETTLEMENT In most discussions of the rôle of larval settlement in soft-substratum community ecology (e.g. Thorson, 1946, 1950, 1957, 1966; Smidt, 1951; Muus, TABLE IV Process-orientated field studies of factors controlling soft-substratum community structure: studies are arranged by the process under investigation; included in this table are studies of processes structuring macrofaunal communities, but not studies of single species populations; this list includes the commonly cited studies in the English literature and is not intended to be comprehensive; NS=some of the samples were not sieved Reference Colonization, succession, response to disturbance Grassle & Grassle (1974) Boesch, Diaz & Virnstein (1976) Dauer& Simon (1976) McCall (1977) Rees, Nicholaidou & Laskaridou (1977) Rhoads, Aller & Goldhaber (1977) VanBlaricom (1982) Woodin (1978) Oliver et al. (1980) Santos& Bloom (1980) Santos& Simon (1980a) Santos & Simon (1980b) Arntz & Rumohr (1982) Zajac & Whitlatch (1982a, b) Gallagher, Jumars & Trueblood (1983) Watzin (1983, 1986) Ambrose (1984b) Levin (1984) McGrorty & Reading (1984) Predation Young, Buzas & Young (1976) Reise (1978) Virnstein (1978) Arntz (1980) Holland et al. (1980) Hulberg& Oliver (1980) Mahoney & Livingston (1982) Animal-sediment relations

Minimum sampling interval

Sieve screen size (µm)

3 days 3 months 1 month 10 days 1 month 2 months 1 wk 1 month 1 month 1 month 1 wk 1 month 2 months 2 wk 2 days 7 days 2 months 3 days 6 months

297 500, 1000 500 297, 1000 1000 300, 1000 500 1000 500 500 250, 500 500 125, 500, 1000 297 63 63 500 250 500

1 month 2 wk 2 months 2 months 2 months 2 months 1 month

1000 500, 1000, NS 500 1000 500 250, 500 500

100

CHERYL ANN BUTMAN

Rhoads & Young (1970) Young & Rhoads (1971) Levinton (1977) Myers (1977a, b) Orth (1977) Wilson (1979) Brenchley (1981) Wilson (1981) Competition Woodin (1974) Peterson (1977) Weinberg (1979) Peterson & Andre (1980) Wilson (1983) Temporal variability Muus (1967) Lie (1968) Dauer& Simon (1975) Peterson (1975) Holland & Polgar (1976) Holland, Mountford & Mihursky (1977) Whitlatch (1977) Buchanan, Sheader & Kingston (1978) Ziegelmeier (1978)

1 month Sampled once 1 yr 1 wk 1 month 1 month 7 days 15 days

500, 1000 1000 2000 500 1000 500 500 500

1 month 1 day 1 month 55 days 1 wk

500, 1000 2300 500 2300 500

1 month 2 months 3 months 4 months 3 months 3 months 1 month 2 months 6 months

700, 1000, NS 1000 500 3200 1000 1000 250 500 3200

1973; Gray, 1974; Woodin, 1976, 1979, 1985; Oliver, 1979; Woodin & Jackson, 1979; Dayton & Oliver, 1980; Watzin, 1986), active habitat selection by larvae is the favoured mechanism for establishing benthic communities. Support for this hypothesis comes primarily from the numerous laboratory experiments where larvae were given a choice of substrata in which to settle (Tables I and V). Post-settlement mortality may also determine (e.g. Levinton & Bambach, 1970; Luckenbach, 1984) or further restrict the distribution of adults (e.g. Thorson, 1966; Muus, 1973; Oliver, 1979; Peterson, 1986; Watzin, 1986). It is not surprising that active habitat selection has been the favoured larval-settlement process because the clear evidence from the early laboratory studies (see Table V) is difficult to ignore. In a few notable discussions, however, reservations were raised regarding the application of these laboratory results, where experiments were conducted at very small scales and in still water, to the field, where the scales of processes are much larger. While Thorson is frequently credited as an early advocate of active habitat selection, because of his observations of settling larvae associated with particular sediment types in “bottle collectors” (Thorson, 1946), he was, in fact, consistently cautious when applying results of laboratory


101

experiments in still water to the field. For example, regarding the choice experiments of Wilson (1952, 1953a, b), Thorson (1966, p. 275) noted that the experiments were done, “… in petri-dishes, where the larvae by swimming 1 or 2 centimeters only had a chance to discriminate between heaps of sand which might be more or less attractive, neutral or more or less repellent to them as a future substratum. In nature, the larvae will not get a similar opportunity to compare a series of substrata by swimming a short distance only…. Far from questioning Wilson’s main thesis: That the larvae may discriminate between attractive and non-attractive substrata, a fact shown so convincingly that it can be accepted as a ‘biological rule’, we have, however, to find out what will happen in nature, when a larval swarm ready to metamorphose and drifting along the bottom will for the first time meet a substratum which they might ‘accept’, although it is far from ideal for their settling. The larvae cannot know, that if they continued to drift over the bottom for perhaps 10–20 kilometers more, they might meet a much more attractive substratum. It seems reasonable to assume, that such larvae, at least if they have already postponed their metamorphosis for some time and are in their less critical phase, will accept and accordingly settle in a bottom substratum much less attractive than the one they would have preferred, had they been given a ‘free choice’. “The consequence of this must be, that the distributional pattern of larvae on the natural bottom substrata must be much less delicate, i.e. much more coarse, than in the experiments undertaken in the laboratory.” Thorson was impressed with the behaviour of dispersing and settling larvae, but he also acknowledged that test sites for larvae on the sea bed were probably dictated by near-bed currents, so he tended to under- rather than over-state the rôle of active habitat selection. The evidence that larvae can test the substratum and have preferred habitats simply indicates that “…their chance of finding a suitable place for settling is much better than hitherto believed” (Thorson, 1950, p. 36). One of the most lucid commentaries on the extent to which active habitat selection determines the distribution of benthic marine organisms is the brief (and infrequently cited) paper by Moore (1975), which was written in response to the views of Meadows & Campbell (1972a, b) and Meadows & Mitchell (1973). Moore (1975) proposed “habitat availability” and “ecological opportunity” as alternative arguments to active habitat selection and animal behaviour for explaining the “local” distribution of organisms in the sea. He reemphasized one of Thorson’s (1966) points, that organisms may not have the same kinds of “free choices” in the field as they have been given in the laboratory. During dispersal, planktonic larvae are restricted to particular localities by passive transport processes so that larvae may never even encounter preferred substrata (as determined in laboratory experiments) in the field. Post-settlement mortality or passive deposition of larvae may then shape species distributions. Moore (1975) also reiterates the postulate of Pratt (1953; discussed more later, see pp. 141–3) that correlations between the composition of softsubstratum communities and sediment type may also result from passive sorting of both larvae and sediments by hydrodynamical processes. Furthermore, Moore (1975) emphasizes the importance of scale in directly applying the habitat-selection results to the field, stating that “local” to a behaviouralist (e.g. Meadows & Campbell, 1972a, b; Meadows & Mitchell, 1973) may refer to a much smaller scale (i.e. onthe-order-of the organism) than the “local” of an ecologist, which generally refers to more geographicaltype scales; the disparity between these scales decreases, however, with increasing organism size and their ability to independently traverse large distances. His concluding remarks (Moore, 1975, p. 100) raise questions that are still relevant, and largely unanswered, today: “Re-examining the generality of Meadows and Campbell’s statement that habitat selection largely determines the local distribution of animals in the sea, a number of issues appear conditional, (i) how is ‘determine’ construed? (ii) how is the concept of ‘local’ envisioned? and (iii) which type of ‘animal’ is involved with reference to (ii)? But in any circumstances, to regard habitat selection as ‘largely’ determining local distribution would seem to be an overstatement of the case.”

102

CHERYL ANN BUTMAN

A small number of benthic studies (e.g. Baggerman, 1953; Pratt, 1953; Fager, 1964; Tyler & Banner, 1977) have favoured the passive deposition, rather than the active habitat selection, hypothesis to account for some or all of the observed patterns of infaunal species distribution. Curiously, the passive deposition hypothesis was suggested to these authors by the same kinds of correlations between sediment and species composition, that led most other authors (cited previously) to conclude that larvae actively select for particular sediment or sedimentary environments. Support for the passive deposition hypothesis was largely correlative in these early studies (but see later discussion of experimental manipulations by Baggerman, 1953). Later studies have, however, shown that hydrodynamical null hypotheses are feasible explanations for the observed patterns of distribution; Jumars & Nowell (1984b) review some of this work. Now that the stage has been set with the alternative hypotheses of active habitat selection compared with passive deposition for creating observed patterns of species distributions, it is fruitful to examine closely the data base substantiating each of these views to determine the plausibility and scales of cause and effect. THE ACTIVE HABITAT SELECTION HYPOTHESIS Laboratory experiments on larval settlement can be divided roughly into two groups: (1) studies of habitat selection (i.e. where larvae were given a choice of substrata) and (2) studies of environmental or biological factors that induce metamorphosis. Results from experiments in the first category can provide direct evidence of habitat selection, while selection is only implied by results from experiments in the second category. There is some confusion in the literature as to which studies actually provide direct evidence of habitat selection (through choice experiments), so these are listed in Table V and will be discussed separately from the metamorphosis experiments. All laboratory studies of active habitat selection (i.e. the choice experiments) were done in still water, except one (Cuomo, 1985), so that relevance of these results to settlement in field flows is at present obscure. The response of settling larvae to water motion was qualitatively investigated in the laboratory for several infaunal larvae and meiofauna species. The polychaete larvae of Ophelia bicornis and Polydora ciliata were stimulated to attach to sand grains when subjected to water motion (“squirting” water on larvae placed in a Petri dish in the case of Wilson) in the studies of Wilson (1948) and Whitelegge (1890), respectively. Wilson (1948) also reported that Ophelia could “use” the current in order to detach from an unpreferred substratum to re-enter the flow. Wilson (1968) and Eckelbarger (1975, 1976) induced settlement of larvae of sabellariid polychaetes which live in habitats subjected to waves as adults, by stirring the water in the experimental container. Boaden (1963, 1968) and Gray (1966b) observed behaviour of meiofauna in water flowing through a small space between parallel plates and through clear tubing. They found that, at low current speeds, some species were rheotactic, moving upstream toward the source of the current, but all of the organisms were simply washed downstream above some higher current speeds. These studies of water motion relative to some aspects of the behaviour of settling infaunal larvae or meiofauna were not designed to mimic a particular, realistic boundary-layer flow regime. At most, the mean current speed (i.e. TABLE V Laboratory experiments on substratum selection by soft-substratum invertebrate larvae, juvenile or adult macrofauna, epifauna and meiofauna: dimensions of treatments and distances between treatments are rough estimates, taken from the information available in the reference; A=archiannelid; B=bivalve; C=cumacean; CR=crab; G=gastrotrich; GA=gammarid amphipod; H=harpacticoid copepod; I=isopod; L=lancelet; LO=lobster; N=nematode;


103

O=ophisthobranch gastropod; P=polychaete; S=shrimp; T=turbellarian; TO=tubificid oligochaete; TP=thin partition between adjacent sediment treatments; NG=information not given in paper Reference

Organism(s) studied

Studies of macrofauna or epifauna larvae Wilson (1948) Ophelia bicornis (P) Wilson (1952, 1953a, Ophelia bicornis (P) b, 1954, 1955) Wilson (1970a) Sabellaria alveolata (P) Wilson (1970b) Sabellaria spinulosa (P) Wilson (1977) Lygdamis muratus (P) Keck, Mauer & Mercenaria Malouf (1974) mercenaria (B) Botero & Atema Homarus (1982) americanus (LO) Cuomo (1985) Capitella sp. I (P) J.P.Grassle (pers. Capitella sp. I and II comm.) (P) McCann (in press) Streblospio benedicti and L.A.Levin (pers. (P) comm.) Studies of meiofauna Wieser (1956) Cumella vulgaris (C) Gray (1966a) Protodrilus symbioticus (A) Gray (1966b) Protodrilus symbioticus (A) Gray (1966c) Protodrilus symbioticus (A) Gray (1967a) Protodrilus rubropharyngeus (A) Gray (1967b) Protodrilus hypoleucus (A) Jansson (1967a) Parastenocaris vicesima (H) Jansson (1967b) Coelogynopora schulzii (T) Aktedrilus monospermatecus (TO)

Maximum dimension Maximum dimension Maximum distance of experimental of treatment (cm) between treatments container (cm) (cm) 9·0 7·0

1·5 0·75

4·5 “a few cm”

6·5

1·7

3·0

6·5

1·7

3·0

6·5

2·3

2·0

NG

NG

NG

55·0

27·5

TP

60·0 13·3

7·5 3·9

3·0 5·5

25·5

7·6

1·0

NG 7·0

NG 1·0

NG 5·0

15·0

NG

NG

7·0

1·0

5·0

7·0

1·0

5·0

7·0

1·0

13·0

NG

0·8

NG

NG

0·8

NG

104

CHERYL ANN BUTMAN

Gray (1968)

Leptastacus constrictus (H) Gray & Johnson Turbanella hyalina (1970) (G) Jensen (1981) Chromadorita tenuis (N) Klauser (1986) Convoluta sp. (T) Studies of macrofauna or epifauna Teal (1958) Uca minax (CR) Uca pugilator (CR) Uca pugnax (CR) Webb & Hill (1958) Branchiostoma nigeriense (L) Williams (1958) Penaeus setiferus (S) Penaeus aztecus (S) Penaeus duorarum (S) Meadows (1964a) Corophium volutator (GA) Corophium arenarium (GA) Meadows (1964b) Corophium volutator (GA) Meadows (1964c) Corophium volutator (GA) Corophium arenarium (GA) Croker (1967) Parahaustorius longimerus (GA) Neohaustorius schmitzi (GA) Lepidactylus dytiscus (GA) Haustorius sp. (GA) Acanthohaustorius sp. (GA) Lewis (1968) Fabricia sabella (P) Sameoto (1969) Haustorius canadensis (GA) Neohaustorius biarticulatus (GA) Acanthohaustorius millsi (GA) Parahaustorius longinerus (GA)

7·0

1·0

5·0

15·0

1·0

13·0

10·0

10·0

6·0

5·0

1·0

3·0

75·0

37·5

TP

24·0

7·5

10·6

243·0

45·7

137·2

9·0

9·0

16·0

34·0

17·0

TP

12·0

6·0

TP

“Large finger bowls”

“Divided in half”

TP

9·0 9·5

6·4 6·7

TP TP


Hadl et al. (1970) Jones (1970)

Morgan (1970) Gray (1971) Phillips (1971)

Protohaustorius deichmannae (GA) Microhedyle milaschewitchii (O) Eurydice pulchra (I) Eurydice affinis (I) Pectenogammarus plancrurus (GA) Scolelepsis fuliginosa (P) Callianassa jamaicense louisianesis (S) Callianassa islandgrande (S)

6·0

1·5

0·2

“Large circular tank”

“Crystallizing dishes”

“At equal intervals around perimeter”

15·2

2·5

10·2

30·0

7·5

15·0

32·0

16·0

TP

105

from the average fluid-discharge rate) was measured. Relevant aspects of the boundary-layer flow regime (e.g. the shear or boundary shear stress, see pp. 145–8) were quantified, relative to settlement, in only one published study to date, that of Crisp (1955) on barnacle cyprids. The experiments were conducted in clear glass tubing and the animals were stimulated to attach over a range of low shear (the change in velocity with distance above the surface), but were prevented from attachment beyond some threshold value. All the water flow compared with attachment or settlement observations mentioned above indicate the potential sensitivity of larvae to moving fluid and the likelihood of passive transport very close to the sea bed, although the limiting values of boundary-layer flow parameters for which this would occur have yet to be quantified for soft-substratum organisms (but see theoretical calculations of Butman, 1986a). LABORATORY STUDIES OF HABITAT SELECTION Laboratory choice experiments of settling larvae were pioneered by Wilson in an extensive series of substratum-selection experiments on Ophelia bicornis (Wilson, 1948, 1952, 1953a, b, 1954, 1955). The studies were done in small Petri dishes (3–9 cm in diameter), where larvae were allowed to choose between small piles (0·75–1·5 cm in diameter) of sediment separated by several centimetres. These are the smallestscale experiments conducted on active habitat selection. A similar experimental design was used by Wieser (1956), Gray (1966a, b, c, 1967a, 1968), Croker (1967), Gray & Johnson (1970), Hadl, Kothbauer, Peter & Wawra (1970), Wilson (1970a, b, 1977), and Klauser (1986) (see Table V). Similar-sized dishes were used in the studies of Wilson (1948), Meadows (1964a), Lewis (1968), and Sameoto (1969), with the various treatments separated into pie-shaped sections by narrow vertical barriers (e.g. glass slides), so they were essentially adjacent. Very small-scale experiments also appear to have been done in the studies of Jansson (1967a, b), but only the treatment dimension (0·5 cm3) is given in the paper. The experiments were done in a “simple alternative chamber” made of plastic tubing, where the treatment patches were placed in either end. In all the studies cited above, the entire experiment was conducted at the scale of centimetres: in containers 9 cm in diameter (except, perhaps, Jansson, 1967a, b), with maximum treatment dimensions of 90% metamorphosis occurred in aqueous extracts from water overlying the sand in which the adult pheromone was released, but only 5% metamorphosis occurred in extracts overlying sand outside the sand dollar bed. Like the direct choice studies and the metamorphosis experiments, the logistics of active habitat selection whether by the tactile chemical sense or by chemotaxis, are poorly explored for infaunal larvae settling in moving fluid; all experiments have been done in still water. If an entirely waterborne cue can elicit the response while the larva is still in the plankton, as the results of Highsmith (1982), Burke (1984), and Cuomo (1985) suggest, then it is particularly critical to do laboratory studies in simulated field flows. As mentioned earlier, by the time cues advected and mixed by flow turbulence are perceived by a planktonic larva, the organism may end up on the sea bed in a habitat from which the cue did not emanate (e.g. Cameron & Rumrill, 1982). Doyle (1975) proposed a settlement model for active habitat selection where the probability of a larva responding to a given cue in the water can be only zero or one, i.e., a threshold level of the stimulus evokes the response. This is an attractive theory, particularly if the competent larvae drift in water very close to (i.e. within centimetres of) the sea bed during the cue-detection stage, because it minimizes errors in site selection and requires a relatively simple behaviour response. While accurate site location would be improved if a larva could swim upstream along a cue concentration gradient, this possibility appears to be limited to very weak near-bed flow regimes, due to the relatively slow swimming capabilities of most infaunal larvae (Mileikovsky, 1973; Mann & Wolfe, 1983; Chia, Buckland-Nicks & Young, 1984) compared with velocities very close to the sea bed (Butman, 1986a). Even if cues must be adsorbed to a surface that the larva can test, as the bulk of the evidence to date suggests, test sites also may be specified by bottom boundary-layer flow conditions (Butman, 1986a; see later discussion, pp. 148–51). Thus, while larvae of many infaunal invertebrates are clearly capable of discriminating between microhabitats and metamorphosing in response to specific cues, the field conditions wherein active habitat selection actually determines patterns of recruitment are unknown.


109

FIELD EXPERIMENTS ABOVE OR ON THE SEA FLOOR Field studies on active habitat selection may have the advantage of being realistic from a fluid-dynamic point of view, but other aspects of field conditions are limiting to experimentation. For example, it is nearly impossible to sample initial settlement onto the sea bed and to measure subsequent early postlarval mortality in soft-substratum systems. Even in hard-substratum systems, the newly settled stages have been identified and followed for only a few species (Connell, 1985). Experimental studies in the field can be classified either as manipulations above the bottom or direct manipulations of the sea floor. The literature is reviewed with the primary emphasis on identifying the spatial scales involved in the experiments and on separating cases where habitat selection actually was demonstrated in the study from indirect or inconclusive evidence of selection. While experiments generally are required to determine processes, direct sampling of the unmanipulated sea floor has also provided useful information on patterns of recruitment and significant correlates, especially when sampling was both frequent (days to weeks) and rigorous (using appropriately small sieves to sample newly settled organisms) and when the water column was sampled simultaneously (e.g. Muus, 1966, 1973; Oliver, 1979; Hannan, 1980; Luckenbach, 1984; Webb, 1984). Results of the detailed field study of Muus (1966, 1973) on bivalve larval availability in the plankton (e.g. from Fosshagen, 1965, which overlapped with the first year of Muus’ study) and recruitment in two localities (at 18 m and 27 m), indicate that both habitat preferences in settling larvae and early postlarval mortality shape adult distributions. This is a somewhat unusual study in that the two field sites were separated by only 1 km and differed markedly in faunal composition, but differences in the bottom sediments (dominance by the 64 to 250-µm fraction at the 18-m site and dominance by the 64 to 125-µm fraction at the 27-m site) probably were not hydrodynamically meaningful (i.e. did not represent a large enough change in bed roughness to alter the structure of the near-bed flow; see pp. 148–54. These relatively small sediment differences certainly may be biologically meaningful. Muus acknowledged that observed patterns of recruitment imply active habitat selection only if the supply of larvae to the two sites was equivalent, but provides reasonable arguments, based on known circulation patterns in the Øresund, that “the same water masses and same larval swarms” probably pass over the two localities (Muus, 1973, p. 103). To avoid problems associated with direct sampling of the sea bed (e.g. processes operating at the sediment-water interface that may obscure initial settlement patterns), several manipulative field studies have been done in structures raised above the sea floor (Table VI). In nearly all cases, larval settlement differed among the various treatments deployed simultaneously and the authors concluded that larvae actively select their settlement sites. All field studies which compared collections in artificial structures with collections from the natural sea bed may, however, have suffered from “trapping artifacts” —physical, chemical, and biological differences between the micro-environment of the trap and the natural bottom— which complicate interpretation of the results (Oliver, 1979; Hannan, 1981). Unless the collection characteristics of the traps for passive inert particles (e.g. sediments) can be defined (Hargrave & Burns, 1979; Gardner, 1980; Butman, 1986b; Butman, Grant & Stolzenbach, 1986), collections resulting from biological processes (e.g. active habitat selection behaviours of the larvae) cannot be separated from collections resulting entirely from hydrodynamical processes. Hannan (1981) was unable to distinguish between these possibilities to account for the differences (orders of magnitude) in numbers of postlarvae collected in traps placed 1 m above the sea bed compared with those in cores of the natural bottom. Oliver (1979; some results are also reported in Dayton & Oliver, 1980) used relatively “tall” and “short” plastic cups filled with the same amount of sediment (to the rim of the short cup) to simulate physical conditions of deposition and resuspension, respectively. Relatively more “Capitella capitata” (the sibling species, sensu Grassle & Grassle, 1976, may be similar to Capitella sp. Ia, as in Hannan’s, 1981, study

110

CHERYL ANN BUTMAN

conducted nearby) were collected in the tall than in the short cups, while another polychaete species, Armandia brevis, was not differentially collected by the two trap designs. Oliver (1979) suggested that Capitella actively selected the depositional environment in the short traps, but that Armandia was less selective in its settlement requirements; both behaviours are consistent with the distributional patterns of the adults and the responses of the populations to disturbance (Oliver, Slattery, Hulberg & Nybakken, 1980). Passive accumulation of Capitella larvae or postlarvae in the depositional environment is, however, also consistent with the results. Differences between species in hydrodynamic properties (e.g. fall velocities), swimming abilities, and periods of larval availability relative to flow processes, could account for the different patterns of collection by the traps for the two species. In the only study (Hannan, 1984a, b) where hydrodynamical properties of larvae and collection characteristics of traps (Butman, 1986b) were defined in the laboratory before field deployments, nearly all of the abundant infaunal organisms (in three invertebrate phyla) were collected in the relative abundances predicted for passive particle collections by traps. It is valid to compare collections in different sediment treatments placed in the same type of artificial structure raised above the sea floor, when the treatments are exposed to the same flow regime (e.g. Oliver, 1979; Levin, 1981, 1984; Watzin, 1983, 1986), but considerations of possible “edge effects” and other differences between treatments due to position within the structure TABLE VI Field experiments on larval settlement or early recruitment of infauna: experimental studies where sampling intervals were 1 month or sieve screen sizes ≥1 mm are not included in this table because initial settlement or early recruitment are unlikely to be detected with these methods; distances and dimensions were estimated, when possible, from the information given in the paper; *=in studies where settlement into one type of box, tray or trap (usually containing defaunated sediment) was compared with settlement onto the natural sea bed, the distance between “treatments” would be between the structure and where the sea bed was sampled; this usually was not given in the study but probably is no greater than tens of centimetres or metres; NG=information not given in the paper; NA=not applicable to this study; OSO=one station only was sampled; P=polychaete; B=bivalve Reference

Experimental Organisms approach studied

Thorson (1946)

“Bottle collectors” moored above sea bed; plankton and bottom samples Manipulatio ns of the sea bed “Sediment bottle collectors” Different substrata placed in suspended

Baggerman (1953) Reish (1961) Hermans (1964)


Sieve screen Minimum size (µm) distance between stations (m)

Maximum treatment dimension (cm)

Maximum distance between treatments (cm)

All infauna

6 wk for bottles, 2 wk for plankton

NG for bottles, 83 for plankton

NG

Bottle dimensions not given

NG*

Bivalves

12 h

500

OSO

45

NG

All infauna

28 days

246

710

NG

2 wk

NG

OSO

“Gallon jar” dimensions not given “Thorson” bottles, dimensions not given

Armandia brevis (P)

NG


Reference

Richter & Sarnthein (1977), but technical layout in Sarnthein & Richter (1974) Grassle & Grassle (1974)

Guérin & Massé (1978), Massé & Guérin (1978)

McCall (1977)

VanBlarico m (1978)


bottle collectors Different substrata placed in trays moored above sea floor Boxes of defaunated sediment made flush with the sea bed; bottom samples Different substrata placed in three designs of collectors moored on the bottom or above the bottom Boxes of defaunated sediment placed on the sea bed; bottom samples Containers with thin layer of sediment moored above the sea bed; bottom samples





Molluscs

2 wk

63

80

71

118

All infauna

3 days

297

4600

100

NG*

Polychaetes and molluscs

1 month

1000 “on diagonal” (=700 if mesh is square)

NG

8·6

180

All infauna

10 days

297

9000

3700

NG*

All infauna

13 days for containers, 1 month for bottom

250 for containers, 500 for bottom

OSO

10 for containers

NG*

111

112

CHERYL ANN BUTMAN

Eckman (1979) Oliver (1979), Dayton & Oliver (1980)

Santos & Simon (1980a)

Williams (1980) Bhaud, Aubin & Duhamel (1981)

Hannan (1981)

Levin (1981)

Manipulatio ns of the sea bed (a) different treatments filled containers held in racks above sea bed; bottom samples (b) manipulatio ns of the sea bed Containers filled with sediment placed in rack above sea bed; plankton and bottom samples Manipulatio ns of the sea bed Collectors filled with sediment placed above sea bed; bottom samples with epibenthic sledge Collectors with thin layer of sediment moored above sea bed; bottom samples Different sediment treatments filled containers

All abundant infauna Polychaetes

11 days

61

OSO

100

700

(a) 6 days

(a) 250 for containers, 500 (250 for “a few cores”) for bottom

(a) OSO

(a) 14 for containers

(a) 42 for containers

(b) 1 month

(b) 500 (250 for “a few cores”)

(b) 1000

(b) 2000

(b) NG

All infauna

7 days for containers, “irregularly ” for plankton, 1 month for bottom

OSO

250 for containers, 144 for plankton, 500 for bottom

5 for containers

NG*

Tapes japonica (B) Polychaetes and bivalves

7 days

149

OSO

150

750

5 days

NG

OSO

“2 litre capacity” collectors, dimensions not given

NG*

Armandia brevis (P), Capitella spp. (P), Nothria elegans (P), Prionospio pygmaea (P) Streblospio benedicti (P) Pseudopoly dora

7 days

250

400

10

NG*

2 wk

Worms were visually counted under

OSO

9

26


VanBlarico m (1982) Zajac & Whitlatch (1982a)

Eckman (1983) Gallagher, Jumars & Trueblood (1983) Watzin (1983, 1986)

Hannan (1984a, b)

placed directly on sea bed; bottom samples Manipulatio ns of the sea bed Buckets of defaunated sediment made flush or protruding above the sea bed; bottom samples Manipulatio ns of the sea bed Manipulatio ns of the sea bed

paucibranc hiata (P)

Different sediment treatments filled containers held in racks raised above sea bed Different sediment trap designs, with known passive particle collection characteristi cs, moored

dissecting microscope

All infauna

7 days

500

OSO

30

NG

All infauna

14 days

297

620

48

NG*

All abundant infauna All abundant infauna

2 days

61

OSO

30

1970

2 days

63

OSO

3·7

500

All infauna

7 days

63

OSO

14

50

All abundant infauna

1 day

100

OSO

14·7

2400

113

114

CHERYL ANN BUTMAN

Reference

Levin (1984)

Luckenbach (1984)

Bonsdorff & Österman (1985)

Whitlatch & Zajac (1985)

Woodin (1985)


above sea bed Manipulatio ns of the sea bed; containers filled with sediment placed on sea bed; plankton samples Sampled four areas of sea bed representing natural sediment treatments; plankton and bottom samples for initial availability Trays filled with sediment placed on sea bed; bottom samples Different sediment treatments in cores held in racks above sea bed Different sediment treatments in cores implanted in bottom





Polychaetes

3 days for manipulated sediments, 1 month otherwise

250 for containers and sediments, 63 for plankton

100

63 for bottom, 9 for containers

100

Mulinia lateralis (B)

1 day for initial availability; 4 days for initial settlement; 1 month for recruitment

105

OSO

10

NG*

All infauna

2 wk

500 for macrofauna ; 500, 200, 63 for meiofauna

OSO

40

NG*

All infauna

10 days

180, 300

OSO

5

100

Spionid polychaetes

9 days

250

OSO

11

1500


115

must be carefully analysed (see Nowell & Jumars, 1984). While it may not be possible to define the hydrodynamic conditions above these sediment treatments, as long as conditions are constant among treatments, between-treatment differences in settlement or recruitment can be assessed. Results of the field manipulations above the sea floor are strongly suggestive of active habitat selection by many infaunal larvae or postlarvae on scales of tens of centimetres to metres (Table VI), with the caveat that hydrodynamic alternative hypotheses usually were not considered or tested. Data from many of these manipulative field studies will remain equivocal until the possibility of differential passive deposition or accumulation between structures or between structures and the bottom can be discounted. The strongest results are for studies where hydrodynamic conditions were held constant among treatments, although the physical characteristics of the flows over these sediments are undefined. Directly manipulating bottom sediments to test the active habitat selection hypothesis alleviates the problems with structures. If the bottom roughness scales remain unchanged between manipulated and control sediments, then flow characteristics should be similar for all treatments (see Nowell & Jumars, 1984). The results of such studies (e.g. Oliver, 1979; Williams, 1980; Gallagher, Jumars & Trueblood, 1983) show that recruited postlarvae are associated with distinct habitats on the scale of metres (Table VI). Because of problems in sampling initially settled larvae and subsequent early mortality, mentioned above, it is not clear if the pattern results from active selection when the larvae first reach the sea floor or from a re-distribution of the postlarvae after initial settlement; it is also possible that observed distributions resulted from very early postlarval mortality of settled larvae that were originally evenly distributed among the sediment treatments. In cases where bottom sediments were manipulated specifically to change the nature of the nearbed flow regime, with accompanying a priori predictions of hydrodynamic effects on recruiting infaunal postlarvae or meiofauna (Eckman, 1979, 1983; Hogue & Miller, 1981), the hydrodynamic null hypotheses could not be falsified (see pp. 141–5). PATTERN OF DISTRIBUTION AND ACTIVE HABITAT SELECTION: A PROBLEM OF SPATIAL SCALES The spatial scales (centimetres to tens of centimetres) for which active habitat selection has been conclusively demonstrated in laboratory experiments in still water (Table V) are one to six orders of magnitude smaller than the spatial scales (tens of metres to tens of kilometres) over which species and sediment composition are significantly correlated in the field (see Table III, pp. 120–1). Thus, the process of active habitat selection, as demonstrated by these laboratory results, cannot account for the observed field distributions, due to the mismatch in spatial scales. Field experiments on processes controlling larval settlement were conducted at spatial scales of tens of centimetres to tens of metres (Table VI); while active habitat selection was strongly implied by the results of many of these studies, this interpretation remains equivocal because the alternative hypothesis of passive deposition was usually neither considered nor tested. When patterns of community composition and structure have been delimited at small spatial scales, e.g. of the order of 1 to 10 m in Jones (1962), of 0·1 to 1 m in Angel & Angel (1967) and Grassle et al. (1975), of 0·01 to 1 m in Reise (1979), of 100 cm in Gärdefors & Orrhage (1968) and Jumars (1976), of 10 m in Gage & Geekie (1973a), and of 10 cm in Olsson & Eriksson (1974), sediment samples were not taken at each infaunal sampling location, except in one case (Angel & Angel, 1967). The entire area sampled in these small-scale dispersion studies was usually considered homogeneous in its bulk sediment characteristics, based on one to a few sediment samples from the area. Thus, the spatial patterns and scales of diversity

116

CHERYL ANN BUTMAN

detected in these studies were usually attributed to processes other than those directly related to bulk properties of sediments. The small-scale patterns of species distribution and diversity detected in these studies are the only patterns to which results of habitat selection experiments can be applied directly. Even though bulk properties of sediments were presumed to be constant within the areas sampled in these studies, Angel & Angel (1967) and Jumars (1976) briefly discussed the potential importance of small-scale variability in sediment characteristics; Jumars & Eckman (1983) provide a more detailed discussion of this topic. Local heterogeneity in sediment topography (e.g. due to geological or biological processes) can cause significant small-scale (centimetres to metres) variations in sediment grain size because of local changes in the nearbottom flow regime; patchiness in infaunal distributions at these small scales can be attributed to this sediment heterogeneity (e.g. Rhoads & Young, 1970; Eckman, 1979, 1983; Thistle, 1983). In addition, detailed analyses of sediment characteristics using micro-scopic methods and staining techniques (e.g. Whitlatch & Johnson, 1974) indicate that bulk sediment analyses obscure variation in sediment properties (e.g. protein, carbohydrate, and lipid contents, as well as grain size) to which organisms may respond (Whitlatch, 1974, 1980). Many laboratory studies of habitat selection have demonstrated that there are chemical and biological substances (e.g. chemical conditioning of sediments by adults or the abundance and composition of bacterial populations) in sediments which augment grain size as attractive factors to stimulate or enhance larval settlement. Thus, within an area of homogeneous sediment type (based on analysis of grain size), larvae may actively select for microhabitats based on these other aspects of sediments. For example, Thistle, Reidenauer, Findlay & Waldo (1984) and Eckman (1985) have shown that there is local enhancement of bacterial abundances around vertical protrusions (seagrass shoots or animal tubes) from the sea bed and that infauna are concentrated in these regions. In summary, larvae may select for microhabitats at small spatial scales (centimetres to tens of centimetres) based on sediment characteristics other than just grain size (as determined from bulk sediment analyses). The capability of larvae to distinguish between and actively select for habitats with distinctly different grain sizes and separated by large distances (tens of metres to tens of kilometres) is yet to be demonstrated. The passive deposition hypothesis may resolve this problem because it specifies that larvae are deposited at the same spatial scales as apply to sediment transport and deposition (see p. 144). At this time, passive deposition of larvae represents the simplest and most feasible mechanism for creating initial large-scale distributions of larvae in the field. Active habitat selection may be confined to only very small spatial scales. THE PASSIVE DEPOSITION HYPOTHESIS To my knowledge, the passive deposition hypothesis was first formally proposed to account for patterns of initial larval settlement or recruitment of infaunal species in the studies of Baggerman (1953) and Pratt (1953). Prior to these, brief, qualitative discussions of the rôle, or potential rôle, of “currents” in controlling larval dispersal and in determining settlement sites were given in Orton (1937), Kreger (1940), Thorson (1946, 1950), and Verwey (1952). For hard-substratum habitats, experiments on the rôle of hydrodynamical processes in settlement occurred much earlier. Observations and experiments on flows which permit or inhibit settlement of fouling organisms date from the 1940s (McDougall, 1943; Smith, 1946; Doochin & Smith, 1951; Crisp, 1955; Wood, 1955) due, at least in part, to the important applied aspects of this problem (i.e. the commercial need for developing methods to inhibit biofouling). Likewise, probably the most extensive studies, to date, of the rôles of both biological and physical processes in the dispersal and settlement of any single species were done on barnacles (Bousfield, 1955; de Wolf, 1973).


117

Strictly speaking, the passive deposition hypothesis stipulates that competent planktonic larvae initially reach the sea floor at sites where passively sinking particulates, with fall velocities similar to larvae, initially settle (Hannan, 1984a, b). As indicated in Hannan (1984b), this hypothesis does not specify that the deposited organisms will accumulate at these locales, as the geological definition of “deposits” implies, but refers only to the process controlling where the larvae will initially come to rest on the sea bed. Then, other biological or physical processes may re-distribute the organisms (see later discussion, pp. 154–5). Note also that, “deposited larvae may or may not have ‘settled’ according to the biological definition of Scheltema” (Hannan, 1984b, p. 1109). The passive deposition hypothesis has never been tested directly because it requires simultaneous sampling of initially deposited larvae and passive particles with fall velocities similar to larvae. This eventually may be possible in a laboratory flume, where realistic field flow regimes could be simulated (see Nowell & Jumars, 1987), and the distributions of inert particles with known fall velocities could be compared with the distributions of larvae or postlarvae when they first reach the bottom. The chances of testing the passive deposition hypothesis in the field seem remote, due to problems in actually sampling initial distributions of larvae and particles prior to interference by benthic biological and physical processes, and to problems of defining the fall velocities of initially settled particulates in their naturally occurring states (e.g. flocculated or biologically aggregated). Support for the passive deposition hypothesis comes from studies of passive accumulation, passive sinking, and passive resuspension and transport of larvae, postlarvae or meiofauna, which are discussed separately below. It is important to distinguish the passive deposition hypothesis from the earlier notion that larvae fall in a “random rain” onto the sea bed. which was once considered the alternative hypothesis to active habitat selection (e.g. see discussion of these early ideas in Thorson, 1957). Random deposition explicitly states that there is an equal probability that individual larvae will fall onto any bed location. This hypothesis is synonymous with the passive deposition hypothesis only for a homogeneous suspension of larvae and particles falling through still water. In moving water, for an infinite water mass with a uniform particle supply distributed homogeneously in the water column, and with a steady and non-varying physical regime, the initial distribution of particles on the sea bed would be random. For temporally and/or spatially varying flow regimes, particle abundances, and particle distributions in the water column, the particles will not, however, fall at random onto the sea bed. In these cases, the sites for initial settlement of particles are determined by the hydrodynamical processes and the particle characteristics. Thus, for the physical regimes of interest in most marine studies, a random rain of larvae to the sea bed is not the appropriate null hypothesis for testing the importance of physical processes, since particle deposition is not expected to be random. In fact, a random pattern of initial larval settlement would, in most cases, falsify the passive deposition hypothesis. If larvae physically behave in a flow like passive particles, then it is their fall velocity and hydrodynamical processes which determine when and where the larvae will reach the sea bed. Thus, passive deposition is the appropriate physical null hypothesis against which biological (i.e. active habitat selection) hypotheses can be tested. PASSIVE ACCUMULATION The correspondence of distributional patterns of a cockle (Baggerman, 1953), two species of bivalve (Pratt, 1953), and several echinoderm species (Tyler & Banner, 1977) with modern fine-sediment distributions was attributed to the passive accumulation of settling larvae and fine sediments in similar locales. Illuminated by discussions with Baggerman, Verwey (1952) and Kristensen (1957) suggested that some bivalve populations may result from passive accumulation in “sheltered” or “weak current” areas. Orton (1937), Segerstråle (1960, but see also 1962), and Carriker (1961) also indicated the potential importance of strong

118

CHERYL ANN BUTMAN

near-bed currents maintaining larvae in suspension (e.g. in gyres or swift tidal channels) and weak currents allowing the spat to settle onto the bed. Fager (1964) attributed the presence of an unusually shallow, very dense, and oddly shaped (elliptical, with the long axis parallel to shore) bed of the polychaete, Owenia fusiformis, to such physical processes. He suggested that larval settlement was concentrated in this locale due to the coincidence of a water mass containing large numbers of competent larvae with a rip current at the site. The hydrodynamics associated with the rip would allow larvae to accumulate passively in the unusual bed configuration. On a much smaller scale, Birkeland & Chia (1971) suggested that early recruitment of sand dollars may be more successful in patches of sand within a cobble field compared with a sand flat, because the cobbles act like “break-waters” for the flow over the sand; it is, however, unclear if the authors were implying passive deposition or enhanced retention in these slow-flow regions. In all of these studies, evidence of the rôle of hydrodynamical processes is largely correlative, but the novelty of these interpretations when other, similar, correlative evidence invoked the active habitat selection hypothesis (see pp. 123–7), is striking. The manipulative field experiments of Baggerman (1953), Eckman (1979, 1983), and Hogue & Miller (1981) provide substantive support for the passive accumulation hypothesis. Baggerman (1953) placed vertical barriers (screens) on the sea bed and sampled for cockle spat near and away from the screens. She also determined that a range in sizes of cockle spat were likely to be transported and deposited like fine sediments by showing that measured gravitational fall velocities of spat were similar to the measured fall velocities of the sediments transported at the study sites. She did not, however, determine, a priori, how the vertical screens would affect the near-bottom flow at her study sites; she assumed that the region of low flow developing in the lee of the screens would be sufficient to trap sediments and passively falling spat. Eckman (1983) made specific a priori hypotheses on how the artificial tubes he placed in sediments would affect both the fluid flux to the bed and the boundary shear stress because he did laboratory flume experiments to measure these physical effects. In another study, Eckman (1979) placed artificial tubes at regular intervals in sediments and, taking contiguous samples over the area, determined the spatial scales of organism distributions and compared them with the spatial scale of the physical effects resulting from this manipulation. Hogue & Miller (1981) repeated Eckman’s (1979) experiments in a different intertidal area, but studied dispersion patterns of nematodes, rather than recruitment of infauna. In all these studies, recruitment patterns were consistent with predictions based on hydrodynamical criteria; that is, the null hypothesis of passive accumulation could not be falsified. Indirect support for passive deposition and accumulation comes from the numerous reports of higher postlarval or adult infaunal abundances in depressions on the sea floor (e.g. Chapman & Newell, 1949; Pratt, 1953; Pamatmat, 1968; Sameoto, 1969; Howard & Dörjes, 1972; Farke, de Wilde & Berghuis, 1979; VanBlaricom, 1982; McLusky, Anderson & Wolfe-Murphy, 1983; Levin, 1984) or in seagrass beds that baffle water motion (e.g. Orth, 1977; Scheibling, 1980; Peterson, Summerson & Duncan, 1984) than in adjacent sandflats. The pattern of distribution for newly settled larvae has, however, yet to be measured. These patterns of enhanced abundances in areas of relatively slow flow need not arise at the time of settlement, but may result from differential post-settlement mortality. To determine at what stage in the life history the pattern of enhanced abundances of the hard clam, Mercenaria mercenaria, in seagrass beds compared with adjacent sandflats is established, Peterson (1986) computed the ratio of organism densities between the two habitats for the 0-year class and all subsequent year classes. Because these ratios were ) for the older year classes, Peterson (1986) concluded that post-settlement considerably larger (by phenomena, such as competition and predation, were at least as important as settlement phenomena in creating the pattern.


119

For all the passive accumulation studies, a problem with unambiguously interpreting the process responsible for the observed pattern of enhanced recruitment in regions of slow flow is that fine sediment and detritus also will accumulate in these areas. Thus, the alternative hypothesis that larvae actively select sites where fine sediments and detritus accumulate or preferentially survive in these areas cannot be discounted. PASSIVE SINKING Organisms are unlikely to be passively deposited onto the sea floor unless they sink through near-bottom waters like passive particles. Hannan (1984a, b) tested this passive sinking hypothesis for larvae falling through turbulent field flows using several groups of geometrically different sediment trap designs (see Table VI, p. 138). A priori predictions regarding the rank order that the various traps would collect larvae in the field were dictated from laboratory flume experiments to determine particle collection efficiencies of the traps in flows dynamically similar to average conditions at the field site studied. The flume flow was seeded with particles having fall velocities similar to those measured in the laboratory for non-swimming polychaete larvae. In these experiments, nearly all of the abundant organisms (polychaete, bivalve, and enteropneust postlarvae) were collected by traps in the patterns predicted for passive particle collections. Thus, the passive sinking hypothesis could not be falsified. PASSIVE RESUSPENSION AND TRANSPORT Indirect evidence that organisms living at the sediment surface may be resuspended and transported comes from studies where the water column and the sea bed were sampled simultaneously, or where the bottom was sampled intensively, throughout storm events (Hagerman & Rieger, 1981; Hogue, 1982; Dobbs & Vozarik, 1983); organisms either were missing from the sea bed or were present in the water column during the storms. The sampling studies of Bell & Sherman (1980) and Palmer & Brandt (1981) suggested that even tidal velocities may be sufficient to resuspend and transport meiofauna (but see also Grant, 1981). Palmer & Gust (1985) quantified this effect by measuring the bottom shear stress over a tidal cycle, when simultaneous water column and bottom samples also were collected. The a priori hypothesis was that meiofauna would be resuspended with the surface sediments only when the bottom shear velocity exceeded the critical erosion velocity for the sediments. They found that organism abundances in the water directly above (within tens of centimetres of) the sea bed were highest when the bottom shear velocity exceeded the threshold value. Furthermore, from laboratory experiments, Palmer (1984) showed that the organisms probably were not actively entering the water, although certain behaviours (i.e. remaining at the sediment surface rather than burrowing) increased a given organism’s probability of being resuspended. Indirect support for passive resuspension and transport of surface- or near surface-dwelling infauna comes from the numerous reports of postlarval and adult organisms in the water column (see Table II, pp. 117–9) and of post-settlement migrations (e.g. Chapman & Newell, 1949; Baggerman, 1953; Kristensen, 1957; Sigurdsson, Titman & Davies, 1976; Farke et al., 1979). SUMMARY The hypotheses that hydrodynamical processes determine accumulation, sinking or resuspension and transport of larvae, postlarvae or meiofauna could not be falsified in the experimental studies conducted thus far. These results provide support for the passive deposition hypothesis, but direct tests for initially

120

CHERYL ANN BUTMAN

settled larvae are lacking. A limitation to interpreting results from the passive accumulation experiments is that fine sediments and detritus tend to accumulate in regions of slow flow so that the observed enhanced abundances of organisms in these areas, presumed to be the result of passive accumulation, could also result from active habitat selection for these detritalrich zones or from enhanced early postlarval survival. Experimental manipulations are needed to distinguish among these possibilities. LARVAL SETTLEMENT IN THE BOTTOM BOUNDARY LAYER Results of the passive accumulation, sinking, and resuspension and transport studies stipulate that physical processes cannot be discounted in considerations of larval settlement phenomena, so it is worthwhile to discuss briefly characteristics of the bottom boundary-layer flow environment, where settlement takes place. The near-bed flow regime determines the spatial scales applicable to passive deposition and also the hydrodynamical constraints for successful active habitat selection. Other discussions of bottom boundarylayer processes relevant to benthic ecology, and written for a general audience, can be found in Wimbush (1976), Vogel (1981), Nowell (1983), Nowell & Jumars (1984), and Butman (1986a); the recent review of Grant & Madsen (1986), written primarily for fluid dynamicists, summarizes many important aspects of boundary-layer flows in continential-shelf environments. The following discussion is limited to steady, uniform (in the horizontal) flow over a bottom which is also uniform over large horizontal distances, relative to the height off the bottom. The purpose is to provide some basic fluid-dynamical perspective on larval settlement, while retaining the essential physics. GENERAL FEATURES OF BOUNDARY-LAYER FLOWS OVER SOFT SUBSTRATA where u is the As water flows over the sea bed, a region of shear (the slope of the velocity profile, horizontal velocity component and z is the perpendicular distance from the bed; see Fig. 1) develops as a result of the retarding effect (drag) of the boundary on the flow. This region of shear near the bed is called the boundary layer. Within the boundary layer, current speed goes from zero at the bed to the mean-stream the boundary-layer thickness). For heights velocity (U) at the top of the boundary layer (where the bottom no longer has a significant effect on the flow; this is called the region of exceeding potential or frictionless flow and, in the absence of other flow processes (e.g. surface wind stress or other in this region. When the shear near the bed is sources of flow turbulence) and for a constant density, sufficiently large, turbulent eddies are generated that mix lower-momentum fluid close to the bed with higher-momentum fluid away from the bed; this thickens the boundary layer and reduces the mean velocities at a given height above the bed (especially close to the bottom). The shape of the velocity profile in the boundary layer depends on flow properties (e.g. the flow Reynolds number, the background turbulence and accelerations), fluid properties (e.g. stratification induced by temperature, salinity and suspended sediment), and boundary characteristics (e.g. the bed roughness and the cohesiveness of sediments). Velocity profiles have been measured for controlled laboratory flows and their characteristics have been determined theoretically under certain conditions. For the steady, uniform flow case considered here, two shapes of the velocity profile are well known, a parabolic shape for laminar boundary layers and a logarithmic shape for turbulent boundary layers. The boundary layer will be laminar or turbulent, depending on the flow Reynolds number, a dimensionless parameter which is the ratio of inertial forces to viscous forces in the flow. The Reynolds number (VL/v) depends on a length (L) and a velocity (V) scale for the flow, as well as on the fluid


121

Fig. 1.—Diagram of a turbulent boundary layer plotted on a linear scale for both axes, showing the relative positions of the viscous sublayer, the log layer, and the log-deficit layer: taken from Butman (1986a).

kinematic viscosity (v). Laminar boundary layers occur at low Reynolds numbers; molecular viscosity dominates as inertial forces are relatively unimportant for these conditions. Laminar boundary layers are very stable in the downstream direction; any disturbance to the layer (caused by flow over a bump, for example) will be quickly dissipated by viscosity, restoring the velocity profile to the undisturbed state. Thus, in laminar boundary layers, the flow is parallel to the bottom. Turbulent boundary layers occur at high Reynolds numbers and thus inertial forces (or turbulence) dominate over molecular viscosity. The velocity is composed of a mean component plus a fluctuating (turbulent) component. Transfer of mass and momentum within the layer is caused by these turbulent eddies. While the time-averaged flow velocity is in the horizontal, as in the laminar case, turbulent eddies have velocity components in all directions. Descriptions of laminar and turbulent boundary layers can be found in Clauser (1956), Schlichting (1979), and Yaglom (1979); features most relevant to problems in benthic ecology are indicated in Nowell & Jumars (1984). Laminar boundary layers are rare in the ocean, so that subsequent discussion will be for the turbulent case. Turbulent flows are classified as smooth, rough, or transitional (e.g. Schlichting, 1979), depending on the which is, again, a dimensionless ratio of inertial to viscous roughness Reynolds number (Re*= forces in the flow, but in this case it depends on the shear in the flow (u*, the bottom shear velocity, which where µ is the molecular viscosity of the fluid) and on the physical bed roughness (kb), as well is as on kinematic viscosity. In the immediate vicinity of the bottom, molecular viscosity is primarily responsible for dissipating flow energy. Outside the viscous sublayer, turbulent eddies mechanically dissipate flow energy as they break down into smaller and smaller eddies until, ultimately, energy is again dissipated by viscosity. A pronounced viscous sublayer (see Fig. 1) may develop in the case of flow over hydrodynamically smooth bottoms occurring at low Re* (e.g. Eckelmann, 1974). Over hydrodynamically rough bottoms (high Re*), viscosity still acts at the boundary, but no distinct well-behaved sublayer forms and eddies may penetrate to within tenths of a millimetre of the bed; thus, in rough-turbulent flow, the velocity structure close to the bed is complicated (e.g. Nowell & Church, 1979) and not well known. For intermediate Re*, transitional flow occurs, with characteristics intermediate between smooth-and roughturbulent. In the field, smooth-turbulent profiles have been measured by Chriss & Caldwell (1982) and W.D.Grant (pers. comm., see Butman, 1986a) and rough-turbulent profiles by Smith & McLean (1977), Cacchione & Drake (1982), Gross & Nowell (1983), and Grant, Williams & Glenn (1984). At a given site, the flow can be smooth-turbulent under one flow condition and rough-turbulent under another, for example, due to changes in bed roughness by rippling during storms or by bioturbation (see Grant & Madsen, 1986).

122

CHERYL ANN BUTMAN

Based on empirical studies and scaling arguments (Clauser, 1956), turbulent boundary layers can be divided into three regions (Fig. 1). Adjacent to the boundary, in the viscous sublayer, velocity (u) varies linearly with distance from the bottom. Above this, u varies with ln z in what is known as the log layer. The region farthest from the boundary is known as the log-deficit layer because the deficit velocity (U— u) varies with In z. The remainder of this discussion will focus on the qualitative and quantitative features of the log layer and the viscous sublayer, because their characteristics are relatively well known (e.g. Clauser, 1956; Yaglom, 1979; Nowell, 1983; Grant & Madsen, 1986) and they are the regions most relevant to larval settlement. The total thickness of the bottom boundary layer depends on the bottom shear velocity (u*) and inversely on the forcing frequency for the flow. On the continental shelf, at a latitude of 40°, for a flow periodicity stipulated by the Coriolis force, a u* of about 1 cm·s−1, and in the absence of stratification, the bottom boundary layer would be about 40 m thick (Grant & Madsen, 1986). The boundary layer grows all the way to the water surface in the smooth-turbulent, tidally driven flows at 10-m depth in Buzzards Bay, Massachusetts (U.S.A.), and for u* between 0.4 and 0.6 cm·s−1 (flow speeds of about 10 to 15 cm·s−1 at z=50 cm); the boundary layer fills half the water column for u*=0·2 cm·s−1 (a flow speed of about 5 cm·s−1 at z=50 cm) (Butman, 1986a). Boundary layers resulting from forcing due to surface waves are very thin (centimetres to tens of centimetres), however, because of the high-frequency nature of these flows (Grant & Madsen, 1986). In the field, the log layer is known to be about 10–15% of the total boundary layer (Clauser, 1956; Nowell & Church, 1979; Grant & Madsen, 1986), so the log-layer thickness varies between centimetres (wave boundary layer) to about a metre (tidal boundary layer) to several metres (planetary boundary layer), in the examples above. For smooth-turbulent flows, the viscous sublayer can be estimated by 10 v/u*; for u* between 0·1 and 1·0 cm·s−1 (typical values for smooth-turbulent flow) and v=0·01 cm2·s−1, the viscous sublayer thickness will be from 0·1 to 1·0 cm. In summary, a larva beginning its descent through the water column in the region of potential flow will and then will experience a sheared flow, experience a constant horizontal velocity until it reaches where the velocity decreases approaching the bed. At some distance close to the bottom, the horizontal velocity becomes vanishingly small (since u=0 at the sea bed), so the organism would be free to manoeuvre in basically still water. A question relevant to larval settlement in general, and active habitat selection in particular, is: in what region above the sea bed are flow speeds sufficiently low such that settling organisms could effectively manoeuvre (e.g. swim among test sites)? Such hydrodynamical constraints for active habitat selection are discussed below. If the larvae sink through the water and are deposited onto the sea bed like passive particles, then parameters of the boundary-layer flow and the gravitational fall velocities of the organisms determine where they will initially reach the sea floor and where they are likely to accumulate. In this case, sediment transport theory can be used to predict depositional or accumulation sites for larvae on the sea floor. Physical considerations involved in such predictions are also discussed below. HYDRODYNAMICAL CONSTRAINTS ON ACTIVE HABITAT SELECTION Hydrodynamical constraints on active habitat selection depend on how settlement cues are perceived by the organisms (see pp. 132–4) and on their swimming behaviours and speeds. If larvae respond to waterborne cues, then the boundary-layer flow determines the extent of mixing (and thus, of dilution) of the cue by the time the larva perceives it. The manner in which the larva responds to the cue (e.g. does it suddenly quit swimming and sink or does it actively swim straight down to the bed?) and the structure of the near-bed flow regime determine how far the larva will be advected downstream before it reaches the bottom. If larvae must make direct contact with the sea bed in order to perceive a settlement cue then, again, potential test


123

Fig. 2.—Turbulent velocity profiles constructed in Butman (1986a), plotted on a log-linear scale: for the roughturbulent profiles, the dashed portion represents (10) (z0); below this level, the accuracy of predictions of velocity by the log-layer function are unknown; for the smooth-turbulent profile, the curved region is the viscous sublayer; the line is dashed at the interface between the log layer and the viscous sublayer because the actual function predicting velocities in this region is unknown; the profiles were constructed for a flow speed of u=15 cm·s−1 at z=50 cm, but for different values of bottom roughness (see Table I in Butman, 1986a).

sites on the bed depend on how they conduct a search (e.g. do they swim horizontally among sites or do they swim or sink down to a site and then reject it by swimming straight up?) and on the boundary-layer flow regime. Most of the existing laboratory data suggest that larvae must make direct contact with a surface bearing the cue in order to perceive it; this sensing mechanism is assumed, for the sake of argument, in the following discussion. To determine the flow velocities that larvae experience as they approach the sea floor, Butman (1986a) constructed boundary-layer velocity profiles, based on near-bottom current observations from a shallow (10m depth), subtidal site in the coastal embayment of Buzzards Bay. The flows at this site are primarily driven by the semi-diurnal tides and current speeds measured one metre above the bottom ranged from 0 to 22 cm·s−1. Velocity profiles in the log layer were calculated, assuming both smooth- and rough-turbulent flow, and for different flow speeds. These velocities were compared with the maximum horizontal swimming speed measured for polychaete larvae (from the review of Chia, Buckland-Nicks & Young, 1984). The surprising result of this study was that horizontal flow velocities considerably exceed larval swimming speeds, even at only several larval body lengths above the bed, for most of the flow conditions used in the analysis. At near-peak ebb or flood tide (when u=15 cm·s−1 at z=50 cm), the flow speed is 1 cm·s−1 at distances of about 300 µm (smooth-turbulent), 500 µm (rough-turbulent, u*=0·82 cm·s−1) and

124

CHERYL ANN BUTMAN

Fig. 3.—Smooth-turbulent velocity profiles for three flow speeds (representing three stages of the tidal cycle in Buzzards Bay), as constructed in Butman (1986a) and plotted on a log-linear scale: only the viscous sublayer is shown on the Figure; Line A is for u=15 cm·s−1 at z=50 cm; Line B is for u=10 cm·s−1 at z=50 cm; Line C is for u=5 cm·s−1 at z=50 cm.

1500 µm (rough-turbulent, u*= 0·98 cm·s−1) above the bed (Fig. 2). Because maximum measured swimming speeds of polychaete larvae are only 5 mm·s−1, they would have a difficult time manoeuvring horizontally (e.g. to swim between potential test sites) in any of these flows. Swimming full-speed against the flow at about two body lengths above the bottom, the larvae would still be advected downstream at 5 mm·s−1! Plots of smooth-turbulent velocity profiles for various current speeds (stages of the tide for the Buzzards Bay case) (Fig. 3) indicate that larvae could effectively manoeuvre via horizontal swimming at distances of several body lengths above the bed during near-slack tide (line C in Fig. 3, where u=5 cm·s−1 at z=50 cm) and for slower forcing flows. Figure 2 also shows that, at a given height above the bed and for the same forcing flow at the top of the log layer, the mean horizontal velocity close to the sea bed will be substantially slower in rough- than in smooth-turbulent flow due to the more efficient mixing of high- and low-momentum fluid by eddies in the rough-turbulent flow. Larvae experience, however, only horizontal flow velocities within the viscous sublayer in smooth-turbulent flow, whereas for rough-turbulent flow, they experience the mean horizontal flow speed plus the fluctuating velocity components in all directions, as eddies regularly penetrate the viscous sublayer. Thus, while a larva may encounter unmanageable flow velocities for effective manoeuvring in the horizontal, it can swim up and down unperturbed by vertical flow velocity within the viscous sublayer for smooth-turbulent, but not for rough-turbulent flow. From this quantitative analysis of bottom boundary-layer velocity profiles in a realistic field flow environment, it appears that polychaete larvae probably do not actively swim horizontally among test sites, except under very low-flow conditions (i.e. around slack tides in the Buzzards Bay case). It seems more likely that larvae test habitats by sinking or swimming down to the bed and reject a site by swimming back up into the water column, although the potential effectiveness of this behaviour for rough-turbulent flow is unclear. Since near-bed velocities would carry the larvae over a suite of potential test sites, the habitats presented for their perusal are hydrodynamically constrained. Note that while the sites that a drifting larva may inspect are hydrodynamically determined, the larva may be carried over a wide range of habitats (at 1 cm·s−1, the larva is carried about 1 km·day−1), much farther than it can swim in the same amount of time. Sinking at a


125

rate of 0·1 to 1·0 cm·s−1 (as measured in Hannan, 1984a, b), the larva would, however, hit bottom after being advected only centimetres, so it would have to swim up at speeds greater than or equal to its fall velocity to stay above the bottom while drifting. It is possible that larvae do not select habitats by swimming among them; once they reach the sea floor, they may simply crawl between microhabitats, in which case the spatial scales for active habitat selection are very small indeed. Given that near-bed flow velocities over a relatively smooth, flat bottom may allow for very limited manoeuvring by larvae, any flow region with substantially lower velocities (e.g. in the lee of a relatively large roughness element, such as a rock or a tube, or in a dense canopy, such as a seagrass bed) may be particularly important to settling larvae. Whether they actively leave the flow (i.e. by swimming down or sinking) to enter such regions, or simply get deposited there (see pp. 151–4), they may be able to investigate actively such areas without significant interference from the flow regime. The velocity profiles constructed in Butman (1986a) are discussed in detail here because they are unique to the present day literature in larval ecology; they represent, however, conditions for but one class of flow environment (steady, uniform, tidally driven flows in shallow, coastal embayments) and for one group of infauna (the polychaetes). Furthermore, the analysis is limited by the lack of biological information, for example, on how larvae actually peruse available sites even in still water (i.e. is it by horizontal or vertical swimming, or some combination of the two, and from what height above the bed?) and on the relative swimming speeds and fall velocities of the organisms and the changes in these speeds over their pelagic life. While it is clear that hydrodynamics may limit the active habitat selection options for settling larvae, the capabilities of the larvae to overcome or utilize these flow obstacles are not clear. PASSIVE PARTICLE TRANSPORT AND DEPOSITION The long-held tenets that larval dispersal is primarily passive, via ocean currents, but that larval settlement is controlled by active larval behaviours have assumed that flows very close to the sea bed (e.g. within the viscous sublayer) were sufficiently slow to allow for searching and active habitat selection by the larvae. The likelihood that larval settlement is controlled, at least in part, by hydrodynamical processes is strengthened by the results of Butman (1986a) that relatively large (compared with larval swimming speeds, but see also Herrmann, 1979, and Lee, 1984) horizontal flow speeds occur within larval body lengths of the sea bed. If boundary-layer flow processes are transporting and depositing larvae, then the body of literature on sediment transport and deposition can be used to generate a priori predictions of depositional and accumulation sites for the organisms on the sea bed. The trajectory of a particle falling from the water surface to the bottom is determined by the horizontal displacement caused by the flow (advection) and by gravitational sinking of the particle. Once it reaches the a force per unit bottom bottom, the particle will settle on the sea bed if the bottom shear stress area, where =fluid density) does not exceed the critical value for suspension of the particle. This critical stress is usually reported in units of velocity as u*s, the critical suspension velocity. The ratio of particle fall 0·8, the particle will fall to the sea bed, but will move as bedload transport if 1972). For where u*c is the critical shear velocity for the initiation of particle motion. Bedload transport involves sliding, rolling or hopping of particles along the sea bed. From detailed laboratory measurements, curves have been constructed which allow prediction of u*c for given particle characteristics (diameter and density) and fluid characteristics (density and viscosity). The most common relationship used is Shields’ curve (Shields, 1936) or a subsequent modification (e.g. Miller, McCave & Komar; Yalin, 1977); these curves

126

CHERYL ANN BUTMAN

were, however, constructed from measurements on abiotic, non-cohesive sediments 100 in size and spread in homogeneous size classes (i.e. not size-class mixtures) on the bottom. Results for initiation of motion or suspension of fine, biotic or cohesive sediments and sediment mixtures (e.g. Nowell, Jumars & Eckman, 1981; Grant, Boyer & Sanford, 1982; Lick, 1982; McCave, 1985; Partheneides, 1986) have not yet been integrated into formal predictive functions, at the level of Shields’ curve, for example. The shear velocity of the flow (u*) and the particle fall velocity (w) are involved in all estimates or predictions of particle deposition and transport. Measurements of w now are reliable and routine for a wide range of non-aggregated particles; as mentioned earlier, determining the fall velocities of naturally occurring aggregates is still, however, troublesome (but see new in situ techniques in Bartz et al., 1985). Estimating u* for the suite of complex flow environments occurring in the field has been a primary focus in sediment-transport modelling over the last decade (Grant, 1977; Smith, 1977; Smith & McLean, 1977; Grant & Madsen, 1979, 1982; Grant & Glenn, 1983; McLean, 1985). From detailed field measurements of velocity at several heights within the log layer, it is possible to estimate u* from the slope of the line relating where z0=the bottom roughness parameter); the u and ln z (because, within the log layer, correlation between the two variables must, however, be extremely high (generally >0·990) for reasonable limits (e.g. Cryptopleura ramosa > Delesseria sanguinea; rs=1·00, P=0·05, Spearman Rank Correlation Coefficient Analysis). When other algal species were included in the comparison for Aplysia punctata, even though they would normally not have been encountered by sublittoral animals, the green alga Enteromorpha intestinalis was preferred over all others, the green alga Ulva lactuca was preferred after Plocamium, and the brown alga Laminaria digitata was consistently refused when any of the other seaweeds were available to be eaten (Table VI). Not only was there good correlation of growth and spawn production in the sea hares with food preference but, for the sublittoral red algae, this order reflected the order of distribution of animals in their sublittoral habitat (Carefoot, 1967c). Thus, of 641 animals collected sublittorally for which the type of seaweed on which each animal was found was recorded, 96% were found on Plocamium, about 2% on each of Heterosiphonia and Cryptopleura, and less than 0·5% on Delesseria. In comparison with these small sublittoral animals, large intertidal Aplysia punctata clearly preferred Enteromorpha and Ulva over other seaweeds. Laminaria, contrary to other reports (e.g. Eales, 1921), was completely refused even though this alga would presumably be encountered regularly in the intertidal habitat (Carefoot, 1967c). Kupfermann & Carew (1974) used SCUBA diving and snorkelling to observe Aplysia californica in its natural habitat to determine which seaweeds were preferred. They found: (1) that certain seaweeds, as for example Laurencia sp. and Gigartina sp., were distinctly favoured, whereas others were eaten only infrequently (see Table V), (2) that the size of an animal determined whether certain brown seaweeds would be eaten (smaller animals tended to avoid these; see also Winkler & Dawson, 1963), and (3) that some common red seaweeds such as Pterocladia sp. were consistently avoided, whereas Laurencia, even when rare, was consistently eaten. In a comprehensive survey of algal foods of Aplysia californica, Winkler & Dawson (1963) noted that different populations had their own distinct preferences, which varied depending on the algae present in a given area. Red algae were observed to be eaten most by A. californica including, surprisingly, several species of corallines (Table V). The preferred foods of several populations were the red algae Hypnea valentiae, Plocamium cartilagineum, Laurencia pacifica, and Ceramium eatonianum. Winkler & Dawson listed a number of red and brown seaweeds which were common in one area (Lunada Bay, near Palos Verdes), but never eaten by Aplysia californica. Other than these few data on A. punctata and A. californica, there have been no field studies which have comprehensively related food choice in sea hares with availability of seaweeds in a species’ habitat. The general preference by sea hares for green algae has been documented in laboratory studies. Winkler & Dawson (1963) observed that A. californica, while feeding almost exclusively on red algae in the field, exhibited a strong preference for Ulva and Enteromorpha in the laboratory. Saito & Nakamura (1961) found that both Aplysia Juliana and A. kurodai preferred these green seaweeds in laboratory situations to what the authors considered were their normal foods of brown and red algae, respectively. (The contention by Saito & Nakamura that A. juliana normally eats the brown alga Undaria pinnatifida is difficult to accept. All other studies on A. juliana have shown that green algae, principally Ulva spp., but including Enteromorpha, and some Cladophora, are the only foods eaten: Frings & Frings, 1965; Carefoot, 1970; Usuki, 1970b; Sarver, 1978, 1979; Switzer-Dunlap & Hadfield, 1979; Vitalis, 1981.) Laboratory-held Aplysia dactylomela in Barbados were observed to eat preferentially Enteromorpha, Ulva, and Cladophora when given a choice of several seaweeds (Carefoot, 1970). Sawaya & Leahy (1971) have noted that field Aplysia brasiliana=willcoxi and A. dactylomela in the São Paulo area of Brazil principally eat Ulva spp. Finally, both Aplysia depilans and A. fasciata appear to favour a diet of Ulva (Jordan, 1917; Susswein, et al., 1984a; Achituv & Susswein, 1985). It is not known what characteristics make sea hares prefer one seaweed over another. Factors such as energy content and nutritional value have doubtless been important in the evolution of optimal diet in

182

THOMAS H.CAREFOOT

Aplysia, but probably no more so than texture, how easily it can be manipulated, its palatability and digestibility, and its availability to the animal. Perhaps it is not surprising that despite a wealth of information on feeding preferences in sea hares and other opisthobranch gastropods (Stehouwer, 1952; Braams & Geelen, 1953; Cook, 1962; Carefoot, 1967a, 1970; Edmunds, Potts, Swinfen & Waters, 1974; and others), there is not one instance where the factor or factors governing the selection of a food can be precisely identified (Carefoot, in press). As an example, the series of dietary preferences in Table VI for A. punctata, which shows such good correlation with growth and spawn production, does not show significant correlation with absorption of total dry matter on each diet, nor with absorption of total nitrogen, specific amino acids, or total carbohydrates (Carefoot, 1967a). Moreover, an experiment using a mechanical device to simulate the triturating action of the A. punctata gizzard showed that the two algal species most susceptible to being broken down by such treatment were the red alga Delesseria sanguinea and the brown alga Laminaria digitata. These were the two seaweeds Aplysia punctata preferred least and represented two of the three species giving poorest growth and spawn production on ad libitum diets (Carefoot, 1967a). Similarly, scrutiny of Niell’s (1977) comprehensive data on foods eaten by A. punctata in the Ria de Vigo area of Spain showed no apparent influence of texture on food choice. The 11 algae (of 44 species in total) found most frequently in the guts of 33 individuals represented seven different morphological categories of seaweeds (characterized by membranous blades, soft filaments, soft branches, rigid branches, firm rounded cylinders, flattened cylinders, and flattened segments with dichotomous branches), and had greatly differing textures. Of these 11 species, one was green, two were brown, and eight were red. It is clear that sea hares’ feeding preferences vary with habitat. This is most TABLE VI Feeding preferences of Aplysia: data were obtained from differences in amounts of seaweeds consumed when animals were presented with a choice of two or more algal species; the exceptions to this were the studies on A. californica by Brady & Young (unpubl.), which represented the difference in distribution of animals and different food substrates when presented with a choice of six different algae, and by Chapman & Fox (1969), which represented qualitative estimates of amounts of each alga consumed; A=angiosperm; C=Chlorophyceae; P=Phaeophyceae; R=Rhodophyceae Species

californic a

P P R P P

P

Food

Rank of Rank of feeding value for preference growth s

Egregia menziesii Macrocyst is pyrifera Gigartina armata Eisenia arborea Pterygoph ora californica Laminaria farlowii

1

3 4 5

6

Rank of value for spawn productio n

Rank of distributio n of animals on each food in the field

Rank of larval settling preference (lab. or field)

Rank of algal abundanc e in the field

References

Leighton, 1966 2

APLYSIA

Species

P

californic a

R

R

R

californic a (animals on a red algae-free diet)

P

P P P P P P P

C californic a

R

R

Food


Cystoseira osmundac ea Laurencia pacifica

7

Plocamiu m cartilagine um Pterocladi a pyramidal e Egregia menziesii

2

Eisenia arborea Laminaria sp. Petalonia debilis Macrocyst is pyrifera Pelvetia fastigiata Zonaria farlowii Dictyopter is zonarioide s Codium fragile Plocamiu m cartilagine um Laurenica pacifica

3·5




1


References

Chapman & Fox, 1969

3

1

Chapman & Fox, 1969

3·5 3·5 3·5 7·5 7·5 7·5

7·5 1·5

1·5

1

183

Brady & Young, unpubl.

184

THOMAS H.CAREFOOT

Species

C

dactylomela

dactylomela

juliana

juliana

kurodai

Food


Enteromor pha intestinalis

5·5

C R R P A C C C R R R C C R R R P P C C P C

Codium fragile Gelidium purpurascens Pterocladia capillacea Macrocystis pyrifera Phyllospadix torreyi Enteromorpha sp. Ulva fasciata Cladophora sp. Laurencia papillosa Galaxaura oblongata Laurencia spp. Enteromorpha sp. Ulva lactuca Plocamium costatum Champia laingii Herposiphonia sp. Undaria pinnatifida Ecklonia cava Enteromorpha sp. Ulva pertusa Endarachne binghamiae Ulva fasciata

5·5 5·5 5·5 5·5 5·5 1 2 3 4 5 1 2 3 4 5 6 1 2 4 4 4 1

C C C C C C R R

Ulva reticulata Enteromorpha sp. Ulva fasciata Enteromorphia sp. Enteromorpha sp. Ulva pertusa Grateloupia filicina G. okamurai

2 3 1 2 1·5 1·5 4 4


1 2 4 3 5 1 2

1 2 4 3 5




References

Carefoot, 1970, 1985 2 1 1

Willan, 1979

3 Saito & Nakamura, 1961

1

1·5

2 3 1 2

1·5 3 2 1

Vitalis, 1981; Switzer-Dunlap & Hadfield, 1977 Carefoot, 1970

1 Saito & Nakamura, 1961

APLYSIA

punctata

R C R C R R R P

G. tsurutsuru Enteromorpha intestinalis Plocamium cartilagineum Ulva lactuca Heterosiphonia plumosa Cryptopleura ramosa Delesseria sanguinea Laminaria digitata

4 1 2 3 4 5 6 7

2 1 3 4 5 6 7

1 2 3 4 6 5 7

185

Carefoot, 1967a, c 1 2·5 2·5 4

1

1 2 3 4

evident from examination of the data given for A. californica in Table VI, representing studies by Brady & Young (unpubl.) and Leighton (1966). Feeding preferences appear to differ markedly in populations from the different areas. Because one study tested the preferences of sublittoral animals for algae prevalent in kelp beds (near La Jolla, California: Leighton, 1966), while the other tested the preferences of shallowwater animals for algae associated with this type of habitat (Catalina Island, California: Brady & Young, unpubl.), perhaps the differences in preference-ranking were, however, predictable. In addition to habitat, feeding preferences could be predicted to vary with season and with changing nutritional needs associated with age, sex, and reproductive state. None of these latter has been investigated in sea hares, nor is it known whether sea hares would show a predictably higher degree of selectivity in their choice of foods when satiated or when food is common, and be less discriminating when starved or when food is scarce (Emlen, 1966). Differences in food selectivity in relation to size have been mentioned by Winkler & Dawson (1963) and Kupfermann & Carew (1974), but as yet no one has investigated this topic. The broad range of feeding modes in sea hares, from almost strict monophagy (e.g. A. juliana) to polyphagy (e.g. A. californica, A. dactylomela, and A. punctata), would provide a challenging basis for a further study of feeding preferences in this group. Food choice in sea hares is also affected by past dietary history. “Ingestive conditioning”, or the enhanced response of an animal to its food based on previous exposure to the same food, was observed in A. punctata in animals maintained for 80 days on single seaweed species (Carefoot, 1967a). The effect of this conditioning was short-lived, and within a few days of exposure to other seaweed foods the conditioned animals reverted to their normal level of preference. In studies on the effect of previous dietary history on food choice by A. californica, Brady & Young (unpubl.) also demonstrated ingestive conditioning. They found that animals previously fed on Plocamium cartilagineum for one month strongly preferred this alga over Laurencia pacifica when tested against both species in a Y-maze. Conversely, animals fed for as short a period as four days on Laurencia showed a measurable (but non-significant) preference for this alga over Plocamium. Internal food stimuli were shown by Susswein, Weiss & Kupfermann (1984b) to enhance feeding behaviour in Aplysia californica. By feeding experimental animals on small amounts of dried seaweeds (species unspecified) and control animals on filter paper, or by just stimulating the control animals by touching seaweed to their mouths then testing biting responses of both experimental and control groups to seaweeds, the authors were able to demonstrate a significant decrease in latency to bite as a response to eating in A. californica. In other words, after having eaten, the animals were quicker to respond on subsequent encounter with food. The effects lasted for up to 80 minutes following the stimuli for all groups. Other types of non-associative learning by Aplysia in connection with food and feeding are: (1) a decrease in feeding response after repeated stimulation with a non-food object (forceps and glass rods: Lickey, 1968; Lickey & Berry, 1966) and, (2) an inhibition of the feeding response following an electrical

186

THOMAS H.CAREFOOT

shock (Kupfermann & Pinsker, 1968). There are, therefore, a number of examples of non-associative learning or habituation responses in connection with food and feeding in Aplysia. In comparison, only a few examples of associative learning have been described for sea hares in this context. Jahan-Parvar (1970) described an instance of classical conditioning in A. californica, using seaweed as the unconditioned stimulus and light as the conditioned stimulus to train animals to show typical food-seeking behaviour, but this could not be successfully repeated by Kupfermann (1974b). Susswein & Schwarz (1983) and Susswein & Markovich (1983) trained A. fasciata and A. californica not to eat Ulva lactuca nor leaves of the lily Hemerocallis fulva which were wrapped in plastic netting. The animals could taste the food through the holes in the net and would initially bite and attempt to swallow the preparation. An essential component of the training exercise was, however, the tendency of the net-bound food to become stuck in the animal’s buccal cavity causing the animal to gag, which may have acted as a kind of punishment. After the animals had learned not to respond to the net-enclosed food, they could still be induced to eat nonnetted food (Schwarz & Susswein, 1982; Susswein & Markovich, 1983; see also Schwarz & Susswein, 1984). Some memory was retained by Aplysia fasciata 24 hours following training. Recently, Cook & Carew (1986) demonstrated a similar type of associative learning in Aplysia, that of operant conditioning. In their experiments the authors showed that A. californica could be operantly trained to change their explorative head-waving, which is normally side-to-side as when they are sensing food, to one side more than another in order to terminate the shining on them of an aversive strong light. The existence of a negative ingestive conditioning, in this case a post-ingestive “learning aversion” to foods, has been investigated in A. dactylomela using artificial diets with large imbalances of amino acids (Carefoot & Switzer-Dunlap, unpubl.). Preliminary results have suggested that individuals previously fed on such a nutritionally poor diet can recognize it on a subsequent encounter, remember their past experience, and avoid it. Such post-ingestive learned responses, both preference and aversion, may have played important rôles in modifying feeding behaviour in sea hares, not only in day-to-day selection of foods, but also in the longterm evolution of optimal diets. FEEDING AND MOVEMENT OF FOOD THROUGH THE GUT The mechanism of action of the radula and jaws is well described in Eales (1921) and Howells (1942), and will not be repeated here. Bite-sized pieces of algae are stored in a voluminous crop which, when full in A. dactylomela, can hold up to 10% of the live body weight (Carefoot, 1985). In A. californica, the total weight of anterior gut contents (oesophagus, crop, and gizzard) may represent 20% of the total body weight in satiated animals (Susswein & Kupfermann, 1975b). Comparable values for weight of anterior gut contents in satiated A. fasciata, A. depilans, and A. oculifera are 7·9, 11·9, and 10·8%, respectively (Susswein & Markovich, 1983). From the crop, the food moves to the two-part gizzard, each part containing a number of chitinized teeth (see Winkler, 1960; Beeman, 1969; Arnould & Jeuniaux, 1977 for information on the chemical characteristics of the teeth). Here, the food is macerated and the slurry, barely recognizable as seaweeds, is passed into the stomach (Howells, 1942). Eales (1921) believes the gizzard teeth to be of little use for grinding, but rather thinks their function is to compress and strain the food in preparation for action by digestive enzymes. During the process of digestion the crop and gizzard show rhythmical contractions which can continue for some time in in vivo preparations (Bottazzi, 1897). The possible rôles of neurotransmitter substances such as acetylcholine and serotonin in controlling these rhythmical movements have been investigated in A. dactylomela by Wells & Hill (1980, 1985). The effect of FMRFamide (Phe-Met-Arg-Phe-NH2) on

APLYSIA

187

contraction of the gizzard has been studied in A. californica (Austin, Weiss & Lukowiak, 1983), as has that of atropine in connection with oesophageal contractions in the same animal (Winkler & Tilton, 1961). Chemical digestion occurs in all regions of the fore- and mid-gut regions, and is regulated by enzymatic secretions produced in the salivary and digestive glands. The food moves from the stomach to the caecum, where digestion is completed, then travels via the intestine to the rectum. The faeces emerge from the anus into the mantle cavity area as loose, irregular pellets. From time to time the animal may forcibly expel water from the mantle cavity, which helps to carry these pellets to the outside. The large volume of food processed each day, combined with its high roughage content, results in a large production of faecal matter, to the extent that animals in confined laboratory conditions sometimes bury themselves totally in their own faeces. Sand may comprise a large portion of the gut contents in some species (e.g. A. californica: Winkler, 1961; A. dactylomela: Carefoot, 1970). In populations of A. dactylomela in Barbados sand can represent up to 28% of the dry weight of the crop contents in freshly collected and recently fed animals, and may account for a large part of the voided faecal residues (Carefoot, 1985). In fact, a minimum estimate of annual turnover of sand by A. dactylomela (of 60 g mean live weight) per linear km of suitable coastal habitat in Barbados is one metric ton (Carefoot, 1985). Its rôle in digestion, if any, is not known. It could act in digestion as a bulk carrier, by aiding in movement of food through the gut, or it could act as an adjunct to the gizzard, by aiding in the mechanical breakdown of algal tissues. Alternatively, its occurrence could be incidental, a simple consequence of feeding in shallow sand-swept areas on tightly meshed, sand-infested algae such as Cladophora sp. In Barbados, a shallow-inshore population of Aplysia dactylomela was observed to feed preferentially in such a Cladophora-dominated habitat. Interestingly, this area, almost in the wave-break part of the shore and characterized by extensive movements of sand, was chosen by the bulk of the population to feed in over a less sand-swept offshore area dominated by red algae. The animals used the latter area for their copulations, as well as to hide in during the day. Based on energetic and nutritional properties of the different algal foods in the two areas, and on the fact that the inshore Cladophora habitat offered few daytime retreats, therefore requiring that most animals undertake a return daily feeding excursion of several metres to and from the feeding area, the animals would have appeared to be better off in the offshore red-algal habitat. A comparison of amount of sand eaten by Aplysia dactylomela in the two areas, however, disclosed little difference (Carefoot, 1985). Since the offshore red algae (mainly Gracilaria sp. and Laurencia papillosa) contained visibly less entrapped sand than did Cladophora, this showed that sand might be important in digestion as the animals may have been eating it directly. FEEDING RATES Feeding rates of sea hares are thought to be influenced by a number of intrinsic and extrinsic factors, only a few of which have been studied. Chief amongst these are likely to be temperature, salinity, body size, type of food, time of day, state of tide, season, density of animals, reproductive status, and past feeding history. Data on feeding rates of sea hares are generally derived from laboratory animals, and the question arises as to whether these rates would differ from rates measured directly in the field. A test of this using bagenclosed Aplysia dactylomela in the field in Barbados showed no significant differences between laboratory and field rates when Cladophora sp. and Ulva fasciata were the foods, but did show a significant difference S.E. equivalent Joules eaten-live g−1·day−1, as when Laurencia papillosa was being eaten (field rate: S.E. equivalent J eaten·live g−1·day−1; Carefoot, 1985). compared with laboratory rate: Ingestion rates for Aplysia are given in Table VII and are shown graphically for a number of species in Figure 4. In this Figure the data are expressed as dry mg food eaten-dry g animal−1·day−1 as a function of live weight for two arbitrary temperature ranges, 15–17°C and 18–28°C. The data are varied but the two

188

THOMAS H.CAREFOOT

sets best fit the relationship where a and b are constants. This is consistent with observations on feeding in other gastropods where consumption (C) is related to body weight (W) by the (Edwards & Huebner, 1977; Bayne & Scullard, 1978). equation, The effect of temperature on feeding rates in A. juliana has been investigated by Saito & Nakamura (1961), although the results are difficult to interpret owing to variable past feeding histories and to variable weights of animals used in the experiment. In general, however, the rates of consumption increased with increasing temperature as expected. At 7–9°C animals on a diet of the brown alga Undaria pinnatifida ate about 1 % of their live body weight in fresh weight of algae per day, increasing to a value of about 11 % at 22–24°C. At temperatures higher than 24°C, feeding rates declined by an order of magnitude, and the animals died three days into the four-day experiment. Little is known of the factors that stimulate sea hares to feed. The only natural phagostimulant known for Aplysia is crude water-extract of seaweed. This was first demonstrated by Frings & Frings (1965) in A. juliana using water-extracts of Ulva lactuca, and confirmed later for Aplysia dactylomela using extracts of Ulva fasciata (Carefoot, 1979, 1980). Susswein et al. (1976b) induced Aplysia californica to bite by applying water extracts of dried laver (Porphyra sp.) to the animals’ mouths. A graded response was obtained depending on the concentration of extract and degree of satiation of the animals, although the chemical identities of these water-extracted phagostimulants are unknown. Frings & Frings (1965) determined that the extract from Ulva lactuca was heat-stable, insoluble in ether, and acted only on receptor sites near or in the mouth and on the oral tentacles, but not on the rhinophores. In later studies, Sakata et al. (1985; 1986) found that ether-extracted substances from U. pertusa, identified as glycerolipids, were highly attractive to Aplysia juliana, and the authors proposed a phagostimulatory rôle for these substances. Phagostimulatory properties of various chemical TABLE VII Feeding rates of Aplysia: C=Chlorophyceae; P=Phaeophyceae; R=Rhodophyceae Species

dactylomela

C C C R

juliana

C

P C

Food

Body weight Temp. °C No. of days Amount of (mean live g) food eaten (fresh g·animal−1 ·day−1)

% of mean body wt eaten·day−1

References

Cladophora sp. Enteromorp ha sp. Ulva fasciata Laurencia spp. Ulva pertusa

61·8

28

14

11·7

18·9

Carefoot, 1970

77·7

28

14

14·5

18·7

82·6

28

15

4·4

5·4

25·5

17

37

3·0

11·9

Willan, 1979

210·0

18–21

6

21·0

10·0

Saito & Nakamura, 1961

336·0

21

4

25·4

7·6

46·4

28

15

3·4

7·3

Undaria pinnatifida Ulva fasciata

Carefoot, 1970

APLYSIA

Species

C

Food

Body weight Temp. °C No. of days Amount of (mean live g) food eaten (fresh g·animal−1 ·day−1)

% of mean body wt eaten·day−1

29·2

28

9

10·2

35·0

110·0 239·0

25 16–20

104 5

23·1 14·8

21·0 6·2

kurodai

C C

Enteromorp ha sp. Ulva lactuca Ulva pertusa

punctata

C

Ulva lactuca

8·0

15

15

0·3

3·7

C

Enteromorp ha intestinalis Plocamium cartilagineu m Plocamium cartilagineu m Plocamium cartilagineu m Plocamium cartilagineu m Plocamium cartilagineu m

13·0

15

15

1·1

8·5

14·4

15

15

0·7

4·9

3·1

15

20

0·4

12·9

7·1

15

20

0·8

11·3

12·5

15

20

0·9

7·2

17·3

15

20

0·9

5·2

R

punctata

R

R

R

R

189

References

Sarver, 1978 Saito & Nakamura, 1961 Carefoot, 1967a

Carefoot, 1967b

substances were also investigated in A. dactylomela and A. kurodai (Carefoot, 1982b). Greatest phagostimulatory responses were elicited by starch, L-glutamic and L-aspartic acids, maltose, oleic acid, and combinations of certain vitamins. Interestingly, combinations of these phagostimulatory materials did not enhance feeding activity in A. kurodai and A. dactylomela; rather, their effects were negatively synergistic, resulting in a “masking” of the phagostimulatory properties of the components (Carefoot, 1982b). An important ramification of this discovery with regard to nutritional studies on sea hares using artificial diets of chemicals set in agar, is that “super diets” of irresistible palatability cannot apparently be created by combining several individually phagostimulatory materials. Satiation can be induced in sea hares in both the laboratory (Kupfermann, 1974b; Susswein & Kupfermann, 1974, 1975a, b; Susswein et al., 1976a, b; Kuslansky et al., 1978; Susswein & Markovich, 1983) and the field (Kupfermann & Carew, 1974). It is not known for certain that sea hares eat to satiation under natural conditions in the field. Because Kupfermann & Carew (1974), however, observed behaviour in field A. californica that was consistent with behaviour of satiated animals in the laboratory, field animals probably do eat to satiation. The quantity of food needed to produce satiation in A. californica appears to be determined by two inputs: (1) an external cue related to palatability (Susswein & Kupfermann, 1974) and,

190

THOMAS H.CAREFOOT

Fig. 4.—Rates of food consumption by sea hares: the data are arbitrarily divided into two groups representing the temperature regimes 15–17°C and 18–28ºC; equation of regressions, for 15for 18–28°C, the number by each point identifies the reference; 1, Saito & Nakamura (1961); 2, Carefoot (1967a); 3, Carefoot (1967b); 4, Carefoot (1970); 5, Sarver (1978); 6, Willan (1979).

(2) an internal cue related to the bulk properties of the food (Susswein & Kupfermann, 1975a, b). To this extent, the situation in Aplysia parallels that in vertebrates; what has not yet been identified in sea hares is the third major input noted by Susswein & Kupfermann (1975a, b) to be present in vertebrates: that of metabolic cues involved in the regulation of long-term energy and nutritional needs. An interesting example of satiation or near-satiation occurs with A. dactylomela in Barbados, where animals fill their crops in pre-dawn bouts of intense feeding, then hide for 12 hours or more during daytime to digest the food (Carefoot, 1985). Studies on A. californica in the field suggest that the animals normally partition their feeding into discrete meals (Kupfermann & Carew, 1974). The meals are separated by periods of time when portions of food are transported out of the crop (Susswein et al., 1978). Because satiation in A. californica can be induced by a non-nutritive bulk material such as filter paper, or silicone- and polyacrylamide-based gels, it is likely that some kind of mechanoreceptor in the crop is responsible for terminating feeding and not a feed-back signal based on chemical characteristics of the food (Susswein & Kupfermann, 1975a, b; Kuslansky et al., 1978). This feed-back signal from presumed stretch receptors in the crop occurs in A. californica only after a delay of some 15–20 minutes following satiation by the artificial introduction of a seaweed mash (laver: Porphyra sp.) or non-nutritive polyacrylamide gel (Kuslansky et al., 1978). As noted by Kuslansky et al. (1978), a delay of this duration opens the possibility that the signal from the gut receptors indicating satiation is hormonally (neurosecretory?) rather than neuronally mediated. LARVAL FEEDING RATES The veliger larva of Aplysia feeds on unicellular phytoplankton which it captures through the filtering action of the velum, a densely ciliated organ which also provides for locomotion. Gallager & Mann (1980) measured the feeding rates of larval sea hares and found that the highest rate of grazing on the flagellate

APLYSIA

191

Isochrysis galbana by veliger larvae of Aplysia californica (about 105·3 cells·ind−1·day−1) occurred at algal cell concentrations of 105·ml−1. At concentrations higher or lower than this optimum level, grazing rates were markedly lower. For example, at a flagellate concentration of 104 cells·ml−1, a level customarily employed in larval cultures of A. californica and other sea hares (Kriegstein, Castellucci & Kandel, 1974; Switzer-Dunlap & Hadfield, 1977, 1981; Otsuka et al., 1981), rates of grazing were about 20 times less than at the optimal concentration. On the other hand, at flagellate densities of 106·ml−1, the lowest concentration employed in the culturing of A. brasiliana=willcoxi by Strenth & Blankenship (1978a), rates of grazing by A. californica veligers as measured by Gallager & Mann (1980) were about six times less than at the optimal concentration of 105 cells·ml−1. DIGESTION AND ABSORPTION Sea hares appear to digest mainly starches and simple sugars. Chemical digestion takes place primarily in the stomach and digestive gland, but may be initiated in the crop when digestive juices are regurgitated from the stomach (Howells, 1942). Howells reported that amylases are active in secretions from the salivary and digestive glands, and that the latter organ secretes a number of enzymes which hydrolyse sucrose, lactose, and maltose. A protease is present in all regions of the gut, including secretions from the salivary glands, and lipase activity has been identified (Howells, 1942; Cho, Pyeun, Byun & Kim, 1983). While Howells noted that a cellulase was absent or only weakly present in A. punctata, this enzyme was later positively identified in the crop and digestive gland secretions of this species by Stone & Morton (1958). Koningsor & Hunsaker (1971) and Koningsor, McLean & Hunsaker (1972) also found a cellulase in the crop juices of A. vaccaria. In none of these instances, however, has an exogenous origin of the enzyme, as for example from symbiotic bacteria in the gut, been entirely ruled out. The fate of digested food particles in sea hares, particularly the sites of absorption and types of transport mechanisms involved for such materials as sugars, peptides, and amino acids, are not well known. In comparison, transport mechanisms of sodium, chloride, and other ions across the intestinal epithelia in sea hares, especially A. californica, have been extensively investigated by Gerencser (1978, 1979a, b, 1981a, b, 1982, 1983, 1984a, b), Gerencser & Hong (1977), Gerencser & White (1980), Gerencser & Loughlin (1983), and Gerencser & Lee (1985). The value of a food to Aplysia should be reflected in the efficiency with which it is digested and absorbed. Highest values would be expected on diets giving the best growth and in instances of monophagy (e.g. A. juliana), where strong selection for traits which would maximize the capacity for digesting and absorbing a (where single foodstuff would be expected. Values for absorption efficiency, expressed as C=food consumed and F=faeces produced), are given in Table VIII for sea hares eating a variety of diets. It may be seen that absorption of total dry matter ranges from 15– 84%, and highest values are predictably displayed by the monophagous A. juliana eating its favoured diet of Ulva spp. (73–84%; Carefoot, 1970; Sarver, 1978). In Aplysia punctata, absorption of total nitrogen varies from 54–79%, with no correlation with diet. Similarly, absorption of carbohydrates varies from 55–84%, again with no correlation with diet, but paralleling the pattern found for absorption of nitrogen. When the actual amount absorbed of each of these materials is, however, calculated over a known period of time for the various diets shown in Table VIII for A. punctata, the values obtained correlate positively and almost perfectly with the quality of each diet as reflected by its growth-promoting ability (Carefoot, 1967a). Uptake of specific amino acids by A. punctata was found to be highly variable, ranging from 0–100% for a variety of seaweed diets (Carefoot, 1967a). For animals in a sublittoral area studied in the Irish Sea, highest mean uptake of 17 amino acids was realized on a diet of the red alga Plocamium cartilagineum

192

THOMAS H.CAREFOOT

(mean of ). This seaweed gave the best growth and was the species on which the animals were most commonly found (Carefoot, 1967a, c). Overall, the highest mean value for absorption of amino acids was found for Aplysia punctata eating three species of red algae used in the study (75% as noted above for for animals eating Heterosiphonia plumosa, and for ones eating Delesseria Plocamium, sanguinea). In comparison, absorption of amino acids from diets of the green seaweeds Enteromorpha and respectively). intestinalis and Ulva lactuca was markedly lower ( A special feature of green algae which may bear relevance to their usefulness as foods for Aplysia, is their generally low content of amino acids (e.g. 3·5 and 5·1% of total dry weight in Enteromorpha intestinalis and Ulva lactuca, respectively, as compared with 18·2–21·2% for the three species of red algae listed above; Carefoot, 1967a, b). When the two factors of poor absorption and low concentration of amino acids are combined, a picture emerges which TABLE VIII Absorption and growth efficiencies for Aplysia: A=absorption; C=Chlorophyceae; CHO=carbohydrates; F=faeces; N=nitrogen; P=Phaeophyceae; R=Rhodophyceae; C=consumption; P=production Absorption efficiency (C−F/C×100%) Species dactylomela

R

dactylomela

C C C R R

juliana

C C C C C C C

punctata

R

punctata

R

N

Growth efficiencies

Food

Temp. °C Total dry mattera

CHO Gross, K1 Net, K2 References (P/C)×100% P/A×100%

Laurencia spp. Cladophora sp. Enteromorph a sp. Ulva fasciata Galaxaura oblongata Laurencia papillosa Enteromorph a sp. Ulva fasciata Cladophora Ulva lactuca Ulva lactuca: low rationb Ulva lactuca: medium ration Ulva lactuca: high ration Plocamium cartilagineum Plocamium cartilagineum

17

79

22

28

Willan, 1979

28

35

29

84

Carefoot, 1970

28

68

45

67

28 28

62 24

27

43

28

67

33

49

28

69

14

20

28 28 25 25

84 15 73

28

33

16 15

22

Sarver, 1978

Carefoot, 1967b Carefoot, 1967a

25

11

25

12

15

67

74

73

21

31

15

65

74

72

23

35

Carefoot, 1970

APLYSIA

Absorption efficiency (C−F/C×100%) Species C C R R R P a b

193

Growth efficiencies

Food

Temp. °C Total dry mattera

N

CHO Gross, K1 Net, K2 References (P/C)×100% P/A×100%

Enteromorph a intestinalis Ulva lactuca Heterosiphoni a plumosa Cryptopleura ramosa Delesseria sanguinea Laminaria digitata

15

59

68

69

17

29

15 15

75 71

79 74

84 76

18 15

24 21

15

71

73

76

11

15

15

45

54

57

33

74

15

53

57

55

20

38

Expressed in Joules in some studies. See Table IV for ration levels used in this experiment.

suggests that green algae may not, in fact, be particularly good sources of amino acids for Aplysia. This may explain why Ulva spp., which are otherwise favoured foods for sea hares and which are eaten well in laboratory studies, actually promote only poor growth in all species but Aplysia juliana (Carefoot, 1967a, 1970; see also pp. 234–236). Enteromorpha seems to be a superior diet for sea hares (Carefoot, 1967a, 1970), but measurements of consumption rates in ad libitum laboratory studies on Aplysia punctata have shown that 50–100% more of this green alga must be eaten to yield growth rates comparable with rates attained on the best red-algal diets (Carefoot, 1967a). It would be interesting to know whether absorption of amino acids is more efficient in A. juliana to match the overall high absorption of dry matter and high rates of growth exhibited by this species on its specialized diet. In addition to food quality, other factors which may affect absorption efficiency in Aplysia are temperature, feeding rate, age, reproductive state, and density of individuals. All are known to affect absorption efficiencies in other gastropods (Carefoot, 1987) but, save for feeding rate, have not been investigated in sea hares. As expected, feeding rate is negatively correlated with absorption efficiency in A. punctata (Carefoot, 1967b). Also, sea hares rapidly eating green algae may digest and absorb their foods so poorly that their faeces are coloured bright green (see also Winkler & Dawson, 1963). This suggests that in such circumstances of “superfluous feeding” (e.g. Ryther, 1954) the faeces may contain substantially more unused nutrients than when animals feed more slowly. After such bouts of quick feeding, A. punctata may occasionally eat its own faeces (Carefoot, 1967a). Coprophagy, although apparently rare in sea hares, may play a rôle in nutrition. NUTRITION Despite considerable research on diets, feeding preferences, and growth in sea hares, knowledge of their specific nutritional requirements is scanty (Carefoot, 1967a, b, 1970; Sarver, 1978). Reasons for this partly relate to difficulty in ascertaining the precise nutritional quality of the seaweed foods and to the absence of any radiotracer studies to monitor uptake and assimilation. In addition, large seasonal changes in chemical composition of seaweeds, diffusion of nutrients from algal tissues during growth, and loss of nutrients

194

THOMAS H.CAREFOOT

during the mechanical processes of feeding create special problems in determining the precise nutritional needs of sea hares. The development of artificial diets, made up entirely of chemicals bound in agar, has solved some of these problems and has provided a means for preliminary assessment of nutritional needs in Aplysia (Carefoot, 1979, 1980). Experimental diets are eaten well by some species, but not by others. They sustain some growth and spawn production in A. dactylomela, maintain constant weight in A. kurodai, but are not eaten at all by A. juliana. When juvenile A. kurodai were fed on such chemically defined diets, each diet deficient in a different single amino acid, no significant weight losses occurred after 24 days on any of the diets (Carefoot, 1981b). Assuming that this period was sufficiently long for amino acid deficiencies to become apparent in these fast-growing animals, the results suggest that either A. kurodai do not require the usually recognized 10 essential amino acids (as shown for the rat), or that they obtain required amino acids from another source. Save for the possibility that uptake of dissolved organic matter might provide some of the missing amino acids, the most likely source is from symbiotic bacteria in the gut. In this regard, Ghiretti, Ghiretti-Magaldi & Tosi (1959) isolated several strains of bacteria from the digestive glands of A. depilans and A. fasciata, and Vitalis (1981) identified some twelve different colony types of bacteria from the crop and stomach regions of A. juliana. Six of these types in A. juliana were abundant and appeared consistently; the others were sporadic and always occurred in small numbers. A preliminary study by Spence (unpubl.) on the effect of diet on the composition of bacterial flora, their distribution through the gut, and the effect of antibiotics on survival of the bacteria in the guts of A. dactylomela showed the following: (1) that up to 23 bacterial strains could be identified in the gut of A. dactylomela, (2) that the bacteria existed more or less evenly throughout the crop, gizzard, and digestive gland, both in numbers of species and numbers of individuals to bacteria·wet g gut tissue plus fluids−1 throughout these regions), (3) (numbers ranged from that animals starved for five days showed a slight, but non-significant, decrease in numbers and types of bacteria in the gut, (4) that numbers and types of bacteria were somewhat greater in animals eating red algae (Spyridia sp. and Laurencia sp.) than in ones eating Ulva sp., (5) that six bacterial types appeared to be ‘resident’ forms, present in the gut regardless of diet and also present in starved animals, (6) that oral administration of antibiotics (ampicillin, neomycin, and tetracycline incorporated into artificial diets of chemical nutrients set in agar) resulted in a one order of magnitude reduction in numbers of gut bacteria over 10 days of treatment and, finally, (7) that in vitro tests of susceptibility of the bacteria to several antibiotics showed that the most effective antibiotics amongst those tested were ampicillin and chloromycetin, and the least effective, erythromycin, kanamycin, neomycin, streptomycin, penicillin, tetracycline, and polymycin B. The fact that 12 types of bacteria were identified in A. juliana (Vitalis, 1981), as opposed to 23 types in A. dactylomela, further suggests that types of bacteria may be somewhat greater in animals eating red algae than in ones eating green algae. Some of the antibiotics employed in the study (e.g. tetracycline) appeared to be distasteful to A. dactylomela in concentrations necessary to kill significant numbers of the bacteria. Also, in several instances, bacterial numbers actually increased after eight days of antibiotic treatment in the in vivo experiments, possibly caused by a secondary invasion of antibiotic-resistant forms following an early reduction in numbers of the primary populations by the antibiotics (Spence, unpubl.). Vitalis (1981) undertook a series of experiments with A. juliana, fed on Enteromorpha sp., Ulva reticulata, and U. fasciata, to assess the nutritional contribution made by gut bacteria. He showed that growth of animals exposed to antibiotics dissolved in the sea-water medium (streptomycin and penicillin each in concentrations of 10 mg·l−1) was 40– 50% less than ones in a sea-water medium without antibiotics over a 30-day period. Counts of gut bacteria in treated animals showed a 98% reduction in number compared with control animals (to bacteria·ml gizzard fluid−1). While it is tempting to conclude that the reduction in

APLYSIA

195

numbers of bacteria caused the decreased growth in these sea hares through a lessened availability of bacterial metabolites to the sea hares, the possibility exists that the antibiotic treatment itself was responsible. Vitalis tested this by exposing A. juliana to high concentrations of antibiotics. He showed that A. juliana survived well, with no obvious effects on feeding or other behaviour, in concentrations of combined streptomycin and penicillin of 1000 mg·l−1, equal to 50 times the dosage used in the experiment. While further work is needed to assess any deleterious effects of antibiotics on growth and metabolism of the host animal, the results of Vitalis’ study are none the less provocative. The possible bacterial source of required amino acids was further tested in a series of experiments with juvenile A. dactylomela (Carefoot, 1981a). Here, animals were given a chemically defined diet deficient in arginine (essential for the rat) and containing antibiotics to suppress the activities of gut bacteria. The sea hares gradually lost weight over a 20-day period on the experimental diet (Carefoot, 1981a). When the arginine-deficient diet was replaced on Day 21 with a diet complete in all nutrients, the sea hares immediately began to gain weight and continued to do so for the remaining 16 days of the experiment. These data, combined with the previous observations on A. kurodai and A. juliana, suggest that sea hares may rely on their bacterial symbionts for provision of essential amino acids and other nutrients just as has been found in other animals (e.g. insects, isopods, sea urchins, and vertebrates). This may explain how sea hares, like sea urchins (Fong & Mann, 1980), can utilize a wide variety of seaweeds as food. If bacteria are involved in this way in sea hares, further nutritional studies using traditional techniques of deletion and augmentation of specific nutrients with chemically defined artificial diets would be pointless unless a way is found to eliminate completely gut bacteria or at least to suppress their metabolic activities. Even if this is done, it must be demonstrated that this technique is affecting only the bacteria in question. Alternatively, animals could be grown axenically. Given our current knowledge of Aplysia larvae and current techniques of culturing them through metamorphosis, it should be a routine matter to grow sea hares with sterile guts. This would open a marvellous range of possibilities for their further nutritional study. FEEDING ECOLOGY The only true way to ascertain the effect of sea hares or any herbivore on an algal community is to remove the animals from an area and to keep them out for a long time. This can be done simply by hand-clearing animals or by the use of “exclosure” cages. The latter have been used in two studies on sea hares: the first on A. dactylomela and A. parvula in New Zealand (Willan, 1979); the second, on A. dactylomela in Barbados (Carefoot, 1985). In neither case were the results completely convincing. The small size of exclosure cages presented problems in both studies, as did too fine a size of mesh employed as screening, the possibility in both studies of unknown “cage effects” and, in the Barbados study, too short a period of study. If any conclusion could be drawn from the Barbados work, it would be that herbivorous fish may in fact have exerted a greater impact on the algal community than did the population of A. dactylomela (Carefoot, 1985). Willan (1979) was able to show significant differences between the biomass TABLE IX Densities of Aplysia in different geographical areas Species

Geographical area

Time of year

Density (no·m−2)

References

californica dactylomela

Ellwood, Southern California Barbados

Apr. 1975– Feb. 1977 summer

0·03 0·6

Sousa, 1979 Carefoot, 1985

196

THOMAS H.CAREFOOT

Species

Geographical area

Time of year

Density (no·m−2)

References

juliana

North Island, N.Z. Hawaii North Island, N.Z.

0·9–1·6 1·0 3·5 3–5

Willan, 1979 Sarver, 1978

parvula

Mar.–July (autumn-winter) summer/autumn winter/spring Oct.–Nov. (spring)

Willan, 1979

of Laurencia spp. within exclosure cages as opposed to without after 14 months (three of four cages ended up with a percentage cover of Laurencia about 40% greater than the uncaged control areas; the fourth cage had 25% less cover of Laurencia than its control). The author expressed a concern about possible “cage effects” in his study, such as shading and accumulation of sand within the cages, as well as how other large herbivores were excluded along with Aplysia spp. Willan did not employ a control cage or cages in his study, although there is a question whether their inclusion would be useful since it is difficult to devise adequate control cages and then correctly interpret the data obtained from them. With their large sizes and concomitantly large appetites, sea hares should exert a profound influence on the seaweed community. It is therefore surprising that so little work has been done on Aplysia in this regard in comparison to, say, sea urchins or limpets (see reviews by Lawrence, 1975 and Branch, 1981, respectively). What we find ultimately is that sea hares are never very numerous in a given area, with densities rarely exceeding 2 animals·m−2 (see Table IX). Compared with densities observed for sea urchins in areas where effects of their grazing may be pronounced (e.g. 60–1000 individuals·m−2: Paine & Vadas, 1969; Foreman, 1977; Chapman, 1981) such low numbers of sea hares, even with their greater comparative rates of feeding (Carefoot, 1981), would seem unlikely to be capable of exerting comparable large-scale effects on the seaweed community. That diet and skin colour are probably related in sea hares has been suggested by several authors (Eales, 1921, 1960; Grigg, 1949; Winkler, 1958b, 1959b, c; Kandel, 1979; Willan & Morton, 1984). The idea arose from an early and somewhat cursory observation by Garstang (1890) on colour change in A. punctata, as well as from biochemical analyses of skin pigments of various species of Aplysia. These analyses suggested that the colours in A. punctata, A. depilans, and A. californica derive from degradation products of the tetrapyrrole molecule of chlorophyll (e.g. porphyrins and bilins: MacMunn, 1899; Schreiber, 1932) or from other algal pigments (e.g. phycoerythrin and phycocyanin: Winkler, 1959b). A porphyrin has also been found in the skin of A. punctata by Kennedy & Vevers (1954) and several carotenoids in A. punctata, A. depilans, and A. fasciata by Czeczuga (1984). Since the skin colour resulting from prolonged feeding on an alga is thought to match the colour of the seaweed, a camouflage function is hypothesized. As attractive as this notion seems, there are several points that should be considered. First, the effect of diet on skin colour has not been convincingly demonstrated in experiments. Winkler (1959b) showed that brown-, dark green-, and grey-coloured A. californica changed over a period of 1–3 months to a uniform light-brown base colour when fed on diets of parsley leaves and celery tops. He further observed that if these blanched specimens were fed on a large amount of Plocamium cartilagineum, the base colour of the skin turned decidedly, but temporarily, pink. A similar pink undertone colouration has been observed also in Aplysia dactylomela in Hawaii from areas where they appear to be eating primarily red algae (Switzer-Dunlap, pers. comm.). Winkler also noticed that certain animals bearing red body streaks and purple foot colourations had been eating predominantly Plocamium cartilagineum, as determined from faecal analysis. Winkler did not attempt to re-establish the original colours in his experimental animals by feeding them subsequently on diets of different coloured seaweeds. When small (2 g live weight) sublittorally collected Aplysia punctata were fed for six weeks on Laminaria, 11 weeks on

APLYSIA

197

Enteromorpha and Ulva, and for similar lengths of time on various red algae, the animals became only a darker shade of red, similar in colour to that of freshly collected sublittoral animals of equivalent size (Carefoot, 1967a). These field animals were eating mainly the red alga Plocamium cartilagineum in their sublittoral habitat (Carefoot, 1967c). A second point with regard to diet and skin colouration is that Aplysia juliana feeds only on various species of green algae, principally Ulva spp., yet is always brown in colour, often of a rich chocolate hue. A brown coloured A. juliana cannot be described as cryptic amongst its bright green coloured food. The notion of adaptive colouration in Aplysia in response to diet arose as part of a theory of migration from deeper offshore areas to the intertidal region during an animal’s life, with a gradual colour change occurring as the animals browsed successively through red, brown, green and, finally, olive green-coloured seaweeds (see pp. 202–203). Although no evidence has been provided in support of this theory, such colour changes were none the less predicted and possibly expected by workers in the field. Despite these comments, however, it seems likely that diet does affect colouration in Aplysia. It remains to be shown convincingly in experiments, although some points support this idea. (1) It is known that eggs are coloured in response to different seaweeds being eaten and the mechanism of pigment transfer and incorporation from food to eggs has been documented (Chapman & Fox, 1969). (2) Switzer-Dunlap (1978) notes that in species of sea hares which as juveniles feed on red algae (A. brasiliana=willcoxi, A. californica, and A. parvula) the overall body colour is pink initially and grows progressively darker with continued feeding and growth. (3) Winkler (1959c) has observed that A. californica fed on Plocamium cartilagineum develop characteristically dark coloured blood and purple subcutaneous pigment deposits, and Chapman & Fox (1969) have noted that Aplysia californica fed artificially on phycocyanin pigment exhibit a distinct blueing of the inner skin. All of these points suggest that skin colour is at least partially regulated by diet in sea hares. An inverse relationship of size of animal to the depth at which it occurs was found for a sublittoral population of A. punctata in the Irish Sea (Carefoot, 1967c). As size was strongly and positively correlated with biomass of Plocamium cartilagineum, it was presumed that the animals were simply eating more, and growing larger, in shallower depths where the alga was more abundant. Willan (1979) noted also that maximum densities of Aplysia dactylomela on Echinoderm Reef, New Zealand, occurred during MarchJune when Laurencia spp., its favoured food, were dominant. The Reef, 2·76 hectares in size, hosted up to 6000 animals during peak densities in June. Willan calculated that during March-June the equivalent energy required in seaweeds by the Aplysia dactylomela population was 4·4 kJ·ha−1·month−1 while the energy available represented by Laurencia spp. at the site was never less than 59·8 kJ·ha−1·month−1. Overall, for all months of the year, the energy represented by the standing crop of Laurencia spp. never fell below six times that required by the Aplysia dactylomela population. While the author noted that other herbivores, including two aplysiid species, also grazed on the same standing crop of Laurencia spp., it would seem unlikely that the Aplysia dactylomela population in this area would ever be food-limited. An interesting aspect of feeding behaviour in A. juliana was identified by Saito & Nakamura (1961) and Frings & Frings (1965), in which animals appeared to eat only the succulent distal portions of the brown alga Undaria pinnatifida or Ulva lactuca plants, respectively, while leaving the thicker, coarser, stem or mid-rib sections. The ecological significance of this behaviour was evident, since the basal portions, at least of Ulva, were observed to produce new fronds within a week or two. Only under conditions of long starvation or if “Ulva-water” (prepared from succulent new growth) were flooded over their mouths, could the sea hares be induced to eat the tough basal sections (Frings & Frings, 1965). The authors conjectured that the phagostimulatory material present in the new growth may be absent or of insufficient concentration in the bases of the plants to stimulate the animals to eat. A similar response to food texture was shown for

198

THOMAS H.CAREFOOT

Aplysia fasciata by Susswein & Schwarz (1983), who made otherwise palatable foods (Ulva lactuca and leaves of the lily Hemerocallis fulva) too tough to swallow by enclosing them in a plastic net. These preparations became stuck in the buccal cavity and the sea hares soon learned not to eat them. In another related experiment on Aplysia oculifera, Schwarz & Susswein (1984) denervated the crop region which destroyed this ability to learn. The authors intimated that A. fasciata in their natural habitat might learn that certain seaweed species, as for example the calcium carbonate-impregnated Jania rubens, are too tough to eat (Susswein & Schwarz, 1983). Susswein, Weiss & Kupfermann (1984b) suggested that when soft food enters the crop it may activate receptors that reinforce feeding upon soft foods, whereas unsuccessful entry of tough foods may act as a negative reinforcer. Whatever the proximate causes of such behaviour, the end result for feeding sea hares would be a cropful of easily digestible soft seaweeds. In addition, if the sparing of the tough bases of Ulva plants by Aplysia juliana were a pattern followed by other Aplysia, then there exists the ultimate prospect of a renewed harvest of fronds as a result of the animals’ unique feeding ecology. THE EFFECT OF FEEDING ON OTHER BEHAVIOUR The effect of contact with food, of feeding, and of varying degrees of satiation on other behavioural activities has been extensively investigated in A. californica. Advokat (1980), for example, has shown that various defensive activities such as siphon withdrawal, locomotion, and inking are measurably attenuated following a meal (see also Advokat, Carew & Kandel, 1976). For the siphon-withdrawal reflex, actual ingestion of food is not required to produce a significant diminution of the response; this can be accomplished simply by applying food to the lips and oral veil regions. Kupfermann (1974b), Kupfermann & Carew (1974), and Susswein & Kupfermann (1974, 1975a, b) have similarly found that after a large meal, satiated A. californica stop moving and fail to bite at other foods. In some instances, animals remain in this “frozen” state for many hours. Conversely, Kupfermann & Weiss (1981) have shown that certain “aversive” stimuli, such as tail-pinching and handling, enhance feeding behaviour in A. californica. The functional significance of such behavioural modifications is unclear as it does not seem advantageous for an animal to ‘shut down’ or reduce its locomotory and defensive capabilities during or after eating. In this regard, the suggestion by Advokat (1980), that attention to food brings with it a concomitant inattention to other stimuli —that animals cannot flee and eat simultaneously—may well explain the situation, but does not clarify its biological rôle. An unusual behaviour, apparently in response to food deprivation, was observed in both the field and laboratory for A. brasiliana=willcoxi by Aspey, Cobbs & Blankenship (1977). The animals swim at the surface and exhibit a characteristic “head-bobbing” activity where the head and anterior portion of the body are stretched to extend the oral tentacles momentarily and repeatedly out of the water (Fig. 5). Observations on field animals showed head-bobbing frequency to be about 40·min−1, with bouts of activity lasting about 47 s on average. Laboratory animals were seen to increase head-bobbing frequency at the approach of their regularly scheduled feeding time, and later experiments to test this showed that the frequency increased in direct relation to the length of time they had gone without food. After 41 h without food, head-bobbing, however, stopped. Aspey et al. (1977) suggested that head-bobbing serves to increase the chances of encountering floating and/or surface food but, after several hours, becomes energetically wasteful. When touched on the oral region with food (dried laver: Porphyra sp.) or with a piece of paper towelling during swimming and/or head-bobbing, the animals dropped immediately to the bottom: a similar response to touch was noted by Hamilton & Ambrose (1975) for swimming individuals of the same species. Despite the strong correlation of frequency of bobbing with duration of food deprivation, this complex behaviour seems

APLYSIA

199

to have a function beyond that of simply finding food. In the first place, the animal’s food is benthic, not floating. Moreover, head-bobbing would be unlikely to enhance the probability of finding food: swimming on the surface would itself serve this function. Hamilton & Russell (1981a, b) and Hamilton (1986) suggest the most likely hypothesis, that while swimming allows the animal to move out of a region of food scarcity, head-bobbing aids navigation. A sexual rôle has not been intimated for this unusual behaviour as yet, even though a common response of opisthobranchs, including Aplysia, is to seek out mates, copulate, and lay eggs in times of food scarcity (see pp. 183–188). GROWTH AND ENERGETICS As might be expected in a large animal with an annual life cycle, growth rates are rapid. An early postmetamorphic animal may attain a rate of size increase of 13000% over a two-week period (Hadfield, 1975). The largest recorded sea hare was a 6·8-kg A. californica collected from Elkhorn Slough, California (MacGinitie, 1935); the largest size estimated, however, was 15·9 kg for a sublittoral A. vaccaria (Limbaugh: in Winkler & Dawson, 1963). High rates of feeding, fast growth rates, and large production of spawn were translated at one densely populated site in Hawaii to 11 kg of algae eaten per day and close to produced per day by a population of A. juliana (Sarver, 1978). one trillion eggs GROWTH The veliger larva represents the first stage of growth in sea hares and spends its life swimming and feeding in the plankton. It is separated from the second, or benthic adult phase, by a metamorphosis. Rates of growth of the veliger larvae are well known through laboratory studies (Krakauer, 1969; Kriegstein et al., 1974; Switzer-Dunlap & Hadfield, 1977). In post-metamorphic animals the pattern of growth is sinusoidal and involves a ‘lag’ phase, followed by an exponential phase, leading eventually to a levelling off (Sarver, 1978; Peretz & Adkins, 1982). Rates vary greatly in each stage and these differences must be accounted for when comparing rates in different species and especially in animals of different sizes. Growth rates of post-metamorphic animals are generally calculated from absolute measurements of live weight, or live weight converted to dry weight or energy equivalents, over time, defined as:

where W represents weight and t represents time. Because of the different rates of growth of animals at different sizes, as noted previously, such data on absolute growth rates may, however, not be useful for interspecific comparisons. Growth may be expressed better in relative terms, such as:

which gives information on growth increment per unit size per unit time, or defined according to the growth coefficient, K:

where m is the slope of a regression line relating size at time t+1 to size at time t in a Ford-Walford plot (Walford, 1946; see also Branch, 1981; for equation see Table X). The value of K is useful in growth

200

THOMAS H.CAREFOOT

Fig. 5.—“Head-bobbing” in swimming. A. brasiliana=willcoxi: figure modified from Aspey, Cobbs & Blankenship (1977).

APLYSIA

201

studies as it can be used as a comparative index of growth that is independent of body size. Moreover, when the regression line in a Ford-Walford plot is extrapolated TABLE X Growth rates of Aplysia: N=number of replicates; the Ford-Walford growth equation, is usually based on units of length L, for purposes of the comparisons below, weight was substituted instead thus, Lt=weight at time t and Lt+1=weight after a given period of time (here=2 wk); values are given for i and m; C=Chlorophyceae; i=intercept; m=slope; P=Phaeophyceae; Pr=spawn production; Ps=somatic production; R=Rhodophyceae; r=correlation coefficient Species

i

Food

N

Startin Temp. g °C weight (mean live g)

Rate of Ford-Walford growth growth equation statistics dw/dt (g live (wt·da y−1)

K growth coeffic ient K= −logem

Live g Comme Referen alga·g nts ces growth −1

(growt h=Ps +Pr)

m r

brasili ana =willc oxi

califor nica

R

califor nica

R

dactyl omela

C

Field seawe eds

9

201

13–28

1·7

16·6

1·04

0·99

−0·04

Laure ncia sp., Polysi phonia sp., Dasya sp., and Porph yra sp. Dried Porph yra sp. and fresh romain e lettuce (Lactu ca saliva longifo lia) Entero morph a sp.

9

59·3

18

6·2

49.4

1·64

0·97

−0·49

15 12 9 11 4

1·0 30 85 310 850

16 16 16 16 16

0·6 1·0 3·3 4·2 5·3

4

52·0

28

2·9

40·2

0·99

0·56

0·01

4·1

Data from a single growth curve (field study) Data from a single growth curve

Kraka uer, 1969

Numb er of days of study: variabl e up to 235

Peretz & Adkin s, 1982

Kriegs tein et al., 1974

Carefo ot, 1970

202

THOMAS H.CAREFOOT

Species

i

Food

C

R

R

Ulva fasciat a Clado phora sp. Laure ncia papillo sa Galax aura oblong ata

R C C C

juliana

C

C C juliana

Startin Temp. g °C weight (mean live g)

Rate of Ford-Walford growth growth equation statistics dw/dt (g live (wt·da y−1)

K growth coeffic ient K= −logem

Live g Comme Referen alga·g nts ces growth

8

66·2

28

1·5

19·3

1·03

0·93

−0·03

2·7

4

49·0

28

1·5

26·3

0·89

0·75

0·12

6·5

4

50·1

28

1·7

−19·2

1·85

0·84

−0·62

4·4

10

56·6

28

−1·1

−1

(growt h=Ps +Pr)

m r C

dactylom ela juliana

N

C

Laurenci a spp. Ulva fasciata Enterom orpha sp. Cladoph ora sp. Ulva fasciata

Ulva reticulata Ulva lactuca Enterom orpha sp.

10

37·4

14–17 0·8

7·12

1·11

0·98 −0·10

10

29·3

28

1·6

18·4

0·82

0·86

0·20

1·9

6

24·9

28

0·3

8·5

0·81

0·96

0·21

11·3

6

23·7

28

−0·5

0·2

0·69

0·93

0·37

10

0·01

25

0·29

0·67

343·5 0·97

−5·84

6

0·01

25

0·11

−0·07

241·7 0·91

−5·49

5

0·01

25

0·36

3·04

162·1 0·98

−5·08

10

1·0

23·5

0·32

−0·11

5·40

−1·69

0·91

Willan, 1979 Carefoot, 1970

Data from a single growth curve for each food species

Sarver, 1979

Data from a single growth

Vitalis, 1981

APLYSIA

203

curve for each food species C C punctata

R

C

C R

R

R

P P

punctata

R

Ulva reticulata Ulva fasciata Plocamiu m cartilagi neum Enterom orpha intestinal is Ulva lactuca Heterosi phonia plumosa Cryptopl eura ramosa Delesseri a sanguine a Laminari a digitata Desmare stia aculeata Plocamiu m cartilagi neum

11

1·1

23·5

0·26

−1·07

5·27

0·93

−1·66

10

1·2

23·5

0·30

−0·96

8·26

0·93

−2·11

26

9·5

15

0·20

3·4

0·94

0·99

0·06

3·0

27

8·4

15

0·19

3·5

0·90

0·96

0·11

5·4

31

6·1

15

0·09

2·1

0·85

0·97

0·16

2·9

20

4·3

15

0·08

1·8

0·84

0·99

0·17

4·4

30

4·0

15

0·03

1·4

0·77

0·97

0·26

6·0

29

3·0

15

0·04

3·3

12

1·8

15

0·02

6·2

10

1·9

15

0·0

18

6·2

15

0·2

2·95

0·99

0·99

0·01

3·2

Spawn included as “growth”

Carefoot, 1967a

Spawn included as “growth”

Carefoot, 1967b

to the ordinate axis and to its intersection with the 45° diagonal (the point at which size at time t is the same as at t+1; i.e. when there is no further growth), growth rates can be estimated for all sizes of the population, assuming that growth is constant over the period considered. Although best used for slow-growing animals (such as limpets and other snails: Ward, 1967; Hughes, 1971a, b, 1972; Balaparameswara Rao, 1976; Branch, 1981), the growth coefficient K can none the less provide a useful index for comparison of growth in other animals. Its usefulness declines, however, when sample numbers are small, when size ranges are narrow, or when animals are young and in a fast-growing phase. Data for a number of growth studies on Aplysia are presented in Table X. Growth is expressed as absolute rates in g live wt·day−1, as regression data for Ford-Walford plots of size at time t+1 against size at time t, and as values for the growth coefficient, K. Absolute growth rates of Aplysia range from 0·2–6·2 g live weight·day−1 for animals eating their optimal seaweed foods (e.g. A. punctata eating Plocamium

204

THOMAS H.CAREFOOT

cartilagineum and Aplysia californica eating Laurencia sp. and other red seaweeds), with the variability, for the most part, probably due to starting sizes of the animals. Major factors affecting growth rates in Aplysia are: (1) temperature and light—as related to season, microgeography, and the tides; (2) size and age; (3) reproductive state; (4) water movement; and (5) food, both quantity and quality. Of these, the effect of food has been studied most completely. The effect of diet quality on growth in Aplysia is shown clearly for A. punctata feeding on eight seaweed diets in the laboratory (Table X; Carefoot, 1967a). The diets represent five common species in the animal’s sublittoral habitat (the red algae, Plocamium cartilagineum, Heterosiphonia plumosa, Cryptopleura ramosa, and Delesseria sanguinea, and the brown alga, Desmarestia aculeata), including Aplysia punctata’s preferred field diet, Plocamium cartilagineum. The three other algae (the green seaweeds, Enteromorpha intestinalis and Ulva lactuca, and the brown alga, Laminaria digitata) represent abundant intertidal forms, are commonly eaten by intertidal animals, and are species thought to be principal food of Aplysia punctata (e.g. Eales, 1921). Best absolute growth occurs on a diet of Plocamium, closely matched by that on a diet of Enteromorpha. Other diets support variable, but poorer rates of growth. Values for K in this study are 0·06 and 0·11 for Aplysia punctata eating the foods which give overall best growth (Plocamium cartilagineum and Enteromorpha intestinalis, respectively). The larger values for K in the series for Aplysia punctata indicate that the regressions of size at time t+1 over size at time t flatten out as the animals increase in size; that is, the relative increment of growth becomes smaller with increasing size. This is evident for animals eating foods that are nutritionally poor (i.e. giving poor growth), such as for A. punctata eating Cryptopleura ramosa, where K=0·26 (also Aplysia dactylomela and A. juliana eating Cladophora sp., where K=0·12 and 0·37, respectively; see Table X). Similar effects of diet quality on growth of sea hares have been shown by Sarver (1979) and Vitalis (1981) in studies of Aplysia juliana. Of the several Ulva diets tested in these studies, U. fasciata proved best nutritionally for Aplysia juliana, as shown by lowest K values (−5·84 as compared with −5·49 and −5·08 for animals eating Ulva reticulata and U. lactuca, respectively; data from Sarver, 1979; see Table X). The negative values for K obtained in these studies indicate that the tiny Aplysia juliana were in their exponential phases of growth, with lowest values being associated with the steepest positive slope of the regression line in the Ford-Walford plot. The effect of diet quantity on growth of Aplysia was investigated by Sarver (1978). He reared A. juliana in laboratory culture for 137 days on “low”, “medium”, and “high” rations of Ulva lactuca (representing 5, 10, and 15 g alga·day−1·ind.−1, respectively). Sarver showed that while P (growth including spawn) as a proportion of C (food consumption) did not vary significantly on the different ration levels (values obtained were 15, 11, and 12% of C for animals eating the “low”, “medium”, and “high” rations, respectively; see Table VIII, p. 222), the absolute amount of growth which each group of Aplysia juliana attained did differ considerably. For example, animals on a “high” ration gained approximately twice as much weight as did animals on a “low” ration and produced nearly three times as much spawn; “medium”-ration animals were intermediate in both respects. In his studies on A. juliana, Sarver (1978) investigated absorption and assimilation of foods and channelling of energy into various body processes in order to compare animals of different “physiological” ages (i.e. size) with ones of different chronological ages. He found that an animal maintained on a low ration of food for several weeks and then returned to an ad libitum diet behaved physiologically like a much younger animal in terms of growth and spawn production, even though normally it would have long since stopped growing at that chronological age. Its new rate of growth corresponded to its physiological age, not to its chronological age. This line of research is highly provocative in that it provides an elegant means of investigating various processes of ageing in sea hares. It also allows comparison of sea hares from different

APLYSIA

205

geographical areas, where relative size may not be a true reflection of actual age. The availability of animals of known age through their culture in the laboratory, combined with the possibility of ascertaining age of field animals from shell size or other growth characteristics of the shell (see p. 236), would help to remove the uncertainty of not knowing an animal’s age in physiological and behavioural studies of sea hares where this knowledge is important. Few of the many environmental factors which could affect growth rates in Aplysia have been investigated, although Sarver (1978) has taken a preliminary look at the effects of light and water current on growth of A. juliana. He found that growth was about 75% faster in animals in slow-moving water than in a strong current (approximately 1 m·s−1), suggesting that currents may interfere with normal feeding or with attachment and movement. In comparison, growth was about 120% greater in animals kept continually in the dark than in those exposed to sunlight. As this nocturnal species responds to light by burying itself during the day and normally feeds only during hours of darkness, a continuous dark regime allowed almost constant feeding and greater subsequent growth (Sarver, 1978). and for K2, net growth efficiency of Values for K1, gross growth efficiency and from 15–84% sea hares are presented in Table VIII (p. 222). They range from 11–45% for Highest values of growth efficiency are generally, but not invariably, associated with foods for giving best growth. These foods, in turn, tend to be the ones favoured, and this trend is shown best for A. dactylomela and A. punctata. For A. punctata, K2 efficiencies decline from high values of 31–35% on the animal’s favoured field diet of Plocamium, to a low of 15% on the less-preferred diet of Cryptopleura (Carefoot, 1967a). Two diets eaten most poorly in the same study and giving poorest growth, the red alga Delesseria and the brown alga Laminaria, actually give the highest net growth efficiencies of 74 and 38%, respectively, suggesting that there may have been some sort of physiological compensation for the poor quality of these diets. The values for K1 and K2 efficiencies in Table VIII are calculated on the bases of either dry weight or energy. When live-weight units are used to relate growth and food consumption, a slightly different picture emerges as to the relative value of different seaweeds as food for Aplysia. This is shown in the final two columns in Table X, where growth is expressed as the number of live g of each alga required to produce a single g of either somatic or spawn tissue. On this “per mouthful” basis, Ulva spp. appear to be the optimal diet for sea hares (1·9–2·9 live g Ulva spp.·g growth−1), followed by some of the red seaweeds, including Plocamium and Delesseria. In actuality, Ulva spp. are not the best foods for most sea hares in the laboratory —only for Aplysia juliana do they promote most rapid growth. Missing from this presentation is the factor of how much of each food is actually eaten and this, in turn, depends on other factors such as palatability, texture, ability of the animal to manipulate the food, and so on. A possible compensatory type of interaction exists between growth efficiency and absorption efficiency in sea hares. Diets which are poorly absorbed are assimilated more efficiently than are ones which are highly absorbed. This negative correlation of growth efficiency and absorption efficiency is a well-known phenomenon in aquatic animals, including fish (Welch, 1968) and other gastropods (Carefoot, in press). From visual inspection of a graphical representation of these data for Aplysia (Fig. 6) there does not seem to be an effect of algal type (i.e. red, green, or brown) on the relationship between net growth efficiency and absorption efficiency. In addition to quantity and quality of diet, other factors of size, age, and reproductive state are known to affect growth efficiency in gastropods (Streit, 1976; Edwards & Huebner, 1977; Huebner & Edwards, 1981; Macé, 1981; Ansell, 1982). Of these, only age has been investigated in Aplysia and is shown by: (1) a decrease in K2 or net growth efficiency from 55 to 17% over an 80-day period for A. punctata growing from 0·9 to 18·4 g live weight (Carefoot, 1967b), and (2) a decrease in K1 or gross growth efficiency from 42 to

206

THOMAS H.CAREFOOT

Fig. 6.—Relationship of net growth efficiency (K2) and absorption efficiency in Aplysia: equation of regression, data from Carefoot (1967a, b, 1970), Sarver (1978), and Willan (1979).

12% over a 103-day period for A. juliana (including spawn as growth; Sarver, 1978). Temperature does not appear to affect either net or gross growth efficiencies in gastropods (Ansell, 1982) and, consequently, has been disregarded in the above comparisons of Aplysia where animals may have been maintained at different temperatures. Aspects of shell growth in Aplysia have been investigated by Winkler (1958a, 1959d), as have allometric relationships of growth of shell and body (Krakauer, 1974; Usuki, 1979, 1981a, b; Willan, 1979), and of growth of various organs (including shell and body) (Peretz & Adkins, 1982). By culturing A. californica to various ages, then dissecting out and measuring the maximum dimension of their shells, Peretz & Adkins (1982) have shown a precise relationship between shell size and body weight and thus, in these cultured animals, between shell size and age. Whether this relationship holds true for Aplysia under more varied conditions in the field has not yet been tested. A characteristic of growth noted for A. kurodai by Nishiwaki, Ueda & Makioka (1975) was a remarkable day-to-day fluctuation in live weight. The authors recorded daily variations of up to 40% in live weight in four individuals of A. kurodai over a two-month period in the laboratory. In some instances losses could be correlated with production of spawn, but these would have accounted for no more than about 13% of the major fluctuations noted. Similar patterns have not been observed, or at least not to the same extent, in other species of Aplysia. One possibility not considered by Nishiwaki et al. (1975) is that the fluctuations may have been a phenomenon associated with senescence, since growth measurements on these four individuals commenced at almost exactly the time of general seasonal decline in size and vigour of the laboratory population. While such fluctuations in live weight are unlikely to be reflected to the same extent in an animal’s dry weight (Carefoot, 1981b), similar periodic daily monitorings would be advisable in studies of other species that require precise and accurate measures of live weight.

APLYSIA

207

Finally, a number of studies have assessed growth of sea hares in the field. Some of these have estimated growth rates indirectly through collections at intervals and/or size-frequency analyses (Carefoot, 1967c; Audesirk, 1979); others have measured growth rates directly from tagged animals (Nishiwaki et al., 1975; Audesirk, 1979; Willan, 1979), from animals in field cages (Sarver, 1978), and from monitoring of distinctively coloured and therefore easily recognizable animals “seeded” into other field populations (Sarver, 1978). In all instances where field data have been compared with laboratory data on growth at similar temperatures, good agreement has resulted (Carefoot, 1967c; Sarver, 1979; Willan, 1979). The pattern of growth in field animals is illustrated by the results of Audesirk’s (1979) study of tagged A. californica at Santa Catalina Island. Settlement occurs in late summer or autumn with slow growth over the winter. Fastest growth is between February and April, prior to the onset of breeding (at this time a doubling in size, from mean live weights of 700 to 1400 g, was recorded by Audesirk during a single month in spring —one specimen gaining 900 g between February and March, and a further 1200 g the following month). Growth slows in late spring as breeding intensifies, and maximum weights of about 3 kg are reached in June or July. Weights decline during the most active time of breeding in August, and the population dies off in October–December. Audesirk (1979) attributes the seasonal weight loss to two main causes: to loss of foraging time due to reproductive activities, and to the large energy requirement for egg production. With some variation, similar patterns of seasonal growth have been noted in other field studies of Aplysia (e.g. A. juliana: Sarver, 1978; A. kurodai: Nishiwaki et al., 1975; A. punctata : Carefoot, 1967c). An apparent exception to this pattern of growth was noted by Krakauer (1969) for A. brasiliana=willcoxi, where the population appeared to show no seasonal decline in weight, but this may have resulted from the author terminating her field collections in late spring—possibly too early to have shown a decline in weight. ENERGY ALLOCATIONS All animals require a mechanism to control energy balance and to allocate food energy or stored energy to various metabolic pathways. In most instances regulation of energy balance is by controlling energy input; only in instances of starvation and aestivation is energy balance regulated by controlling energy output. In Aplysia, as in any animal, allocation of food energy C is to production of somatic tissue Ps and gametes Pr. Initially, all energy allocated to production is devoted to growth of somatic tissue; the first ‘decision’ comes at reproductive maturity with the shunting of some growth energy into production of eggs and sperm. Later, all energy of production save for that required for tissue maintenance is allocated to reproduction. The sum of Ps and Pr, representing an animal’s “scope for growth” (Warren & Davis, 1967), has been used by various authors as an indication of diet quality and effect of “stress” in marine invertebrates, including a few gastropods (e.g. Stickle, 1985). As a by-product of assimilation of nutrient materials into new tissue and of various maintenance costs, energy is lost in the form of urinary excretions U and as heat of respiration R. After absorption, unused food energy is egested as faeces F. The assembled energy budget is represented as the familiar equation:

Not included in energy budget models for sea hares are a number of special investments such as: (1) the shell, which represents about 0·1% of the total live weight of the animal (Peretz & Adkins, 1982; to obtain overall energy values for Ps, the shell is routinely ground up and combusted with the other body tissues, and thus actually contributes a small energy loss through endothermy: Paine, 1966); (2) opaline and ink-gland secretions, when they are lost from the body; (3) radular teeth broken off and lost during feeding, or consumed

208

THOMAS H.CAREFOOT

and lost along with the faeces; (4) mucus lost as sheathing of the faeces, and from the skin especially during locomotion (see Denny, 1980); and (5) sperm, following copulation, except in instances of simultaneous copulation when incoming sperm would ‘compensate’ to some extent for the energy loss represented by the animal’s own outgoing sperm. Of these costs, all save mucus lost during locomotion would be predicted to be small. None, however, has been reliably estimated in studies of energetics of sea hares. Energy budgets of sea hares are shown in Table XI. Several features are evident from these models. First, the energy lost in urinary secretions U, has never been determined for Aplysia. Its omission in studies of energy budgets of gastropods is usually thought to introduce only a small error. Where it has been measured in gastropods, as for example in the freshwater snail Hydrobia ventrosa by Kofoed (1975), it has, however, been shown to account for 11–19% of consumed food energy (this value also includes mucus and dissolved organic carbon losses; and it may not reflect accurately the extent of loss under the different osmotic conditions of a sea-water habitat). Secondly, where each component of the budget has been separately determined, large imbalances in the budgets are apparent (e.g. −3 to +155% for Aplysia punctata: Carefoot, 1967a, b; −46% for A. juliana: Sarver, 1978). Negative imbalances can result from failure to account for urinary and other losses such as mucus and dissolved organic molecules as previously noted, but can also result from: (1) over-estimating C during feeding by not taking into account possible loss of food materials such as plant juices or seaweed fragments, (2) under-estimating F due to loss or dissolution of, or leaching from, the faeces, or (rarely) to coprophagy (representing a double source of error), and (3) under-estimating R through “flask-effects” during respirometry measurements (Wightman, 1981). These and other errors in budgeting are discussed in more detail elsewhere (Carefoot, in press). The magnitude of one of these potential sources of error, namely the possible loss of food materials through ‘sloppy’ eating habits, was estimated in a field study of A. dactylomela using nylon-mesh bags to enclose single animals in overnight feeding experiments (Carefoot, 1985). The results showed that such where W=wastage and C=consumption, losses never exceeded 8% (calculated as expressed in Joules) in all feeding trials in the field. This was considerably less than expected in view of the rough manner of biting and tearing by which sea hares graze, which tends to fragment extensively branched or delicate algae. How much of the negative budgetary imbalances would be accounted for by such wastage in laboratory studies would depend on the degree to which uneaten remnants of food were reclaimed by the experimentor after a feeding session by Aplysia and the extent to which food bits were allowed to escape in flow-through systems. Even though such wastage probably accounts for only a small part of the missing energy in budgetary models, it does represent a possibly significant contribution to detrital food chains in areas populated by Aplysia. Allocation of ingested food energy C to somatic Ps and spawn Pr production, and to respiration R was monitored in A. juliana by Sarver TABLE XI Energy budgets for Aplysia: C=Chlorophyceae; F=faeces; P=Phaeophyceae; Pr=spawn production; Ps=somatic growth; R=Rhodophyceae; U=urine; C=consumption; R=respiration Species

Laboratory dactylomel a

R

Food

C

=Ps +Pr +R

+F +U Balance “Scope for growth” (Ps+Pr)

Remarks

References

Laurencia spp.

100

21

21

No spawn produced by these

Willan, 1979

0

41

–

−17

21

APLYSIA

Species

Food

C

=Ps +Pr +R

+F +U Balance “Scope for growth” (Ps+Pr)

juliana

C

Ulva lactuca

100

5

11

11

27

–

−46

16

punctata

R

Plocamium cartilagine um

100

14

9

16

35

−

−26

23

C

Enteromor pha intestinalis Ulva lactuca Heterosiph onia plumosa Cryptopleu ra ramosa Delesseria sanguinea Laminaria digital a Plocamium cartilagine um

100

9

9

13

41

–

−28

18

100

10

7

19

25

–

−39

17

100

10

5

16

29

–

−40

15

100

8

3

25

29

–

−35

11

100

17

10

54

55

–

+36

27

100

19

4

185

47

–

+155

23

C R

C R P punctata

R

100

22

4

14

33

–

−27

26

Field dactylomel a

Mixed red algae 100

25

6

20

49

–

0

31

Remarks

juvenile animals F estimated from a separate experiment R for animals on Plocamium diet applied to animals on other diets

209

References

Sarver, 1978

Carefoot, 1967a

Carefoot, 1967b

R estimated by difference to get perfect balance

Carefoot, 1985

(1978). These data, shown in simplified form in Figure 7, were derived from animals grown in laboratory culture from soon after metamorphosis (1 mg live weight) to almost end of life (220 g). Sarver found that at Day 5 in culture the animals were allocating about 40% of energy of C to somatic growth. This diminished to 0% after Day 60 and remained at this level to the end of the study. The first allocation of energy to spawn production was recorded at Day 17. This increased to a maximum of about 14% of C during Days 50 to 60, then diminished slowly thereafter to a level of about 10% at Day 103. Two points of major interest emerge from these data. First, when the animals initially reach reproductive maturity the two allocations to somatic

210

THOMAS H.CAREFOOT

and spawn production did not appear to ‘compete’ instantly for common energy to an extent that Ps was shut down completely. Rather, reproduction only gradually replaced somatic growth, and for about one-third of this species’ normal life span, both functions occurred together (over Days 20–50). At the same time, it should be noted that the smoothed curves in Figure 6 (simplified for the present paper) obscure an important feature noted by Sarver. This was that peaks of egg production by A. juliana were always correlated with low points for somatic growth, indicating expected short-term energy trade-offs. Secondly, diminished over the life of the animal (from 42–12%), possibly correlated with “scope for growth” an increase in percentage allocation of energy to R during the animal’s life to a maximum of about 10% of C on Day 80 (Fig. 6). The reason for this increase in R, as explained by Sarver, was that food consumption measured as a percentage of body weight actually decreased steadily and uniformly over the animal’s life, while the animal itself continued to grow larger. Only at about Days 60 to 70, when Ps became zero and allocation to Pr began to decline, did energy allocation to R begin to level off. The effect of dietary rationing on the pattern of energy allocation in sea hares was also investigated by Sarver (1978). He found that reproductively mature A. juliana maintained on low rations “favoured” egg production over weight gain, whereas on intermediate rations both somatic growth and egg production were supported about equally, and at ration levels approaching ad libitum a higher percentage of ingested energy was again allocated to reproduction. Energy flow in a coral rubble-inhabiting population of A. dactylomela in Barbados is shown in Figure 8. The data represent two groups of animals feeding in two habitats: an area dominated by Cladophora sp., and situated near the wave-break part of the sublittoral habitat; and the other, an area dominated by the red algae Gracilaria sp. and Laurencia papillosa and located a few metres offshore in slightly deeper water. The animals normally feed at night in the Cladophora area and rest during the day in the GracilariaLaurencia area, where crevices for hiding are more plentiful. These data on energy flow, representing combined field and laboratory studies (Carefoot, 1970, 1985), show that the red alga-dominated area is more energy-rich, provides more total food energy, and ultimately yields more energy represented by production than does the green alga-dominated area (by about 40%). Therefore, it is puzzling that selection favours a behaviour involving a feeding excursion each night to the shallow-water area with a return at dawn to the deeper-water resting areas. Any change in behaviour costs energy, and it would be presumed that the new activity would have to yield some advantage, either energetically, nutritionally, or whatever, to make the change worthwhile (cf. Larkin & McFarland, 1978). The ingestion of sand, as an aid to digestion, and availability of daytime hiding places, are factors that have already been considered with respect to this feeding excursion. It may be that the animals simply find Cladophora more palatable than the red seaweeds, even though Laurencia spp. are known to be preferred foods in populations of Aplysia dactylomela in other geographical areas (Morton & Miller, 1968; Switzer-Dunlap & Hadfield, 1979; Willan, 1979). In an earlier study of food choice in a field population of A. dactylomela in the same area in Barbados, Cladophora sp. was found to represent 60–70% of the foodstuffs in the dissected crops (Carefoot, 1970). LOCOMOTION Crawling is the major mode of locomotion in Aplysia, although a few species burrow and swim. The characteristic, slow movements during browsing alternate with periods of hunched immobility. Swimming has been studied most extensively in A. brasiliana=willcoxi, where it functions at least partly to enable animals to move out of shallow areas where they may be stranded by the tides, and involves a degree of navigation through celestial cueing.

APLYSIA

211

Fig. 7—Allocation of consumed food energy (C) to somatic production (Ps), spawn production (Pr), and respiration (R) in A. juliana: the animals were fed ad libitum on Ulva lactuca; “scope for growth” represents data from Sarver (1979).

CRAWLING Crawling is accomplished by a combination of muscular movements and hydraulic extensions of the anterior part of the body, the latter being involved more at high speeds. At low speeds the propulsive force is strictly by muscular waves which move from front to back along the animal’s foot (a monotaxic retrograde pattern; but see Pilsbry, 1951), one or two waves passing along it at any given time (Parker, 1917; Bebbington & Hughes, 1973). The waves lift part of the foot locally and temporarily from the substratum, enabling it to move forward while the rest anchors the animal in place (Parker, 1917). Muscles move blood into and out of lacunar spaces in the foot allowing contraction and expansion of various parts (Bebbington & Hughes, 1973). In addition to the longitudinal wave of contraction, Hening, Walters, Carew & Kandel (1979) describe a transverse contraction involving part of the body wall as well as the foot-sole which tends to constrict and narrow the foot. At higher rates of locomotion the anterior part of the body is extended by muscular antagonism against a hydrostatic skeleton, created by blood in various haemocoelic spaces. The anterior part of the foot-sole attaches to the substratum by a combination of mucous adhesion and suction, and the posterior part of the body is drawn up by muscular contraction. This produces the humping or “inchworm” style of locomotion so characteristic of a crawling sea hare. At highest speeds, as for example during escape behaviour or after full arousal following various noxious stimuli (see Wachtel & Impelman, 1973), the anterior body extends maximally and locomotion becomes a two-phase pattern of anterior extension and attachment, followed by release of the tail and pulling up of the body (Jahan-Parwar & Fredman, 1979a). In this “galloping” mode of locomotion, the body arches quite markedly. During vigorous locomotion the parapodial flaps may open and close and the mantle shelf and siphon may contract, producing a coordinated ventilatory pumping (Hening et al., 1979). Observation of such behaviour has led to the study of

212

THOMAS H.CAREFOOT

Fig. 8—Comparison of daily energy budgets for A. dactylomela feeding in two seaweed communities in Barbados: all values not otherwise indicated are in J·live g−1·day−1 for animals 60 g in live weight; the data show the ‘standing energy’ in a Cladophora- and a Gracilaria/Laurencia-dominated algal community, how much is eaten and wasted, and the allocation of consumed energy to various body processes in A. dactylomela; data from Carefoot (1970, 1985).

neural control of locomotion, initially by Jordan (1901) in A. fasciata and, more recently, by Hening et al. (1979) and Jahan-Parwar & Fredman (1978a, b, c, 1979a, b, 1980) in A. californica. Locomotory rates in some of the larger species, such as A. dactylomela, range from 10 cm·min−1 during slow crawling, to 50–150 cm·min−1 at full gallop. Strumwasser (1967) recorded a mean velocity of 1 m·h−1 in A. californica over a 12-h period of normal daytime activity, while Jacklet (1972) and Hening et al. (1979) recorded maximum velocities of 10–18 m·h−1 in the same species. Average distances moved by A. californica in laboratory studies were 35 m in 12 h (in a cycle of 12 h light: 12 h dark; Kupfermann, 1968), and 35–40 m in 24 h (in 24 h constant light; Jacklet, 1972). Kupfermann (1965) recorded movements of up to 200 m·day−1 in A. californica in the laboratory, while Willan (1979) measured normal crawling rates of 1–5 m·h−1 in A. dactylomela, with no observable correlation between size and rate. Aplysia rarely travel far in the field. Kupfermann & Carew (1974) recorded daily movements of tagged A. californica in the field of up to 100 m, although when they followed a single individual slowly grazing for one hour on its seaweed foods, it moved a distance of only 6·3 m. In studies of movements of tagged A. californica in sites at Santa Catalina Island, Audesirk (1979) noted that nine animals spent at least 4–5 months in an area not larger than 5600 m2. In comparison, tagged A. dactylomela in Barbados moved no more than 10–20 m·day−1, and many individuals remained several days in an area of about 100 m2 (Carefoot, pers. obs.). For a species like A. dactylomela which inhabits daytime hiding places, the question arises as to whether there is any fidelity to specific home-sites after nightly feeding excursions. Monitoring of movements of tagged individuals and observing known resting spots from day to day has shown none; the animals just seem to select crevices and under-rock hiding sites opportunistically as dusk nears (Carefoot, pers. obs.). Crawling can be elicited in Aplysia by several stimuli including: (1) food (Frings & Frings, 1965; Preston & Lee, 1973; Jahan-Parwar, 1972a; Audesirk, 1975a, b; Jahan-Parwar & Fredman, 1979a), (2) pheromones (Jahan-Parwar, 1976; Audesirk, 1977; Audesirk & Audesirk, 1977), and (3) various noxious stimuli (salt,

APLYSIA

213

heat, pinching, electrical shock: Wachtel & Impelman, 1973; Walters, Carew & Kandel, 1978; salt: JahanParwar & Fredman, 1979a, b). In addition, light intensity appears to affect locomotory rates. Jacklet (1972) showed in one experiment that A. californica subjected to 40 lux intensity of light over a continuous 24-h period moved a mean distance of 39 m, while those subjected to 280 lux moved a mean distance of 60m. A generous endowment of mucous glands in combination with the musculature of the foot provides Aplysia with the ability for suction attachment over the full area of the foot (Parker, 1917), but in most species development of anterior and posterior grouped mucous glands enables the animal to cling better by its front and back portions than by the remaining surface (Engel & Eales, 1957). Some species, as for example A. dura and A. juliana, even possess a sucking disc on the posterior part of the foot. This enables the animal to cling especially well by the posterior portion of the foot while raising the anterior part of the body from the substratum. Sea hares are adept at gripping with the anterior part of the foot. This is noticeable during copulation in some species (e.g. A. dactylomela) where a small pinch of the sperm-recipient’s skin may be gripped tightly by its partner. It is also evident during feeding, where an animal may use the foot to hold seaweeds, or during dislodgment, where re-attachment may be facilitated through gripping or by suction action of the anterior part of the foot. Kupfermann & Carew (1974) noted in A. californica what they described as particularly strong adhesive properties of mucous “glue” secreted by animals dislodged in strong surge conditions. Unless this species produces mucus of varying adhesive properties, as for example depending on the degree of water movement, it is likely that the observers’ fingers were stuck not be a mucous glue as described, but simply by suction generated by prehensile portions of the foot. Under normal circumstances the foot mucus in sea hares is not at all sticky. A circadian rhythm of locomotory activity is present in A. californica (Lu, Strumwasser & Gilliam, 1966; Kupfermann, 1968; Jacklet, 1972, 1974; Block & Roberts, 1981), a species that is normally active during daylight hours. When kept in constant darkness it will, however, maintain an abbreviated form of the same activity cycle for at least 48 h (Kupfermann, 1965, 1968). A remarkable pacemaker system in the eyes is thought to be involved in the circadian rhythm of locomotion. Considerable research has been done on this pacemaker system with respect to: (1) the circadian activity of the ocular pacemaker system and its entrainment by light cycles (Jacklet, 1969, 1974; Eskin, 1979; Current, Eskin & Kay, 1982), (2) its proposed relationship to locomotory activity (Jacklet, 1972, 1976), (3) the photoreceptors responsible for the entrainment to light, both within the eye itself (Eskin, 1971, 1979) and through possible activity of extraocular photoreceptors (Block, Hudson & Lickey, 1974; Jacklet, 1980), (4) the effect of removing the eyes or cutting the optic nerve on locomotion and other behavioural activities (Lickey et al., 1977; Lickey & Wozniak, 1979), (5) the possible modes of synchronization of the various pacemaker systems (Lickey, Hudson & Hiaasen, 1983; Jordan, Lickey & Hiaasen, 1985), and (6) the fine structure of the eye itself (Jacklet, Alvarez & Bernstein, 1972). There are actually three circadian rhythms in A. californica: in the abdominal ganglion (Lickey, 1969; see also Strumwasser, 1973); in the eye; and locomotory. The current view is that the circadian pacemaker in the eye is at least one of the controls for locomotory rhythm (Jacklet, 1972; Strumwasser, 1973; Lickey et al., 1977; Block & Roberts, 1981; Lickey et al., 1983), yet some extraocular regulation must be involved because eyeless animals maintain a normal diurnal pattern of activity for some time (Block, 1971; Block et al., 1974; Lickey et al., 1977; Lickey & Wozniak, 1979), although with deteriorated fidelity (Strumwasser, 1973; Lickey et al., 1977). The extraocular photoreceptors could be in the central neurons (Arvanitaki & Chalazonitis, 1961; Block & Smith, 1973; Brown & Brown, 1973), or in the siphon, rhinophore, or other areas of the skin (Lukowiak & Jacklet, 1972a; Chase, 1979b; Jacklet, 1980). Lickey et al. (1977) put

214

THOMAS H.CAREFOOT

forward the possibility, although admitted remote even by the authors, that the extraocular oscillator which drives the system in eyeless A. californica may reside in some symbiotic or parasitic organism (e.g. crabs). Assuming that the eye and optic nerve behave in the same way in vivo as in isolated preparations in the laboratory, a normal day for A. californica begins about an hour before dawn with a burst of impulses from the optic nerve, with frequency peaking just after mid-day, then attenuating through the remainder of the day (Jacklet, 1969). Locomotory activity is never completely absent during the night, but similarly shows a pronounced burst at dawn and reaches a peak in intensity by mid-afternoon, By dusk, or shortly thereafter, the animals are mostly quiescent (Kupfermann, 1968; Jacklet, 1972), and the impulse frequency from the optic nerves is diminishing at this time (Jacklet, 1969). Just as the locomotory rhythm maintains its circadian pattern in animals in total darkness, so the discharge frequency of the optic nerve retains its circadian rhythm when isolated eyes are kept in darkness. Interestingly, the free-running cycle for isolated eyes in vitro in darkness is about 26 h, with some variation, and can be maintained for a week or more in culture (Jacklet, 1974). No work has been done on a nocturnally active species, such as A. dactylomela or A. juliana, to test whether the supposed ocular or other pacemaker systems operate in these species on a 12-h phase shift. BURROWING Burrowing in sea hares may function as a mechanism to avoid light (specifically ultraviolet radiation), wave action, or intertidal exposure (see p. 173). In A. brasiliana=willcoxi, an habitual burrower, burrowing is accomplished in two phases (Aspey & Blankenship, 1975, 1976a). In the first phase, a sequence of shovelling movements with the head and oral tentacles serves to bury the front part of the body. In the second phase, following immediately after the first, a series of swelling and forward heaving movements of the entire body ensues, easing the animal into the substratum until it is fully covered save for the rhinophore tips, the opening to the mantle cavity area, and the siphon. An interesting variant on this pattern has been described for A. geographica (proposed as a new species: formerly Siphonota geographica: Willan, pers. comm.). In this animal, digging is accomplished by muscular (hydraulic?) thrusts of the head. First, the head is turned vertically and thrust into the substratum, and then is simultaneously dilated and returned to the horizontal plane, thereby parting the sand and creating a deeper hole. The oral tentacles act as wedges in the first phase and as lateral ploughs in the second, while the animal rocks back and forth throughout. Eventually, all that is visible of A. geographica are the tips of its rhinophores, the centre field of its parapodia, and its anal siphon. The animal can draw clean water for gas exchange into the mantle area through apertures created by the anterior edges of the parapodia and expel it though the siphon. Burrowing has been studied in A. brasiliana=willcoxi in relation to weight and general condition of the animal, time taken to burrow, time spent under the substratum and degree of subsequent coverage, latency of re-burrowing, degree of coverage when buried, orientation during burrowing, vigour of burrowing, propensity to ink when removed from the burrow, and behaviour following emergence (Aspey & Blankenship, 1976a). Among other things, these authors found that smallest animals burrow the fastest, are most reponsive to disturbance when they are burrowing, and are most likely to engage in reproductive activities following emergence. Animals remain under the substratum for periods ranging from one hour to several weeks. Further studies showed that buried animals induce burrowing in swimming conspecifics in aquaria, presumably through a pheromonal mediator (Aspey & Blankenship, 1976b). The emergence of one buried group member from its burrow causes other animals to follow, and mass copulation often ensues (Aspey & Blankenship, 1975). The authors proposed that burrowing in A. brasiliana=willcoxi, at least in small individuals, serves as some form of preparation for subsequent reproductive activity. In this respect, the

APLYSIA

215

situation resembles the tendency of A. dactylomela to undertake copulation following emergence from burrows and crevices in which they hide overnight (Carefoot, 1985). SWIMMING Seven species of Aplysia are known to swim (A. brasiliana=willcoxi, A. depilans, A. extraordinaria, A. fasciata, A. morio, A. pulmonica, and A. tanzanensis), and another two may (A. maculata and A. winneba; Eales, 1960). The observations by Allan (1941), MacNae (1955, 1957), Eales (1960), and Kay (1964) that A. dactylomela can swim, by Allan (1941) that A. nigra and A. parvula can swim, by Eales (1921) and Haefelfinger & Kress (1967) that A. punctata can swim, and by Kandel (1976, 1979) that A. juliana can swim, do not appear to be supported by fact. Sea hares swim by a similar parapodial flapping in all species, but with possibly different mechanisms of propulsion through the water (Neu, 1932; Pruvot-Fol, 1954; Farmer, 1970; Bebbington & Hughes, 1973; von der Porten et al., 1980, 1982). The propulsive force is a wave of muscle contraction passing from front to back along each parapodium, beginning in A. fasciata in the right parapodium slightly in advance of the left, and commencing from a “starting” position in which the flaps are folded, right over left, to cover the mantle area (Bebbington & Hughes, 1973). Parapodial flaps are fully extended momentarily at the end of this part of the propulsive stroke. The “recovery” stroke begins at the front of each parapodium and, through a wave of contraction moving from front to back, serves to fold the flaps once more over the mantle cavity. The way in which this propulsive wave translates into a swimming motion in Aplysia is unclear. Sculling and jet propulsion are the most popular theories (the latter proposed initially by Neu, 1932, and later reconsidered by Kandel, 1979), yet appear too simplistic to explain movement in A. brasiliana=willcoxi. Rather, von der Porten et al. (1982) suggest that in this species the thicker leading edge of each parapodium presents an “airfoil” which generates “lift”. This lift force is thought to be produced on both downstroke (extension) and upstroke (flexion) phases of the flapping cycle of the parapodia, and thus would explain the smooth forward progression of sea hares as they flap along, rather than the cyclical jerking which would accompany either sculling or jet propulsion. The authors propose that thrust is generated by the anterior onethird of each parapodium, while the posterior two-thirds may operate for attitude control. This theory offers an intriguing and novel approach to the study of swimming in sea hares and is bound to generate fresh interest in the subject. Neuronal control of swimming has been studied by Jahan-Parwar & Miller (1978), Pinsker et al. (1978), von der Porten et al. (1980, 1982), and others. There appears to be a neuronal oscillator in each pedal ganglion that regulates the flapping frequency. Through lesion studies, the activation “command” for swimming has been found to arise in the cerebral ganglion (Jahan-Parwar & Miller, 1978; von der Porten et al., 1980, 1982). A complete flapping cycle, including both extension and flexion, in an A. fasciata of 20-cm length takes about three seconds (temperature unspecified: Bebbington & Hughes, 1973) and carries the animal about one body length. This rate compares favourably with the rate of one cycle·2·6 s−1 recorded for A. brasiliana=willcoxi by von der Porten et al. (1982) at a temperature of 13·5°C, probably similar to the temperature used in the previous study on A. fasciata. The authors showed a marked effect of temperature on flapping frequency and swimming speed in A. brasiliana=willcoxi (0·4 beats·s−1 gives a swimming speed of 4·9 cm·s−1 at 13·5°C, while 0·6 beats·s−1, gives a speed of 6·1 cm·s−1 at 18°C). Interestingly, while temperature affects the frequency of parapodial flapping and swimming speed, it does not appear to affect rate of progression of the wave of contraction along the parapodia (von der Porten et al., 1980). Although Aplysia regulates its swimming speed (as in a current: Hamilton & Ambrose, 1975), the mechanism by

216

THOMAS H.CAREFOOT

which this is done is not known. Since the frequency of flapping does not, however, seem to vary in A. brasiliana=willcoxi when swimming at different levels of effort in currents, it may be that flap amplitude can be regulated (Hamilton & Ambrose, 1975). Most of our knowledge of swimming in sea hares comes from the work of Hamilton and colleagues on A. brasiliana=willcoxi. This species lives in intertidal and shallow subtidal areas, often in seagrass meadows (Krakauer, 1969) from the Gulf of Mexico to Martha’s Vineyard in the northeastern U.S.A. Animals often swim at the ocean surface, especially during trips of more than 4 m (Hamilton & Ambrose, 1975), where they may exhibit the peculiar “head-bobbing” behaviour described by Aspey, Cobbs & Blankenship (1977; see pp. 229–231) and, by being in the upper part of the water column, they become prone to stranding themselves on the shore (Hamilton & Ambrose, 1975). The species can swim continuously for almost two hours in a single bout and may swim for distances of up to one kilometre (Hamilton, 1985). When released in shallow water the animals generally swim in a direct line offshore. In one release of 20 sea hares in a shallow lagoon area in southwestern Florida, Hamilton (1986) recorded a median swimming duration of 9·9 min, a median distance covered of 52 m, and a median “ocean floor” speed of 5·3 m·min−1. One energetic individual swam continuously for 114 min and travelled a distance of 953 m. In comparison, another actively swimming species, A. fasciata, was observed by Susswein et al. (1984a) to swim only in calm water and then only for short periods (in the laboratory about 50% of all swimming bouts lasted for 30 s or less; one unusual animal in the field swam for 35 min.). A. brasiliana=willcoxi is able to modulate its swimming effort according to the direction it is swimming relative to current direction (Hamilton & Ambrose, 1975; Hamilton, 1984). Swimming in a current may, however, be excessively energy-demanding. For example, individuals released in an area of strong current swam for less than half the time of ones released in an area of weak current (7·8 and 21·0 min, respectively; Hamilton, 1986). The eyes of Aplysia are tiny and simple. Each has a relatively large lens and has about 7000 receptor cells in the retina (Jacklet et al., 1972). They may be sensitive to ultraviolet light (Waser, 1968). While the eyes do not appear to have image-resolving capability, Hamilton (1986a) in fact thinks that A. brasiliana=willcoxi may be able to see objects above the water’s surface. Both swimming at the surface and the “head-bobbing” behaviour described by Aspey et al. (1977) are thought by Hamilton & Russell (1982a) to aid in celestial navigation. In their studies of swimming in A. brasiliana=willcoxi, Hamilton (1979) and Hamilton & Russell (1982a, b) found that straight-line navigation is mediated through a combination of orientation to waves and celestial cueing. When animals are released from points near the shore, they swim offshore initially towards waves. This behaviour is thought to be mediated through the rhinophores, since their surgical removal brings about rapid disorientation. Since animals move in an offshore direction even in the absence of waves, however, some perception of horizon (e.g. the tree-line) may be involved in setting the initial direction (Hamilton & Russell, 1982b). For animals swimming without visual reference points on the horizon, celestial cues are thought to be used. One of the points of evidence in favour of this idea is that animals swimming under a translucent white cover become quickly disorientated (Hamilton & Russell, 1982a) suggesting that an unblocked view of the sky is important for navigation. The cues involved in celestial navigation are presumably the sun and possibly polarized light, but neither these nor the integrating mechanism involved in navigation are known. Blinded animals, however, become disorientated when swimming, so it is thought that an actual visual detection of celestial cues is responsible (Hamilton & Russell, 1982a, b). The statocyst is probably also important in the orientation of swimming Aplysia. Its structure and neural connections to the cerebral ganglion have been described by Coggeshall (1969), Dijkgraaf & Hessels (1969), Wolff (1973), Gallin & Wiederhold (1977), and Janse (1983). If the nerve from one of the paired statocysts to the cerebral ganglion is severed unilaterally there is no effect on posture or movement;

APLYSIA

217

bilateral severing of the statocyst nerves, however, markedly affects posture and causes somersaulting in swimming A. fasciata (Dijkgraaf & Hessels, 1969). The simply structured statocysts (containing only 13 cells each in A. californica and A. fasciata: Coggeshall, 1969; Dijkgraaf & Hessels, 1969) only provide information on gravity perception and not on rotation or on acceleration or deceleration (Dijkgraaf & Hessels, 1969; Wolff, 1973). Neither the function of swimming nor the need to navigate are well understood in Aplysia. In A. brasiliana=willcoxi, swimming has been implicated in reproduction (Hamilton et al., 1982), food-finding (Aspey et al., 1977), as well as in moving between eelgrass beds in search of mates, food, or more suitable habitats (Hamilton, 1985, 1986). Susswein (1984) reports that the urge to swim intensifies in A. fasciata just after dusk whether food is present or absent, but that in this noctural species swimming remains at a high level through the night only when food is absent. As A. brasiliana=willcoxi usually swim offshore from shallow inshore release points, strandings could be minimized by this behaviour (Hamilton & Russell, 1982b). A point in favour of this idea is that A. brasiliana=willcoxi are stranded more often on beaches after night-time high tides than after daytime high tides (Hamilton & Russell, 1982b; Hamilton et al., 1982; Krakauer, 1969), but whether this means that this nocturnal species just swims more at night is not known. No studies appear to have been done on the navigational abilities of A. brasiliana=willcoxi during night-time bouts of swimming. PREDATORS AND DEFENCE Aplysia has few predators. Its possible defences include ink, opaline secretions, and toxins in the digestive gland. In addition to the distinctive smell of the opaline secretions, the body often has a fruity odour, which in some way may be related to defence. As members of a molluscan group in which common or popular names are rare, sea hares are distinguished in areas of Brazil as “inkwells”, and in the Gulf of Mexico as “inkfish”, through their propensity to release clouds of reddish-coloured ink when disturbed. Interest in the chemical defences of sea hares and in their ability to sequester secondary metabolites from their foods has led to a vast outpouring of work by ‘natural products’ chemists. PREDATORS No animals are known to prey solely on sea hares, nor are ones known for which sea hares account for even a regular portion of the diet. A number of examples of predation have been reported, and are listed in Table XII, but these appear to be ‘low intensity’ interactions: examples of opportunism, or otherwise sporadic encounters leading to predation. Of these examples, predation on the eggs and juveniles predominate (e.g. MacGinitie, 1934; Winkler & Tilton, 1962; Sarver, 1979; Willan, 1979). MacGinitie (1934) notes that only after A. californica reach a size of 3–4 mm in length do they become distasteful, and it may be that their defences do not reach full operating status until a minimum size is attained. This may be equivalent to the “size refuge” attained by A. dactylomela, which protects it from predation by the starfish Coscinasterias calamaria in New Zealand (Willan, 1979), and by Aplysia juliana in response to certain carnivores in Hawaii (Sarver, 1979). Predation by the great green sea anemone Anthopleura xanthogrammica on several Aplysia californica (Winkler & Tilton, 1962) was probably a unique occurrence and one unlikely to exert a steady pressure on the population. The reasons for this are twofold. First, in a comprehensive field study on A. californica, Kupfermann & Carew (1974) never observed predation by this species of anemone on A. californica, nor has it been reported by other authors working on this species (e.g. Audesirk, 1979). Secondly, the anemone prefers a habitat of wave-swept surge channels where it feeds opportunistically on

218

THOMAS H.CAREFOOT

food materials washing by, but such channels are not favoured by A. californica (see Table I, p. 174; although it is possible that the two species may meet in tide-pools). There is also a possibility that the sea hares in Winkler & Tilton’s study were dying or already dead before being caught up in the tentacles of the anemones, as the authors did not actually witness their capture. Finally, although no data are available, it is presumed that predation TABLE XII Predators of Aplysia Species

Predator species

Aplysia spp.

Bailer shell: Melo amphora Green sea turtles: Caretta caretta Aeolid nudibranch: Favorinus japonicus Starfish: Asterina Starfish Great green sea anemone: Anthopleura xanthogrammica Cephalaspidean opisthobranch: Navanax inermis Flatworms, nemertines, annelids, crabs, isopods, hermit crabs, fish Starfish: Coscinasterias calamaria Starfish: Patiriella regularis Hermit crab: Dardanus sp., cone shell: Conus pennaceus, crab: Calappa sp., also several wrasses, two species of flatworms

brasiliana =willcoxi californica californica

dactylomela

juliana

Remarks

References

Eats the eggs

Coleman, 1975 Felger & Norris (cited in Fenical, 1975) Kay, 1979

Young individual captured Eats the eggs Only juveniles eaten

Sawaya & Leahy, 1971 MacGinitie, 1934 Winkler & Tilton, 1962

Eats sea hares in both field and laboratory

Paine, 1963

Only juveniles eaten

MacGinitie, 1934

Preys on juveniles

Willan, 1979

Eats the eggs All prey on juveniles

Sarver, 1979

on the veliger larvae of Aplysia by filter-feeding animals such as tube worms, bivalves, some anemones, and fish, is enormous. In this regard, Krakauer (1969) has observed predation on A. brasiliana=willcoxi veligers by an unidentified hydroid and the filter-feeding crab Petrolisthes armatus in the laboratory. Few animals have been seen to attack healthy adult sea hares. Susswein et al. (1984a) described an attack on Aplysia fasciata by an unidentified crab in the field, but were unable to tell if the sea hare had been previously injured. In feeding experiments with fish, Thompson (1960) noted that no live A. punctata were eaten. In tests of palatability of A. dactylomela to various fish, Russell (1966) showed that portions of the foot from otherwise healthy specimens were consistently refused, and Krakauer (1969) and Kinnel et al. (1979) showed that portions of the body of A. brasiliana=willcoxi were unpalatable to various fish and sharks (save for the buccal mass, which was consistently eaten: Kinnel et al., 1979). Similarly, Ambrose, Givens, Chen & Ambrose (1979) and DiMatteo (1981a) reported that various portions of the bodies of A. brasiliana=willcoxi and A. dactylomela were distasteful to laughing gulls, Larus atricilla, a species known as a fairly indiscriminate general scavenger.

APLYSIA

219

Defence mechanisms in sea hares can be categorized as active or passive. Active defences include: (1) escape by crawling, (2) withdrawal of the gill and siphon, and (3) inking and release of opaline secretions. Passive defences include: (1) crypsis, and (2) distastefulness. Another way to separate them, however, is by their behavioural and chemical characteristics, and it is according to these broad distinctions that the defences of sea hares will be treated. CRYPSIS While some authors would disagree (Thompson, 1960; Russell, 1966; Krakauer, 1969), sea hares are not particularly cryptic in their behaviour or in their colour. They tend to be large, bulky in shape and, to the human eye, fairly conspicuous. Their predominant colour is brown or shades of reddish-brown, often with lighter fleckings or patterns of white spots. Against their favoured foods of red and green algae, only a few species seem to be camouflaged. Yet, in opposite view, their apparent distastefulness is not reflected by any sort of flamboyant warning colouration. This led Thompson (1960) to question the concept of adaptive colouration as it applied to animals such as Aplysia which, in his view, are both cryptically coloured and distasteful. In many other opisthobranch molluscs, warning colouration is associated with various defences, such as spicules or acid secretions. Ambrose et al. (1979) have even speculated that the reddish-purple ink of sea hares functions, not as a toxin, but in warning. The function of the ink will be considered more fully in a later section. Batesian mimicry has been proposed as an explanation for the similarity in appearance of post-larval juveniles of the tropical burrfish Chilomycterus antennatus and Aplysia dactylomela (Heck & Weinstein, 1978). The two species occur sympatrically in seagrass meadows along the Caribbean coast of Panama. Heck & Weinstein point out that neither species is apparently eaten by Caribbean fishes, despite a presumed palatability of the juvenile burrfish. Quite apart from the remarkable similarity in colour patterns in the two species, Heck & Weinstein’s theory is supported indirectly by a feature of the fish’s maturation. The burrfish appears to develop unusually rapidly from the juvenile stage, where it is protected by the proposed mimicry, to the relative safety of the inflatable, spiny, adult stage, thereby possibly reducing its period of vulnerability to predators. ESCAPE Sea hares escape from predatory animals by crawling. Willan (1979) noted that A. dactylomela crawl faster on contacting the predatory starfish Coscinasterias calamaria. Kandel (1979) also cited an unpublished observation by Dieringer & Koester that when Aplysia californica was pinched by pedicellariae of the starfish Astrometis sertulifera, it brought about a galloping escape response in the sea hare. DiMatteo (1981b) recorded an increased frequency of withdrawal by Aplysia dactylomela from contact with various predators, including the gastropod Fasciolaria tulipa and the starfish Echinaster sentus, over controls (empty Fasciolaria shells and live specimens of the herbivorous gastropod Strombus gigas). The author noted that Aplysia dactylomela also seemed to flee in response to pinches by the starfish’s pedicellariae. This same species of sea hare also withdrew from the jellyfish Cassiopea xamachana (Lederhendler, Bell & Tobach, 1975; DiMatteo, 1981b), although Lederhendler et al. (1975) were unsure whether the defensive withdrawal was to a chemical released by the jellyfish or to its nematocysts, or both. DiMatteo (1981b), however, demonstrated that withdrawal was at least partly due to the nematocysts. A common inference in such investigations is the likelihood of an evolved predator-prey interaction between the animals in question. Of course, this cannot be determined from such experiments and, in this

220

THOMAS H.CAREFOOT

regard, it is unfortunate that so little is known at present about sea hares in their natural habitats. Interpretation of the results in such studies has sometimes also been confounded by apparent escape behaviour of the sea hares to non-predatory animals. This was evident in the work of DiMatteo (1981b), where Aplysia dactylomela fled as much from contact with the detritivorous holothuroid Astichopus multifidus, as it did from contact with the predatory starfish Echinaster, and fled two and one-half times more from contact with the holothuroid than from contact with the molluscivorous snail Fasciolaria. Burrowing is probably too slow to be an effective means of escape from would-be predators, and has never been observed as such in sea hares (Aspey & Blankenship, 1976a). Hamilton & Russell (1982b) and Hamilton (1986b) similarly doubt that Aplysia brasiliana=willcoxi swim to escape from predators, although Willan (1979) has observed A. extraordinaria (Willan, pers. comm.) swimming (presumably from some threat), where the first flapping strokes have been accompanied by release of ink. Finally, Dieringer, Koester & Weiss (1978) recorded an increase in rate of heartbeat of up to 67% above baseline level in A. californica actively escaping from the apparently noxious stimulus of salt being applied to its parapodia. INKING Interest in the toxicity of Aplysia dates back to ancient times. Dioscorides in the first century AD was possibly the first to observe that sea hares were poisonous (Halstead, 1965). The naturalist, Pliny, also in the first century, noted that if taken with food or drink the sea hare may be poisonous to some, while to others its very sight causes death (Bostock & Riley, 1857). Interest specifically in the toxicity of the ink of Aplysia dates from the work of Flury (1915) on A. fasciata. The ink is produced in special glands (called Blochmann’s glands after their discoverer, or ink or purple glands) opening into the mantle cavity on the underside of the free edge of the mantle shelf. All species of sea hares possess ink glands save for A. cedrosensis, A. depilans, A. dura, A. juliana, A. nigra, and A. vaccaria. Early studies on Aplysia ink which refer to A. depilans, as for example that of Schreiber (1932), presumably involve a misidentification, possibly of A. fasciata. No comparisons have been made between species with and without ink glands on any aspect of defensive behaviour. The chemical composition of the ink has been investigated by De-Negri & De-Negri (1876), Moseley (1877), MacMunn (1899), Flury (1915), Schreiber (1932), Fontaine & Raffy (1936a, b), Lederer & Huttrer (1942), Christomanos (1955), Winkler (1959a), Nishibori (1960), Rüdiger (1967a, b, 1968), and Chapman & Fox (1969). In A. fasciata it consists mostly of water and other volatile materials with about 1·6% organic substances and 4·6% minerals (Flury, 1915). The violet pigment portion is bound to a protein (Christomanos, 1955), and has been identified by Rüdiger (1967a, b) and Rüdiger, Carra & hEocha (1967) as a bile pigment (a monomethylester of a biladiene dicarboxylic acid) with the formula C34H40N4O7. Studies by Chapman, Cole & Siegelman (1967) and Chapman & Fox (1969) have shown that the ink is a monomethyl ester of phycoerythrobilin and is derived from the bilin chromophore of phycoerythrin contained in the red algae eaten by the sea hares. These authors further discovered that when A. californica are maintained on a diet of brown algae, which lack the essential phycoerythrin pigments, they become facultatively de-inked. The animals regain the ability to produce ink only when returned to a diet of red algae or to an artificial diet with phycoerythrins added. The ink of Aplysia has been referred to as aplysiopurpurin by MacMunn (1899; who also called it aplysine) and by Nishibori (1960). It has been described as consisting of two component chromoproteins aplysiovioline and aplysiorhodine by Lederer & Huttrer (1942), with the designation by Winkler (1959a) of a third component aplysioazurin (this last being

APLYSIA

221

equivalent to phycocyanobilin: Chapman & Fox, 1969). Finally, the ink has been termed aplysioviolin by Rüdiger (1967a, b) and Chapman & Fox (1969), and is the name which appears to be most commonly used. The release of ink is a high-threshold, all-or-none response (Carew & Kandel, 1977a, b, c; Shapiro, Koester & Byrne, 1979; Byrne, 1981). Rough handling in some species will often, but not always, elicit release of ink, as will separation of copulating animals. Tobach, Gold & Ziegler (1965) obtained inconclusive results in field tests of the factors causing ink to be released in A. brasiliana=willcoxi and A. dactylomela. Only sometimes did the animals release ink when they were handled or pricked with a pin, and there was no consistency in response to the different levels of stimulation. The authors did, however, observe that animals previously found in pairs or in larger groups in the field showed a slight, but significantly greater, propensity to release ink than those which were previously solitary. The significance of this observation is not known. Kupfermann & Carew (1974) never observed release of ink under natural conditions in field A. californica, but they could induce release of ink by separating copulating animals (in which case the sperm-donor frequently inked) or by vigorously hand-stimulating the mantle cavity. Interestingly, in their field studies of A. californica, Carew & Kupfermann (1974) showed a significant difference in the propensity of animals from different habitats to release ink after being stuck with a pin. Thus, sea hares from rough-water habitats were much less likely to release ink than were animals from calmwater habitats. Susswein et al. (1984a) noted that A. fasciata never released ink spontaneously, and would do so only occasionally when separated during mating. Finally, Advokat (1980) demonstrated that recently fed animals were less likely to release ink than were animals that had been starved for 24 hours. It is not known how long it takes for a sea hare to replenish its ink supply after spontaneous release, although an estimate can be made from data given by Chapman & Fox (1969) in their experiments on deinking of A. californica. The authors rubbed the ink gland and adjacent tissues with their fingers daily, and showed that an animal on a bilin-free diet (e.g. the brown alga Egregia laevigata) could be effectively deinked after five days (to about 1% of normal level), and completely de-inked after 14 days. They believe that under normal circumstances some residual ink may always be retained in the ink gland, even under conditions of severe handling. From these results it was estimated that normal animals on a diet of red algae could regenerate their ink in at least two days. Experimental animals that were kept on a bilin-free diet and manually de-inked to facultative exhaustion of the ink supply, then fed on a normal diet of Laurencia sp., required three days to begin releasing new ink (Chapman & Fox, 1969). In some species the ink is released along with considerable mucus (Schreiber, 1932; Sawaya & Leahy, 1971; Willan, 1979; DiMatteo, 1982), which may vary in amount between different animals in a given species, such as A. dactylomela, and in the same animal at different times (Carefoot, pers. obs.). If this is generally true for other species of Aplysia, it will add to the already difficult task of making quantitative assessments of the effect of ink in studies of predator-prey interactions. Release of ink in response to predation has been rarely witnessed. Willan (1979) noted that a field A. dactylomela released its ink while attempting to escape from the starfish Coscinasterias calamaria. The ink had no apparent effect on the starfish. In contrast, ink released by Aplysia parvula in response to attack by the same starfish under laboratory conditions seemed to retard the locomotory movements of the predator, thus allowing the sea hares to escape (Willan, 1979). Susswein et al. (1984a) observed a crab attacking an A. fasciata which did not release its ink, although they noted that the sea hare may have been unhealthy even before the attack. Finally, in a study on the effects of the toxin ‘holothurin’ from the sea cucumber Actinopyga agassizi on Aplysia dactylomela and of the reciprocal effects of the ink from A. dactylomela on the sea cucumber, Lederhendler et al. (1975) found that A. dactylomela released its ink in response to the presence of holothurin in the water. The sea hare’s ink did not, however, appear to affect the sea cucumber in a reciprocal manner.

222

THOMAS H.CAREFOOT

Despite the apparent defensive rôle of ink in sea hares, contradictory results have been obtained in experiments to demonstrate it. For example, Flury (1915) directly injected or added the ink of A. fasciata dropwise to the sea-water medium of a variety of animals including crabs, fishes, and frogs, and showed that it had little or no effect. Krakauer (1969) reported that pieces of shrimp soaked in the ink of A. brasiliana=willcoxi were readily eaten by blue crabs Callinectes sapidus. Similarly, Ambrose et al. (1979) soaked pieces of fish in a mixture of ink, blood, and mucus from Aplysia brasiliana= willcoxi and showed that the mixture was not unpalatable to laughing gulls Larus atricilla. On the basis of these experiments the authors concluded that Aplysia ink was possibly important as a signal of unpalatability to potential predators, rather than for its supposed toxic properties. Contrasting results were, however, obtained by DiMatteo (1981a) using the ink of A. dactylomela in similar tests with laughing gulls. He demonstrated that gulls clearly found the ink unpalatable when it was injected into pieces of fish rather than simply coated on to, or soaked into, the fish. The author suggested that Ambrose et al. (1979) may have added insufficient ink to the gull’s treated food to elicit a rejection. Further studies by DiMatteo (1982) on the possible defensive rôle of Aplysia ink showed that several species of crabs living sympatrically with A. dactylomela in their seagrass habitat (Panopeus herbstii, Mithrax sculptus, Portunus spinimanus, and Callinectes sapidus) were repelled by otherwise edible pieces of fish coated with Aplysia dactylomela ink. Willan (1979) tested the effects of A. dactylomela ink in solution on survival of a blenny Tripterygion varium, a species occupying the same habitat as the sea hare. Preliminary tests showed that ink from a single Aplysia dactylomela (80 g live weight) in three litres of sea water was toxic to the blenny, although the author could not be sure that the mucus released with the ink was not also deleterious to the fish. The blenny appeared to die of respiratory distress. It is surprising that with all the interest shown in the effect of Aplysia ink on various potential predators, no one has seriously investigated the effect on Aplysia of its own ink. In fact, Lederhendler et al. (1975) did test for this in a single A. dactylomela but found no apparent response. There is also no evidence at present that the ink of sea hares acts as a “cryptic odour” (Kittredge, Takahashi, Lindsey & Lasker, 1974) in the manner described for the ink of Octopus by MacGinitie & MacGinitie (1968). On the basis of these and other studies, the following are the possibilities for the function of ink in sea hares: (1) to rid the animal of unwanted bile pigments from biliproteins consumed in its diet (Chapman & Fox, 1969); (2) to function in defence as a “smoke-screen”, thus enabling the sea hare to escape (Eales, 1921; Halstead, 1965; Hyman, 1967), although this may only be effective in tide-pool inhabiting Aplysia (Carew & Kandel, 1977a; Kandel, 1979); (3) to function in defence through its unpalatable qualities (Beeman, 1961; DiMatteo, 1981a, 1982); and (4) to function as a warning to would-be predators of the sea hare’s other toxic properties (Ambrose et al., 1979). The only function the ink certainly fails to fulfil is that of a clothes dye: the pigment is not colour-fast (Sanford, 1922) and turns a dirty brown upon exposure to air and sunlight (Winkler, 1959a). OPALINE SECRETION In comparison with the extensive work done on the ink of Aplysia, little is known of the other secretion from the mantle cavity, that of the opaline glands. Although a defensive function is generally assumed for the opaline secretion, there is only indirect suggestion in the literature to support this. All sea hares produce the secretion in a collection of subepidermal gland cells known as the opaline glands or glands of Bohadsch, located in the mantle floor near the gonopore (Eales, 1960; Hyman, 1967). The milky fluid is described as having a musky, acrid, nauseating, odoriferous, foetid, or otherwise strong smell in various species (Flury, 1915; Ando, 1952; Eales, 1921, 1960; Kay, 1964; Hyman, 1967; Krakauer, 1969; Willan, 1979). Little is

APLYSIA

223

known of the composition of the opaline secretion. The only analysis of it seems to be that done in Flury’s (1915) early studies on A. depilans. He showed that the secretion consisted of 94·8% water and other volatile substances, with 1·6% organic matter, and 3·6% mineral material. There is little information on factors affecting release of these opaline secretions in sea hares. At the biochemical level, Tritt & Byrne (1980, 1982), and Tritt, Zigmond & Byrne (1980) investigated the effects of various neurotransmitter substances on their liberation in A. californica. The authors concluded that dopamine was most probably the substance discharged by the motor neurons of the opaline gland to effect release of the secretions. Almost no information exists on the factors leading to natural liberation of opaline secretions in field animals, although when Carew & Kupfermann (1974) placed a juvenile A. californica into the tentacles of the great green sea anemone, Anthopleura xanthogrammica, the sea hare released its opaline secretions after several minutes of entanglement in the tentacles. The sea hare later broke free of the tentacles and seemed to be unharmed by its encounter. The rôle of opaline secretion in defence is generally inferred from the study of Flury (1915). The author subjected a variety of animals, including coelenterates, annelids, molluscs, arthropods, echinoderms, fish, frogs, and rabbits, to opaline secretions from Aplysia depilans by either injecting them with it or adding it dropwise to their sea-water medium. The test animals manifested various degrees of immotility and paralysis, which often led to death. Flury prepared a distillate of the residue remaining after alcohol extraction of the opaline secretion from A. depilans and concluded that its main toxic or nerve-paralysing property was in a nitrogen-free, water-distillable, volatile oil, which behaved physically and chemically like a terpene. When injected into octopuses, fish, and frogs, the oily substance caused death. Ando (1952) similarly demonstrated a toxic effect of the opaline secretions of A. kurodai on several invertebrates and thought that the toxic component may have come from the animal’s diet of the red alga Laurencia spp. Sure enough, steam distillation of Laurencia nipponica produced an oily fraction containing terpene-like substances which were found by Ando to be toxic to certain invertebrates. These appear to be the only accounts in the literature involving tests of the toxicity of the opaline secretions and thus, indirectly, of their possible defensive rôle in sea hares. SIPHON- AND GILL-WITHDRAWAL Although not involved in protection as such, the siphon- and gill-withdrawal reflex is a characteristic response by sea hares to weak tactile stimulation, such as touch or water currents, to the siphon and mantleshelf area. The siphon is a tubular fold of mantle tissue which normally protrudes from between the parapodia. Its function is to direct exhalant water from the mantle cavity away from the body to minimize contamination of the inhalant stream. Passage over the gill alters the carbon dioxide and oxygen content of the circulating water and at various times the exhalant stream contains excretory and faecal wastes. The siphon or mantle-shelf area responds to a low-threshold of touch, such that the siphon, mantle shelf, and gill contract and withdraw into the mantle cavity according to a graded response that is proportional to the intensity of the stimulus (Carew & Kandel, 1977a). Aplysia exhibit spontaneous siphon-withdrawal (Kupfermann & Kandel, 1969; Pinsker, Kupfermann & Castellucci, 1970), an action which may not be a defensive reflex but, rather, a way that the animal uses to clear the mantle cavity of debris including faecal material. Because of ease of preparation and monitoring, most studies have involved the gill-withdrawal part of the defensive reflex. Only short-term effects can be studied from dissected preparations where the animals have been restrained and the mantle region opened and pinned out. For long-term studies, the siphon-withdrawal response must be monitored in animals that are intact and healthy. The overall response is sensitive, easy to observe and to quantify, and has featured in a number of studies of habituation and

224

THOMAS H.CAREFOOT

sensitization in Aplysia. These studies on Aplysia have contributed greatly to our understanding of the neuronal and chemical mechanisms involved in learning in sea hares and, thus, in animals in general. The first detailed study of the gill-withdrawal response was by Kupfermann & Kandel (1969) who identified in A. californica the neuronal pathways involved in the reflex response. A feature of the system later remarked on by Pinsker et al. (1970) is the relative simplicity of the neuronal circuitry. Only a few synaptic relays are involved, which increases the speed and effectiveness of the reflex and thus fulfils one of the prime requisites of a good defensive response. Later work by these and other workers on the gillwithdrawal reflex in A. californica showed: (1) the existence of short-term habituation and dishabituation responses (Pinsker et al., 1970); (2) a causal relationship of habituation to a decrease in the amplitude of excitatory synaptic potentials produced at the motor neurons of the gill, and dishabituation to an increase of these synaptic potentials (Kupfermann, Castellucci, Pinsker & Kandel, 1970); (3) that true habituation was involved and not simply fatigue of the gill-withdrawal muscles or changes in response of sensory receptors (Kupfermann et al., 1970); (4) that habituation and dishabituation were independent events using the same neuronal pathways but caused by different (opposite) synaptic events (Castellucci, Pinsker, Kupfermann & Kandel, 1970); (5) that both the purple gland and siphon provide independent afferent pathways each capable of eliciting the gill-withdrawal reflex and each capable of becoming habituated independently (Carew, Castellucci & Kandel, 1971); (6) that short-term (a few hours only) habituation and dishabituation can occur in the absence of central nervous system control, apparently involving terminations of central motor neurons in the gill itself (Peretz, 1970; Lukowiak & Jacklet, 1972a, b; Peretz & Howieson, 1973; Peretz, Jacklet & Lukowiak, 1976; Lukowiak, 1977); (7) that the central nervous system interacts with the peripheral system in the gill to mediate the adaptive behaviour of the gill-withdrawal reflex (Lukowiak & Jacklet, 1972a, b; Peretz & Howieson, 1973; Peretz et al., 1976; Lukowiak & Peretz, 1977); (8) that longterm habituation of the response can be induced, lasting for several days (Carew, Pinsker & Kandel, 1972); and (9) that through application of a second stimulus to another part of the body, a long-term sensitization of the siphon-withdrawal reflex can be induced which lasts for up to three weeks following training (Pinsker, Hening, Carew & Kandel, 1973). Pinsker et al. (1973) noted that because sensitization of the withdrawal response involves the presentation of a second stimulus, usually to a different location on the body, it resembles to some extent classical conditioning (one neuronal pathway enhancing activity in another) and, hence, may be the basic adaptive mechanism from which true associative learning evolved. Classical conditioning of the siphon- and gillwithdrawal reflex using a weak tactile stimulus to the siphon (which produces only a weak withdrawal) as the conditioned stimulus, and a strong shock to the tail (which produces a strong withdrawal) as the unconditioned stimulus, was demonstrated in A. californica by Carew, Walters & Kandel (1981) and Carew, Hawkins & Kandel (1983). A similar type of associative learning was also demonstrated in in vitro preparations of the siphon, mantle, gill, and abdominal ganglion of A. californica by Lukowiak & Sahley (1981), in which they used light as the conditioned stimulus and a tapping on the gill as the unconditioned stimulus. In addition, Walters, Carew & Kandel (1979a, b) showed for A. californica that the locomotory escape response could be learned in aversive response to a chemosensory stimulus consisting of exposure to an extract of shrimp (the conditioned stimulus), and an electrical shock to the head (the unconditioned stimulus). The authors later showed that a suite of defensive responses, including withdrawal of the head and siphon, release of ink, and escape locomotion, could be conditioned in A. californica in response to shrimp juice paired with electrical shock (Walters, Carew & Kandel, 1981). In contrast, feeding behaviour was depressed in these animals in response to conditioning. Walters and co-workers likened these combined effects in conditioned sea hares to a functional equivalent of “conditioned fear” seen in mammals and other higher animals—a choice of terminology that to some will seem inappropriate, but is none the less

APLYSIA

225

an apt analogy. In a manner similar to vertebrates, Aplysia learns to associate a stimulus with possible danger and to prepare itself for defensive action. Long-term habituation of the siphon-withdrawal reflex has been shown to last for several weeks in both laboratory (Carew et al., 1972) and field. Carew & Kupfermann (1974) directed jets of sea water from a syringe at the siphon of A. californica and measured the magnitude of the response according to how long the contracted siphon remained hidden within the parapodia following its withdrawal. The authors were able to demonstrate that animals from turbulent habitats, which were regularly exposed to water-mediated tactile stimulation, exhibited a weaker siphon-withdrawal to the water-jet stimulation and habituated significantly faster than did animals from calm-water environments. The latter animals habituated at about the same rate as control animals in the laboratory and demonstrated a ‘brisk’ withdrawal reflex. The study was especially significant in that it was the first to show that habituation learning of this sort can occur as a natural event in a sea hare’s life. A number of factors are known to influence the siphon-withdrawal response, such as food, sexual activity, and age. Contact with food (including feeding) and sexual activity generally have a modulating effect on the response (Advokat, Carew & Kandel, 1976; Advokat, 1980; Lukowiak, 1980; Lukowiak & Freedman, 1983). Similarly, age appears to affect the withdrawal reflex by: (1) increasing the rapidity of habituation (Peretz & Lukowiak, 1975), but at the same time, (2) impairing long-term retention of habituation and preventing acquisition of sensitization of the reflex (Bailey et al., 1983; see also pp. 201–202), and (3) generally suppressing the sensitivity of the response (Peretz & Lukowiak, 1975; Lukowiak, 1979; Lukowiak & Peretz, 1980; Rattan & Peretz, 1981). Overall, young Aplysia seem to be less adaptable in their response by being less able to discriminate between stimuli of varying intensities, by habituating less easily, and by having generally less suppressive control of this behaviour than older animals (Lukowiak, 1977, 1979; Lukowiak & Peretz, 1980). Lukowiak (1980) also noted that the completion of a meal did not suppress the gillwithdrawal response in young A. californica to the extent that it did in old animals. These age-related differences in the siphon-withdrawal reflex may relate to the “size-refuge” idea noted earlier in this section, in which younger animals may have less well-developed (chemical) defensive capabilities than older ones; hence, would be generally more responsive to all types of stimulation. Older animals with their diverse and well-developed armoury of alternative defences can afford to “relax” their sensitivity to various tactile stimuli (Lukowiak, 1980). CHEMICAL DEFENCES The sea hare’s main line of defence appears to reside in the toxic properties of the digestive glands (also called midgut glands, hepatopancreas, or ‘liver’). The first indication in recent times of this toxicity came with the discovery by Winkler & Tilton (1962) that the digestive glands of juvenile A. californica eaten by the sea anemone Anthopleura xanthogrammica were largely undigested. In most instances the organs were egested by the anemones before the protective membranes enclosing the glands could be perforated by the action of the digestive juices. Preliminary tests, in which water- and acetone-extractions of the digestive glands of Aplysia californica were injected intraperitoneally into mice and other small mammals, and subcutaneously into frogs and baby chicks, indicated a high level of toxicity (Winkler, 1961). The test animals entered into various states of paralysis and respiratory distress, leading to death. Winkler, Tilton & Hardinge (1962) named the toxin aplysin and noted that its effect was reminiscent of cholin esters, such as acetylcholine and succinylcholine (see also Winkler & Tilton, 1962; Langlais & Blankenship, 1972). In later studies on A. pulmonica and other related opisthobranchs in Hawaii, Watson (1973) and Watson & Rayner (1973) identified two lethal substances in digestive glands originally extracted with acetone, one

226

THOMAS H.CAREFOOT

termed water-soluble toxin, the other, ether-soluble toxin. On intraperitoneal injection into mice, the ethersoluble portion produced irritability, visciousness, and flaccid paralysis. The water-soluble portion, in contrast, caused convulsions and respiratory distress. Both toxins ultimately killed the mice, although similar extracts from A. juliana were not lethal (Watson, 1973). Watson & Rayner (1973) compared the pharmacological effects of the water-soluble toxin from A. pulmonica and other aplysiids described by Watson (1973) with the effects of A. californica aplysin described by Winkler et al. (1962) and found some similarities between the two substances. At about the same time, a number of brominated compounds were being isolated from sea hares, mainly from the digestive gland. In the first of these studies, Tanaka & Toyama (1959) identified several brominated compounds in A. kurodai. Later, two bromine-containing sesquiterpenes were isolated from the same species by Yamamura & Hirata (1963). The authors independently named the substances aplysin and aplysinol. A third brominated material aplysin-20 was later isolated from A. kurodai by Matsuda, Tomiie, Yamamura & Hirata (1967) and Yamamura & Hirata (1971). The aplysin (from A. kurodai) and a related structure debromoaplysin have since been synthesized by Yamada, Yazawa, Toda & Hirata (1968), and the optically active forms, (—)-aplysin and (—)-debromoaplysin, by Ronald, Gewali & Ronald (1980). Winkler (1969) investigated the distribution of organic bromine compounds in A. californica and concluded that over 90% of them were present in the digestive gland, with most of the remainder being located in the foot, body wall, and skin. Further studies by Stallard & Faulkner (1974a) on A. californica showed that greater than 99% of the bromine was located in the ether-soluble fraction of the digestive gland. Bromine compounds were absent from the opaline gland in A. californica (Winkler, 1969). Some confusion therefore existed concerning the toxic compounds in Aplysia, not just in the designation of the name aplysin to two different substances, but in the chemical nature of the substances and their possibly differing pharmacological effects. As noted by Blankenship, Langlais & Kittredge (1975), the cholinomimetic aplysin of Winkler (1961) was fairly well characterized in terms of its action, but its chemical identity was unknown. At the same time the halogenated terpenoids described for A. kurodai (including the “aplysin” of Yamamura & Hirata, 1963) were well known chemically, but their pharmacological action, if any, was unknown. Although Watson (1973) did not test for brominated compounds in the ether extracts he obtained from A. pulmonica and other aplysiids, he did suggest that these might be similar to the brominated substances isolated from A. kurodai. In fact, a later study by Kato & Scheuer (1974) on Steilocheilus longicauda, a related aplysiid and one included in Watson’s study, showed that the ether-soluble toxin isolated by Watson from various opisthobranchs was probably a mixture of terpenoid-like substances found also in Steilocheilus, one of which was brominated (see also Scheuer, 1975, 1977). These substances were designated by the authors as aplysiatoxin and debromoaplysiatoxin (the latter substance was later isolated from the blue-green alga Lyngbya majuscula, where it has been shown to cause dermatitis in humans similar to “swimmer’s itch” and to have antileukemic properties: Mynderse, Moore, Kashiwagi & Norton, 1977; and to be a weak tumour promoter: Fujiki et al., 1982, 1984, 1985; see also Moore, 1982, and Willey, Moser & Harris, 1984). Furthermore, the cholinomimetic toxin first described by Winkler (1961), which possessed properties similar to the water-soluble toxin described by Watson (1973), was later isolated from A. californica and identified as urocanylcholine by Blankenship et al. (1975). Urocanylcholine and related choline esters belong to a family of toxic compounds known as murexine, more familiarly known from prosobranch gastropods in the families Muricidae and Thaisidae. These compounds mimic the effects of acetylcholine, are resistant to cholinesterase, and are fairly heat stable (Blankenship et al., 1975). These authors also reviewed the literature on the effects of murexine on animals and concluded that its pharmacological effects (mainly neuromuscular blocking action) were similar to

APLYSIA

227

those described for the original aplysin by Winkler (1961) and Winkler et al. (1962). Also, the activity of Winkler’s aplysin could not be eliminated by cholinesterase, which is similar again to the response of urocanylcholine and related choline esters. It therefore appears that the cholinergic aplysin originally described by Winkler may have been urocanylcholine, a member of the murexine family of toxins. Blankenship and his colleagues suggested that since the common name “aplysin” is most frequently applied to the halogenated terpenes, it should be used in this respect only, and not in reference to the cholinomimetic compound. From the time of these first studies in the 1960s, secondary metabolites of sea hares have been intensively investigated by ‘natural-products’ chemists. This has led to a rich harvest from Aplysia spp. of mono-, sesqui-, and diterpenoid substances, as well as other compounds, most of them halogenated, and most of them from the digestive glands (e.g. Faulkner & Stallard, 1973; Faulkner, Stallard, Fayos & Clardy, 1973; Faulkner, Stallard & Ireland, 1974; Kato & Scheuer, 1974; Schmitz, Hollenbeak & Vanderah, 1978a; Schmitz, McDonald & Vanderah, 1978b; and many others). The number and variety of these substances is quite amazing. Table XIII lists them and provides information on the body part sampled, on the chemical type, molecular formula, and special names, if any, given the substances, and on their pharmacological effects, if known. Little work has been done on the biological effects of these substances, particularly with regard to their rôle in the natural biology of Aplysia. The seaweed foods of Aplysia, with their wide variety of secondary metabolites, are thought to be the source of these various compounds. In the genus Laurencia alone, over 250 natural products have been isolated, many of them representing new structural types (Erickson, 1983). Winkler (1961) was perhaps the first to suggest that the toxins may derive from Aplysia’s seaweed diet, and he later considered that various red algae known to contain brominated compounds (as Laurencia spp.; see also Augier & Mastagli, 1956; Irie, Suzuki & Masamune, 1965a; Irie et al., 1965b; Irie, Suzuki, Kurosawa & Masamune, 1966; Irie, Suzuki & Hayakawa, 1969; Craigie & Gruenig, 1967; Bhakuni & Silva, 1974; Fenical, 1975) might be the source of the brominated organic compounds (Winkler, 1969). While Darling & Cosgrove (1966) were unable to identify a possible brominated precursor for these compounds in the red alga Plocamium cartilagineum, several lines of evidence support strongly the notion of a dietary source for these materials. First of all, Yamamura & Hirata (1963) initially noted that the kind and content of brominated substances in Aplysia kurodai depended on when and where the animals were collected. Mynderse & Faulkner (1978) also recognized a possible effect of diet on the content of halogenated substances in Aplysia. These authors were able to relate variations in content of certain halogenated monoterpenes in the digestive gland of A. californica to variations in the levels of these substances in the foods eaten, especially in the red alga Plocamium cartilagineum. By isolating and identifying 12 compounds, most with new structures, from Aplysia dactylomela collected near La Parguera, Puerto Rico, Schmitz et al. (1981) also provided evidence that the digestive gland components may vary with habitat. They did not isolate any of these same compounds from A. dactylomela collected at Bimini (Schmitz et al., 1978a, b). Further indirect support for the idea of a dietary source of the compounds was provided by Watson (1973). He showed that extracts and homogenates of the digestive gland of A. juliana were not lethal. This species eats mainly the green alga Ulva which lacks brominated compounds. A number of additional studies have related substances isolated from Aplysia with similar or even identical compounds in seaweeds, often from actual foods eaten by the sea hares. For example, various sesquiterpenes isolated from Laurencia spp. were found to be structurally similar to the sesquiterpenes of Aplysia (Irie et al., 1965a, b, 1966, 1969; see Scheuer, 1971 for review). In fact, Irie et al. (1969) isolated aplysin, debromoaplysin, and aplysinol from Laurencia okamurai, which suggests that some sea hares might simply ingest their brominated terpenes intact, rather than perform chemical changes on precursor

228

THOMAS H.CAREFOOT

compounds found in the seaweeds (Scheuer, 1971). Stallard & Faulkner (1974a) have also proposed that there may be a dietary source of the halogenated substances in the digestive gland of A. californica. These authors showed that a number of halogenated terpenoids including aplysin, debromoaplysin, laurinterol, pacifenol, johnstonol, and pacifidiene, and two monoterpenes, were stored in the digestive gland. Not only are these brominated terpenes known to be found in the algal foods of A. californica (Irie et al., 1966, 1969; Sims, Fenical, Wing & Radlick, 1972; Kazlauskes, Murphy, Quinn & Wells, 1976), but the digestive glands of animals maintained solely on diets of Laurencia pacifica and Plocamium cartilagineum were shown on later analysis to contain the terpenes of each respective alga (halogenated sesquiterpenes from Laurencia and halogenated monoterpenes from Plocamium: Stallard & Faulkner, 1974a). Two of these algal metabolites, laurinterol and pacifenol, were further shown to undergo transformations to aplysin and pacifidiene, respectively, within the digestive glands of Aplysia californica (Stallard & Faulkner, 1974b). This latter finding removed some of the doubt that the algal substances were not just being identified amongst the undigested food constituents in the ducts of the digestive gland. In addition, Faulkner et al. (1974) identified a halogenated sesquiterpene prepacifenol epoxide in extracts of the digestive gland of A. californica and showed its interconversion to “johnstonol”, the latter apparently an artifact formed during isolation of the epoxide from the red alga, Laurencia johnstonii, and previously identified as being present in the alga (Sims et al., 1972). Pacifenol, in turn, is known to be present in both the digestive glands of A. californica and in at least one of its seaweed foods Laurencia pacifica (Sims, Fenical, Wing & Radlick, 1971; Stallard & Faulkner, 1974a). Imperato, Stallard & Faulkner (1977) identified several polyhalogenated monoterpenes in the digestive glands of Aplysia fasciata and showed that one of these was present in the red alga Plocamium cartilagineum. In an interesting example of investigatory chemical ecology, the authors deduced that even though the crops of the test animals were filled mainly with the red alga Gracilaria verrucosa, the sea hares must at one time have eaten Plocamium cartilagineum, because Gracilaria does not contain halogenated monoterpenes while Plocamium does. Similarly, two diterpenes, dictyol A and dictyol B, were shown by Minale & Riccio (1976) to be present in the digestive glands of Aplysia depilans which were eating mainly the brown alga Dictyota dichotoma (known to contain these substances), but not in Aplysia fasciata or A. punctata which were eating mainly red algae and no Dictyota. Several other diterpene substances, including amongst them pachydictyol A, were found by Minale and colleagues (Minale & Riccio, 1976; Danise et al., 1977) in Aplysia depilans, and by Vanderah & Faulkner (unpubl. obs., cited by Fenical et al., 1979) in A. vaccaria. These compounds are known also from various brown algae, including Pachydictyon coriaceum (the source of pachydictyol A: Hirschfeld et al., 1973) and other members of the family Dictyotaceae (e.g, Dictyota dichotoma and Dilophus ligulatus: Fattorusso et al., 1976; Danise et al., 1977). Two bromosesquiterpenes discovered in the digestive glands of Aplysia brasiliana=willcoxi, brasudol and isobrasudol, were also found in the red alga Chondria cnicophylla on which the sea hare feeds (Dieter, Kinnel, Meinwald & Eisner, 1979). Finally, a diterpene monoacetate from Aplysia dactylomela (Schmitz et al., 1981) closely resembled the diterpene diacetate isolated from the red alga Laurencia obtusa (Higgs & Faulkner, 1982). In summary, the evidence from these studies on the chemical ecology of

APLYSIA

TABLE XIII Chemical substances in the digestive glands or whole body of Aplysia Species

Part of body sampled

Chemical type

brasiliana =willcoxi

Digestive gland

Choline (possibly acetylcholine and urocanylcholi ne)

brasiliana =willcoxi brasiliana =willcoxi

Digestive gland Digestive gland

Linear C15

C15H15O2Br

Panacene

Sesquiterpene

C15H26O (two isomers: brasilenol and epibrasilenol)

Brasilenol Epibrasilenol


Digestive gland

Sesquiterpene

Brasudol Isobrasudol


Digestive gland

Linear C15

C15H25BrO (two isomers: brasudol and isobrasudol) C15H19ClO (brasilenyne), C15H20BrClO (two isomers: cisdihydrorhodo phytin and cisisodihydrorho dophytin)

californica and vaccaria

Digestive gland

Cholinomimet ic substance

Chemical formula

Name given to Pharmacologic substance al effects

References

Cholinergic action on lantern muscles of the sea urchin, Echinometra lucunter, and muscles of the sea cucumber, Ludwigothuria grisea; decreased frequency and amplitude of beating in the toad heart (Bufo ictericus) and stopped the heart in diastole

De Freitas, 1977

Brasilenyne

Aplysin

When painted on the crab, Callinectes sapidus, the extract elicited avoidance by Octopus sp. Brasudol is a potent feeding deterrent to several fish Both substances unpalatable to swordtail fish (Xiphophorus helleri)

Neuromuscula r effects on injection into frogs, baby chicks and

229

Kinnel et al., 1977 Stallard et al., 1978

Dieter et al., 1979

Kinnel et al., 1979

Winkler, 1961; Winkler et al., 1962; Winkler & Tilton, 1962

230

THOMAS H.CAREFOOT

Species


Chemical type

Chemical formula

californica

Digestive gland

Urocanylcholi ne or murexine

californica

Digestive gland Digestive gland Digestive gland and skin

Monoterpene

C10H16OBr3Cl

Monoterpene

C10H12Br3Cl3

Sesquiterpene and monoterpene

californica

Digestive gland

Monoterpene and sesquiterpene

C15H21O3Br2C l (prepacifenol epoxide), C15H21Br2ClO 3, and C15H21OBr2Cl ; also C10H13Br2Cl3 and C12H15O2Cl3 C10H16OBr3Cl and C10H12Br3Cl3 (see also Faulkner et al., 1973); also C15H19OBr (aplysin) and C15H20O (debromoaply sin), and C15H19OBr2Cl (pacifidiene), and pacifenol, johnstonol, and laurinterol

dactylomela

Whole animal (?)

dactylomela

Whole animal

californica californica

Name given to Pharmacologic substance al effects mice: caused paralysis and death Depolarizing blockage of frog rectus and rat diaphragm muscle; block of myoneural junctions

C15H24O

Langlais & Blankenship, 1972; Blankenship et al., 1975

Faulkner & Stallard, 1973 Faulkner et al., 1973 Faulkner et al., 1974; Ireland et al., 1976

Prepacifenol epoxide

Stallard & Faulkner, 1974a

Amelioration of leukemia P-388 in mice Sesquiterpene

References

DactyloxeneB

Sigel et al., 1970 Schmitz & McDonald, 1974

APLYSIA

dactylomela

Whole animal or digestive gland

Linear C15

dactylomela

Whole animal (?)

Sesquiterpene

dactylomela

Whole animal

dactylomela

Whole animal and digestive gland

C15H19OBr2Cl (two isomers: dactylyne and isodactylyne)

231

Dactylyne Isodactylyne

Intraperitoneal injection of dactylyne into mice and rats produced CNS depresssion as shown by decrease in spontaneous and locomotory activities, and decreased body temperature; also inhibits the metabolic elimination of pentobarbiton e in mice

McDonald et al., 1975; Vandereh & Schmitz, 1976; Kaul et al., 1978a, b; Kaul & Kulkarni, 1978

C15H21BrO3

Aplysistatin

Antileukemial properties against P-388 lymphocytic mouse leukemia; (−)aplysistatin has been synthesized by Shieh & Prestwich (1982) and Tanaka et al. (1984), and (+)aplysistatin by Hoye et al. (1982) and White et al. (1982)

Pettit et al., 1977; Von Dreele & Kao, 1980

Diterpene

C20H30Br2O3

Angasiol

Sesquiterpene

C15H26O

Dactylol

Pettit et al., 1978 Schmitz et al., 1978a

232

THOMAS H.CAREFOOT

Species


Chemical type Chemical formula

Name given to substance

dactylomela

Whole animal and digestive gland

Sesquiterpene

Dactyloxene Dactylenol

Schmitz et al., 1978b

dactylomela

Digestive gland

Sesquiterpene

Deodactol Isodeodactol

dactylomela

Digestive gland

Diterpene

Hollenbeak et al., 1979; Gopichand et al., 1981 Schmitz et al., 1979

dactylomela

Digestive gland

Sesquiterpene

dactylomela

Digestive gland

Linear C15

dactylomela

Digestive gland Sterol

dactylomela

Digestive gland (animals from La Parguera, Puerto Rico)

Diterpene, indole, sesquiterpene

C15H24O (dactyloxene and dactylenol) C15H25O2Br2C l (two isomers: deodactol and isodeodactol) C20H35O2Br

Some cytotoxicity shown against the National Cancer Institute’s cell cultures P-388 lymphocytic leukemia and L-1210 lymphoid leukemia

C17H27O5Br2C l C15H20OBrCl and C15H20O2BrCl 5α, 8αepidioxy sterols C22H33BrO5 (two isomers: parguerol and isoparguerol) and C22H33BrO4 (deoxypargue rol); also parguerol 16acetate and isoparguerol 16-acetate; also Nmethylindole, elatol, allolaurintol acetate, and isoobtusol acetate

Pharmacologic al effects

References

Schmitz et al., 1980 Gopichand et al., 1981 Gunatilaka et al., 1981

Parguerol Deoxyparguer ol Isoparguerol

Schmitz et al., 1981, 1982

APLYSIA

dactylomela

Digestive gland

Diterpene

C20H34O3

dactylomela

Digestive gland

Sesquiterpene

C15H22Br2O (two isomers) C9H7Br2N and C9H6Br3N C20H30O2 (dictyol A), C20H32O2 (dictyol B), and C20H32O (pachydictyol A)

dactylomela and juliana depilans

Indole Digestive gland

Diterpene

depilans

Digestive gland

Diterpene

depilans

Whole animal (?) Digestive gland

Diterpene

fasciata

kurodai

Whole animal

kurodai

Whole animal

Possesses some antimicrobial activity

C20H34O2 (dictyol C) and C20H32O2 (two isomers: dictyol D and E) C20H30O2

Monoterpene

233

Gonzalez et al., 1983a

Gonzalez et al., 1983b Kato, 1984 Pachydictyol A has mild antibiotic activity vs. Staphylococc us aureus (Hirschfeld et al., 1973)

Minale & Riccio, 1976

Danise et al., 1977

Dictyolactone

Finer et al., 1979 Imperato et al., 1977

C10H11Cl15 and C10H11Cl3Br2 C16H21OBr

kurodai

Whole animal

Diterpene

C15H19OBr Aplysin (aplysin), Aplysinol C15H20O Debromoaply (debromoaply sin sin), and C15H19O2Br (aplysinol) C20H35O2Br Aplysin-20

kurodai

Whole animal (?)

Diterpene

C20H35O2Br

Tanaka & Toyama, 1959 Yamamura & Hirata, 1963

Sesquiterpene

Isoaplysin-20

Debromoisoap lysin-20 has been synthesized by

Matsuda et al., 1967; Yamamura & Hirata, 1971 Yamamura & Terada, 1977

234

THOMAS H.CAREFOOT

Species


Chemical type Chemical formula

Name given to substance

kurodai

Digestive gland

Monoterpene

C10H15OBr3Cl2

Kurodainol

oculifera

Digestive gland

Linear C15

oculifera

Whole animal

Linear C15

C15H20Br2O2 (two isomers) C15H20BrClO

Srilankenyne

parvula

Digestive gland (?)

Diterpene

pulmonica

Digestive gland

Ether-soluble toxin

Water-soluble toxin

vaccaria

Digestive gland

Diterpene

C20H30O4 (dihydroxy crenulide), C22H32O5, C20H30O3, and C22H32O4

Pharmacologica References l effects Imamura & Ruveda, 1980 Katayama et al., 1982 Schulte et al., 1981 Dilip de Silva et al., 1983 Fenical & Howard, unpubl. obs. (cited in Fenical et al., 1979) Intraperitoneal Watson, 1973; Watson & injection into Rayner, 1973 mice causes vasoconstrictio n and irritability Intraperitoneal injection into mice causes hypotension and bradycardia; convulsions, respiratory distress, death Midland et al., 1983

Aplysia seems to offer convincing support for the theory that the diverse array of brominated compounds in their digestive glands are derived from dietary seaweeds. The majority of these compounds can be divided into four categories consisting of the monoterpenes, the sesquiterpenes, the diterpenes, and the linear Cl5 compounds, based on their carbon skeletons (see Table XIII). This simple classification is useful as a guide to the algal origin of the substances. In general, the monoterpenes come from Plocamium spp., the sesquiterpenes from Laurencia spp., the diterpenes from Laurencia spp. and various brown algae such as the Dictyotaceae, and the linear Cl5 compounds exclusively from Laurencia spp. (Andersen, pers. comm.). Data are sparse on the possible function of these materials in the seaweeds from which they are derived (e.g. possible antibacterial and antiviral activity; see Fenical, 1975; Finer et al., 1979), and almost nothing is known of their potential function in Aplysia. It is interesting to note that several sea hares which subsist primarily on green algae, such as Ulva spp., have no dietary source of brominated terpenes. Yet, these sea

APLYSIA

235

hares, of which Aplysia juliana is probably the best example, appear to be as predator-free as any other species possessing the supposedly defensive brominated compounds. Do the opaline secretions take on the major rôle of deterring predators in such species, or do these animals rely on cholinomimetic or other substances as their first line of defence? There appear to be several unexplored lines of research in the area of predatorprey interactions of sea hares, which would richly repay investigation. INTERNAL DEFENCES Research into possible internal defensive mechanisms in sea hares has focused on whether haemagglutinins are present in the haemolymph (McKay, Jenkin & Rowley, 1969; Pauley, Granger & Krassner, 1971a; Pauley, 1971; see also Bevelaqua, Kim, Kumarasiri & Schwartz, 1975), their possible function as opsonins (Pauley, Krassner & Chapman, 1971b) and, in a related way, their possible participation in macrophagemediated cytolysis of tumours in mice (Yamazaki et al., 1983a, b). Haemagglutinins appear to be present in A. californica (McKay et al., 1969; Pauley et al., 1971a, b) where they may act as opsonins (factors which enhance phagocytosis). Pauley (1971) and Pauley et al. (1971b) tested in vivo clearance rates of several species of marine bacteria and red blood cells of pigs and chickens injected into A. californica, and noted that serum agglutinin titres decreased initially after exposure to the bacteria and red blood cells. Titres later returned to normal. The authors also noted that clearance of a bacterial species was accelerated by previous exposure of the sea hare to the bacterium. The agglutinin was, however, apparently non-specific, as treatment with one bacterial species completely prevented the serum agglutinating several other bacterial species or red blood cells (Pauley, 1971; Pauley et al., 1971a). Enhanced clearance activity was retained as long as one month after the primary injection, suggesting the presence of a memory response (Pauley et al., 1971b). The authors further noted that in vitro phagocytosis of chicken red blood cells was enhanced by opsonic factors in the haemolymph of the sea hare, possibly by the same molecule or group of molecules involved in agglutination (see also Pauley, 1974). Furthermore, while the haemolymph of A. californica was found to be sterile (Pauley et al., 1971b), no bactericidal activity was evident (Johnson & Chapman, 1970; Pauley et al., 1971b); hence, elimination of marine bacteria from the haemolymph was not apparently due to a lytic effect. Sea hare eggs may possess antibiotic factors, as the egg masses appear to be free of bacteria. Kamiya & Shimizu (1981) demonstrated potent agglutinins in extracts of A. kurodai egg masses which could agglutinate mammalian erythrocytes and marine bacteria. In a later study, Kamiya, Muramoto & Ogata (1984) discovered that eggs of A. kurodai do exhibit antibacterial activity and suggested that antibacterial factors were produced in the albumen gland, such that each egg was coated with antibacterially active albumen before passing down the oviduct. Although Kamiya et al. (1984) could not actually demonstrate haemagglutinin activity in the albumen gland of A. kurodai, Gilboa-Garber, Mizrahi & Susswein (1984) and Gilboa-Garber, Susswein, Mizrahi & Avichezer (1985) found that gonadal extracts from A. californica, A. depilans, A. fasciata, and A. oculifera contained considerable haemagglutinating activity. This activity was lectin-mediated. The authors suggested a number of possible functions for the lectins (a protein that specifically binds sugars), including binding of the fertilized eggs to the capsules, and protection. Finally, Yamazaki et al. (1984, 1985) identified a cytolytic factor in the eggs of A. kurodai which lyses tumour cells in vitro and inhibits tumour growth in vivo in mice. In general, this picture of the internal defences of Aplysia conforms with that of other gastropods (see review by Bayne, 1983).

236

THOMAS H.CAREFOOT

PARASITES A number of parasites and other animal associates of Aplysia have been reported. These are: (1) unidentified microorganisms in the abdominal ganglion of A. californica (Coggeshall, 1967), possibly the same as a new species of microsporidan found also in the abdominal ganglia of A. californica and named Microsporidium aplysiae (Krauhs, Long & Baur, 1979); (2) a metacercarian larva in the left pedal ganglion of Aplysia fasciata (Vicente, 1962); (3) colonies of the hydroids Obelia and Pedicellina attached to the mantle of young Aplysia punctata (Eales, 1921); (4) the copepods Strongylopleura pruvoti, in the mantle cavity of Aplysia sydneyensis and on the ctenidium of Australian specimens of A. juliana (Eales, 1960), and Anthessius variedens var. aplysiae associated with Aplysia brasiliana=willcoxi (Cipolli & Sawaya, 1967; Sawaya & Leahy, 1971); and finally, (5) an unidentified crab in the mantle cavity of A. californica (Lickey et al., 1977), possibly the same as, or similar to, the pinnotherid crab Opisthopus transversus observed in the mantle cavity of Aplysia vaccaria (Beondé, 1968). At present, there is insufficient information to determine whether the copepod and crab associates are parasites, or some sort of commensals, or indeed whether the other parasites listed are localized or widespread in their occurrence. ACKNOWLEDGEMENTS I thank Elizabeth Carefoot, Gerri Cheng, Kathy Gorkoff, David Huggard, Vita Janusas, John McNicol, and Mark Roberts for technical help in the preparation of the manuscript; Sarah Smith for editorial comments; Anthony Barrett, Horacio de la Cueva, Tom Mommsen, and Sam Gopaul for their assistance in translating papers; Mike Hawkes and Julie Oliveira for advice on seaweeds; and Raymond Andersen, Chris Bayne, Bob Blake, Peter Hochachka, Al Lewis, Sandra Millen, and Marilyn Switzer-Dunlap for helpful scientific comments on various parts of the manuscript. The work was supported in part by an Operating Grant from the Natural Scientific and Engineering Research Council of Canada. REFERENCES Achituv, Y. & Susswein, A.J., 1985. J. exp. mar. Biol. Ecol., 85, 113–122. Advokat, C., 1980. Behav. neur. Biol., 28, 253–265. Advokat, C., Carew, T. & Kandel, E., 1976. Neurosc. Abstr., 2, 313 only. Allan, J.K., 1932. Rec. Aust. Mus., 18, 314–320. Allan, J.K., 1941. Victoria Nat., 57, 178–182. Ambrose, III, H.W., Givens, R.P., Chen, R. & Ambrose, K.P., 1979. Mar. Behav. Physiol., 6, 57–64. Ando, Y., 1952. Kagaku, 22, 87–88. Ansell, A.D., 1982. J. exp. mar. Biol. Ecol., 61, 1–29. Arch, S. & Smock, T., 1977. Behav. Biol., 19, 45–54. Arch, S., Smock, T., Gurvis, R. & McCarthy, C. 1978. J. comp. Physiol., 128, 67–70. Arnould, C. & Jeuniaux, C., 1977. Cah. Biol. mar., 18, 465–473. Arvanitaki, A. & Chalazonitis, N., 1961. In, Nervous Inhibition, edited by E.Florey, A Pergamon Press Book, The Macmillan Co., New York, pp. 194–231. Aspey, W.P. & Blankenship, J.E., 1975. Fed. Proc., 34, 418 only. Aspey, W.P. & Blankenship, J.E., 1976a. Behav. Biol., 17, 279–299. Aspey, W.P. & Blankenship, J.E., 1976b. Behav. Biol., 17, 301–312. Aspey, W.P., Cobbs, J.S. & Blankenship, J.E., 1977. Behav. Biol., 19, 300–308. Audesirk, T., 1975a. Am. Soc. Zool., 15, 796. Audesirk, T.E., 1975b. Behav. Biol., 15, 45–55.

APLYSIA

237

Audesirk, T.E., 1976. Ph.D. thesis, University of Southern California, Los Angeles. Audesirk, T.E., 1977. Behav. Biol., 20, 235–243. Audesirk, T.E., 1979. Biol. Bull. mar. biol. Lab., Woods Hole, 157, 407–421. Audesirk, T.E. & Audesirk, G.J., 1977. Comp. Biochem. Physiol., 56A, 267–270. Augier, J. & Mastagli, P., 1956. C.r. hebd. Séanc. Acad. Sci, Paris, 242, 190–192. Austin, T., Weiss, S. & Lukowiak, K., 1983. Can. J. Physiol. Pharmacol, 61, 949– 953. Baba, K., 1936. J. Dep. Agric., Kyūshū Imper. Univ., 5, 1–50. Baba, K., 1937. Dobutsu. Zasshi, 49, 57–63. Baba, K., 1970. Collect. Breed., 32, 94–96. Baba, K., Hamatani, I. & Hisai, K., 1956. Publs Seto mar. biol. Lab., 5, 209–220. Bailey, C.H., Castellucci, V.F., Koester, J. & Chen, M., 1983. Behav. new. Biol., 38, 70–81. Balaparameswara Rao, M., 1976. J. moll. Stud., 42, 136–145. Bandel, K., 1976. Stud. Neotrop. Fauna Environ., 11, 87–118. Barash, A. & Danin, Z., 1971. Israel J. Zool., 20, 151–200. Barash, A. & Zenziper, Z., 1980. Veliger, 22, 299–317. Bardach, J.E., 1975. In, Olfaction and Taste V, edited by D.A.Denton & J.P. Coghlan, Proc. Fifth int. Symp., Univ. Melbourne, Aust., Academic Press, New York, pp. 121–132. Barnes, D.M., 1986. Science, 231, 1246–1249. Bartsch, P. & Rehder, H.A., 1939. Smithson. misc. Collns, Vol. 98, 1–18. Bayne, B.L. & Scullard, C., 1978. J. exp. mar. Biol. Ecol., 32, 113–129. Bayne, C.J., 1983. In, The Mollusca. Volume 5. Physiology, Part 2, edited by A.S.M.Saleuddin & K.M.Wilbur, Academic Press, New York, pp. 407–486. Bebbington, A., 1972. Pubbl. Staz. zool. Napoli, 38, 25–46. Bebbington, A., 1974. Zool. J. Linn. Soc., 54, 63–99. Bebbington, A., 1975. Pubbl. Staz. zool. Napoli, 39, 121–128. Bebbington, A., 1977. Trans. zool. Soc. Lond., 34, 87–147. Bebbington, A. & Brown, G.H., 1975. J. Conchol, 28, 329–333. Bebbington, A. & Hughes, G.M., 1973. Proc. malac. Soc. Lond., 40, 399–405. Bebbington, A. & Thompson, T.E., 1969. Malacologia, 9, 253 only. Beeman, R.D., 1960. Bull. Sth. Calif. Acad. Sci., 59, 144–152. Beeman, R.D., 1961. Veliger, 3(suppl.), 87–102. Beeman, R.D., 1963. Veliger, 5, 145–147. Beeman, R.D., 1969. Biol. Bull. mar. biol. Lab., Woods Hole, 136, 141–146. Beondé, A.C., 1968. Veliger, 10, 375–378. Bethe, A., 1929. Pflügers Arch. ges. Physiol., 221, 344–362. Bethe, A., 1930. J. gen. Physiol. 13, 437–444. Bethe, A., 1934. Pflügers Arch. ges. Physiol., 234, 629–644. Bevelaqua, F.A., Kim, K.S., Kumarasiri, M.H. & Schwartz, J.H., 1975. J. biol. Chem., 250, 731–738. Bhakuni, D.S. & Silva, M., 1974. Botanica mar., 17, 40–51. Blankenship, J.E., 1980. In, The Role of Peptides in Neural Function, edited by J.L. Barker & T.G.Smith Jr, Marcel Dekker, Inc., New York, pp. 159–187. Blankenship, J.E., Langlais, P.J. & Kittredge, J.S., 1975. Comp. Biochem. Physiol., 51C, 129–137. Blankenship, J.E., Rock, M.K., Robbins, L.C., Livingston, C.A. & Lehman, H.K., 1983. Fed. Proc., 42, 96–100. Blankenship, J.E., Rock, M.K. & Schlesinger, D.H., 1982. Prog. clin. biol. Res., 79, 159–177. Block, G. & Smith, J.T., 1973. Comp. Biochem. Physiol. 46A, 115–121. Block, G.D., 1971. Physiol., 14, 112 only. Block, G.D., Hudson, D.J. & Lickey, M.E., 1974. J. comp. Physiol., 89, 237–249. Block, G.D. & Roberts, M.H., 1981. J. comp. Physiol., 142, 403–410. Bostock, J. & Riley, H.T., 1856. The Natural History of Pliny. Vol. V. Henry G. Bohn, London, 523 pp.

238

THOMAS H.CAREFOOT

Bostock, J. & Riley, H.T., 1857. The Natural History of Pliny. Vol. VI. Henry G. Bohn, London, 529 pp. Bottazzi, F., 1897. Arch. Ital. Biol., 28, 81–90. Braams, W.G. & Geelen, H.F.M., 1953. Arch. Néerl. Zool., 10, 241–264. Branch, G.M., 1981. Oceanogr. Mar. Biol. Ann. Rev., 19, 235–380. Breuer, J.P., 1962. Publ. Inst. mar. Sci. Univ. Texas, 8, 153–183. Bridges, C.B., 1975. Ophelia, 14, 161–184. Brown, A.M. & Brown, H.M., 1973. J. gen. Physiol., 62, 239–254. Bruce, J.R., Colman, J.S. & Jones, N.S., 1963. Marine Fauna of the Isle of Man and Its Surrounding Seas. Mem. No. 36, Liverpool University Press, Liverpool, 307 pp. Burky, A.J., 1971. Ecol. Monogr., 41, 235–251. Byrne, J.H., 1981. Brain Res., 204, 200–203. Capo, T.R., Perritt, S.E. & Berg, Jr, C.J., 1979. Biol. Bull. mar. biol. Lab., Woods Hole, 157, 360 only. Carazzi, D., 1900. Anat. Anz., 17, 77–102. Carefoot, T.H., 1967a. J. mar. biol. Ass. U.K., 47, 565–590. Carefoot, T.H., 1967b. Comp. Biochem. Physiol. 21, 627–652. Carefoot, T.H., 1967c. J. mar. biol. Ass. U.K., 47, 335–350. Carefoot, T.H., 1970. J. exp. mar. Biol. Ecol., 5, 47–62. Carefoot, T.H., 1979. Can. J. Zool., 57, 2271–2273. Carefoot, T.H., 1980. J. exp. mar. Biol. Ecol., 42, 241–252. Carefoot, T.H., 1981a. In, Proc. Fourth int. Coral Reef Symp. Vol. 2, pp. 665–669. Carefoot, T.H., 1981b. Can. J. Zool., 59, 445–454. Carefoot, T.H., 1982a. Proc. 2nd int. Conf. Aquacult. Nutr: Biochem. Physiol. Approaches Shellfish Nutr., pp. 321–337. Carefoot, T.H., 1982b. Mar. Biol., 68, 207–215. Carefoot, T.H., 1985. In, Proc. Fifth int. Coral Reef Congress, Tahiti, Vol. 4, pp. 9– 16. Carefoot, T.H., in press. In, Animal Energetics, edited by T.J.Pandian & F.J. Vernberg, Academic Press, New York (in press). Carew, T.J., Castellucci, V.F. & Kandel, E.R., 1971. Int. J. Neurosc., 2, 79–98. Carew, T.J., Hawkins, R.D. & Kandel, E.R., 1983. Science, 219, 397–400. Carew, T.J. & Kandel, E.R., 1977a. J. Neurophysiol., 40, 692–707. Carew, T.J. & Kandel, E.R., 1977b. J. Neurophysiol., 40, 708–720. Carew, T.J. & Kandel, E.R., 1977c. J. Neurophysiol., 40, 721–734. Carew, T.J. & Kupfermann, I., 1974. Behav. Biol., 12, 339–345. Carew, T.J., Pinsker, H.M. & Kandel, E.R., 1972. Science, 175, 451–454. Carew, T.J., Walters, E.T. & Kandel, E.R., 1981. Science, 211, 501–504. Carlton, J.T., 1985. Oceanogr. mar. Biol. Ann. Rev., 23, 313–371. Castellucci, V., Pinsker, H., Kupfermann, I. & Kandel, E.R., 1970. Science, 167, 1745–1748. Chapman, A.R.O., 1981. Mar. Biol., 62, 307–311. Chapman, D.J., Cole, W.J. & Siegelman, H.W., 1967. J. Am. chem. Soc., 89, 5976– 5977. Chapman, D.J. & Fox, D.L., 1969. J. exp. mar. Biol. Ecol., 4, 71–78. Chase, R., 1979a. Can. J. Zool., 57, 781–784. Chase, R., 1979b. Can. J. Zool., 57, 698–701. Cho, D.M., Pyeun, J.H., Byun, D.S. & Kim, C.Y., 1983. Bull. Korean fish. Soc., 16, 216–224. Christomanos, A., 1955. Nature, Lond., 175, 310 only. Cipolli, I.N. & Sawaya, P., 1967. Cienc. Cult. (São Paulo), 19, 432 only. Clark, K.B., 1984. Nautilus, 98, 85–97. Cobbs, J.S. & Pinsker, H.M., 1982. J. comp. Physiol. 147, 523–535. Coggeshall, R.E., 1967. J. Neurophysiol., 30, 1263–1287. Coggeshall, R.E., 1969. J. Morph., 127, 113–131. Coleman, N., 1975. Sea Front., 21, 344–346.

APLYSIA

239

Cook, D.G. & Carew, T.J. 1986. Proc. natn. Acad. Sci. U.S.A., 83, 1120–1124. Cook, E.F., 1962. Veliger, 4, 194–196. Cooper, J.G., 1863. Proc. Calif. Acad. Nat. Sci., 3, 56–60. Corrent, G., Eskin, A. & Kay, I., 1982. Am. J. Physiol., 242, R326–332. Craigie, J.S. & Gruenig, D.E., 1967. Science, 157, 1058–1059. Czeczuga, B., 1984. Comp. Biochem. Physiol., 78B, 259–264. Danise, B., Minale, L., Riccio, R., Amico, V., Oriente, G., Piattelli, M., Tringali, C., Fattorusso, E., Magno, S. & Mayol, L., 1977. Experientia, 33, 413–415. Darling, S.D. & Cosgrove, R.E., 1966. Veliger, 8, 178–180. De Freitas, J.C., 1977. Comp. Biochem. Physiol, 56C, 57–61. De-Negri, A. & De-Negri, G., 1876. Rc. Accad. Lincei, Atti (Roma), Ser. 2, 3, 394– 449. Denny, M., 1980. Science, 208, 1288–1290. Dieringer, N., Koester, J. & Weiss, K.R., 1978. J. comp. Physiol., 123, 11–21. Dieter, R.K., Kinnel, R., Meinwald, J. & Eisner, T., 1979. Tetrah. Lett., 19, 1645– 1648. Dijkgraaf, S. & Hessels, H.G.A., 1969. Z. vergl. Physiol., 62, 38–60. Dilip de Silva, E., Schwartz, R.E. & Scheuer, P.J., 1983. J. org. Chem., 48, 395–396. DiMatteo, T., 1981a. Mar. Behav. Physiol., 7, 285–290. DiMatteo, T., 1981b. Veliger, 24, 72–75. DiMatteo, T., 1982. J. exp. mar. Biol. Ecol., 57, 169–180. Downey, P. & Jahan-Parwar, B., 1972., Physiol., 15, 122 only. Dudek, F.E., Injeyan, H.S., Soutar, B., Weir, G. & Tobe, S.S., 1980a. Can. J. Zool., 58, 2220–2229. Dudek, F.E., Soutar, B. & Tobe, S.S., 1980b. Can. J. Zool., 58, 2163–2166. Eales, N.B., 1921. Liverpool Marine Biology Committee Memoirs, No. 24, edited by W.A.Herdman & J.Johnstone, Liverpool University Press, Liverpool, 84 pp. Eales, N.B., 1944. Proc. malac. Soc. Land., 26, 1–21. Eales, N.B., 1957. Bull. Mus., Sér. 2, 29, 246–255. Eales, N.B., 1960. Bull. British Mus. (Nat. Hist.), Zoology, 5, 269–404. Eales, N.B., 1970. Proc. malac. Soc. Land., 39, 217–220. Edmunds, M., Potts, G.W., Swinfen, R.C. & Waters, V.L., 1974. J. mar. biol. Ass. U.K., 54, 939–947. Edmundson, C.H., 1946. Reef and Shore Fauna of Hawaii. Bernice P.Bishop Museum Spec. Publ. 22, Honolulu, Hawaii, 381 pp. Edwards, D.C. & Huebner, J.D., 1977. Ecology, 58, 1218–1236. Emery, D.G., 1976. Am. Zool., 16, 241 only. Emery, D.G. & Audesirk, T.E., 1978. Neurobiol., 9, 173–179. Emlen, J.M., 1966. Am. Nat., 100, 611–617. Engel, H., 1928. Proc. malac. Soc. Land., 18, 147–151. Engel, H. & Eales, N.B., 1957. Beaufortia, 6, 83–114. Erickson, K.L., 1983. In, Marine Natural Products, Vol. 5, edited by P.J.Scheuer, Academic Press, New York, pp. 131–257. Eskin, A., 1971. Z. vergl. Physiol., 74, 353–371. Eskin, A., 1979. Fed. Proc., 38, 2573–2579. Farmer, W.M., 1970. Veliger, 13, 73–89. Fattorusso, E., Magno, S., Mayol, L., Santacroce, C., Sica, D., Amico, V., Oriente, G., Piattelli, M. & Tringali, C., 1976. J. Chem. Soc., No. 13, 575–576. Faulkner, D.J. & Stallard, M.O., 1973. Tetrah. Lett., 14, 1171–1174. Faulkner, D.J., Stallard, M.O., Fayos, J. & Clardy, J., 1973. J. Am. chem. Soc., 95, 3413–3414. Faulkner, D.J., Stallard, M.O. & Ireland, C., 1974. Tetrah. Lett., 40, 3571–3574. Feinstein, R., Aspey, W.P. & Schmale, M.C., 1976. Behav. Biol., 17, 271–278. Feinstein, R., Pinsker, H., Schmale, M. & Gooden, B.A., 1977. J. comp. Physiol., 122, 311–324.

240

THOMAS H.CAREFOOT

Fenical, W., 1975. J. Phycol., 11, 245–259. Fenical, W., Sleeper, H.L., Paul, V.J., Stallard, M.O. & Sun, H.H., 1979. Pure appl. Chem., 51, 1865–1874. Ferguson, G.P., Parsons, D.W., ter Maat, A. & Pinsker, H.M., 1984. Neurosc. Abstr., 10, 150 only. Finer, J., Clardy, J., Fenical, W., Minale, L., Riccio, R., Battaile, J., Kirkup, M. & Moore, R.E., 1979. J. org. Chem., 44, 2044–2047. Flury, F., 1915. Arch. exp. Pathol. Pharmakol., 79, 250–263. Fong, W. & Mann, K.H., 1980. Can. J. Fish, aquatic Sci., 37, 88–96. Fontaine, M. & Raffy, A., 1936a. C. r. Séanc. Soc. Biol., 121, 735–736. Fontaine, M. & Raffy, A., 1936b. Bull. Soc. zool. Fr., 61, 49–57. Foreman, R.E., 1977. Helgol. wiss. Meeresunters., 30, 468–484. Fox, R.S. & Ruppert, E.E., 1985. Shallow-water Marine Benthic Macroinvertebrates of South Carolina. Species Identification, Community Composition and Symbiotic Associations. Belle W.Baruch Institute for Marine Biology and Coastal Research. University South Carolina Press, Columbia, S.C., 329 pp. Frings, H. & Frings, C., 1965. Biol. Bull. mar. biol. Lab., Woods Hole, 128, 211–217. Fujiki, H., Ikegami, K., Hakii, H., Suganuma, M., Yamaizumi, Z., Yamazato, K., Moore, R.E. & Sugimura, T., 1985. Jpn J. Cancer Res. (Gann), 76, 257–259. Fujiki, H., Suganuma, M., Nakayasu, M., Hoshino, H., Moore, R.E. & Sugimura, T., 1982. Gann, 73, 495–497. Fujiki, H., Suganuma, M., Tahira, T., Yoshioka, A., Nakayasu, M., Endo, Y., Shudo, K., Takayama, S., Moore, R.E. & Sugimura, T., 1984. In, Cellular Interactions by Environmental Tumor Promoters, edited by H.Fujiki et al., Japan Sci. Soc. Press, Tokyo/VNU Science Press, Utrecht, pp. 37–45. Gallager, S.M. & Mann, R., 1980. Mar. Biol. Lett., 1, 341–349. Gallin, E.K. & Wiederhold, M.L., 1977. J. Physiol., 266, 123–137. Garstang, W., 1890. J. mar. biol. Ass. U.K., 1, 399–457. Gerencser, G.A., 1978. Comp. Biochem. Physiol., 61A, 203–208. Gerencser, G.A., 1979a. Comp. Biochem. Physiol., 63A, 519–522. Gerencser, G.A., 1979b. Comp. Biochem. Physiol., 64A, 251–255. Gerencser, G.A., 1981a. Am. J. Physiol., 240, R61-R69. Gerencser, G.A., 1981b. Comp. Biochem. Physiol., 68A, 225–230. Gerencser, G.A., 1982. Comp. Biochem. Physiol., 72A, 721–725. Gerencser, G.A., 1983. Am. J. Physiol., 244, R412-R416. Gerencser, G.A., 1984a. Comp. Biochem. Physiol., 78A, 607–611. Gerencser, G.A., 1984b. Biochim. Biophys. Acta, 775, 389–394. Gerencser, G.A. & Hong, S.K., 1977. Comp. Biochem. Physiol., 58A, 275–280. Gerencser, G.A. & Lee, S.-H., 1985. Biochim. Biophys. Acta, 816, 415–417. Gerencser, G.A. & Loughlin, G.M., 1983. Comp. Biochem. Physiol., 74A, 701–704. Gerencser, G.A. & White, J.F., 1980. Am. J. Physiol., 239, R445-R449. Gev, S., Achituv, Y. & Susswein, A.J., 1984. J. exp. mar. Biol. Ecol., 74, 67–83. Ghiretti, F., Ghiretti-Magaldi, A. & Tosi, L., 1959. J. gen. Physiol., 42, 1185–1205. Gilboa-Garber, N., Mizrahi, L. & Susswein, A.J., 1984. Mar. Biol. Lett., 5, 105– 114. Gilboa-Garber, N., Susswein, A.J., Mizrahi, L. & Avichezer, D., 1985. FEBS Lett., 181, 267–270. González, A.G., Martin, J.D., Norte, M., Pérez, R. & Weyler, V., 1983a. Tetrah. Lett., 24, 1075–1076. González, A.G., Martin, J.D., Norte, M., Pérez, R. & Weyler, V., 1983b. Tetrah. Lett., 24, 847–848. Gopichand, Y., Schmitz, F.J. & Shelly, J., 1981. J. org. Chem., 46, 5192–5197. Grahame, J., 1973. J. anim. Ecol., 42, 391–403. Grigg, U.M., 1949. J. mar. biol. Ass. U.K., 28, 795–805. Gunatilaka, A.A.L., Gopichand, Y., Schmitz, F.J. & Djerassi, C., 1981. J. org. Chem., 46, 3860–3866. Hackney, A.G., 1944. Nautilus, 58, 56–64. Hadfield, M.G., 1975. Lab. Anim., 4, 17 only.

APLYSIA

241

Hadfield, M.G. & Switzer-Dunlap, M., 1984. In, The Mollusca, Vol. 7, edited by K.M.Wilbur, Academic Press, New York, pp. 209–350. Haefelfinger, H.R. & Kress, A., 1967. Rev. Suisse Zool., 74, 547–554. Halstead, B.W., 1965. Poisonous and Venomous Marine Animals of the World. U.S. Government Printing Office, Washington, D.C., 994 pp. Hamilton, P.V., 1979. Am. Zool., 19, 992 only. Hamilton, P.V., 1984. Anim. Behav., 32, 367–373. Hamilton, P.V., 1985. In, Migration: Mechanisms and Adaptive Significance, edited by M.A.Rankin, Suppl. to Contr. Mar. Sci., Vol. 27, pp. 212–226. Hamilton, P.V., 1986. Veliger, 28, 310–313. Hamilton, P.V. & Ambrose, III, H.W., 1975. Mar. Behav. Physiol., 3, 131–144. Hamilton, P.V. & Russell, B.J., 1982a. J. exp. mar. Biol. Ecol., 56, 145–152. Hamilton, P.V. & Russell, B.J., 1982b. J. exp. mar. Biol. Ecol., 56, 123–143. Hamilton, P.V., Russell, B.J. & Ambrose, H.W., 1982. Malacol. Rev., 15, 15–19. Hardy, Sir A., 1959. The Open Sea: Its Natural History. Part II. Fish and Fisheries. Collins, London, 322 pp. Heck, Jr, K.L. & Weinstein, M.P., 1978. Biotropica, 10, 78–79. Hening, W.A., Walters, E.T., Carew, T.J. & Kandel, E.R., 1979. Brain Res., 179, 231–253. Higgs, M.D. & Faulkner, D.J., 1982. Phytochemistry, 21, 789–791. Hirsch, H.R. & Peretz, B., 1984. Mech. Ageing Dev., 27, 43–62. Hirschfeld, D.R., Fenical, W., Lin, G.H.Y., Wing, R.M., Radlick, P. & Sims, J.J., 1973. J. Am. chem. Soc., 95, 4049–4050. Hollenbeak, K.H., Schmitz, F.J., Hossain, M.B. & van der Helm, D., 1979. Tetrahedron, 35, 541–545. Howells, H.H., 1942. Q. Jl microsc. Sci., 83, 357–397. Hoye, T.R., Caruso, A.J., Dellaria, Jr, J.F. & Kurth, M.J., 1982. J. Am. chem. Soc., 104, 6704–6709. Hsü, K.J., 1972. Scient. Am., 227, 27–36. Huebner, J.D. & Edwards, D.C., 1981. Mar. Biol., 61, 221–226. Hughes, G.M. & Tauc, L., 1962. J. exp. Biol., 39, 45–69. Hughes, R.N., 1971a. J. exp. mar. Biol. Ecol., 6, 167–178. Hughes, R.N., 1971b. Mar. Biol., 11, 12–22. Hughes, R.N., 1972. Oecologia (Berl.), 8, 356–370. Hughes, R.N. & Roberts, D.J., 1980. Oecologia (Berl.), 47, 130–136. Hunt, A.R., 1878. Devonsh. Ass. Adv. Sci. Lit. Art, Rep. Trans., 10, 611–617. Hyman, L.H., 1967. The Invertebrates. Vol. VI, Mollusca I. McGraw-Hill Book Company, New York, 792 pp. Imamura, P.M. & Ruveda, E.A., 1980. J. org. Chem., 45, 510–515. Imperato, F., Minale, L. & Riccio, R., 1977. Specialia, 33, 1273–1274. Ireland, C., Stallard, M.O., Faulkner, D.J., Finer, J. & Clardy, J., 1976. J org. Chem., 41, 2461–2465. Irie, T., Suzuki, M. & Hayakawa, Y., 1969. Bull. chem. Soc. Jpn, 42, 843–844. Irie, T., Suzuki, M., Kurosawa, E. & Masamune, T., 1966. Tetrah. Lett., 17, 1837– 1840. Irie, T., Suzuki, M. & Masamune, T., 1965a. Tetrah. Lett., 16, 1091–1099. Irie, T., Yasunari, Y., Suzuki, T., Imai, N., Kurosawa, E. & Masamune, T., 1965b. Tetrah. Lett., 40, 3619–3624. Jacklet, J.W., 1969. Science, 164, 562–563. Jacklet, J.W., 1972. J. comp. Physiol., 79, 325–341. Jacklet, J.W., 1974. J. comp. Physiol., 90, 33–45. Jacklet, J.W., 1976. In, Biological Rhythms in the Marine Environment, edited by P.J.DeCoursey, The Belle W.Baruch Library in Marine Science No. 4, University S.Carolina Press, Columbia, S.C., pp. 17–31. Jacklet, J.W., 1980. J. comp. Physiol., 136, 257–262. Jacklet, J.W., Alvarez, R. & Bernstein, B., 1972. J. Ultrastr. Res. 38, 246–261. Jahan-Parvar, B., 1969. Am. Zool., 9, 1097 only. Jahan-Parvar, B., 1970. Am. Zool., 10, 287 only.

242

THOMAS H.CAREFOOT

Jahan-Parwar, B., 1972a. Am. Zool., 12, 525–537. Jahan-Parwar, B., 1972b. Physiol., 15, 180 only. Jahan-Parwar, B., 1975. In, Olfaction and Taste V, edited by D.A.Denton & J.P. Coghlan, Proc. Fifth int. Symp., Univ. Melbourne, Aust., Academic Press, New York, pp. 133–140. Jahan-Parwar, B., 1976. Physiol., 19, 240 only. Jahan-Parwar, B. & Fredman, S.M., 1978b. J. Neurophysiol., 41, 600–608. Jahan-Parwar, B. & Fredman, S.M., 1978c. Comp. Biochem. Physiol., 60A, 459–465. Jahan-Parwar, B. & Fredman, S.M., 1978a. J. Neurophysiol., 41, 609–620. Jahan-Parwar, B. & Fredman, S.M., 1979a. Behav. neur. Biol., 27, 39–58. Jahan-Parwar, B. & Fredman, S.M., 1979b. Brain Res. Bull., 4, 407–420. Jahan-Parwar, B. & Fredman, S.M., 1980. Brain Res. Bull., 5, 169–177. Jahan-Parwar, B. & Miller, D.L., 1978. Biol. Bull. mar. biol. Lab., Woods Hole, 155, 447 only. Jahan-Parwar, B., Smith, M. & von Baumgarten, R., 1969. Am. J. Physiol., 216, 1246–1257. Jahan-Parwar, B., Wells, L.J. & Fredman, S.M., 1974. Proc. int. Union physiol. Sci., 26th int. Congr. physiol. Sci., X, New Delhi, p. 230 only. Janse, C., 1983. J. comp. Physiol., 150, 359–370. Johnson, P.T. & Chapman, F.A., 1970. J. Invert. Path., 16, 259–267. Jordan, H., 1901. Z. Biol., 41, 196–238. Jordan, H., 1917. Biol. Zbl., 37, 2–9. Jordan, W.P., Lickey, M.E. & Hiaasen, S.O., 1985. J. comp. Physiol. A, 156, 293– 303. Kamiya, H., Muramoto, K. & Ogata, K., 1984. Experientia, 40, 947–949. Kamiya, H. & Shimizu, Y., 1981. Bull. Jap. Soc. scient. Fish., 47, 255–259. Kandel, E.R., 1976. Cellular Basis of Behavior. An Introduction to Behavioral Neurobiology. W.H.Freeman & Co., San Francisco, 727 pp. Kandel, E.R., 1979. Behavioral Biology of Aplysia. A Contribution to the Comparative Study of Opisthobranch Molluscs. W.H.Freeman & Co., San Francisco, 463 pp. Kandel, P. & Capo, T.R., 1979. Veliger, 22, 194–198. Katayama, A., Ina, K., Nozaki, H. & Nakayama, M., 1982. Agric. biol. Chem., 46, 859–860. Kato, Y., 1984. Ph. D. thesis, University of the Ryūkyūs, Japan. Kato, Y. & Scheuer, P.J., 1974. J. Am. chem. Soc., 96, 2245–2246. Kaul, P.N. & Kulkarni, S.K., 1978. J. pharm. Sci., 67, 1293–1296. Kaul, P.N., Kulkarni, S.K. & Kurosawa, E., 1978a. J. Pharm. Pharmac., 30, 589– 590. Kaul, P.N., Kulkarni, S.K., Schmitz, F.J. & Hollenbeak, K.H., 1978b. In, Drugs and Food from the Sea. Myth or Reality?, edited by P.N.Kaul & C.J. Sindermann, University of Oklahoma, Norman, Oklahoma, pp. 99–106. Kay, E.A., 1964. Proc. malac. Soc. Land., 36, 173–190. Kay, E.A., 1979. Hawaiian Marine Shells. Reef and Shore Fauna of Hawaii. Section 4: Mollusca. Bernice P. Bishop Museum Spec. Publ. 64(4) , Bishop Museum Press, Honolulu, Hawaii, 652 pp. Kazlauskas, R., Murphy, P.T., Quinn, R.J. & Wells, R.J., 1976. Aust. J. Chem., 29, 2533–2539. Kempf, S.C., 1981. Mar. Ecol. Prog. Ser., 6, 61–65. Kennedy, G.Y. & Vevers, H.G., 1954. J. mar. biol. Ass. U.K., 33, 663–676. Kinnel, R., Duggan, A.J., Eisner, T. & Meinwald, J., 1977. Tetrah. Lett., 44, 3913– 3916. Kinnel, R.B., Dieter, R.K., Meinwald, J., Van Engen, D., Clardy, J., Eisner, T., Stallard, M.O. & Fenical, W., 1979. Proc. natn Acad. Sci. U.S.A., 76, 3576– 3579. Kittredge, J.S., Takahashi, F.T., Lindsey, J. & Lasker, R., 1974. Fish. Bull. NOAA, 72, 1–11. Koch, U.T. & Koester, J., 1982. J. comp. Physiol., 149, 31–42. Koch, U.T., Koester, J. & Weiss, K.R., 1984. J. Neurophysiol, 51, 126–135. Kofoed, L.H., 1975. J. exp. Mar. Biol. Ecol., 19, 243–256. Kohn, A.J., 1961. Am. Zool., 1, 291–308. Koningsor, Jr, R.L. & Hunsaker, II, D., 1971. Veliger, 13, 285–289.

APLYSIA

Koningsor, Jr, R.L., McLean, N. & Hunsaker, II, D., 1972. Comp. Biochem. Physiol., 43B, 237–240. Krakauer, J.M., 1969. M. Sc. thesis, University of Florida, Gainesville, Florida, 89 pp. Krakauer, J.M., 1971. Nautilus, 85, 37–38. Krakauer, J.M., 1974. Veliger, 16, 396–398. Krauhs, J.M., Long, J.L. & Baur, Jr, P.S., 1979. J. Protozool., 26, 43–46. Kriegstein, A.R., 1977a. Proc. natn Acad Sci. U.S.A., 74, 375–378. Kriegstein, A.R., 1977b. J. exp. Zool., 199, 275–288. Kriegstein, A.R., Castellucci, V. & Kandel, E.R., 1974. Proc. natn Acad. Sci. U.S.A., 71, 3654–3658. Kupfermann, I., 1965. Physiol., 8, 214 only. Kupfermann, I., 1967. Nature, Lond., 216, 814–815. Kupfermann, I., 1968. Physiol. Behav., 3, 179–181. Kupfermann, I., 1970. J. Neurophysiol., 33, 877–881. Kupfermann, I., 1972. Am. Zool., 12, 513–519. Kupfermann, I., 1974a. Behav. Biol., 10, 89–97. Kupfermann, I., 1974b. Behav. Biol., 10, 1–26. Kupfermann, I. & Carew, T.J., 1974. Behav. Biol., 12, 317–337. Kupfermann, I., Castellucci, V., Pinsker, H. & Kandel, E., 1970. Science, 167, 1743– 1745. Kupfermann, I. & Kandel, E.R., 1969. Science, 164, 847–850. Kupfermann, I. & Pinsker, H., 1968. Commun, behav. Biol., Part A, 2, 13–17. Kupfermann, I. & Weiss, K.R., 1976. J. gen. Physiol., 67, 113–123. Kupfermann, I. & Weiss, K.R., 1981. Behav. new. Biol., 32, 126–132. Kuslansky, B., Weiss, K.R. & Kupfermann, I., 1978. Behav. Biol., 23, 230–237. Lance, J.R., 1967. Veliger, 9, 412 only. Lance, J.R., 1971. Veliger, 14, 60–63. Langlais, P.J. & Blankenship, J.E., 1972. Fed. Proc., 31, 822 only. Larkin, S. & McFarland, D., 1978. Anim. Behav., 26, 1237–1246. Laverack, M.S., 1968. Oceanogr. Mar. Biol. Ann. Rev., 6, 249–324. Lawrence, J.M., 1975. Oceanogr. Mar. Biol. Ann. Rev., 13, 213–286. Lederer, E. & Huttrer, C., 1942. Trav. Membr. Soc. Chim. Biol., 24, 1055–1061. Lederhendler, I., Bell, L. & Tobach, E., 1975. Veliger, 17, 347–353. Lederhendler, I.I., 1977. Veliger, 20, 173–178. Lederhendler, I.I., Herriges, K. & Tobach, E., 1977. Anim. Learn. Behav., 5, 355– 358. Lederhendler, I.I. & Tobach, E., 1977. Nature, Lond., 270, 238–239. Leighton, D.L., 1966. Pacif. Sci., 20, 104–113. Lickey, M.E., 1968. J. comp. physiol. Psychol., 66, 712–718. Lickey, M.E., 1969. J. comp. physiol. Psychol., 68, 9–17. Lickey, M.E. & Berry, R.W., 1966. Physiol., 9, 230 only. Lickey, M.E., Hudson, D.J. & Hiaasen, S.O., 1983. J. comp. Physiol., 153, 133– 143. Lickey, M.E. & Wozniak, J.A., 1979. J. comp. Physiol., 131, 169–177. Lickey, M.E., Wozniak, J.A., Block, G.D., Hudson, D.J. & Augter, G.K., 1977. J. comp. Physiol., 118, 121–143. Ligman, S.H. & Brownell, P.H., 1985. J. comp. Physiol. A, 157, 31–37. Littler, M.M., Littler, D.S., Blair, S.M. & Norris, J.N., 1985. Science, 277, 57–59. Lo Bianco, S., 1888. Mitt. zool. Stn Neapel, 8, 385–440. Lo Bianco, S., 1909. Mitt. zool. Stn Neapel, 19, 513–761. Lombardini, J.B., Pang, K.T. & Griffith, R.W., 1979. Calif. Acad. Sci., Occ. Papers, 134, 160–169. Longley, A.J. & Longley, R.D., 1984. Can. J. Zool., 62, 8–14. Lu, C., Strumwasser, F. & Gilliam, J.J., 1966. Calif. Inst. Tech. Biol. Div. Ann. Rep., p. 153 only. Lukowiak, K., 1977. Brain Res., 132, 553–557. Lukowiak, K., 1979. Can. J. Physiol. Pharmacol., 57, 987–997.

243

244

THOMAS H.CAREFOOT

Lukowiak, K., 1980. J. Neurobiol., 11, 591–611. Lukowiak, K. & Freedman, L., 1983. Can. J. Physiol. Pharmacol, 61, 743–748. Lukowiak, K. & Jacklet, J.W., 1972a. Science, 178, 1306–1308. Lukowiak, K. & Jacklet, J.W., 1972b. Fed. Proc., 31, 405 only. Lukowiak, K. & Peretz, B., 1977. J. comp. Physiol., 117, 219–244. Lukowiak, K. & Peretz, B., 1980. J. Neurobiol., 11, 425–433. Lukowiak, K. & Sahley, C., 1981. Science, 212, 1516–1518. Macé, A.-M., 1981. Téthys, 10, 117–120. MacFarland, F.M., 1966. Mem. Calif. Acad. Sci., No. 6, 546 pp. MacGinitie, G.E., 1934. Biol. Bull. mar. biol. Lab., Woods Hole, 67, 300–303. MacGinitie, G.E., 1935. Am. Midl. Nat., 16, 629–765. MacGinitie, G.E. & MacGinitie, N., 1968. Natural History of Marine Animals. McGraw-Hill Book Company, New York, 523 pp. MacMunn, C.A., 1899. J. Physiol., 24, 1–10. Macnae, W., 1955. Ann. Natal Mus., 13, 223–241. Macnae, W., 1957. S. Afr. J. Sci., 54, 289–292. Magnus, O., 1555. Historia de Gentibus Septentrionalibus. Rosenkilde and Bagger, Copenhagen, 1972, 815 pp. Marcus, E., 1958. Bolm Inst. oceanogr. São Paulo, 7, 31–78. Marcus, E., 1961. Veliger, 3 (suppl.), 1–84. Marcus, E. & Marcus, E., 1955. Bolm Inst. oceanogr. São Paulo, 6, 3–48. Marcus, E. & Marcus, E., 1958. Bolm Inst. oceanogr. São Paulo, 8, 3–22. Marcus, E. & Marcus, E., 1960. Bull. mar. Sci. Gulf Caribb., 10, 141–203. Marcus, E. & Marcus, E., 1962. Bull. mar. Sci. Gulf Caribb., 12, 450–488. Marcus, E.d.B.-R., 1972. Bull. mar. Sci., 22, 841–874. Marcus, E.d.B.-R., 1976. Stud, neotrop. Faun. Environ., 11, 119–150. Marcus, E.d.B.-R., 1979. Ann. Inst. Oceanogr., 55, 131–137. Marcus, E.d.B.-R. & Hughes, H.P.I., 1974. Bull. mar. Sci., 24, 498–532. Matsuda, H., Tomiie, Y., Yamamura, S. & Hirata, Y., 1967. Chem. Comm., 898– 899. McDonald, F.J., Campbell, D.C., Vanderah, D.J., Schmitz, F.J., Washecheck, D.M., Burks, J.E. & van der Helm, D., 1975. J. org. Chem., 40, 665–666. McKay, D., Jenkin, C.R. & Rowley, D., 1969. Aust. J. exp. Biol. med. Sci., 47, 125– 134. Merriman, D., 1937. Nautilus, 50, 95–97. Midland, S.L., Wing, R.M. & Sims, J.J., 1983. J. org. Chem., 48, 1906–1909. Mileikovsky, S.A., 1974. Mar. Biol., 26, 303–311. Miller, M.C., 1960. Proc. malac. Soc. Land., 34, 165–167. Minale, L. & Riccio, R., 1976. Tetrah. Lett., 31, 2711–2714. Moore, D.R., 1961. Gulf Res. Rep., 1, 1–58. Moore, R.E., 1982. Pure appl. Chem., 54, 1919–1934. Moritz, C.E., 1936. Carnegie Inst. Wash. Year Book, No. 35, 90 only. Morton, J. & Miller, M., 1968. The New Zealand Sea Shore. Collins, London-Auckland, 638 pp. Moseley, H.N., 1877. Q. Jl microsc. Sci., 17, 1–23. Mynderse, J.S. & Faulkner, D.J., 1978. Phytochem., 17, 237–240. Mynderse, J.S., Moore, R.E., Kashiwagi, M. & Norton, T.R., 1977. Science, 196, 538–540. Nagle, G.T., Painter, S.D., Kelner, K.L. & Blankenship, J.E., 1985. J. comp. Physiol. B, 156, 43–55. Neck, R.W., 1976. Veliger, 19, 107 only. Neu, W., 1932. Z. vergl. Physiol., 18, 244–254. Niell, F.X., 1977. Malacologia, 16, 207–209. Nishibori, K., 1960. Publs Seto mar. biol. Lab., 8, 327–335. Nishiwaki, S., Ueda, H. & Makioka, T., 1975. Mar. Biol., 32, 389–395.

APLYSIA

245

Oren, O.H., 1970. Underw. Sci. Technol. J., 2, 222–229. Ostergaard, J.M. 1950. Pacif. Sci., 4, 75–115. Otsuka, C., Oliver, L., Rouger, Y. & Tobach, E., 1981. Hydrobiol., 83, 239–240. Otsuka, C., Rouger, Y. & Tobach, E., 1980. Veliger, 23, 159–162. Paige, J.A., 1981. In, Advances in Invertebrate Reproduction, edited by W.H.Clark & T.S.Adams, Proc. Second int. Symp. of the int. Soc. Invert. Repr. (ISIR), Elsevier/North Holland, New York, p. 387 only. Paine, R.T., 1963. Veliger, 6, 1–9. Paine, R.T., 1966. Limnol. Oceanogr., 11, 126–129. Paine, R.T. & Vadas, R.L., 1969. Limnol. Oceanogr., 14, 710–719. Papka, R., Peretz, B., Tudor, J. & Becker, J., 1981. J. Neurobiol, 12, 455–468. Parker, G.H., 1917. J. exp. Zool., 24, 139–145. Parry, G.D., 1982. Mar. Biol., 67, 267–282. Pauley, G.B., 1971. Proc. natn Shellfish Assoc., 61, 11–12. Pauley, G.B., 1974. In, Biomedical Perspectives of Agglutinins of Invertebrate and Plant Origins, edited by E.Cohen, Ann. N.Y. Acad. Sci., 234, 145–160. Pauley, G.B., Granger, G.A. & Krassner, S.M., 1971a. J. Invert. Path., 18, 207– 218. Pauley, G.B., Krassner, S.M. & Chapman, F.A., 1971b. J. Invert. Path., 18, 227– 239. Peretz, B., 1970. Science, 169, 379–381. Peretz, B. & Adkins, L., 1982. Biol. Bull. mar. biol. Lab., Woods Hole, 162, 333–344. Peretz, B. & Howieson, D.B., 1973. J. comp. Physiol., 84, 1–18. Peretz, B., Jacklet, J.W. & Lukowiak, K., 1976. Science, 191, 396–399. Peretz, B. & Lukowiak, K.D., 1975. J. comp. Physiol., 103, 1–17. Perron, F.E., 1982. Mar. Biol., 68, 161–167. Pettit, G.R., Herald, C.L., Allen, M.S., Von Dreele, R.B., Vanell, L.D., Kao, J.P.Y. & Blake, W., 1977. J. Am. chem. Soc., 99, 262–263. Pettit, G.R., Herald, C.L., Einck, J.J., Vanell, L.D., Brown, P. & Gust, D., 1978. J. org. Chem., 43, 4685–4686. Pilsbry, H., 1951. Not. Nat. (Philadelphia), 240, 1–6. Pinsker, H., Kupfermann, I., Castellucci, V. & Kandel, E., 1970. Science, 167, 1740– 1742. Pinsker, H., Redmann, G., von der Porten, K., Heine, D. & Rothman, B., 1978. Fed. Proc., 37, 527 only. Pinsker, H.M. & Dudek, F.E., 1977. Science, 197, 490–493. Pinsker, H.M., Feinstein, R. & Gooden, B.A., 1974. Fed. Proc., 33, 361 only. Pinsker, H.M., Hening, W.A., Carew, T.J. & Kandel, E.R., 1973. Science, 182, 1039–1042. Pinsker, H.M. & Parsons, D.W., 1985. J. comp. Physiol. B, 156, 21–27. Poizat, C. & Vicente, N., 1977. Bull. Mus. natl Hist, not., Ser. 3, 34, 2–15. Preston, R.J. & Lee, R.M., 1973. J. comp. physiol. Psych., 82, 368–381. Pruvot-Fol, A., 1954. Mollusques Opisthobranches. Faune de France, Paris, No. 58, 460 pp. Rattan, K.S. & Peretz, B., 1981. J. Neurobiol., 12, 469–478. Rey, J.R. & Stoner, A.W., 1984. Estuaries, 7, 158–164. Ricketts, E.F. & Calvin, J., 1968. Between Pacific Tides. Stanford University Press, Berkeley, CA, 4th edition, 614 pp. Ronald, R.C., Gewali, M.B. & Ronald, B.P., 1980. J. org. Chem., 45, 2224–2229. Rosen, S.C., Weiss, K.R. & Kupfermann, I., 1982. Behav. new. Biol., 35, 56–63. Rothman, B.S., Weir, G. & Dudek, F.E., 1983. Gen. comp. Endocrinol., 52, 134– 141. Rüdiger, W., 1967a. Hoppe-Seyler’s Z. Physiol. Chem., 348, 129–138. Rüdiger, W., 1967b. Hoppe-Seyler’s Z. Physiol. Chem., 348, 1554 only. Rüdiger, W., 1968. Umschau, 68, 405–406. Rüdiger, W., Carra, P.O. & hEocha, C.O., 1967. Nature, Land., 215, 1477–1478. Russell, E., 1966. Pacif. Sci., 20, 452–460. Rutowski, R.L., 1983. Biol. Bull. mar. biol. Lab., Woods Hole, 165, 276–285. Ryther, J.H., 1954. Ecology, 35, 522–533.

246

THOMAS H.CAREFOOT

Saito, Y. & Nakamura, N., 1961. Bull. Jap. Soc. scient. Fish., 27, 395–400. Sakata, K., Tsuge, M. & Ina, K., 1986. Mar. Biol., 91, 509–511. Sakata, K., Tsuge, M., Kamiya, Y. & Ina, K., 1985. Agric. biol. Chem., 49, 1905– 1907. Sanford, S.N.F., 1922. Nautilus, 35, 80–82. Sarver, D.J., 1978. Ph. D. thesis, University of Hawaii, Honolulu, Hawaii, 151 pp. Sarver, D.J., 1979. Mar. Biol. 54, 353–361. Saunders, A.M.C. & Poole, M., 1910. Q. Jl microsc. Sci., 55, 497–539. Sawaya, P. & Leahy, W.M., 1971. Bol. Zool. Biol. mar., N.S., 28, 1–17. Scheller, R.H., Jackson, J.F., McAllister, L.B., Schwartz, J.H., Kandel, E.R. & Axel, R., 1982. Cell, 28, 707–719. Scheuer, P.J., 1971. Naturwissenschaften, 58, 549–554. Scheuer, P.J., 1975. Lloydia (Cincinnati), 38, 1–7. Scheuer, P.J., 1977. Accounts chem. Res., 10, 33–39. Schmekel, L., 1971. Z. Morph. Ökol. Tiers, 69, 115–183. Schmitz, F.J., Gopichand, Y., Michaud, D.P., Prasad, R.S., Remaley, S., Hossain, M.B., Rahman, A., Sengupta, P.K. & van der Helm, D., 1981. Pure appl. Chem., 51, 853–865. Schmitz, F.J., Hollenbeak, K.H., Carter, D.C., Hossain, M.B. & van der Helm, D., 1979. J. org. Chem., 44, 2445–2447. Schmitz, F.J., Hollenbeak, K.H. & Vanderah, D.J., 1978a. Tetrahedron, 34, 2719– 2722. Schmitz, F.J. & McDonald, F.J., 1974. Tetrahedron, 29, 2541–2544. Schmitz, F.J., McDonald, F.J. & Vanderah, D.J., 1978b. J. org. Chem., 43, 4220– 4225. Schmitz, F.J., Michaud, D.P. & Hollenbeak, K.H., 1980. J. org. Chem., 45, 1525– 1528. Schmitz, F.J., Michaud, D.P. & Schmidt, P.G., 1982. J. Am. chem. Soc., 104, 6415– 6423. Schreiber, G., 1932. Pubbl. Staz. zool. Napoli, 12, 291–321. Schulte, G.R., Chung, M.C.H. & Scheuer, P.J., 1981. J. org. Chem., 46, 3870– 3873 . Schwarz, M. & Susswein, A.J., 1982. In, Behavioral Models and the Analysis of Drug Action, edited by M.Y.Spiegelstein & A.Levy, Proc. 27th OHOLO Conf., pp. 479–486. Schwarz, M. & Susswein, A.J., 1984. Brain Res., 294, 363–366. Shapiro, E., Koester, J. & Byrne, J.H., 1979. J. Neurophysiol., 42, 1223–1232. Shieh, H.-M. & Prestwich, G.D., 1982. Tetrah. Lett., 23, 4643–4646. Sigel, M.M., Wellham, L.L., Lichter, W., Dudeck, L.E., Gargus, J.L. & Lucas, A.H., 1970. In, Food-drugs from the Sea, edited by H.W.Youngken, Jr, Marine Technology Society, Washington, D.C., pp. 281–294. Sims, J.J., Fenical, W., Wing, R.M. & Radlick, P., 1971. J. Am. chem. Soc., 93, 3774–3775. Sims, J.J., Fenical, W., Wing, R.M. & Radlick, P., 1972. Tetrahedron Lett., No. 3, 195–198. Smith, S.T. & Carefoot, T.H., 1967. Nature, Land., 215, 652–653. Sousa, W.P., 1979. Ecol. Monogr., 49, 227–254. Stallard, M.O. & Faulkner, D.J., 1974a. Comp. Biochem. Physiol., 49B, 25–35. Stallard, M.O. & Faulkner, D.J., 1974b. Comp. Biochem. Physiol., 49B, 37–41. Stallard, M.O., Fenical, W. & Kittredge, J.S., 1978. Tetrahedron, 34, 2077–2081. Stehouwer, H., 1952. Arch. Néerl. Zool., 10, 161–170. Stickle, W.B., 1985. In, Marine Biology of Polar Regions and Effects of Stress on Marine Organisms, Proc. 18th Europ. Mar. Biol. Symp., edited by J.S.Gray & M.E.Christiansen, John Wiley and Sons, Chichester, U.K., pp. 601–616. Stinnakre, J. & Tauc, L., 1969. J. exp. Biol., 51, 347–361. Stone, B.A. & Morton, J.E., 1958. Proc. malac. Soc. Lond., 33, 127–141. Streit, B., 1976. Oecologia (Berl.), 22, 261–273. Strenth, N.E. & Blankenship, J.E., 1976. Bull. Am. malac. Union, 42, 48 only. Strenth, N.E. & Blankenship, J.E., 1977. Veliger, 20, 98–100. Strenth, N.E. & Blankenship, J.E., 1978a. Veliger, 21, 99–103. Strenth, N.E. & Blankenship, J.E., 1978b. Bull. Mar. Sci., 28, 249–254.

APLYSIA

247

Strumwasser, F., 1967. In, The Neurosciences. A study program, edited by G.C. Quarton et al., Rockefeller University Press, New York, pp. 516–528. Strumwasser, F., 1973. Physiol, 16, 9–42. Strumwasser, F., Jacklet, J.W. & Alvarez, R.B., 1969. Comp. Biochem. Physiol., 29, 197–206. Susswein, A., Kupfermann, I. & Weiss, K.R., 1976a. Neurosci. Abstr., 2, 336 only. Susswein, A.J., 1984. Behav. new. Biol., 42, 127–133. Susswein, A.J., Gev, S., Achituv, Y. & Markovich, S., 1984a. Behav. neur. Biol., 41, 7–22. Susswein, A.J., Gev, S., Feldman, E. & Markovich, S., 1983. Behav. neur. Biol., 39, 203–220. Susswein, A.J. & Kupfermann, I., 1974. Soc. Neurosci. 4th Ann. Meet. St. Louis, Missouri, p. 445 only. Susswein, A.J. & Kupfermann, I., 1975a. Behav. Biol., 13, 203–209. Susswein, A.J. & Kupfermann, I., 1975b. J. comp. Physiol., 101, 309–328. Susswein, A.J., Kupfermann, I. & Weiss, K.R., 1976b. J. comp. Physiol., 108, 75– 96. Susswein, A.J. & Markovich, S., 1983. Israel J. Zool., 32, 103–112. Susswein, A.J. & Schwarz, M., 1983. Behav. neur. Biol., 39, 1–6. Susswein, A.J., Weiss, K.R. & Kupfermann, I., 1978. J. comp. Physiol., 123, 31–41. Susswein, A.J., Weiss, K.R. & Kupfermann, I., 1984b. Behav. neur. Biol., 41, 90– 95. Swennen, C., 1961. Zoöl. Meded., Leiden, 38, 41–75. Switzer-Dunlap, M., 1978. In, Settlement and Metamorphosis of Marine Invertebrate Larvae, edited by F.-S.Chia & M.E.Rice, Elsevier/North-Holland Press, Amsterdam, pp. 197–206. Switzer-Dunlap, M. & Hadfield, M.G., 1977. J. exp. mar. Biol. Ecol, 29, 245–261. Switzer-Dunlap, M. & Hadfield, M.G., 1979. In, Reproductive Ecology of Marine Invertebrates, edited by S.E.Stancyk, Belle W.Baruch Library in Marine Sciences, No. 9, University of South Carolina Press, Columbia, S.C., pp. 199– 210. Switzer-Dunlap, M. & Hadfield, M.G., 1981. In, Marine Invertebrates, Report of the Committee on Marine Invertebrates, Institute of Laboratory Animal Resources, National Research Council, Nat. Academy Press, Washington, D.C., pp. 199–216. Switzer-Dunlap, M., Meyers-Schulte, K. & Gardner, E.A., 1984. Int. J. Invert. Repr. Devel., 7, 217–225. Tanaka, A., Otsuka, S. & Yamashita, K., 1984. Agric. biol. Chem., 48, 2535–2540. Tanaka, T. & Toyama, Y., 1959. J. Chem. Soc. Japan, Pure Chem. Sec. (Nippon Kagakukai Zasshi), 80, 1329–1332. Thompson, T.E., 1960, J. mar. biol. Ass. U.K., 39, 123–134. Thompson, T.E., 1976. Biology of Opisthobranch Molluscs, Vol. 1. Ray Soc., Land., No. 151, 206 pp. Thompson, T.E. & Bebbington, A., 1969. Malacologia, 7, 347–380. Thorson, G., 1946. Medd. Komm. Danm. Fisk. Havunders., 4, (1), 1–523. Tobach, E., 1978. Veliger, 20, 359–360. Tobach, E., Gold, P. & Ziegler, A., 1965. Veliger, 8, 16–18. Toevs, L., 1966. Calif. Inst. Technol., Biol. Div., Ann. Rep., p. 154 only. Tritt, S.H. & Byrne, J.H., 1980. J. Neurophysiol., 43, 581–594. Tritt, S.H. & Byrne, J.H., 1982. J. Neurophysiol., 48, 1347–1361. Tritt, S.H., Zigmond, M. & Byrne, J.H., 1980. Neurosc. Abstr., 6, 703 only. Tunnell, Jr, J.W. & Chaney, A.H., 1970. Contr. mar. Sci., 15, 193–203. Usuki, I. 1970a. Mem. Sado Mus., Niigata Pref., 19, 1–10. Usuki, I. 1970b. Sci. Rep., Niigata Univ., Ser. D. (Biol.), 7, 91–105. Usuki, I. 1979. J. gen. Educ. Depart., Niigata Univ., 9, 81–84. Usuki, I. 1981a. Venus, 39, 212–223. Usuki, I. 1981b. Venus, 40, 41–49. Vanderah, D.J. & Schmitz, F.J., 1976. J. org. Chem., 41, 3480–3481. Van Weel, P.B., 1957. Z. vergl. Physiol., 39, 492–506. Vicente, N., 1962. Recl. Trav. Stn mar. Endoume, Bull. No. 26, Fasc. 41, 303–305. Vicente, N. & Poizat, C., 1977. Bull. Mus. natn. Hist, not., Paris, Sér. 3, No. 34, 19– 26.

248

THOMAS H.CAREFOOT

Vitalis, T.Z., 1981. B. Sc. thesis, University of British Columbia, Vancouver, B.C., 39pp. Von der Porten, K., Parsons, D.W., Rothman, B.S. & Pinsker, H., 1982. Behav. new. Biol., 36, 1–23. Von der Porten, K., Redmann, G., Rothman, B. & Pinsker, H., 1980. J. exp. Biol., 84, 245–257. Von Dreele, R.B. & Kao, J.P.Y., 1980. Acta Cryst., B36, 2695–2698. Wachtel, H. & Impelman, D., 1973. Fed. Proc., 32, 368 only. Walford, L.A., 1946. Biol. Bull. mar. biol. Lab., Woods Hole, 90, 141–147. Walters, E.T., Carew, T.J. & Kandel, E.R., 1978. Neurosc. Abstr., 4, 209 only. Walters, E.T., Carew, T.J. & Kandel, E.R., 1979a. Neurosc. Abstr., 5, 264 only. Walters, E.T., Carew, T.J. & Kandel, E.R., 1979b. Proc. natn Acad. Sci. U.S.A., 76, 6675–6679. Walters, E.T., Carew, T.J. & Kandel, E.R., 1981. Science, 211, 504–506. Ward, J., 1967. Bull. mar. Sci., 17, 299–318. Warmke, G.L. & Abbott, R.T., 1961. Caribbean Seashells A Guide to the Marine Mollusks of Puerto Rico and other West Indian Islands, Bermuda and the Lower Florida Keys. Livingston Publishing Company, Narberth, Penn., 348 pp. Warren, C.E. & Davis, G.E., 1967. In, The Biological Basis of Freshwater Fish Production, edited by S.D.Gerking, Blackwell Sci. Publ., Oxford, pp. 175–214. Waser, P.M., 1968. Comp. Biochem. Physiol., 27, 339–347. Watson, M., 1973. Toxicon, 11, 259–267. Watson, M. & Rayner, M.D., 1973. Toxicon, 11, 269–276. Weiss, K.R., Koch, U.T., Koester, J., Mandelbaum, D.E. & Kupfermann, I., 1981. In, Neurobiology of Invertebrates. Adv. Physiol. Sci., Vol. 23, edited by J.Salanki, Pergamon Press, Oxford, pp. 305–344. Welch, H.E., 1968. Ecology, 49, 755–759. Wells, D.W. & Hill, R.B., 1980. Comp. Biochem. Physiol., 67C, 97–106. Wells, D.W. & Hill, R.B., 1985. Comp. Biochem. Physiol., 80C, 337–345. Wells, L.J., Jahan-Parwar, B. & Fredman, S.M., 1974. Physiol., 17, 357 only. White, J.D., Nishiguchi, T. & Skeean, R.W., 1982. J. Am. chem. Soc., 104, 3923– 3928. Wightman, J.A., 1981. Oecologia (Berl.), 50, 166–169. Willan, R. & Morton, J., 1984. In, Marine Molluscs, Part 2, Leigh Marine Laboratory, University of Auckland, pp. 27–34. Willan, R.C., 1979. Ph. D. thesis, University of Auckland, Auckland, New Zealand, 198 pp. Willey, J.C., Moser, Jr, C.E. & Harris, C.C., 1984. Cell Biol. Toxicol., 1, 145–154. Winkler, L.R., 1955. Bull. Sth. Calif. Acad. Sci., 54, 5–7. Winkler, L.R., 1958a. Pacif. Sci., 12, 348–349. Winkler, L.R., 1958b. Bull. Sth. Calif. Acad. Sci., 57, 105–106. Winkler, L.R., 1958c. Bull. Sth. Calif. Acad. Sci., 57, 106–107. Winkler, L.R., 1959a. Pacif. Sci., 13, 357–361. Winkler, L.R., 1959b. Pacif. Sci., 13, 63–66. Winkler, L.R., 1959c. Bull. Sth. Calif. Acad. Sci., 58, 8–10. Winkler, L.R., 1959d. Nautilus, 72, 73–75. Winkler, L.R., 1960. Pacif. Sci., 14, 304–307. Winkler, L.R., 1961. Pacif. Sci., 15, 211–214. Winkler, L.R., 1969. Veliger, 11, 268–271. Winkler, L.R. & Dawson, E.Y., 1963. Pacif. Sci., 17, 102–105. Winkler, L.R. & Tilton, B.E., 1961. Pacif. Sci., 15, 557–560. Winkler, L.R. & Tilton, B.E., 1962. Pacif. Sci., 16, 286–290. Winkler, L.R., Tilton, B.E. & Hardinge, M.G., 1962. Arch. int. Pharmacodyn., 137, 76–83. Wolff, H.G., 1973. Mar. Behav. Physiol., 1, 361–373. Wright, D.L., 1960. Ph. D. thesis, University of California, Los Angeles, 104 pp. Wright, H.O., 1960. Veliger, 2, 63–64.

APLYSIA

249

Yamada, K., Yazawa, H., Toda, M. & Hirata, Y., 1968. Chem. Comm., No. 22, 1432 only. Yamamura, S. & Hirata, Y., 1963. Tetrahedron, 19, 1485–1496. Yamamura, S. & Hirata, Y., 1971. Bull. chem. Soc. Jap., 44, 2560–2562. Yamamura, S. & Terada, Y., 1977. Tetrah. Lett., No. 25, 2171–2172. Yamazaki, M., Esumi-Kurisu, M., Mizuno, D., Ogata, K. & Kamiya, H., 1983a. Gann, 74, 405–411. Yamazaki, M., Ikenami, M., Komano, H., Tsunawaki, S., Kamiya, H., Natori, S. & Mizuno, D., 1983b. Gann, 74, 576–583. Yamazaki, M., Kisugi, J., Ikenami, M., Kamiya, H. & Mizuno, D., 1984. Gann, 75, 269–274. Yamazaki, M., Kisugi, J., Kimura, K., Kamiya, H. & Mizuno, D., 1985. FEBS Lett., 185, 295–298. Yonge, C.M., 1947. Phil. Trans. Roy. Soc. Land., Ser. B, 232, 443–518. Yonge, C.M., 1949. The Sea Shore. Collins, London, 311 pp. Zinn. D.J., 1950. Nautilus. 64, 40–47.


A REVIEW OF THE COMPARATIVE ANATOMY OF THE MALES IN CIRRIPEDES WALTRAUD KLEPAL Institut für Zoologie der Universität Wien, Althanstraße 14, A-1090 Wien, Österreich

INTRODUCTION Secondary sexual dimorphism as distinct from the primary sexual dimorphism (which is expressed in the gonads, the germ cells and the copulatory organs) is a common feature in the animal kingdom. Usually there is a difference in size and in the armament in males and females. Often the females are larger as in molluscs, spiders, insects, fishes, frogs, snakes, turtles, and birds of prey. On the other hand, the males are more imposing and larger, e.g. in gallinaceous birds, ostriches, iguanas, beasts of prey, deer, horses and apes. Sexual dimorphism is also a common feature in crustaceans. It is found in the Conchostraca, Cladocera, and Anostraca and is expressed as a difference in size and shape of males and females and a difference in the formation of their appendages. In the Ostracoda it affects the carapace. In the epicaridean isopods the males are small and the females are large and sac-like without any legs. In other crustaceans the males are smaller than the females, e.g. in Gnathia. Advanced types of crustaceans like the Malacostraca show only slight differences in the appendages of males and females (Newman, Zullo & Withers, 1969). The males in the cirripedes are always considerably smaller than the females or hermaphrodites. This dwarfism is an extreme case of secondary sexual dimorphism. It is found in some parasitic crustaceans as the Rhizocephala and in cirripedes which live at the edge of some spatial horizontal or vertical distribution. The males of these cirripedes are generally referred to as dwarf males. Well-known dwarf males in other groups of animals are those of the worm Bonellia viridis and of the deep-sea fish Edriolychnus sp. In cirripedes Darwin (1851) distinguished between dwarf males, associated with a female and complemental males associated with a hermaphroditic cirripede. Crisp (1983) introduced the term “apertural male” for the males which are attached around the opercular opening of Chelonobia. These males are potential hermaphrodites which stopped in their development at the protandric stage. Males, as distinct from females and/or hermaphrodites are found in all three orders of cirripedes: Acrothoracica, Thoracica, and Rhizocephala. The Acrothoracica and the Rhizocephala are obligatory gonochoristic. Within the Thoracica both dwarf- and complemental males have been found. Whilst in the Lepadomorpha either dwarf- or complemental males are common, in the Balanomorpha so far only complemental males are known. Up to now no separate males were found in the Verrucomorpha. A number of papers have dealt with the question of sexuality in cirripedes and also with complemental males. But so far no comprehensive study has been made. This paper is an attempt to consider and to compare the main features of males in the Rhizocephala, Acrothoracica, and Thoracica as far as they are known up to now and as far as they are described in any detail. An attempt is made to show the major trends rather than to describe the male of every single species. Thus, all males will be omitted that are not relevant to the considerations in this paper.

COMPARATIVE ANATOMY OF MALES IN CIRRIPEDES

251

First the descriptions of the males will be reviewed. This will be followed by an effort to elucidate the trends of evolution within the single orders and then an attempt to show the general trend of evolution of the males and of sexuality within the cirripedes in general. DESCRIPTIONS OF THE MALES ACROTHORACICA The Acrothoracica are free-living burrowing cirripedes with separate sexes in all known species. The females are about ten times as big as the males. In many species only the females have been described and the males are not known. The metamorphosis is very complex and the morphology of the males is very variable. The male may be even smaller than the cypris larva. It is motile. With the exception of Berndtia it is buried within a pocket of the female’s mantle tissue in the immediate area of the ovary. Thus its sexual development is synchronized with that of the female. It is even assumed that the male matures under some chemical mediation of the female. The embedding of the male in this way was observed in Cryptophialus melampygos, Lithoglyptes indicus, Trypetesa lateralis and Trypetesa lampas (Tomlinson, 1969). The anatomy of the males has been described by only a few authors. Three types of males may be distinguished in the Lithoglyptidae, supposedly the most primitive family of Acrothoracica (Tomlinson & Newman, 1960; this statement is based on characters of the females). (1) Pear-shaped males with a penis, homologous to the thorax of the cypris. These males are similar in most species of Lithoglyptes and in Weltneria. There are no special differentiations in the larval antennae for the attachment. (2) Males of variable shape. These are often polygonal, they have a penis and a specialized attachment organ. The latter is either a long and thin peduncle as in Kochlorine hamata or it is an orchid lobe as in K. floridana. (3) Males without a penis as in K. bocqueti or perhaps in K. ulula (Tomlinson, 1973). There is a great similarity between the adult male of K. ulula and the juvenile form of K. bocqueti. In the adult male of K. bocqueti there is no penis, but there is a specialized attachment organ which assures the attachment of the male at the side of the female. The genus Weltneria is supposed to be the most primitive one amongst the Acrothoracica. Newman (1971) described W. hessleri but he did not observe any males. He himself calls this unfortunate since Tomlinson (1969) suggested that the males in the genus have a more or less characteristic cyprid-like form with a blunt posterior end. In 1974 Newman described the male of W. exargilla; the female of the species is mm big. They have a single testis, a seminal vesicle, very similar to W. hessleri. The males are about a vas deferens and a penis. The penis seems to be in a channel, which is supposedly a rudiment of the mantle cavity. At the base of the penis there are numerous muscles. Turquier (1985b) studied the larvae of W. zibrowii very carefully as well as the metamorphosis from the cypris to the male. Three major events may be distinguished in the metamorphosis of the larvae. The beginning of the transformation is marked by some cement being secreted. One major step in the metamorphosis is the formation of a rudimentary peduncle. Then the muscles of the antennules and the cement glands degenerate. Remnants of these two organs may be seen during the whole metamorphosis. The second step is the condensation and rotation of the visceral mass. The front head region of the larva becomes the rudimentary peduncle and the visceral mass carries out a characteristic morphogenetic movement in the sagittal plane

252

WALTRAUD KLEPAL

around the tendon of the adductor muscle of the valves. Thus the mouthcone is moved backwards and the cerebral ganglion is moved posteroventrally. The lateral eyes are lost and the optic ganglion connected to the protocerebrum degenerates. The thorax withdraws into the mantle cavity and its muscles dedifferentiate. The nervous system becomes simpler and its volume decreases. The visceral mass is bent to a U-shape. The mouthcone, the cerebral ganglion and the “yellow organ” are moved towards the sternite of the cypris, thus marking the rostral face of the future male. The “yellow organ” may be a stomach rudiment with yolk-absorbing function as in Pollicipes (Batham, 1946) or derived from degenerating endodermal cells (Turquier, 1971). The third important step in the course of metamorphosis is the organogenesis of the genital apparatus. The posterior mantle cavity is now closed. The gonad on the ventral side of the larva is removed little by little from the thorax and totally occupies the head metameres. In the juvenile male the gonad takes up the central space of the body. Now there is intensive sperm production and thus the seminal vesicle is enlarged. The thorax grows vastly and re-differentiates. The thoracal muscles form a double layer of longitudinal and circular strands. The adult male of Weltneria zibrowii (Fig. 1A) has an ovoid body of three distinct regions. The long. The body proper is globular, about mm. In the rudimentary peduncle is about 50 posterior part there are the testes and at the same time it is something like a sheath for the penis. In the testes the sperms develop synchronously and they accumulate in the central part of the gonad. The pearshaped vesicle is curved to an S backwards and towards the rostral face. Then it leads into the ductus ejaculatorius and further into the penis. In the adult male the nervous system is along the ventral side of the animal. The nauplius eye and the “yellow organ” may be easily seen but the mouthcone and the first pair of cirri have disappeared. mm without antennules (Berndt, 1907). It resembles a degenerate sac In W. spinosa the male is in which a long annulated penis may be seen curling inside the body. The body is streamlined and has a mm small, deep purple nauplius eye. A “yellow organ” was not detected. In W. hirsuta the male is big. It has the usual antennae and blunt corrugated posterior projections. No cuticular structures were seen in this species. In W. reticulata the males are rare. They have paired antennules and a well-developed penis is coiled up within the body. There is a curious rounded projection from the end of the body opposite the point of attachment. Scattered clusters of cells of glandular function are seen. No size of the male is given. Tomlinson (1963) described the male of Lithoglyptes hirsutus but was not sure whether the male has a penis. In its general features the dwarf male resembles a cypris larva. Tomlinson (1969) reported that the presence of a penis in some species of Lithoglyptes was an unresolved question. According to him most species have a definite penis which can be pulled out to many times the body length. In L. spinatus Tomlinson & Newman (1960) were again not sure about the presence of a penis. The body of the male in this species is bulbous and it is connected to the antenna by a long stalk. This stalk arises from a T-shaped connection with the two normal appearing antennules and terminates in an annulated attachment to the body. In L. indicus, on the other hand, Aurivillius (1894) describes a large penis as well as a testis and a vesicula seminalis. This male is about 0·5 mm long and has the shape of a coil. The nervous system was seen by Aurivillius (1894) but the alimentary canal is missing. A penis may be present. The In L. mitis Tomlinson (1969) gave the size of the male as mm. posterior end is annulated. It does not have any heavy teeth. The antennae have very short stalks. The male of L. scamborachis was described as typical in general aspect (Tomlinson, 1969). It has a penis. The body is finely annulated. No long stalk was found between the body and the antennules. One of the males had a peculiar projection on the body. This appeared to be stuck to the body of the female. The male of L. wilsoni is a reduced bag of gametes (Fig. 2A). Only the reproductive system is well developed. There is a penis and


253

Fig. 1. —Anatomy of the adult males of Acrothoracica. A, Weltneria zibrowii (after Turquier, 1985b); B, Berndtia purpurea (after Utinomi, 1961); a, antenna; cg, cerebral ganglion; g, ganglion; p, penis; ps, penis sheath; rb, reserve bodies; t, testis; ta, terminal ampulla; th, thorax; vs, vesicula seminalis; yo, yellow organ; scale bars, A=150 B=200

also the antennae. The long penis sheath is faintly annulated. It has rows of very fine spines which are bifid at the tip. The cuticle of the male has rows of fine spines which are then scattered around the antennae. In L. habei the adult male is about 0·5 mm long. It has a pair of antennae but no long stalks, which are seen in some species. There is a penis. The males of Lithoglyptes are, according to Aurivillius (1894), similar to those of Alcippe (=Trypetesa). In both genera they are attached to the outer side of the female. The males do not have any appendages. There is a difference in the position of the antennae in the two genera. In Alcippe the antennae are in the middle of the body. In Lithoglyptes they are in that part of the body distal of the penis. In both genera the productive organs are well developed with a single testis, a single vesicula seminalis and a penis (penis is doubtful in some species of Lithoglyptes, see above). The nervous system consists of an elongated cerebral ganglion which is at about the height of the vesicula seminalis. From this ganglion nerves arise. From the posterior part of this ganglion a nerve goes to a second ganglion, which is nearly as long as the cerebral ganglion but narrower than the latter. From there a nerve goes to the black pigmented eye. Close to the cerebral ganglion there is a small rounded “yellow organ” whose contents are granular. This formation is reminiscent of the “gland of unknown nature” in the male of Scalpellum regium (Thoracica). mm. It has a nauplius eye, testis, vesicula seminalis The male of Kochlorine hamata is about and a penis. The penis extends into one of the “wings”, which presumably is elongated into a penis sheath.

254

WALTRAUD KLEPAL

The antennae have a stalk and together (antenna+stalk) they may be nearly 0·3 mm long. The body has two protuberances (“wings”) from the stalk; one is more pointed than the other. The male of K. floridana (Fig. 2B) attaches to the horny disk formed by the remains of the female exuviae. Young males look much like a cypris larva (Tomlinson, 1969). The mature male has an obvious cellular testis and a vesicula seminalis leading to an annulated penis. There seems to be a penis retractor muscle connected to one of these lobes projecting from the main body of the male, with the testis in another lobe; the third lobe has a few glandular-appearing cells. There is something like the “yellow organ” and the nauplius eye. The outer surface of the male is covered with rows and patterns of extremely fine dots. The mm (smaller than the cypris). This reduction in length is typical of the size of the male is about maturation of males in the Order. There are no trophic structures. It is assumed that after fertilization the male is expended. The male cypris of K. bocqueti is reminiscent of the larvae of the Trypetesidae, although that of Whilst the Kochlorine is a little bigger. The juvenile male of Kochlorine is pear-shaped and anterior region gets wider the peduncle regresses. The thorax also regresses and the posterior pallial cavity is closed. The pallial cavity is reduced to a narrow furrow and it communicates with the exterior at about the height of the distal extremity. In contrast to most males of the Acrothoracica there is no penis-Anlage developed at that stage. Whilst the visceral mass rotates the gonad grows fast and gametogenesis begins. The structure of the gonad is similar to that of other species. The globular testis is hollow. The axial cavity is full of spermatozoids or spermatids. Apart from the testis there is the vesicula seminalis whose outermost end is the remnant of the pallial cavity. Beginning with a pear-shape the body of the male becomes more and more globular. The posterior end of the mantle, which narrowed in the juvenile male disappears in the adult. A voluminous cylindrical projection differentiates arising from the antennal projection in ventral position. This organ of attachment in the oldest males. In develops at the cost of the capitulum the size of which is reduced to addition, the shape of the capitulum changes. Now it looks like a semi-cone and it is plump (Fig. 2C). The tip of the cone gets prolonged by the attachment organ which becomes long and thin, with a diameter of 30– 35 and whose wall increases in thickness. The capitulum is partly attached to the female, with the attachment organ. This consists of rows of short spines on the body wall of the male. The testis shifts into the posterior body region and the vesicula seminalis into the anterior. The ductus ejaculatorius points into the area of the attachment organ. The migration of organs also affects the nervous system and the eye spot. At the same time the trace of the posterior pallial cavity disappears. It probably gets blocked. Kochlorinopsis discoporellae has a triangular male with a single, remarkable two-jointed antenna. The mm. The posterior end is blunt; this is presumably the penis sheath. The penis size of the male is appears to swell out and to fill the posterior part of the organism. The testis is lobed, the vesicula seminalis is tubular. Utinomi (1961) gave a detailed description of the male of Berndtia purpurea (Fig. 1B). The body is 1·2– 1·5 mm long and about 0·2 mm wide in the anterior position and up to 0·06 mm wide in the posterior part. Utinomi (1961) states that he always found a remnant of the cypris sac in contact with the anterior end around the antennules. From this fact the author concludes that the male, once attached, does not moult again after the final metamorphosis. Within the cypris sac there are remnants of the cypris such as the compound eye. The body wall of the adult male consists of a thin and transparent cuticle. On the outside of this cuticle there are thorns arranged in transverse rows and usually directed posteriorly. Utinomi (1961) did not see any epithelial cell layer beneath the cuticle which is presumably due to bad preservation. He mentions strong longitudinal muscles in the middle part of the body. Up to 19 of these muscles are along the dorsal surface


255

and about 16 on the ventral side. The number of muscles varies with the size of the body. The length of the muscles is also variable. Utinomi did not see any cross striation in these muscles (which could also be due to the state of preservation, see above). There are no transverse muscles or fibres. In the posterior prolongation of the body there are no muscles at all. The reproductive organs are the most conspicuous internal organs of the male. They consist of a single tube which may be divided into the testis, vesicula seminalis, vas deferens and ductus ejaculatorius. The tube is coiled (because it is about twice as long as the body). The testis is large and globular. Spermatogonia are found in the basal part and spermatozoa in the portion leading into the vesicula seminalis. The spermatozoon is described as being filiform with a pinhead-like or pyriform head. The vesicula seminalis is marked by a slight swelling; it is a dilation of the proximal part of the vas deferens. The vas deferens enters into the ductus ejaculatorius. At the base of the penis there are several bundles of powerful striated muscles, one group running longitudinally and the other shorter one running obliquely. These muscles enter into the penis sheath and together form an outer circular layer of longitudinal muscles. The annulated penis is very long and greatly contractile. It runs freely, coiled up, within the widest portion of the channel-like body cavity, the penis tube. At the end of the penis there is a brush of fine setae. In the penis there are finely striated longitudinal muscles which form, together with the ringed cuticular covering, an outer sheath for the penis to enclose the ductus ejaculatorius within it. There are no circular muscles as in the penis of the males of Trypetesa and Cryptophialus. The penis tube is usually regarded as the vestige of the mantle cavity. In Berndtia sp. Utinomi (1961) never found any cuticular lining nor any epithelial cell layer which line the mantle cavity in ordinary cirripedes. Therefore Utinomi concludes that at least in B. purpurea the penis tube is a specialized lacunar channel in the body tissue. In the male of B. purpurea there is a terminal sac or ampulla which may be a kind of tactile proboscis serving in the act of fertilization. The nervous system resembles that of Trypetesa lampas (see later). Where the vas deferens communicates with the ductus ejaculatorius there is a large dark-coloured ganglion. This corresponds well with the “Hauptganglion” in Trypetesa. It has a median constriction and thus it seems to consist of two parts. Utinomi (1961) assumes that this is the fused mass of the ventral nerve ganglia. In the male of Berndtia purpurea no optic ganglion or eye has been found. There is a “large yellow peculiar body lying usually in close contact with the main ganglion” (Utinomi, 1961, p. 439). This is homologous with the “gerundete Organ” of Berndt found in the male of Trypetesa lampas. Utinomi (1961) interprets this “yellow organ” as a kind of nutritive organ which originates from the larval alimentary canal. There are ovoid globular cell masses on either side of the vas deferens and one near the proximal end of the penis. These are reminiscent of the cement glands of the males of Scalpellum velutinum and S. regium. They are not found in any other acrothoracican. There is no digestive tract nor were any excretory organs found. In larval development the male shows a close resemblance to that of Cryptophialus, whilst the female resembles Trypetesa. The male of Berndtia nodosa resembles that of B. purpurea. It is found primarily on the exuviae remains of the female; in B. purpurea the males are invariably found attached to the wall of the burrow. Information on the males of Cryptophialus sp. is very scarce and in most cases is restricted to a description of the point of attachment and on the presence of a penis. According to Tomlinson (1960) C. coronatus forms a penis late in life. Several examined males had no penis and some had a small penis only. mm. On the integument there are peg plates (polygons bearing numerous dark The animal is about spots). In C. melampygos the same author observed a long penis. Utinomi (1961) states that in C. minutus the longitudinal muscles are distinctly striated. The penis of the Cryptophialus sp. is known to have circular muscles and according to Berndt (1903a) the retractor muscles lying at its base become weaker towards the

256

WALTRAUD KLEPAL

penis and ultimately merely form an outer layer of finer longitudinal muscle fibres. Most ordinary cirripedes (supposedly the Thoracica) have an outer layer of strong longitudinal muscles in the penis (Klepal, Barnes & Munn, 1972). The circular muscles lie inside the longitudinal muscles and are formed as the outer covering of the ductus ejaculatorius. C. minutus was described by Darwin (1854) who pointed out the great similarity between the males of Cryptophialus and Alcippe inspite of the great dissimilarity of their pupae. The male of Cryptophialus minutus is about 0·3 mm long. No eye was seen, no mouth, no thorax, no cirri and no other organs except the testis, vesicula seminalis and an immensely elongated probosciformed penis. The penis can be stretched to 3–9 times its resting length. C. minutus striatus was described by Berndt (1906) as having a male which is about 0·4 mm long. The shape of the male is that of a bulged bottle with a rounded bottom and a straight neck which is compressed sideways. There is a large testis and a pear-shaped vesicula seminalis which eventually leads into a long vermiform penis. In the “neck of the bottle” there is a characteristic slit-shaped opening. There is a membranous canal which leads the penis and keeps it in its position by connective tissue fibres (and not by muscles). Around the vesicula seminalis there is a thin layer of circular muscles. The cerebral ganglion is big. The ganglion opticum and the eye are missing (but this lack may be due to bad preservation). In the mantle wall there is a system of strong cross-striated longitudinal muscles which seem to have the function of directing the penis in the stretched condition. There are also rings of small transverse muscles which may help to push out the extended penis. The penis is very long and it may be assumed that it reaches nine times its resting length. From the outside to the inside there are small circular muscles, followed by longitudinal muscles forming a tube. There are lacunae and in the centre there is the ductus ejaculatorius. At the tip of the penis there are hooks (no setae) which help to attach the penis on to the integument of the female during the ejaculation of the spermatozoa. The spermatozoa themselves resemble those of Alcippe. Batham & Tomlinson (1965) described the male of Cryptophialus melampygos [now Australophialus mm, only slightly larger than the melampygos (Berndt, 1907)]. The size of the adult male is about cyprid. It is pear-shaped, the broad end being anterior. The posterior end of the male is slightly forked, one fork carrying several multi-spiked chitinous knobs. Internally the body wall shows bands of longitudinal muscles. Young males still show some yolk globules. A conspicuous “yellow organ” is seen in the male as well as in the cyprid and the metamorphosing female. Anterior to it there is the testis and the seminal vesicle, in both of which there are elongate sperms. The long muscular penis is coiled. It is very long, several times the length of the male. The penis is annulated throughout its length and broader distally. At its end there are five small setae. In Cryptophialus heterodontus the male has a pair of antennae without any stalks. The body of the male is rounded. No penis was seen (the present author thinks that the males could have been immature specimens). No eyes, nor any “yellow organ” were noticed. C. wainwrighti has a male with a prominent penis, and paired antennae (Fig. 2D). The hyaline mantle of the male has no peg plates. The posterior end of the animal is bifid with about four small teeth. Whether eyes and “yellow organ” are present could not be made out with certainty. C. variabilis have males which, when young have a flattened cypris carapace with distinct peg plates, dorsal hairs and posterior long bristles. In addition, there is an articulated posterior plate. As usual there are the antennules; when the males are spent the antennae remain on the females. Mature males are smaller than the larvae. They have an elongated sheath for the penis. The sheath terminates in a toothed bilobed process. Other males described by Tomlinson (1969) are those of C. newmani, C. lanceolatus (which has an indication of a stalk between the body and the antennae and a cuticular single “beak” in the central line on the ventral side of the body), and C. unguiculus (whose antennae are often in a spread position). In all cases


257

a penis is well developed. For C. cordylacis the author gave drawings of the males but there was no description. Turquier (1985a) states that the morphology and the size of the males varies with the age of the animal; he observed this in Australophialus pecorus. Whilst soon after the metamorphosis of the cypris the males ) they become more elongated (length: width ratio=1·8 appear globular or pear-shaped (size as morphogenesis proceeds (Fig. 2E). In that particular in mature animals) and they reach a size of 350 species the testis is in the middle of the body and the seminal vesicle is orientated towards the rear end of the body. The ductus ejaculatorius turns to the ventral side before it extends into the penis. The penis itself stems from the metamorphosed larval thorax as in all males of the Acrothoracica. It is coiled in the ventral part of the body and is in a postero-dorsal position within the mantle cavity. The genital opening is in the posterior part of the body and it is armoured with teeth. The two ganglia of the nervous mass are close together, the nerves extending into the rear end of the body. Close to the nervous system there is the “yellow organ”, vesicular with a spherical opaque centre. mm. It has paired antennae and a long coiled penis. In A. turbonis the size of the male is about The end distal to the antennules has two to three teeth on the mantle surface. They look like the boring teeth of the female mantle surface. There are about 12 males per female but up to 17 were seen. Berndt (1903b) found up to 12 males in Alcippe lampas (=Trypetesa lampas). The shape of the males is that of a bulgy bottle (capitulum, Darwin, 1854) with a sideways highly compress neck (peduncle, Darwin, 1854). At the bottom of the bottle neck there are two lateral lobes of the peduncle. Down to these lobes the peduncle is buried in the disk of the female. The epithelium of the female forms a deep pocket around the peduncle of the male. The peduncle is about 0·6 mm long as is the capitulum. The whole inside of the body is freely open to the water. At the bottom of this cavity the penis arises. The basal part of the penis is the body proper of the animal. The mantle cavity forms the penis sheath. The narrow end part of the penis sheath forms a structure leading the penis during copulation. A large single testis is at the bottom of the peduncle, near the ventral side. From there the club-shaped vesicula seminalis arises and continues into the vas deferens which eventually leads into the penis. The penis may be extruded from the capitulum to 3·5 times its length. The arrangement of elements in the testis of dwarf males seems to agree with that in the testis of hermaphroditic cirripedes. There is one vesicula seminalis whose wall consists of a layer of connective tissue with bundles of circularly arranged extremely fine muscle fibres. There muscles must serve to press the spermatozoa into the penis. Thus in Alcippe (Trypetesa) the spermatozoa are ejected actively whilst in the Thoracica the ejection was described as a passive process. Within the penis there is an outer circular and an inner longitudinal layer of muscles. In the centre there is the ductus ejaculatorius. On the cuticle there are setae, and on the tip of the penis there are two long and four short setae. There are also lacunae within the penis. In a live male the penis is continuously moving. Between the ventral and the dorsal edge of the capitulum there are regularly arranged parallel muscle bundles which give this region a striated appearance. A second group of muscles is arranged as circular bars which are open on the central side. Berndt (1906) thinks that these muscles contract and thus help to press out the penis. The nervous system was first described by Aurivillius (1894). On the ventral side of the seminal vesicle there is a large elongated ganglion, with a deep indentation in its middle. The lower part of this ganglion is pear-shaped and smooth, the upper part shows a low indentation. A thin nerve arises from the lower part and reaches into the area of the eye which lies above the junction between the testis and the seminal vesicle. Here is the big ganglion opticum which is club-shaped and is about half the size of the main “ganglion”. The histology of the nervous system corresponds with that of the female.

258

WALTRAUD KLEPAL

Fig. 2.—Morphology of the mature males of various genera of Acrothoracica. A, Lithoglyptes wilsoni, reduced male with long penis sheath (after Tomlinson, 1969); B, Kochlorine floridana with an orchid lobe and a long annulated penis (after Tomlinson, 1969); C, Kochlorine bocqueti, plump male with long and thin attachment organ (after Turquier, 1977); D, Cryptophialus wainwrighti, male with prominent penis and antennae (after Tomlinson, 1969); E, Australophialus pecorus, elongated male with prominent antennae (after Turquier, 1985a); F, Trypetesa spinulosa, male with a big penis sheath, orchid lobe and lateral posterior lobes (after Turquier, 1967); key to symbols as in Fig. 1; lp, lateral posterior lobe; ol, orchid lobe; scale bars C, E and F=150 µm, in A, B and D 1 cm=140 µm, 80 µm, and 70 µm, respectively.


259

The antennae arise below the two lateral lobes. They are exactly like those of the larval antennae of the Lepadidae, with the cement glands and the cement ducts. The males of T. habei are essentially like those of T. lampas. They have paired antennae. The testes extend into a sac descending into the mantle pit, while the penis sheath extends forward towards the aperture. The male may be up to 0·9 mm long, although only about 0·76 mm extends beyond the mantle surface. It is about 0·12 mm wide at the penis sheath. There are several bulbous projections at the level of the entrance into the mantle pit. The testis is at the deepest portion of the lobe in the mantle pit. It is in close proximity to the assumed nauplius eye (Tomlinson, 1969). There is a penis sheath, but whether there is also a penis is doubtful. The outer surface of the male is not ornamented. Kühnert (1934) gave a good description of the development of Alcippe lampas. Good descriptions of the anatomy of Acrothoracica are given by Turquier. This author has described Trypetesa lampas, T. nassarioides, T. spinulosa, T. habei, T. lateralis, Kochlorine bocqueti etc. (Turquier, 1970a, b, 1976, 1977). He has described the cypris of Trypetesa spinulosa and states that there is no difference from the cypris of other species such as T. lampas and T. nassarioides. The adult male of T. spinulosa is also very similar to these two species. Their size is about 0·9 mm. They have a very big penis sheath which is nearly twice as long as the “lobe orchidien” (orchid lobe). There are well-marked lateral posterior lobes which are covered by strong and short spines (Fig. 2F). These cuticular structures are important for the attachment of the male on the side of the rostral tuberculum of the female. There is no peduncle and there are no antennae in the male of T. spinulosa. In T. nassarioides the males have a short penis sheath, as long as the orchid lobe and the only dorsal lobe (Turquier, 1967). The male of T. lateralis is highly reduced and there is no penis sheath. Turquier & Pochon-Masson (1969) investigated the spermatozoa of Trypetesa nassarioides. They came to the conclusion that the development of the spermatozoa in the male is a normal process. The gamete is a sperm with a flagellum. From the original condition with two centriols only one, that of the flagellum, continues to exist. There is no trace of the proximal centriol. The loss of one of the elements of the diplosome has also been noted in some other invertebrates (see Idelman, 1967). The structure of the acrosome is not original. The flagellum has the 9+2 structure, which is, according to Baccetti, Dallai & Rosati (1968) a primitive type. There is a rigid formation on the flagellum which may slow down or weaken the flagellar movement. It is also possible that this formation is the site of important metabolic activity. Thus (with this rigid formation) the gametes can compensate the reduction of their mitochondrial apparatus. It is interesting to note that such an impoverishment of the chondriome occurs also in the vesicular spermatozoa of decapods (Pochon-Masson, 1968a, b). There is also a vesicle, which is a rare formation in spermatozoa with a flagellum. Investigations of the spermatozoa of the operculate cirripedes showed that the vesicle changes in the mature spermatozoon. It is a metabolic reserve for the gamete and could in some way be a replacement for the lacking trophic cells in the testis of Trypetesa. In Trypetesa the cyprids of both sexes and the first stages of metamorphosis are identical. Later there are sexual differences. The cypris of both sexes use cement for their attachment. Then the visceral mass moves into the sagittal plane, the complex eyes histolyse and the peduncular region remains rudimentary. Now the major part of the animal develops from the capitulum and the larval cement apparatus disappears completely. The anterior part of the larval mantle cavity disappears early, so that the upper part of the posterior mantle cavity is the only one which still exists in adult animals. From this stage onwards the male and the female cyprids develop differently. When an apparently undifferentiated cypris larva of T. lateralis settles on a female it degenerates into a bag of testicular tissue (Tomlinson, 1955). After attachment the male cypris develops into a dwarf organism of very simple anatomy with only a nervous system and a genital apparatus. Its morphology differs greatly from that of the female. The male of T. lateralis loses its

260

WALTRAUD KLEPAL

carapace and becomes rounder, averaging mm. The mature male becomes progressively smaller, wrinkled, and variable in shape. This size decrease suggests that there is destructive degeneration in this form with the male discharging its gametic contents and then dying. The phenomenon of the metamorphosis of the male cypris indicates that the sexually differentiated individual is a highly evolved and specialized organism (Turquier, 1971). This could agree with the singularity of the testis and of the vesicula seminalis and the fact that the male does not have a penis. In other species of the same genus the thorax is transformed into a penis. The male moults only once during its metamorphosis whilst in the female there are two exuviations. First in the juvenile phase of the male the rotation of the visceral mass extends and the penisAnlage is there. The “Anlage” of the testis also develops further and the vesicula seminalis becomes independent. In the adult phase the animal grows, gametogenesis continues and the genital tract is completed. The male “Anlage” differentiates very early and its evolution sets in immediately after the attachment of the cypris. Thus the male is able to reproduce about two weeks after hatching from the nauplius. The female matures only several months later. The acceleration of the evolution of the genital apparatus and the lack of a second moulting in the male (as it is there in the female) may be interpreted as a process of condensation of metamorphosis. In the adult male there is only a hypertrophied genital apparatus and a reduced visceral mass which contains only the nervous system, consisting of two ganglia, and the “yellow organ”, which may be interpreted as a rudiment homologous to the digestive tract of the female. There is no trace of any metameres nor of any appendages. The larval muscles are completely de-differentiated. The penis sheath is formed by the mantle of the posterior body region. In T. lateralis Tomlinson (1969) noticed the lack of a copulatory organ. (The majority of the males have a long extensible penis.) The semen is therefore deposited in the rostral part of the pallial opening of the female. It is probable that the spermatozoids get into the pallial cavity of the water current, caused by movements of the cirri and the rhythmic retraction of the body of the female. When Tomlinson (1969) states that the “cross structure of the male seems somewhat acellular. The cell boundaries appear to have broken down and the developing spermatozoa appear to be dispersed in the central region of the organism” it seems that he worked on a deteriorating animal. Turquier (1971) did not find anything of that sort in T. nassarioides nor in T. lampas (Turquier & Pochon-Masson, 1969). It would therefore be very surprising if T. lateralis showed such an aberrant structure. The anatomical simplicity within the males of Trypetesa is relatively constant. Thus, the males of T. lampas and of the much more primitive Berndtia purpurea are very similar. In B. purpurea the male is regularly pear-shaped which is caused by the lack of the orchidien lobe, in which there is the welldeveloped genital apparatus in Trypetesa. THORACICA Balanomorpha The sedentary Thoracica are usually free-living or commensal cirripedes. Most of them are hermaphroditic with pseudo-copulation. Cross fertilization is generally the rule. In the scalpelliform cirripedes of the Lepadomorpha males have been known for a long time. As late as 1965 Henry & McLaughlin found for the first time complemental males within the genus Solidobalanus (Henry & McLaughlin, 1965, 1967), one of the most highly evolved genera of the sessile barnacles. The males are degenerate, with vestigial cirri and the typical antennules but they do not have a mouth. The reproductive organs are well developed. Since then other Balanus species with males have been found. McLaughlin & Henry (1972) compared the morphology


261

of four species of Balanus (Fig. 3A-C). These were B. (Solidobalanus) masignotus, B. (Conopea) galeatus, B. (C.) merilli, and B. (C.) calceolus. In all species mentioned the authors noticed a deep depression in the rostrum near the sheath. It is therefore assumed that all species with a convex rostrum may have complemental males. The external morphology of the males of the four species is similar. They have a semiglobular basis, a narrow lateral band, the opercular surface with prominent opercular valves and opercular lips. Only B. calceolus is different in having separated scuto-tergal valves and raised opercular lips. Concerning the internal morphology there are differences in the degree of degeneration in the four species. All of them have a penis-thorax. The setation of the penis varies. The more setae there are on the penis the more vestigial are the cirri. Thus in B. calceolus the penis has only six to thirteen long spines. On the thorax are six pairs of prominent cirri. In B. merilli there are 25 to 46 long spines on the penis and the six pairs of cirri are vestigial. Whilst in B. calceolus the mouthparts are relatively well developed, those of B. galeatus and B. merilli are considerably reduced. Usually the seminal vesicles are paired. In B. masignotus the vesicula seminalis may be paired or simply bilobed. In most cases the testis is a diffuse bilobed structure. All four species have a strong muscular support for the thorax and the testis. There are also muscular connections to the basis in the region of the antennules. The Musculus adductor scutorum is well developed in B. calceolus and B. merilli but it is weak or lacking in B. galeatus and B. masignotus. A progressive degeneration in cirral structure and mouthparts may be traced from B. calceolus through B. masignotus to B. galeatus and B. merilli. It is evident that close examination of other species is needed before the phylogenetic significance of complemental males in the balanids can be ascertained. Dayton, Newman & Oliver (1982) found that Bathylasma corolliforme had small individuals on the opercular valves of the hermaphrodite which were acting as males. These small individuals are not just protandric; they are definitive males settling predominantly on the terga and the exposed sheath of the carina and they remain small (see also Foster, 1980). Similar results were obtained in B. alearum (Foster, 1983) (Fig. 3D). A larva destined to become a complemental male must not only settle on an established hermaphrodite but it must also select a site from which it can effect fertilization. A larva encountering an isolated receptive hermaphrodite would have a better chance of living long enough to pass its genes on to the next generation if it settled in an appropriate position on the hermaphrodite and became a precocious male. The males of B. alearum are laterally compressed. The parietes are often cracked or deformed. The opercula are embryonic and also often deformed. Mouthparts, cirri, and penis are well developed. The penis is longer than the sixth cirrus. The reproductive organs are well developed. The alimentary canal is there but in the specimen described by Foster (1983) there were no food remains in the gut. Hui & Moyse (1984) found a complemental male in Chionelasmus darwini (Fig. 3E). In Chelonobia patula Crisp (1983) discovered what he called “apertural males” (Fig. 3F-H). These individuals settle in the opercula region of the hermaphrodites and are themselves really hermaphrodites being arrested in development. In contrast to most complemental males and dwarf males the males of Chelonobia are capable of feeding and growing and in contrast to all males described earlier they have the potential to become hermaphrodites. Crisp (1983) assumes that the hermaphrodite has the capability of reducing the growth rate of surrounding complemental males by high competition for food. Somatic growth is limited by food supply, whereas gonad and penis development proceeds more or less independently as a function of age. C. patula seems to be a pointer to the evolution of the complemental male (Crisp, 1983). It is a protandrous hermaphrodite. As long as the individuals are smaller than 2 mm in diameter they do not have a penis. The penis then develops and when the individuals are between 4 and 7 mm in diameter they are protandric males and at greater than 7 mm basal diameter they are simultaneous hermaphrodites. Individuals with a basal diameter between 2 and 4 mm have testes, vesicula seminalis and penis, but no

262

WALTRAUD KLEPAL

ovary. Their penis is abnormally well developed so that it can be assumed that these small individuals would be capable of inseminating the hermaphrodite. Lepadomorpha Within the Lepadomorpha species of the Thoracica there are a great number of dwarf or complemental males. Up to 1908, 170 species of Scalpellum had been described and intermediate types of males had been found (Pilsbry, 1908). Since then several more species of Scalpellum and other Lepadomorpha genera have been described. Repeatedly authors tried to group the various genera according to their systematic positions. Pilsbry (1908) was the first who suggested that the characters of the males should be considered just as those of the females and the hermaphrodites for the elucidation of the systematic position of the various scalpelliform species. Nilsson-Cantell (1921) follows essentially Pilsbry’s classification of the species, but he points out the great variability of the dwarf males. Therefore Nilsson-Cantell states that Pilsbry’s (1908) classification cannot be regarded as being satisfactory for all species. Where Pilsbry talks of “genera”, Nilsson-Cantell uses the term “groups”. Pilsbry (1908) distinguishes four different genera by the plates, by the degree of separation between capitulum and peduncle, by the presence of a mouth and alimentary system and by the condition of the cirri. Calantica has the most original complemental males. This genus comprises C. villosa, C. trispinosa, C. eos, C. calyculus, C. falcata, C. gemma, C. superba, and C. grimaldi. The complemental males of the genus Smilium are like those of Calantica only the females or hermaphrodites may be distinguished by the elevation of a pair of latera above the basal whorl in Smilium. This genus comprises S. peronii, S. uncus, S. pollicipedoides, S. aries, S. sexcornutum, S. scorpio, S. acutum, and S. longirostrum. The genus Euscalpellum differs from the two preceding ones by having more degenerate males. The males are saclike, not distinctly divided into capitulum and peduncle and they have only three plates, the scuta being larger than in Scalpellum. They have six pairs of articulated cirri and a mouth. The genus Euscalpellum comprises E. rostratum, E. renei, E. bengalense, E. stratum, E. squamuliferum. In the genus Scalpellum the males are very degenerate, sac-like without a peduncle or mouth, vestigial cirri or digestive tract. The plates are absent or scuta and terga are extremely small. Within the genus Scalpellum Pilsbry (1908) distinguishes three groups. (1) Group of S. scalpellum. To this group belong S. stearnsi, S. inerme, S. calcaratum, S. hamatum, S. scalpellum, S. patagonicum, and S. salartioe. (2) Group of S. californicum with S. californicum and S. osseum. (3) Group of S. stroemii. This group comprises S. stroemii, S. stroemii obesum, S. s. luridum, S. s. aduncum, S. s. septentrionale, S. s. substroemii, S. s. latirostrum, S. pressum, S. groenlandicum, S. angustum, S. nymphocola, and S. cornutum. Within the genus Scalpellum the subgenus Arcoscalpellum with two sections may be distinguished. To the stock of Arcoscalpellum belong S. velutinum, S. idioplax, and S. carinatum. The section Mesoscalpellum includes S. intermedium, S. nipponense, S. laccadivicum, S. japonicum, S. larvale, S. gruvelii, S. imperfectum, and S. sanctaebarbarae. The section Neoscalpellum comprises S. edwardsi, S. dicheloplax, S. phantasma, and S. marginatum. Both Calantica and Smilium have complemental males that look like free living juveniles (Fig. 4A, B, Table I). Capitulum and peduncle are well separated from each other. They have six large primary capitular plates. In addition to them there may be a number of small latera so that altogether there may be 15 plates


263

Fig. 3.—Morphology of adult males in Balanomorpha. A, generalized male of Balanus; B, Balanus calceolus, penis-thorax with cirri and mouthcone; C, Balanus masignotus, penis-thorax with vestigial cirri; (Figs A-C after McLaughlin & Henry, 1972); D, Bathylasma alearum, male in tergal niche of a larger specimen looking like a young hermaphrodite (after Foster, 1983); E, Chionelasmus darwini, male with right cirri and shell removed (after Hui & Moyse, 1984); F, Chelonobia patula, arrows indicating pits within junction of radius and paries in which cyprids settle; G, hermaphrodite with apertural males, looking like young hermaphrodites; H, apertural male with long penis, removed from shell; (Figs F, G, H, after Crisp, 1983); a, antenna; am, apertural male; ap, aperture; b, basis; c, cirri; me, mouth cone; ol, opercular lips; p, penis; pt, penis thorax; s, scutum; t, tergum; scale bars, A=0·1 mm, B=0·05 mm, C=0·1 mm, D=2 mm, E=0·4 mm; F=5 mm; G=0·5 mm, H=1 mm.

264

WALTRAUD KLEPAL

on the capitulum. Both genera have six pairs of articulated cirri and at least in Calantica there is a long extensible penis that may be beset with setae as in C. trispinosa. The penis is relatively longer than in a juvenile hermaphrodite of the same size, e.g. in C. villosa, C. studeri (Foster, 1978). The digestive tract is functioning (Nilsson-Cantell, 1931). The males are relatively big, e.g. 3 mm long and 1·8 mm wide as in Smilium sexcornutum or 1·8 mm long and 1·3 mm wide as in Calantica trispinosa. In Smilium the capitulum may be up to 5·8 mm long. Thus it exceeds in size many free-living juveniles attached to penduncles of larger specimens. In Smilium sexcornutum of the large plates the one scutum consists of two well-separated pieces (Krüger, 1911), which may point to the original condition. Krüger (1911) found up to three males attached to one female of Calantica, on its integument between the scuta, below the adductor muscle. In general the occurrence of males is rare in Calantica. Foster (1978), however, found up to 16 males between the scuta of large hermaphrodites of C. spinilatera. In the main features described above the males of some species of Scalpellum resemble those of Calantica and Smilium. The males of Scalpellum squamuliferum, S. peronii, S. villosum (now Calantica villosa), Scalpellum pilsbryi, S. scorpio, and S. longirostrum all have peduncle and capitulum separated from each other (Fig. 4C). They are relatively large (at least 1 mm long and sometimes up to 1 mm wide), they have six larger plates on the capitulum (there may be additional smaller ones as the five in S. pilsbryi). These males have six pairs of cirri, usually articulated. They may be well developed as in S. squamuliferum (Darwin, 1851) or they may be relatively short as in S. peronii, S. villosum, S. pilsbryi, S. scorpio, and S. longirostrum. In these last five species at least the first cirrus may be well separated from the following ones. All males of Scalpellum species have a penis which is well developed and extensible as in S. squamuliferum and in S. pilsbryi. In S. villosum the penis is described as being blunt and in S. longirostrum it is short. With the exception of S. villosum all males of the Scalpellum species already discussed have caudal appendages. All these species mentioned above may really belong to the genus Calantica as e.g. Scalpellum villosum (now Calantica villosa) or Scalpellum pilsbryi which is according Bocquet-Védrine (1971) identical with Calantica calyculus, or they may belong to Smilium as does Scalpellum peronii. In all cases the males are complemental males. They are attached to the integument of the hermaphrodite in a fold in the central line between the scuta, a little below the Musculus adductor scutorum. Thus, they are some way below the umbones of the scuta, and they get protected by them when they are closed. Many species of Scalpellum are known but even when males have been found they have often only been described very superficially. Some authors like Annandale (1910) and Stewart (1911) have described the males more accurately. The anatomy of the complemental male of S. squamuliferum was studied by Stewart (1911). This author concentrated on the digestive tract, the reproductive system and the tissues of the peduncle. Annandale (1910) gave an account on the “dwarf” males of several scalpellids. His main interest was the morphology. According to Stewart (1911) the adult male is 1–1·4 mm long and a maximal 0·7 mm wide. The peduncle is short, stout and well separated from the capitulum. On the capitulum there are six calcified plates; rostrum, scuta, terga, and carina. There are no peduncular plates and no latera as in the hermaphrodite. Annandale gave a precise description of the plates. The tergum is broadly triangular, the base of the triangle is rounded and the apex is pointing directly downwards. The scutum is much larger than the tergum and more narrowly triangular, with the apex pointing upwards. The carina is triangular, has a rounded base, is not quite as broad as the tergum, a little larger than the tergum, but not reaching upwards as high as the upper margin of this plate. The base is slightly lower than that of the scutum and above the apex of the rostrum. The rostrum is of about the same length as the tergum. It is rather broader than the carina and with the base produced to a point.


265

The outer surface of the male is covered with simple hairs. They consist of an outer cortical portion (staining intensely with iron haematoxylin) and a core which does not stain. The cirri and penis are well developed, caudal appendages are present. The mouthparts resemble those of the hermaphrodite. They are generally smaller, there is no labrum (this may be doubted—it was probably lost in preparation) and the inner teeth of the mandibles are not so distinctly separated. Thus the male belongs to the type most commonly found in the subgenus. The alimentary canal is a functioning narrow tube. Around the oesophagus there is a layer of circular muscles, at the anterior part of the narrow stomach there are two simple epithelial tubes, the coeca, which run for a short distance forward along the oesophagus. Stewart (1911) did not describe any special muscles of the rectum. The anus is dorsal to the base of the penis. The reproductive organs are tubular and consist of the testes, the seminal vesicles and the ductus ejaculatorius with the penis. The different regions are mainly distinguished by the nature of their contents. In all cases the wall consists of a fine layer of endothelium, only close to the external aperture is there a sphincter. The histology and development of the organs of the peduncle were described in more detail. There is a rostral duct in the peduncle which in the male reaches only from the root of the prosoma to the upper quarter of the peduncle. This system of spaces forms an erectile tissue by which the animal can move its peduncle in a swaying manner or by which it elongates or shortens its peduncle. The connective tissue cells within the peduncle are often large with a rounded nucleus. They may be full of reserve granules (“yolk granules”). In this case Stewart did not find it easy to distinguish them from the cement cells. Reserve material is also found in cells around the stomach and around the ventral nerve cord in the prosoma. The cement cells in the adult male are almost spherical with a spherical nucleus per cell. In their protoplasm there are large granules of irregular shape. Because the staining reaction of these granules was identical with that of yolk Stewart (1911) poses the question whether these granules may be yolk granules which are in the process of conversion to form cement. In unstained males these cells are yellow and thus they could be the “cellules jaunes de la pedoncule” of Gruvel (1905, p. 448). Stewart (1911) also gave a general outline of the development of the male. TABLE I Comparison of the males of Calantica, Smilium, Scalpellum and Pisiscalpellum: under cirri, 6=six pairs normally spaced, 1/2– 6=space between first cirrius and the rest, 1/2–6 R=space between first cirrius and the rest but all reduced; D=distinct; J.d.=just distinct; N.d.=not distinct; E=extensible; Fl=functionless; O.f.=open functional; L=large; S=small; P=plates normally developed; p=plates reduced; R=reduced; += present; –=no information; ? =doubtful. Species Calantica trispinosa Smilium sexcornutum Scalpellum squamuliferum S. peronii S. villosum S. pilsbryi

Size (mm) CapitulumPeduncle

1 1

Alimentary Canal Plates Cirri

Penis Caudal appendages

D

O.f.

6P

6

+E

–

D

O.f.

6P

6

?

–

D

O.f.

6P

6

+E

L

D D ?

O.f. O.f. O.f.

6P 6P 6P

1/2–6 + 1/2–6 + 1/2–6 +E

S − S

266

WALTRAUD KLEPAL

Species

Size (mm) CapitulumPeduncle

Alimentary Canal Plates Cirri

Penis Caudal appendages

S. scorpio S. longirostrum S. bengalense S. gigas S. hoeki S. striatum S. luteum S. intermedium S. kurchatovi S. chiliense S. galapaganum S. (S.) elongatum

? ? ? ? ? ?

O.f. R O.f. O.f. O.f. O.f. O.f. ? ? ? ? ?

+ + ? ? ? ? ? ? ? ? ? ?

S. ornatum S. rutilum S. stearnsi S. discoveryi S. gracile S. tritonis S. wood-masoni S. retrieveri S. vulgare S. rostratum S. alcockianum S. chitinosum S. compactum S. compression S. condensum S. convexum S. crinitum S. distinctum S. fissum S. gibberum S. gruvelianum S. hexagonum S. javanicum S. projection S. regium S. sessile

0·75 ? ? ? ?

J.d. J.d. N.d. N.d. N.d. N.d. N.d. N.d. N.d. N.d. N.d. J.d.

? ? ? 0·8 ? thorax 0·4 µm 1·23 1 0·67 1.3 0·67 0·8

1·17 ?

2·2

N.d. ? N.d. N.d. N.d. ? N.d. N.d. N.d. D N.d. N.d. N.d. N.d. N.d. N.d. N.d. N.d. ? N.d. N.d. N.d. N.d. N.d. ? J.d.

6P 1/2–6 6P 1/2–6 4P ? 4p 6R 4p 6R 2P2p 6R 4p 6R 4p 6 4p ? 2P2p ? 2P2p ? 4p ? – ? ? ? ? ? ? ? – O.f. ? ? ? ? ? ? ? ? ? ? ? ? ? ? Fl ?

4p 4p 4p 2P2p 2P2p 4p 4p 4p 4p 3p – – – – – – – – – – – – – – – –

4 ? ? ? ? ? ? ? 4 1/2–6R – ? ? ? ? ? ? ? ? ? ? ?

L L ? S ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? + ? ? ? ? ? ? ? ? ? ? ? ? ? ? – ?

Setae ? ? ? ? ? ? ? ? S ? ? ? ? ? ? ? ? ? S ? ? ? ? – ?


Pisiscalpellum withersi

0·4

N.d.

?

–

?

?

267

?

After fixation the cypris of the male is shorter than high. Its peduncular region is greater in comparison with the capitular region. Its total length is 0·75 mm. The mouth is open but there is no anus. The stomach contains excretory matter. The six valves are clearly visible through the shell. In the young adult the capitulum is much longer than in the cypris. The testes of the hermaphrodite may be seen lying at the sides or ventral of the stomach. In the cypris of the hermaphrodite the testes appear “rudimentary”. Stubbings (1936) states that the larvae of S. squamuliferum are of two sizes. The shorter one is said to be the male, whilst the longer one is that of the hermaphrodite. (This is in contrast to the cyprids of the Rhizocephala, in which the larger cypris will become the male.) Thus, as Stewart (1911) states “maleness is therefore not the result of the position of attachment” (Stewart, 1911, p. 38) as was assumed by Smith (1906). Smith considered a position on the margin of the pallial aperture to be the cause for the nondevelopment of the hermaphrodite character in the “males”. There is no description of the nervous system of the male. Stewart (1911) only states that the hairs on the outside of the capitulum are supplied with nerves. Whilst the hermaphrodites of S. squamuliferum and S. bengalense resemble each other closely, the males differ to a high degree. In the male of S. bengalense there is no marked outward boundary between the peduncle and the capitulum. There are only four valves present. The capitulum and the peduncle are covered by hairs, there is a broad band of larger hairs (having a bifid tip) on either side of the carinal midline. The mouth cavity is small in comparison with the capitulum, whose wall is relatively thick. Therefore the true body is also small. The adductor scutorum muscle persists in its usual situation, although in the majority of the specimens there are no scuta. The muscle may be able to narrow the opening of the pallial cavity. The alimentary canal has the usual V-shape. The mouth and the anus are open. The stomach is a fairly large sac, two coeca arise from the anterior end of the alimentary canal. The nervous system consists of a pair of large cerebral ganglia and a massive ventral cord. The hairs covering the outer surface fulfil a sensory function. There is also a small ganglion in the outer wall of the pallial cavity in the carinal midline. It consists of a single row of cells which are continuous with the epidermis. A thin layer of intensely staining nervous matter spreads out from the nuclei on either side under the bands of larger hairs. The cuticle is described as being either entirely absent or very much thinned over the ganglion. Stewart (1911) did not describe the reproductive system. He only states that the testes are on either side of the stomach. At the root of the prosoma there is a somewhat indefinite space which may possibly be the equivalent of the rostral duct. The cement glands consist of two groups of cells. Many of them are full of large irregular yolk-like granules. Stewart could not find any ducts. As in S. squamuliferum there is a great deal of reserve material (“yolk”) in the peduncle. It is more concentrated above the cement glands. The material is within vesicular cells with flattened nuclei. The adult male of S. gruvelii is pear-shaped. The animal Stewart (1911) described has the anterior half thicker than the posterior. The males are about 1 mm long and 0·5 mm wide. The antennae are in the same position as in the larva. The pallial cavity, lined by a fine epidermis, and the prosoma are much reduced. The alimentary tract consists of a small hollow ball of cells containing some cuticular and excretory matter (the cuticular matter could come from some prey). The nervous system is much reduced. The cerebral ganglion is on the opposite side of the stomach from the reduced ventral nerve cord. The sensory hairs are supplied with nerves in the

268

WALTRAUD KLEPAL

Fig. 4.—Morphology of adult males in Lepadomorpha. A, Calantica calyculus (after Bocquet-Védrine, 1971); B, Smilium peroni (after Krüger, 1914); C, Scalpellum squamuliferum (after Annandale, 1910); D, Scalpellum rostratum (after Darwin, 1851); E, Scalpellum wood-masoni (after Nilsson-Cantell, 1931); F, Scalpellum regium (after Thomson, 1873); in Figs A–C capitulum and peduncle may be distinguished; in Figs D–F the shap is ovoid and the plates get more and more reduced, at the same time the cuticular structures on the outer surface become denser and they are arrenged regularly; a, antenna; ca, capitulum; cs, cuticular structures; pe, peduncle; pl, plates; scale bars, A=1mm, C=0.5mm, D=0.2mm, E=0.5mm, F=1mm.


269

same way as the hairs of S. squamuliferum. Muscle fibres run from the anterior to the posterior pole of the animal. The testes are heart-shaped, and the vesiculae seminales were not seen by Stewart (1911). It is thought that the seminal vesicles may empty themselves by contraction of the longitudinal muscles. At the same time the prosoma carrying the ejaculatory duct is thrust through the opening of the pallial cavity. Thus the tension in the body is raised. The cement glands consist of two or three small spherical masses of cells lying posterior to the testis. As in the species described above the entire surface of the capitulum and the peduncle is covered with hairs. These are bilaterally distributed. They are parted in the carinal and rostral midlines, the hairs of each side being directed towards the rostral line. The hairs in S. gruvelii are simple and not so rigid as in S. squamuliferum and S. bengalense. Stewart (1911) also describes the post-larval development of the male of S. gruvelii. Whilst the cypris larva of this species resembles closely that of S. regium, the pupa resembles in outer form the adult male of S. velutinum (Gruvel, 1902a). Thomson (1873) described the (complemental) male of S. regium in his notes from the Challenger expedition. This author found five to nine males attached to the occludent margins of the scuta of the fully developed form. (According to Thomson the fully developed form is a female, whilst according to Hoek (1884) it is a hermaphrodite.) Thomson describes the males as being oval and sac-like whilst Hoek (1883) talks about a cylindrical shape of the males. The size is up to 2·5 mm in length and 0·9 mm in extreme width. There is a peduncular and a capitular pole, the antenna being placed on the first one, a little distant from the extremity on the ventral surface. There are no plates and the thorax is not jointed (Fig. 4F). The body wall is described as being “chitinous” (Hoek, 1884), thin and delicate. It is beset with spines in transverse rows. The spines are narrow and pointed where they attach to the wall of the body and they are broadest at the other extremity. The free margin is deeply toothed and thus the spines resemble the scales of Lepidoptera. In other places they look like combs. The epithelium consists of flat cells “with indistinct limits” (Hoek, 1884) and with conspicuous nuclei. Under the hypodermis cells a well-developed layer of muscle fibres is present everywhere. These fibres are transversely striated. The striations are in part indistinct which could have to do with the fibres being nearly functionless and rudimentary. The muscular fibres have an irregular oblique direction which in some parts approaches to a transverse, in other parts to a longitudinal position. The connective tissue consists of fibres and of delicate and finely granulated membranous plates which form the partitions between the large meshes. The occurrence of a well-developed mass of connective tissue between the different organs within the body is the rule in all cirripedes. Hoek (1884) described the mouth as functionless. According to him the oesophagus is a narrow tube which widens and passes into the stomach, whilst Thomson (1873) did not see any oesophagus or stomach. (This difference in the descriptions could be due to a possible variability in the animals.) When the stomach is present it is a pyriform pouch closed on all sides with a rudimentary intestine. In a fully grown male the stomach is almost empty, in a younger individual it is filled with a yellowish brown coloured mass of fatty nature, probably reserve material. The supraoesophageal ganglion is situated against the oesophagus. The commissure unites this ganglion with the large thoracic ganglion. The latter represents the whole ventral nerve cord and is attached before the supraoesophageal ganglion. It is elongated ovoid. The only distinct nerve coming from the thoracic ganglion arises terminally. Two stronger nerves arise from the commissures very close to the supraoesophageal ganglion. All other nerves given off from both ganglia are extremely delicate and hardly recognizable as such.

270

WALTRAUD KLEPAL

The testis is heart-shaped with the incision being directed towards the capitular extremity. This incision is the only sign of the original duplicity of the male genital gland. The vesicula seminalis is an irregular globular vesicle of a diameter of 0·3 mm. It is very closely pressed against the testis. The vesicula seminalis is only a dilation of the vas deferens at the place where it corresponds with the testis. The cylindrical terminal position of the thorax may be called a “penis”. There are six pairs of cirri. Four of them, the posterior ones are well developed and have two rami. In the first and second pair of cirri only one short branch is left. The rami of the fourth to the sixth cirrus are relatively long and narrow and they terminate in two or three very long spines. Each male has a pair of cement glands. They are of an ovoid shape and they are situated a little above the vesicula seminalis. The glands are comprised of very large cells with granular contents and a nucleus. The large granules in the cells are placed at the periphery. Cement ducts come off the glands as thread-like appendages. The male of S. regium has a highly degenerated organization. The elongated body has the shape of a bag with a slit representing the opening between the two scuta in other species. Only the antennae show their original condition, the cirri are straight and functionless, the mouthparts have disappeared. The intestine is rudimentary and has become functionless. The nervous system consists of a relatively small supraoesophageal ganglion, of a not very stout oesophageal ring and of a large thoracic ganglion. Probably the latter alone regulates the functions of the genital apparatus. The peripheral part of the nervous system is not much developed. There are no eyes or other sensory organs. The genital apparatus is, apart from the cement glands in the young males, the only well-developed organ system. The female organs are lost and the male organs show a great deal more concentration than do the same in ordinary hermaphrodite cirripedes. There is only a single testis and a single vesicula seminalis. In all these respects the males of other deep-sea species of Scalpellum, investigated by Hoek (1884) exactly correspond to the male of S. regium. So does the male of S. vulgare (a specimen from the Mediterranean) with the exception of the presence of rudimentary plates, which in that species, as in some of the deep-sea species, represent the so-called primordial valves of the young capitulum of pedunculated cirripedes. S. regium is in many respects identical with S. gruvelii. Darwin (1851) described the complemental male of S. vulgare as flask-shaped, whilst Gruvel (1898) stated that it had the shape of a “Punica granatum” and in his more precise description (Gruvel, 1899) he stated that the complemental male had the shape of a more or less rounded sac. Each male is attached to the hermaphrodite on the inside of the scuta in something like a fold which Darwin calls a “transparent spine-bearing chitine border”. There may be up to 15 males on one side, in which case they are attached so closely to each other than they have no longer their ordinary shape. The males are never found anywhere else on the hermaphrodite as Gruvel (1899) stressed. On the anterior side of the male there are eight lobes arranged in two concentric rows. There are four rudimentary, bead-shaped and calcareous plates. They are believed to be the scuta and terga, although they are placed considerably below the orifice. The eight lobes are simply evaginations of the outer cuticle of the pear-shaped body. The most backward ones are encrusted with calcium. At their base they are covered by short chitinous spines. These are homologous to the calcareous plates of the hermaphrodite. On the cuticle of the males there are small setae which form little hooks; they are united into groups of two to six, rarely more, and arranged irregularly over the whole external surface. These cuticular structures are very numerous in the anterior part of the animal, as well as on the lateral lobes. There are no cuticular structures on the antennae. In the body proper three regions may be distinguished: the head region, the thorax, and the abdomen. The head region is very much reduced. It is just a simple projection without any mouth or mouthparts. Darwin


271

(1851) describes the thorax as being within an inner sac or tube, about 0·4 mm long. It is transversely wrinkled and so extensible that it stretches to twice its former length. Darwin describes four pairs of short limbs on the thorax, whilst Gruvel (1899) talks about six pairs of cirri. This difference is due to the variability of the males. In the animals Darwin (1851) describes the bases of the limbs almost touch each other. The limbs have no articulations except where they unite with the thorax. The anterior smallest limb supports two or sometimes only a single spine. The second pair of cirri has three spines a little shorter than in the first pair. The two upper pairs of cirri are alike. They have two spines on their tips and a third lower down on a notch on the outer side. It is possible that the spine on the notch marks the point where in the larva there is an articulation. In the animals Darwin investigated there were 11 pairs of spines on the thoracic limbs. The spines are straight, long and not plumose. According to Darwin (1851) the four pairs of cirri must correspond to the four posterior cirri as may be inferred from their proximity to the abdominal lobe and from the three posterior pairs, closely resembling each other and differing a little from the first pair. The first pair corresponds with the third pair in the hermaphrodite. Gruvel (1899) describes the six pairs of cirri as looking all alike. There is a basal part (presumably the pedicel) on to which one or two setae-like cirri insert, which get smaller from the first to the last one. The first pair of cirri carry a single long flexible seta, absolutely smooth and ending in a fine tip. The second and third pairs have three setae, two of which are equally long and next to each other and one is a little more lateral and very short. The fourth pair of cirri does not have the two first setae of the previous pairs and the fifth and sixth pair do not have the smooth setae like the first one, but a very short one, flat and bigger at its base. At the level of the fifth and sixth pairs of cirri there are on the ventral surface three cuticular spines which are simply ornaments. The abdomen is reduced to two small cones (Gruvel, 1899). Darwin (1851) talks about one square abdominal lobe. According to the latter author there are three pairs of spines on the ventral surface, which may probably mark the three segments, which are distinct on the abdomen of the larva in the last stage of its development in Lepas and in other genera. On each of the posterior angles of the abdominal lobe there are three moderately long, very sharp spines with the tips of the outer pair bent a little inwards. Gruvel (1899) states that on each abdominal cone there is a strong short seta, similar to those of the fifth and sixth pairs of cirri. Within the thorax there are some longitudinal muscles without transverse striations which enter the short limbs but not the abdomen. At their lower ends these muscles terminate abruptly. They extend a short way beneath the lower pair of limbs and are attached to the outer integument of the animal near the base. Gruvel (1899) describes distinct muscle bundles which are orientated longitudinally. They are united by connective tissue. The arrangement of these compact muscles varies a little depending on whether they insert on the posterior or the anterior part of the sac, where some of them sometimes insert with a broad base. The contraction of these muscles causes a pulling together of the anterior and posterior part of the sac and the closure of the orifice. When the contraction ceases the orifice opens through the elasticity of the sac. There are no circular and no oblique fibres. The inner mantle surface is covered by a very thin cuticle. There is neither a mouth nor a digestive tract and no mouthparts. Darwin (1851) did not see any anus but he thinks that it may exist. The genital apparatus consists of a well-developed testis and the vesicula seminalis. Both organs are only single. (In the hermaphrodite these organs are paired.) There are no muscles in the genital tract, nor are there any muscles in the ejaculatory duct. The wall of the latter is a simple epithelium. The genital opening is between the caudal appendages on the ventral side of the body. The sperms of the male are identical with those of the hermaphrodite. The thorax serves for the emission and first direction of the spermatozoa and

272

WALTRAUD KLEPAL

“hence perhaps its singularly extensible structure” (Darwin, 1851). The longitudinal and transverse muscles lining the upper part of the outer integument of the whole animal could perhaps cause the violent expulsion of the spermatozoa, thus causing them to reach the appropriate area within the hermaphrodite. There are two ganglia in the male: a double one before the head projection and a single one in the median line of the ventral side of the thorax (Gruvel, 1898). From the latter small nerves lead into the appendages. The two ganglia are united by lateral connectives. From the cervical ganglion there lead two small connectives towards the anterior side of the body. They widen into a small nerve cell in which a nucleus may be seen and then they lead into a pigmented mass, which Gruvel (1898) believes to be the eye. Within the eye there is an anterior-posterior partitioning which indicates that originally the eye consisted of two parts. In the anterior part there are two nuclei of the pigment cells and enclosed in these cells there are two small elongated vesicles which surround a certain number of refringent rods which are orientated towards the outer opening. Gruvel (1899) states that the eyes and the setae are the only sensory organs. Darwin (1851) talks about a single eye only. According to this author the eye has a pointed oval form, consisting of an outer capsule, lined with purple pigment cells and surrounding something that looks like a lens. Gruvel (1899) describes the cement glands as being found on either side of the animal. There are two big glands of short and stout cells with big nuclei. In the histological sense they resemble the pancreatic glands described in other cirripedes. There is a big lumen in the gland from which a canal arises and which penetrates the base of the antenna. The antennae with the cement ducts are well developed. Each antenna consists of two limbs, the basal one is long and has a lateral seta, the other is very short and has a very small external appendix with very small and simple setae. The second segment of the prehensile antenna, the disk, is pointed and hoof-like (Darwin, 1851). There is a simple backward pointing spine, attached on the under side, nearly opposite the articulation of the ultimate segment. At the apex there are some excessively minute hairs or down. The ultimate segment projects rectangularly outwards as usual and has on its inner side a conspicuous notch, which bears two or three long, non-plumose spines. On the outside of the large basal segment there is a single spine curving backwards. The male of S. ornatum is very similar to that of S. vulgare. Darwin (1851) described the complemental male of S. peronii, as did Gruvel (1901, 1902a) and also Krüger (1914). Krüger talked about Smilium peronii. There are up to three complemental males within each hermaphrodite. The largest male found is 0·9 mm long and about 1·0 mm wide (from the tip of the tergum to the tip of the rostrum) and it is about 0·6 mm wide from the lowest corner of the rostrum to the corresponding one of the carina. Peduncle and capitulum may be easily distinguished. On the capitulum there are six plates: two terga, two scuta, a carina, a rostrum. All plates are well developed, and they are united by a finely-villose membrane, furnished near the orifice with much longer and thicker spines. The capitulum has the orifice not in the same line with the peduncle but almost transverse to it and therefore it is almost parallel to the surface of attachment. According to Darwin (1851) the scuta and terga are broadly oval with the primordial valves very plain at their upper ends. The carina is straight, triangular and internally slightly concave. The rostrum is shorter and internally more concave than the carina. The capitulum and peduncle are covered by fine setae. The narrow and very short peduncle commences a little below the scuta. The base is flat and truncated. There are longitudinal and oblique muscle strands in two layers. In the capitulum there is the M. adductor scutorum. The animal has a well-developed mouth and well-developed mouthparts. The labrum is highly bullate, as in the hermaphrodite. It is far removed from the M. adductor scutorum. The palpi are small and triangular. Their apices are blunt and clothed with very few scattered bristles. The mandibles have only three teeth, the lower angle is slightly pectinated. The first tooth is some distance from the second and larger than it. The maxillae bear only a few spines, furnished with a long apodeme. Beneath the upper large pair, there is a


273

notch under which there are two spines of considerable size and a small tuft of fine bristles. The relative size of the maxillae and mandibles are the same in the male and in the hermaphrodite. The outer maxilla is blunt, triangular with a few thinly scattered bristles on the inner face, those on the outside being longer. The digestive tract consists of the mouth, a short oesophagus, the stomach, a very narrow intestine and an anus. It is functional; diatoms were found inside the gut. The well-developed prosoma has six pairs of cirri. The first pair is far removed from the second, the rami are very short, barely exceeding the pedicel in length. They are formed of four segments, each bearing a pair of spines. On the end of the terminal segment there are three spines, the central one of these is very long. The second pair of cirri resembles the first pair and is also short. The third, fourth, fifth, and sixth pairs of cirri get gradually longer. All cirri have setae. In the sixth pair five of the six elongated segments have three pairs of long spines. The dorsal tufts are large. The cirri are furnished with transversely striated muscles just as in the hermaphrodite. The number of podomeres in the cirri of the male is variable, and considerably smaller than in the hermaphrodite. Krüger (1914) gave the number of podomeres in the cirri of the male as between three and five. The caudal appendages have no articulations, they are minute plates with a few bristles at their apices. The testes are paired and very irregular in their shape. The vesiculae seminales are long; they have muscles which help to extrude the sperms. There is a short penis, extending only up to the pedicel of the sixth pair of cirri. It has four bristles at its end. Gruvel (1902a) believes that he may have been seen immature ovarian cells, but nobody before or after him has seen such cells. The nervous system consists of a dorsal mass, situated at the base of the mouth cone. This is the cerebral ganglion which consists of two lobes. From this two small nerves arise. These go past the stomach and they are probably the two peduncular nerves. Inside these there arise two other nerves, which are certainly optical nerves. The cerebral ganglion is united with a long thoracal mass, the thoraco-abdominal chain, which is very much condensed and serving the mouthparts and the cirri. The eye in the central line between the scuta is a single and simple pigmented mass, which contains refringent bodies. It resembles very much that of Scalpellum vulgare. The cement apparatus consists of two lobed glands with large nuclei and several nucleoli. The cells are very granular and they have a free central space into which the secretory products are passed. According to Gruvel (1902a) the complemental male of S. peronii resembles closely the hermaphrodite. The hermaphrodite itself resembles closely individuals of the genus Pollicipes, an ancient type of cirripedes. Scalpellum villosum is broader and considerably higher than the male of S. peronii. The orifice of the capitulum is placed obliquely; the membrane connecting the plates is finely villose, furnished with spines, conspicuously thicker and longer than those on the male of S. peronii. The peduncle is naked, narrow and short. The capitulum is about 0·1 mm long and wide. Of the six plates the scuta and terga are much more elongated than in S. peronii. The carina descends only just below them. The rostrum is a little broader and more arched than the carina. The primordial valves are seated on the tips of the scuta, terga, and carina, but not on the rostrum. Scuta, terga, and carina of the male resemble the same plates in the hermaphrodite much more closely than in S. peronii. In both species the rostrum is of large relative size in the complemental male. This is a remarkable character which is difficult to explain (Darwin, 1851). There are six pairs of cirri. The first pair is short with only three or four podomeres in each ramus. The second cirrus has the basal segment not very thickly clothed with spines. The sixth cirrus, has seven segments, not protuberant in front, each bearing four pairs of spines without intermediate tufts. There are no caudal appendages as in the hermaphrodite. The disk of the prehensile antennae is narrower than the basal segment, only slightly pointed (different in S. peronii). At the distal end on the inner side there are two or three spines at the same place

274

WALTRAUD KLEPAL

where there are usually excessively minute hairs in some or all other species of Scalpellum and Ibla. The digestive tract is functional. The labrum is bullate with teeth on the crest, the palpi are blunt and spinose. The mandible has three teeth. Its inferior point is rather strongly pectinated. The maxilla has a considerable notch under the upper pair of large spines. The inferior part of the edge is not prominent. On the inner edge of the outer maxilla spines are arranged in two groups. The penis is thick, exceeding the length of the pedicel of the sixth cirrus. It is square at the end and furnished with some spines. The complemental males of Scalpellum villosum and S. peronii closely resemble each other, more so than the corresponding parts in the two hermaphrodites. S. pilsbryi is probably identical with Calantica calyculus (see above). The size of the male is less than 1 mm and there are 11 plates on the capitulum. These are one carina, two terga, two scuta, one rostrum well developed, one subcarina and two pairs of lateral plates. There are six pairs of cirri, the separation between the first and the subsequent pairs of cirri being less clear than in the hermaphrodite. The caudal appendages are not articulated, with one very long seta and four very short ones. The digestive tract is functional. The mouthcone is like that of the hermaphrodite. The penis is beset by setae, less numerous than in the hermaphrodite. In Scalpellum scorpio the peduncle and capitulum can be distinguished, both being beset by fine hairs. There are no real segments. The six plates are scuta, terga, rostrum and carina; the carina is the longest and is weakly bent, the rostrum is smallest, but bent more. The first pair of cirri is far away from the second. Each ramus consists of five podomeres; the rami are shorter than in the hermaphrodite. The caudal appendages reach only to the middle of the proximal segment of the protopodite and they are like those of the hermaphrodite. The digestive tract is functional. The mandible has four teeth, the next to the outermost is smallest. The mouthparts are similar to those of the hermaphrodite except that the maxilla has less spines. The penis has nearly half the length of the sixth pair of cirri. At its distal end it is much thicker than the cirri at that height. This is a difference between the male and the hermaphrodite (Aurivillius, 1894). The male of S. longirostrum is similar to S. peronii and S. villosum (Gruvel, 1902a). It has a laterally compress capitulum. The cuticle on the capitulum has short setae, arranged irregularly. The peduncle is short without any setae, or the setae are here very much reduced. Since the peduncle is very short the cement glands are entirely at the base of the capitulum. The shape of the six plates is very different from those of S. peronii or S. villosum. The cirri are very similar to those of S. peronii and S. villosum. Each ramus consists of three podomeres and there are three to four setae on the basal limbs. The caudal appendages consist of one limb only which is cylindrical and there are two terminal setae. The alimentary canal is reduced to a short oesophagus and a very globular stomach. The mouthcone is less prominent than in S. peronii and S. villosum. The labrum is big with chitinous nodules. The labial palps are cylindro-conical, very short with some setae. Mandible and maxillae are well developed. The testes are also well developed, the vesicula seminalis is reduced, leading into the ductus ejaculatorius. The penis is very short and probosciform with a few terminal setae and without any distinct annulation. The males of several species of Scalpellum are sac-like, so that capitulum and peduncle cannot be distinguished morphologically (Table I). One exception is e.g. S. (Scalpellum) elongatum which has a short peduncle. These males are smaller (

Oceanography And Marine Biology

Oceanography And Marine Biology

Oceanography and Marine Biology: An Annual Review, Volume 45 (Oceanography and Marine Biology)