Advances in the Study of Behavior, Volume 24

Advances in

THE STUDY OF BEHAVIOR VOLUME 24

This Page Intentionally Left Blank

Advances in

THE STUDY OF BEHAVIOR Edited by

PETERJ.B. SLATER School of Biological and Medical Sciences University of St. Andrews Fife, Scotland

JAY S. ROSENBLATT Institute of Animal Behavior Rutgers University Newark, New Jersey

CHARLES T. SNOWDON Department of Psychology University of Wisconsin Madison, Wisconsin

MANFREDMILINSKI Zoologisches lnstitut Universitat Bern Hinterkappelen, Switzerland

VOLUME 24

ACADEMIC PRESS San Diego New York Boston London Sydney Tokyo Toronto

This book is printed on acid-free paper.

@

Copyright 0 1995 by ACADEMIC PRESS, INC. All Rights Reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Academic Press, Inc. A Division of Harcourt Brace & Company 525 B Street, Suite 1900, San Diego, California 92101-4495 United Kingdom Edition published by Academic Press Limited 24-28 Oval Road, London NWl 7DX International Standard Serial Number: 0065-3454 International Standard Book Number: 0- 12-004524-9 PRINTED IN THE UNITED STATES OF AMERICA 95 96 9 7 9 8 99 O O Q W 9 8 7 6

5

4

3 2 1

Contents ....................................... ...........................................

ix xi

Contributors.

Preface

Is the Information Center Hypothesis a Flop? HEINZ RICHNER AND PHILIPP HEEB I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Colony-Based Group Foraging . . . . . . . . . . . . . . . . . . . 111, Colonial Lifestyle: Parasitism, Mutual Benefit, or Cooperation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. A Critical View of Some Predictions of the Hypothesis. . V. Evaluation of the Empirical Evidence for Information Transfer at the Colony. . . . . . . . . . . . . . . . . . . . . . . . . VI. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1 7 15 21 25 38 39

Maternal Contributions to Mammalian Reproductive Development and the Divergence of Males and Females CELIA L. MOORE I. Introduction.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Adaptation and the Nature of Sex Differences . . . . . . . . 111. Developmental Processes. . . . . . . . . . . . . . . . . . . . . . . IV. Maternal Contributions to the Development of Reproduction in Mammals . . . . . . . . . . . . . . . . . . . . . . V. Maternal Contributions to Offspring Reproductive Success.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI. Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

47 48 54

55

103 104 106

Cultural Transmission in the Black Rat: Pine Cone Feeding JOSEPH TERKEL I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. The Ability of Naive Adult Rats to Learn to Open Cones . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V

119 126

vi

CONTENTS

I11. The Contribution of Genetics and/or Early Experience to the Ability of Young Rats to Learn to Open Cones . . . . . IV . Alteration of the Cones to Motivate and/or Assist the Rats to Learn to Open Them .................... V. Shaving versus Spiral Method of Opening Pine Cones . . . VI . The Energetics of Pine Cone Opening . . . . . . . . . . . . . . VII . Opening of Two Types of Cones in Sympatric Populations of Black Rats ...................... VIII . Discussion and Conclusions . . . . . . . . . . . . . . . . . . . . . IX . An Afterword . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . X. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

129 132 135 140 144 146 151 151 152

The Behavioral Diversity and Evolution of Guppy. Poecilia reticulata. Populations in Trinidad A . E . MAGURRAN. B . H . SEGHERS. P . W . SHAW. AND G. R . CARVALHO I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I1 . Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111. Adaptive Variation as a Direct Consequence of Risk . . . . IV . Indirect Consequences of Predation Risk . . . . . . . . . . . . V . Evolution in the Wild . . . . . . . . . . . . . . . . . . . . . . . . . VI . Genetic Divergence of Populations . . . . . . . . . . . . . . . . VII . Historical and Stochastic Influences on Behavior . . . . . . VIII . Links between Behavior and Divergence . . . . . . . . . . . . IX . Future Directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . X . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

155 157 159 167 172 179 184 188 192 194 195

Sociality. Group Size. and Reproductive Suppression among Carnivores SCOTT CREEL AND DAVID MACDONALD I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I1. Aspects of Social Organization . . . . . . . . . . . . . . . . . . . 111. The Evolution of Social Breeding and Group Size. . . . . . IV . The Evolution of Reproductive Suppression . . . . . . . . . . V . Comparative Analyses of Sociality and Reproductive Suppression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI . Phylogenetic Regression Analyses of Sociality. Group Size. and Reproductive Suppression . . . . . . . . . .

203 204 213 225 228 231

CONTENTS

VII . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

vii 246 247

Development and Relationships: A Dynamic Model of Communication ALAN FOGEL I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I1. Relationships . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111. Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV . Research Implications . . . . . . . . . . . . . . . . . . . . . . . . . V . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

259 264 277 284 285 286

Why Do Females Mate with Multiple Males? The Sexually Selected Sperm Hypothesis LAURENT KELLER AND HUDSON K . REEVE I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I1. The Sperm Sexual Selection Hypothesis . . . . . . . . . . . . 111. Predictions and Evidence . . . . . . . . . . . . . . . . . . . . . . . IV . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix: Oscillating Evolution of Multiple Sperm Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

291 294 298 305 307 308 310

Cognition in Cephalopods JENNIFER A . MATHER I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I1 . Background on Cephalopods .................... I11. Learning Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV . Other Demonstrations of Capacity . . . . . . . . . . . . . . . . V . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Index

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

317 319 324 335 345 346 355

This Page Intentionally Left Blank

Contributors

Numbers in parentheses indicate the pages on which the authors’ contributions begin.

G . R. CARVALHO ( 1 5 9 , School of Biological Sciences, University of Wales, Swansea, Swansea, SA2 8PP, Wales, United Kingdom

SCOTT CREEL (203), Department of Zoology, University of Oxford, Oxford OX1 3PS, United Kingdom ALAN FOGEL (259), Department of Psychology, University of Utah, Salt Lake City, Utah 84124 PHILIPP HEEB ( l ) , Zoology Department, University of Bern, CH-3032 Hinterkappelen, Switzerland LAURENT KELLER (291), Zoologisches Institut, Ethologische Station Hasli, University of Bern, CH-3032 Hinterkappelen, Switzerland DAVID MACDONALD (203), Department of Zoology, Wildlife Conservation Research Unit, University of Oxford, Oxford OX1 3PS, United Kingdom A. E. MAGURRAN (155), School ofBiologica1 and Medical Sciences, University of St. Andrews, Fife KY16 9TS, Scotland JENNIFER A. MATHER (3 17), Department of Psychology, The University of Lethbridge, Lethbridge, Alberta T1K 3M4, Canada CELIA L. MOORE (47), Department ofPsychology, University of Massachusetts, Boston, Massachusetts 02125 HUDSON K. REEVE (291), Section of Neurobiology and Behavior, Cornell University, Ithaca, New York 14853 HEINZ RICHNER (l), Zoology Department, University ofBern, CH-3032 Hinterkappelen, Switzerland B. H. SEGHERS (1551, Department of Zoology, University of Oxford, Oxford OX1 3PS, United Kingdom

ix

X

CONTRIBUTORS

P. W . SHAW (155), School of Biological Sciences, University of Wales, Swansea, Swansea, SA2 8PP, Wales, United Kingdom

JOSEPH TERKEL (1 19), Department of Zoology, Tel Aviv University, Ramat Aviv 69978, Israel

Preface

The aim of Advances remains unchanged since the series began: to serve the increasing number of scientists who are engaged in the study of animal behavior by presenting their theoretical ideas and research to their colleagues and to those in neighboring fields. We believe the series will continue its “contribution to the development of cooperation and communication among scientists in our field,” as its intended role was phrased in the preface to the first volume in 1965. Since that time traditional areas of animal behavior research have achieved new vigor by the links they have formed with related fields and by the closer relationship that now exists between those studying animal and human subjects. Scientists studying behavior today range more widely than ever before: from ecologists and evolutionary biologists to geneticists, endocrinologists, pharmacologists, neurobiologists and developmental psychobiologists, as well as the ethologists and comparative psychologists whose prime domain is the study of behavior. It is our intention, not to focus narrowly on one or a few of these fields, but to publish articles covering the best behavioral work from a broad spectrum. The skills and concepts of scientists in such diverse fields necessarily differ, making the task of developing cooperation and communication among them a difficult one. But it is one that is of great importance, and one to which the editors and publisher of Advances in the Study ofBehavior are committed. We will continue to provide the means to this end by publishing critical reviews, inviting extended presentations of significant research programs, encouraging the writing of theoretical syntheses and reformulations of persistent problems, and highlighting especially penetrating research that introduces important new concepts. The wide range of topics dealt with in the present volume illustrates these aims well, with a mixture of psychological and biological approaches, as well as laboratory and field studies. What they all have in common is that they tackle important topics, and come up with insights of wide significance to those interested in the study of behavior. Readers will be saddened to learn of the death, in a plane crash, of Dr. Walter Heiligenberg. Walter contributed some classic papers, primarily on the causation of behavior in cichlids and crickets, during his early career in Germany and then, after he moved to California, on the neural basis of electric fish behavior. He wrote chapters on these two areas of interest in Volumes 5 and 18 of Advances.

xi

This Page Intentionally Left Blank

ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 24

Is the Information Center Hypothesis a Flop? HEINZRICHNERAND PHILIPPHEEB ZOOLOGY DEPARTMENT UNIVERSITY OF BERN HINTERKAPPELEN, SWITZERLAND

I.

INTRODUCTION

The sight of some thousands of crows chattering and fighting at sunset in the canopy of a small wood is spectacular, and evokes our feelings of the fun and complexity that surround social life. Birds arrive there from their daily feeding sites many miles away, and return to them the next morning, or perhaps go somewhere else. We are impressed by the view of a breeding colony of tens of thousands of gannets, stuck to a small rock in the North Sea, commuting over long distances to find food for their hungry chicks. Likewise, communal roosting or breeding is typical for many species of birds, mammals, and insects. Two decades ago, in a stimulating paper, Ward and Zahavi (1973) reached the provocative conclusion that “communal roosts, breeding colonies and certain other bird assemblages have been evolved primarily for the efficient exploitation of unevenly-distributed food sources by serving as information centres.” This has become known as the information center hypothesis. As one can see, the provocation is double, as it is first implied that roosting in birds serves to gather information about location of food, and second, that the benefits arising from this information transfer have been the driving evolutionary force that led to communal roosting or breeding. The general view at the time was that the function of communal roosting or breeding is rather found in an efficient protection from predators (e.g., Lack, 1968). Ward and Zahavi’s proposition has been fruitful in generating many studies attempting to test the hypothesis, and their paper became the citation classic of the journal in which it was published. 1

Copyright Q 1995 by Academic Press. Ioc. All rights of reproduction in any form reserved.

2

HEINZ RICHNER AND PHILIPP HEEB

A. HISTORY OF THE HYPOTHESIS Fisher (1954) first proposed the idea that group living in birds may be beneficial for feeding. He reviewed the literature back to the study of the griffon vulture (Gyps fuluus) by Tristram (1859), who wrote: The Griffon who first descries his quarry descends from his elevation at once. Another, sweeping the horizon at a still greater distance, observes his neighbour’s movements and follows his course. A third, still further removed, follows the flight of the second; he is traced by another; and so a perpetual succession is kept up so long as a morsel of flesh remains over which to consort.

Fisher (1954) summarizes his own observations on feeding fulmars as follows: some very mobile species of seabirds, such as fulmars, often give the impression of being individually dispersed when in fact the individuals are all part of a “stretched flock”-a great network of beaters spread to the limit of practical neighbour watching, so that the discovery of one can become the prey of all the hunters of a wide sea area,

concluding that a flock can detect and exploit a swarm more efficiently and thoroughly than individuals.

Both Tristram’s and Fisher’s observations support the simplest concept of a food-related grouping behavior, called “local enhancement” (Hinde and Fisher, 1951; Fisher, 1952). The understanding of this concept has some relevance here, since the findings of many studies claiming evidence in favor of the information center hypothesis are better explained by this more parsimonious mechanism. Thorpe ( 1956) defined local enhancement as “an apparent imitation resulting from directing an animal’s attention to a particular part of the environment,” and Mock, Lamey, and Thompson (1988) refer to local enhancement as “cueing on other already foraging birds for food information.” The widest definition is given by Galef (1988) as “a tendency on the part of naive individuals to approach conspecifics, alterations conspecifics have made in the environment, or objects they have contacted.” Ward (1965) first contrasted the idea of local enhancement with his new hypothesis (to which he did not attach a name at the time): “local enhancement” . . . is practicable only over a limited area within which birds can see each other. It seems likely that the main function of the roost is to extend the benefits of this kind of feeding, so that social feeding may be practiced by a population

INFORMATION CENTERS AND THE ALTERNATIVES

3

together exploiting an area of hundreds of square kilometers. It seems reasonable to suppose that when the members of a roost fly out at dawn, their behaviour will depend partly on their success during the previous day. Those individuals which have left a good feeding place the evening before probably return to the same area, while those that have been less fortunate do not. It would obviously benefit the latter if, instead of going on a random search of new feeding grounds, they could simply join a group whose behaviour indicated that they were heading for an area where food was to be had.

Horn (1968) observed that nestling Brewer’s blackbirds (Euphagus cyanocephalus) in the center of a colony gained more weight per day than young of the same age in peripheral nests, and suggested that this is so because central birds have more neighbors from which they could learn the location of food patches. In support, he described three cases of birds following successful foragers to new food sources. Zahavi (1971) later proposed that the predator avoidance function of bird roosts and breeding colonies is of minor importance compared to the foraging advantage through the mechanism proposed by Ward (1965), for which Zahavi coined the term information center. If anything, the spectacular aerial displays at the roost or at preroost gatherings as are typical for many species (e.g., swallows, Hirundo rustica; bee-eaters, Merops superciliosus; starlings, Sturnus vulgaris; pink-footed geese, Anser brachyrhynchus) may not only advertise the roost to conspecifics, but also to predators (Zahavi, 1971). Zahavi recognizes the existence of adaptations against predation in the communal roost, but clearly views these antipredator adaptations as a consequence of the increase in predation pressure at roosts. The antipredator behavior seen in many roosting species should therefore not be interpreted as a proof of the hypothesis that roosts evolved to minimize predation. Ward and Zahavi (1973), in a major review, concluded enthusiastically that the evidence from many studies clearly favors the information center hypothesis, and that even the exceptions (e.g., solitary feeders with communal roosts, or flock feeders roosting solitarily) could be explained by the same mechanism. Somewhat disenchanting was the extensive review of the evidence for the information center hypothesis by Mock et al. (1988): of the many studies covering the fifteen years between the two reviews, only a few provided support for the hypothesis that individuals at roosts or colonies exchange information about location of food patches. In particular, local enhancement (Ward, 1965; Hinde, 1961; Thorpe, 1956) could often not be excluded. The most important difference between the concept of local enhancement and the concept of an information center concerns the location of information transfer between individuals: local enhancement occurs at the food patch and the increase of group size arises as a consequence of

4

HElNZ RICHNER AND PHILIPP HEEB

birds cuing on already foraging animals. An information center is localized away from the food patch and supposedly arises as a consequence of animals seeking information from successful foragers about the location of the distant food source. The information center mechanism requires communal nesting or roosting, whereas local enhancement does not. We do not discuss here the several hypotheses of colonial behavior based on local enhancement (e.g., Morrison and Caccamise, 1985; Caccamise and Morrison, 1986).

B. SOMECONDITIONS UNDERLYING THEINFORMATION CENTER MECHANISM The many papers that have investigated the information center hypothesis often assume implicitly or explicitly that a number of conditions must be fulfilled for a roost or breeding colony to function as a center where unsuccessful foragers can obtain information for food finding. In particular, if a test of the information center hypothesis provides negative results, it cannot be used as evidence against the hypothesis unless it was shown that these conditions were fulfilled. For simplicity, and unless otherwise specified, we will from now on refer to both a breeding colony and a communal roost as a colony. The following are the necessary conditions: 1. The food is patchily distributed in space and/or shows ephemeral

occurrence in time. This creates a need for information about its location and time of occurrence. A larger number of birds will then have a higher chance to detect a food source. 2. Food patches are rich in order to reduce competition within the patch.

3. The duration of a food patch allows at least one return trip to the colony and back to the patch. 4. Successful foragers return to the colony after having located and exploited a food patch.

5. The colony members can distinguish between successful and unsuccessful foragers. The discrimination is possible on the basis of the feeding success achieved on the immediately preceding foraging trip. 6. After a visit to the colony, the successful foragers return to the previously discovered feeding site. 7. Differences in foraging success between individuals arise by chance

INFORMATION CENTERS AND THE ALTERNATIVES

5

in the localization of food or by differing abilities to localize food, not as a consequence of competitive ability or differences in exploitation rates. 8. Unsuccessful foragers that follow others are more successful on their next feeding excursion than unsuccessful foragers that do not follow. Most of these conditions are obvious, but little is usually said about the circumstances that make them likely to be fulfilled. We attempt to provide insight (Section 11) into the circumstances that make it more likely that some conditions may be met, and will also analyze a few predictions (Section IV) that pertain to some of the above conditions. C. THEKEYPROBLEM The key problem for the functioning of colonies as information centers concerns the question of why a successful forager should return to the colony and thereby have to pay the time and energy cost of the trip from the food patch to the colony and back, and furthermore suffer from the costs of being followed by parasitic, previously unsuccessful foragers. Ward and Zahavi did not explicitly address this problem. Mock et al. (1988)have proposed a graphic model of the cost-benefit balance of leaders and followers that distinguishes between parasitic and mutualistic relationships (Fig. 1). They state that “an information center mechanism can evolve only when the Follower (f-axis) benefits and the Leader (I-axis) does not suffer a great net cost relative to the Follower’s net gain. The shaded area shows this set of conflicting interests.’’

FIG. 1. Model showing the zone of conflicting interests of leaders and followers for which the information center mechanism has been proposed. Adapted from Mock et al. (1988).

6

HElNZ RICHNER AND PHILIPP HEEB

The problems with this interpretation are the following: First, without further assumptions of cooperation or kin benefits, the shaded area is not a zone of conflicting interests, since it is no advantage to the successful forager to suffer a cost, even if the unsuccessful forager gets a relatively higher benefit than the costs it presents to the successful one. Therefore the returning successful forager suffers a reduction in energy intake arising from being followed to the patch and, for example, having to share the food. This energetic cost (L,) per se raises the question of why a successful forager should return to the colony. Section I1 illustrates some of the factors that increase or reduce the magnitude of group size-related foraging cost, or factors that can even make group foraging beneficial (i.e., where individuals can derive mutual benefits). Second, even if the costs to the successful forager of being followed are small or absent, the question remains of why the successful forager should pay the cost of spending time and calories (i.e., L2 in terms of net energy costs) to return to a colony and back to the food patch. A nonreturning individual would at least save these costs of the return flight and, in terms of fitness, do better than a returning one. The information center mechanism, as originally proposed by Ward and Zahavi, rests on the assumption that the individuals participating in an information center benefit and lose in turn, and by chance. The information center hypothesis therefore assumes that the altruistic act of an individual is reciprocated by others in the future.

I f

FIG. 2. Conditions where reciprocal cooperation is expected to evolve. Returning to the colony is beneficial to both leaders and followers under the conditions (1) that they strictly alternate roles, and (2) that the costs to the leader are smaller than the benefits to the follower, that is, the slope of the solid line is steeper than - 1.

INFORMATION CENTERS AND THE ALTERNATIVES

7

Reciprocal cooperation is beneficial if the benefits (F) to the receiver of the cooperative act are higher than the costs (L) to the actor (Trivers, 1971; see Fig. 2 for an example. L is composed of L, + Lz, as explained above). Slopes of the line depicted in Fig. 2 steeper than - 1 will therefore satisfy the condition F > -L. Therefore, reciprocal cooperation is necessary if the successfulforagers are expected to return to a colony. However, theory rather predicts defection as the best strategy for the cases where defection has a high initial payoff (e.g., free riders in large groups), or if cheaters cannot be identified, or also for highly mobile animals (Enquist and Leimar, 1993). Without reciprocated cooperation, the originally proposed information center mechanism rests on group selection. This view is discussed further in Section 111. The need for cooperation renders the information center hypothesis, as initially proposed, even more unlikely as an explanation for the evolution of coloniality. Models that are not based on the need for cooperation are also presented in Section 111.

D. INFORMATION TRANSFER AND

THE

EVOLUTION OF COLONIALITY

As pointed out above, the information center hypothesis claims, first, that information about location of food patches passes from successful to unsuccessful foragers at the communal site, and, second, that this information transfer was the main selective force for the evolution of coloniality. It is important to realize that ( I ) the observation of information transfer at the colony does not prove the information center hypothesis because several other models also predict information transfer at the colony, and (2) the observation of information transfer does not prove that selection for colonial behavior is acting. These points are illustrated in Section 111. Of the many studies that have been undertaken, most have failed to provide unequivocal evidence for the process of information transfer, as documented in the review by Mock et al. (1988) and by others (e.g., Wittenberger and Hunt, 1985; Weatherhead, 1987; Richner and Marclay, 1991). Several predictions of the information center hypothesis will be analyzed in the light of more recent theories in Section IV. The relevant literature concerned with the information transfer is reviewed in Section V. 11. COLONY-BASED GROUPFORAGING

If animals use roosts and colonies as a center to obtain information about the location of food, this will necessarily lead to an increase of

8

HEINZ RICHNER AND PHILIPP HEEB

group size at the food patch. This in turn will lead to increasing competition for food. The costs of group size in colony-based foraging are, however, reduced if food patches are rich and ephemeral. For a further understanding of the information center hypothesis and its alternatives, it is useful here to consider the relationship between individual feeding rate and foraging group size, and explore from a theoretical point of view the effects of group size on the mean and variance of the feeding rate, on risk sensitivity and on an individual’s decision to join a colony, and the effect of other factors in promoting coloniality. The aim of Section I1 is not to review the extensive literature on these topics, but rather to provide a conceptual frame that will highlight the complexity involved in the evolution of information transfer and coloniality. Does group size affect an individual’s feeding rate? Let F(n) be the feeding rate of an individual as a function of group size n, A is the amount of food available to the foragers in a patch, B is the patch exploitation time for one individual, and C is the number of patches discovered in a given time. Thus l/C is the time taken to locate a patch. On average an individual in a group of size n will achieve a feeding rate of F(n) =

B/n

Aln + 1/C-n

that is, an individual will eat a fraction I/n only of patch A per unit time. This time span is given by a patch exploitation time shortened n times, and the time taken to locate a patch, which also becomes shortened n times. It is assumed that animals spend all their time either feeding in a patch or locating a food path. In this equation group size n cancels out and it can be seen that feeding rate is independent of group size

F(n) =

A

B

+ 1/C

that is, a group of size n locates a patch n times faster than an individual, but since the patch has to be divided among n individuals, the individual feeding rate is not affected by group size. Equations (1) and (2) are based on a model presented by Clark and Mange1 (1986). Here we extend the basic model by considering the effects of interactions between prey, predators, and resource characteristics on A, B, and C. Equations (1) and (2) show that group foraging as a consequence of information center-based foraging does not, without other effects arising from changes in group size, increase the mean feeding rate of individual. The benefits accruing from a larger number of animals searching for food and transmitting the information of food location at the colony vanish through the larger number of animals that the food has to be shared with. Only if A, B, or C becomes itself a function of group size n, will the

INFORMATION CENTERS AND THE ALTERNATIVES

9

individual feeding rate be affected. Such mechanisms are not uncommon, as shown later, and some of them actually lead to an increase in the mean feeding rate. If A, B, or C becomes a function of group size n, Equation (1) can be rewritten as

F(n) =

A(n)/n B(n)/n + l/C(n).n'

(3)

The functions A(n), B(n), and C(n)may take the form An', Bnj, and Cnk, respectively. It becomes clear that the effect of group size on mean feeding rate only cancels out if exponents i, j, and k are of the same numerical value. This is most unlikely. Interference between individuals exploiting the patch, social factors affecting the consumption speed, information transfer within group members, and a number of other factors determine the precise value of the exponents (i, j, k). In order to understand which conditions facilitate colony-based foraging, it is useful to consider how these factors finally influence individual feeding rates in groups of different sizes, and how ephemerality and food abundance of patches affect the costs of group foraging. A.

THEEFFECTSOF PATCHRICHNESS A N D EPHEMERALITY

If a patch contains a large amount of food compared to the number of individuals (n)exploiting it, and the patch disappears long before competition between individuals becomes significant, an individual in a group will enjoy the same feeding rate as if feeding alone (see also Pulliam and Caraco, 1984). Patch duration is in this case primarily determined by prey ecology or other predator-independent factors, and not by the number of individuals exploiting the patch. Since food abundance is high and does not significantly decrease through the predators' actions or by their numbers (i = 1, and therefore, Ani/n + A), there will be no interference between predators, and also travel times between food items within the patch stay constant ( j= 1, and therefore Bn'/n + B). For rich and ephemeral food patches, in a simplified form, Equation (1) then becomes F(n) =

A

B

+ 1K.n

(4)

and the colony-based foragers benefit from a higher rate of patch location without an apparent reduction in the quantity of food available to the individual, and without interference between predators. B.

PREDATOR-PREY INTERFERENCE

The effect of interference between predators and prey on the amount of food available (A) in a patch is described by the value of i in An'. If

10

HEINZ RICHNER AND PHILIPP HEEB

i = 0, the total amount of food available in a patch is independent of group size, and the amount available per individual is Aln. If i < 0, there is (negative) interference between predators and prey. Resource A is not fully available and consequently the food available per individual is less than Aln. This is the case, for example, in waders (Goss-Custard, 1980) feeding upon invertebrates (e.g., Nereis, Corophiurn).These quickly withdraw into their burrows when they sense the footsteps of waders, and the effect increases with group size of the predators. If i > 0, the resource A available to the predators increases with group size. As an example, this may occur if group hunting allows the capture of larger prey. Values of i > 0 will therefore reduce the effect of group size on the quantity of food available to each individual in a group, and values of i < 0 will amplify the negative effects of group size. Colony-based group foraging is more likely to occur when the negative effects of group size on the prey available per individual are low, that is, when i is rather large and positive. WITHIN C. PREDATOR-PREDATOR INTERFERENCE

THE

PATCH

Interference between predators can affect patch duration B. Thus B itself can become a function of group size, that is B(n)ln, where B(n) may take the form Bn'. Some examples where Bn' becomes a decreasing function (i.e., j < 0) of group size, and therefore leads to an increase in the feeding rate, include the cases in which animals are learning from each other where the food is to be found (Krebs, MacRoberts, and Cullen, 1972) or exploited within the patch, individuals increasing their feeding rate as a result of competition, or individuals in larger groups being able to reduce vigilance in favor of feeding time (for a review, see Elgar, 1989). B can also become an increasing function (i.e., j > 0) of group size. Predator-predator interactions are more common in patches of low food densities in relation to predator density. This effect might simply appear because the predators increasingly meet the same prey item. Overall this will lead to a decrease in feeding rate with increasing group size. In the most extreme case there may be competitive exclusion at the food patch (Prior and Weatherhead, 1991b).

D. INFORMATION EXCHANGE AND PREDATOR-PREDATOR INTERFERENCE AMONG PATCHES Information transfer at a colony will affect C by the fact that food patches are discovered faster if many individuals search. C itself can become a function of group size, that is, C(n), which may take the form Cnk. In the case where the information of the discoveries is diffused to

INFORMATION CENTERS AND THE ALTERNATIVES

11

all group members (i.e., k = 0), each individual can derive the full benefit of group searching. Incomplete information transfer (k < 0) will decrease the group benefit of rate of patch location. Predator-predator interference when searching for patches will also increase patch location time. For example, birds dispersing from a large colony in search of food patches in a relatively small area will increasingly interfere with each other (e.g., find the same patch), and therefore at large colonies the group size benefit on rate of patch location will decrease (i.e., k < 0). E. GROUPSIZE,VARIANCE IN FEEDING RATE, AND RISKSENSITIVITY The rate of patch location C increases with foraging group size and hence the time to find the next patch (I/C) decreases with group size. Large colonies will locate many patches and, assuming that information flow between individuals at a colony is high, the variance in feeding rate among individuals foraging from a large information center will be smaller than that of individuals foraging from a small one (see also Pulliam and Millikan, 1982). At an extreme, a single forager may discover a food patch rarely, but once a patch is found, food may be practically unlimited. How will variance affect an individual’s decision to participate in colony-based group foraging? If the mean expected reward of a bird in a colony of a given size meets its energy requirements, the bird is best to remain a member of the colony, that is, to be risk averse (Caraco et af., 1990; Cartar and Dill, 1990). However, the individual that cannot satisfy its energy requirements should choose the risk-prone option and switch to a smaller colony, or even forage alone, as demonstrated for greenfinches by Ekman and Hake (1988). F. INDIVIDUAL DIFFERENCES IN LOCATING A N D EXPLOITING FOOD Variance among individuals in the ability to locate patches, and variance in the ability to exploit patches, will influencethe benefits of an individual that participates in colony-based group foraging. Differences in these abilities will influence an individual’s decision whether to stay in a particular colony, or to leave it. As Brown, Stutchbury, and Walsh (1990) state, “certain individuals within a colony who appear to be particularly adept foragers may avoid large colonies and their attendant costs altogether, instead settling in extremely small colonies” (C. Brown, unpublished data, cited in Brown et af., 1990). In brief, an individual that can locate patches efficiently may do better

12

HEINZ RICHNER AND PHILIPP HEEB

foraging alone, whereas an individual with a weak ability to locate will benefit by joining the colony and parasitizing the other individuals’ ability to find patches. Information centers will be interesting for good locaters only if other benefits at the colony or the food patch outweigh the cost of information parasitism by other individuals. Similarly, an individual that is a faster exploiter than the average will benefit more from participating in colony-based group foraging than will a poor exploiter. If such differences between individuals exist, at the stable state, a colony will therefore be composed of poor locators and highly competitive exploiters, which would obviously not be a very useful information center. G. COLONY SIZEAND STABILITY The information center hypothesis relies on the general assumptions that the food patches are scarce but contain an abundant food source, and that they are ephemeral (Waltz, 1982). Variance in food abundance among patches and in ephemerality will determine whether an optimal colony size can exist or not, and also influence the longer-term stability of a colony of a given size. Further, the stability of a colony also depends on how food intake, and ultimately fitness, varies with colony size. Low variance and low ephemerality will favor a stable group size and will not require much exchange of information among foragers. For the typical conditions that have been identified for the information center hypothesis, however, we do not expect colony size to be optimal or stable. High variance in food abundance among patches, plus high ephemerality of patches, will, without strong information exchange between individuals, prevent an ideal-free distribution of foragers among the patches (for further discussion, see Milinski and Parker, 1991). These conditions would therefore favor information sampling at the communal site. This, however, requires a mechanism by which the information-sampling individuals at the colony can recognize the level of benefits that they may be able to derive at a food patch. Signaling by the successful foragers would be ideal, but is stable only if the successful forager can also derive a benefit through signaling. Such benefits can be expected and may include safety-bynumbers at the food patch, prey flushing, and others, as pointed out in Section 1I.C.

H. To SEARCH OR TO FOLLOW? A central decision to be made by each individual in a colony is whether it should leave the colony in search of a food patch, or rather find a successful forager at the colony that could be followed to its previously discovered food patch. Clearly, the benefit of the follower strategy de-

INFORMATION CENTERS AND THE ALTERNATIVES

13

pends on the proportion of successful searchers in a given group and is therefore frequency dependent. If all birds in a group search for food and return to the colony, a bird that decides to wait and follow instead of searching will experience short waiting times and will therefore do well. If most birds decide to become followers, only a few will search and therefore the waiting times will become unprofitably long. In contrast, ideally and without interference between searchers, the searching time for finding a patch does not depend on how many other birds search (Fig. 3A). Therefore, if an individual’s net energy gain from searching (i.e., energy content of prey [E,], minus energy used up by searching and consuming prey [E,] minus energy used for returning [E,] to the information center) is higher than the energy it would have expended during waiting [E,] at the information center,

E,

-

E,

-

Ef > E,,

it should decide to search; otherwise it should become a follower. At a given proportion of followers and searchers the benefits of the two strategies are equal (evolutionary stable strategies, ESS; Fig. 3B); in large information centers it would not matter much whether an individual decides to search or to follow once the ESS is reached. A fundamental difference between a roost and a breeding colony is that in a breeding colony the waiting time of a follower depends on the proportion of individuals searching, whereas in a roost the waiting time is fixed by external factors, such as tides in marine habitats, or daylength. It may be noted that the currency need not be energy for both the searcher and the follower strategy. Also, it should include other costs such as predation risk, and then be expressed in terms of Darwinian fitness. In a recent model, Barta and Sz6p (1992) investigated the effects of resource characteristics on information transfer. However, in their model they fixed the ratio of searchers to followers at 1 : 1. Obviously the payoffs of the two strategies change with both their frequency and the resource characteristics, and fixation of this ratio in a model may not lead to correct predictions about how resource qualities may affect information transfer. It may be noted that the waiter-searcher model is open to cheating by non-returning searchers, and therefore does not solve the problem outlined in Section I.C. Additional benefits to searchers are necessary to compensate for the return costs.

I. BENEFITSA N D COSTSAT

THE

COLONY

Direct benefits or costs arising at the colony will influence the decision to stay single or to join a colony. Benefits may accrue through safety from predators (Kruuk, 1964; Lack, 1968; Hamilton, 1971; Hoogland and

14

HEINZ RICHNER AND PHILIPP HEEB a

surehem followem 1

0

Proportion of colony membera searching

b

8

e

2

1

0

Proportion ofcolony members searching

FIG.3. (a) Waiting times of followers as a function of the proportion of colony members searching. The crossing of the curves (*) indicates the proportion at which searching and waiting times are equal. (b) Net energy gain of individuals as a function of the proportion of colony members searching. (*) indicates the stable equilibrium proportion where both strategies enjoy equal benefits. Both figures assume limited information transfer within the colony and no interactions between searchers away from the colony.

Sherman, 1976; Pulliam and Millikan, 1982; Pulliam and Caraco, 1984), thermoregulation (Yom-Tov, Imber, and Otterman, 1976), mate finding, extra-pair copulations (Morton, Forman, and Braun, 1990; Birkhead and Mbller, 1992; Wagner, 1993), and breeding synchrony (Emlen and Demong, 1975). Costs might result from competition for mates, parasite and disease transmission (Alexander, 1974; Brown and Brown, 1986; Mbller, Dufva, and Allander, 1993), and intraspecific brood parasitism and infanticide (Hoogland and Sherman, 1976; Mbller, 1987; for general reviews, see Wittenberger and Hunt, 1985; Brown et al., 1990).

INFORMATION CENTERS AND THE ALTERNATIVES

15

111. COLONIAL LIFESTYLE: PARASITISM, MUTUALBENEFIT, OR COOPERATION

As pointed out in the Introduction, the key problem of the functioning of colonies as information centers concerns the question of why a successful forager should return to the colonial site and thereby have to pay the time and energy cost of the trip from the food patch to the colony and back, and furthermore suffer from the costs of being followed by parasitic, unsuccessful foragers. A.

MODELSTHATDo NOT SOLVETHE KEYPROBLEM

The No-Cost Model This model assumes that being followed has no costs to a successful forager, and that all individuals are free to join or not to join a colony. Since returning to the colony is costly, a nonreturning individual would, however, do better than a returning one. Thus, even ifinformation transfer has no costs to successful foragers, the costs of returning will prevent communal roosting or breeding behavior to evolve. 1.

2 . The Unavoidable Parasite Model This model assumes that a successfulforager has to go back to the colony to bring food to its offspring and hence cannot avoid being parasitized by followers. The driving force for the evolution of communal roosting or breeding behavior is not the information transfer. Information transfer occurs simply because it cannot be economically prevented. This form of parasitism may be of importance in breeding colonies. As an example, a bird may start to breed at a site and at a later breeding stage be joined by parasitic individuals. Giving up at this stage may be more costly than being parasitized.

B. MODELSTHATCANSOLVETHE KEYPROBLEM The Reciprocal Cooperation Model Three basic conditions are important in the promotion of cooperative behavior: ( I ) the costs to a successful forager of giving information have to be small compared to the benefits to the unsuccessful forager of receiving information (i.e., F > -L in Fig. 2); (2) individuals that give information on one occasion must be likely to receive information on the following occasion; and (3) identification and exclusion of nonreciprocators is possible (Trivers, 1971, 1985). 1.

16

HElNZ RICHNER AND PHlLlPP HEEB

We therefore expect that cooperating individuals enjoy a higher benefit than noncooperating ones, and that in this case the superior benefits from reciprocated information transfer would promote communal roosting or breeding behavior. It has recently been demonstrated that the parameter values for which stable cooperation can be expected are limited under conditions of high mobility of individuals (Dugatkin and Wilson, 1991 ; Enquist and Leimar, 1993; Houston, 1993), as is typical for roosts. In contrast, the exchange of information could be favored if individuals are forced into a spatial association over a longer time period, as is the case for birds in a breeding colony. The possibility that nonreciprocators could not be identified and discriminated against in large and even small colonies may exclude stable cooperation. Although the information center hypothesis has been proposed to explain the evolution of colonial roosting, stable cooperation will therefore be most unlikely to occur in roosts. However, it may arise with a low probability in breeding colonies. Although reciprocal cooperation could potentially solve the key problem, it should therefore not be expected to be an important mechanism in the evolution of coloniality.

2 . A General View of Models That Do Not Require Cooperation For an individual that discovered a food patch, returning to the colony would pay only if the payoffs at the colony (Pc) minus the costs of the flight between the food patch to the colony (LJ and back to the patch plus the payoffs of group foraging (PG)at the patch after the return exceed the payoffs of staying at the patch as a single forager (P& P C - Lz

+ P G > Ps.

(6)

For breeding, the options open to an individual are to breed singly or in the colony. The decision in favor of one or the other option is based on the exact value of both sides in Equation 6, and the colonial benefits will most likely be frequency dependent. A single forager that discovered a food patch may additionally benefit or suffer from the effect of local enhancement at the patch (PL), and Equation 6 then becomes: Pc - Lz

+ P G > Ps + P L .

(7)

Many factors can affect the payoffs at the colony or the payoffs from communal feeding at the food patch, as outlined in Section 11: benefits at the colony arise from predator safety, thermal advantages, extra-pair copulations, and so on, and will render Pc large and positive. Benefits at the food patch arise from reduced vigilance, predator safety, social learning, and so on, and will render PG large and positive.

INFORMATION CENTERS AND THE ALTERNATIVES

17

Under the conditions (1) that recruiting unsuccessful foragers at the colony is more efficient than waiting for birds to join the food patch (i.e., local enhancement) or attracting other birds to the patch (i.e., local recruitment), and (2) that waiting at a colony to be recruited is more efficient than finding food alone, colonial behavior is predicted. Two models are relevant in this context, and can offer a solution to the key problem: the generalized two-handed strategy model, and the recruitment center model. The first model promotes coloniality by rendering Pc large and positive, and the second one by increasing PG in Equation 6. An entirely different model is based on the benefits arising from informing genetically related individuals. This model could explain the evolution of information centers through kin selection. The functioning of these three models is presented below. 3. The Generalized Two-Handed Strategy Model Successful foragers go back to the communal site for reasons such as predator protection, opportunities for extra-pair copulations, and thermal advantages. This possibility requires that the benefits from communal roosting outweigh the costs of being parasitized by individuals in search of a foraging site. Weatherhead (1983, 1987), in the specific two-handed strategy model, has proposed that information transfer presents the principal benefit of communal roosting for inferior foragers (e.g., subordinates), and “that the consistently successful foragers gain primarily by establishing a central roosting position buffered from predators by the surrounding subordinate birds.” Therefore, the inferior foragers “buy” the information on food location by giving up safe positions in the colony to the information holder, thereby accepting a higher risk. In this model both participants can benefit, and the question of whether the net benefits accrueing to the better foragers (benefits from predator safety minus twice the flight costs to the colony), or the benefits accrueing to the poorer foragers (finding good foraging sites) are the more important selective force will depend on the relative magnitude of these two benefits. The model does not require information transfer at the roost. Some phenotypes (e.g., adults) may have a consistently higher foraging success than others (e.g., juveniles), and therefore there is no need for the poorer foragers to be able to identify the birds that enjoyed high feeding success before arriving at the colony. They can simply adopt the rule to follow the members of the successful phenotype and by that strategy will increase their chances of finding food. Information transfer is therefore a byproduct of following particular phenotypes.

18

HEINZ RICHNER AND PHILIPP HEEB

4 . The Recruitment Center Model

Successful foragers can derive advantages if feeding in a larger group and go to the colony for recruiting co-feeders (Evans, 1982a).The communal site therefore does function as a recruitment center for the successful foragers, and as an information center for the unsuccessful ones (Richner and Heeb, in press). Although information is transferred between birds, it is the recruitment center function of the colony that is at the origin of the evolution of communal roosting or breeding. Mechanisms that can enhance an individual’s feeding rate in a group include prey flushing, prey confusion, less need to scan for predators, and others as shown in Section 11. An individual that has newly discovered a rich food patch could then increase its feeding rate by recruiting other birds. An increase in feeding rate is, however, not necessary, since it can also be beneficial to recruit other individuals to the food patch, if the predation risk at the patch thereby decreases, for example, through the dilution effect. This could be done either by recruitment at the food patch or at a colony. An individual should recruit others at the colony if this is more efficient than recruiting them at the patch. The less successful foragers should wait at the colony if this is more efficient than finding food by themselves. The successful foragers therefore use a roost as a recruitment center; the less successful ones use it as a center to obtain information about food location. Our proposition that the successful foragers use a colony as a recruitment center overcomes the difficulty of explaining why they should go to the communal roost after the discovery of an abundant food source, and explains the evolution of colonial behavior through the recruitment center function of colonies. 5 . The Kin Model In the kin model the successful forager passes the information on food location to relatives, for example, siblings, cousins, and offspring, and may therefore derive direct or indirect genetic benefits. Hamilton’s rule shows that kin selection is favored when rB-C>O

(8)

where r is the coefficient of relatedness between donor and recipient of an action, B is the benefits to the recipient, and C the costs to the donor (Hamilton, 1964). Within colonies, we identified F as the benefits to the follower and -L as the costs to the successful forager arising from the time and energy expenses of the return flight plus the costs arising from passing the information to unsuccessful foragers. Informing relatives should be favored when rF - (-L) > 0,

(9)

INFORMATION CENTERS AND THE ALTERNATIVES

19

that is, r F > -L, which becomes F > (-l/r)L. (1 1) The area where kin benefits arise is proportional to the degree of relatedness. Depicted in Fig. 4A is a relatedness of r = 0.5 (e.g., offspring or sisters in diploids). The slope of the line that distinguishes between parasitism and kin benefits is determined by -l/r. The cost that the successfulforager is willing to pay for a given benefit to a relative increases with the degree of relatedness (Fig. 4B). As an example, the successful forager informing its sister ( S ) (r = 0.5) is willing to pay four times the

a

mi

parasitism

b cousin

f

mutualism

s

c

c

FIG.4. (a) The costs arising to the leader can be Compensated by kin benefits. (b) The costs a leader is willing to accept increase with the relatedness between leader and follower.

20

HEINZ RICHNER AND PHILIPP HEEB

cost (Fig. 4B) it is willing to pay for informing a cousin (C) (r = 0.125). Kin benefits are probably an important reason why hymenopteran colonies function as a center of information exchange about the location of distant food patches (but see section V1,C). Among the various dispersal patterns, philopatry occurs in both colonial and noncolonial birds (Greenwood, 1980), and informing relatives could therefore provide genetical benefits to the returning bird. Inbreeding may further enhance this potential. Relatedness may also favor reciprocal cooperation between individuals as shown in roosting vampire bats (Desmodus rotundus) (Wilkinson, 1984). Kinship and spatial association of individuals within the roost predicted the occurrence of sharing blood meals. On any given night the number of vampires that failed to find food was high and the roles of donor and recipient frequently alternated. Wilkinson also found that the costs for the donor, in terms of time left before starvation, were smaller than the benefits obtained by the recipient.

c.

LEVELS OF

SELECTION

Among the models presented above, information transfer at the colony occurs in the no-cost model, the unavoidable parasite model, the recruitment center model, the cooperation model and the kin model. In the generalized two-handed strategy model information transfer may occur, but is not required. The no-cost model and the unavoidable parasite model cannot explain why a successful forager should return to a colony, and cannot promote selection for colonial behavior. Given the fact that information transfer may occur in all the other models, it is surprising that a great number of studies have attempted to test the hypothesis by using evidence of information transfer at the colony as support for the one and only information center hypothesis. The recruitment center model, the kin model, and the generalized two-handed strategy model are based on individual selection and can, theoretically, select for colonial behavior. D. OBJECTS OF SELECTION

Concerning the four models that can potentially explain the evolution of coloniality, the objects of selection are not identical between them, and not necessarily identical between the successful and the unsuccessful foragers. Generalized Two-Handed Strategy Model In the generalized two-handed strategy model, the successful foragers are dominant birds that benefit at the colonial site by having access to positions safe from predators, or that obtain other benefits as outlined above. However, it requires that the successful foragers benefit more than 1.

INFORMATION CENTERS AND THE ALTERNATIVES

21

the unsuccessful, subordinate ones. In this mutual benefit model, the benefits accrue in two different currencies to successful and unsuccessful foragers, and it is therefore partly the benefits to the successful foragers at the colony, and partly the foraging benefits to inferior foragers that may select for colonial behavior. As it may be sufficient to discriminate at the colony the dominants from the subordinates, cognitive abilities to discriminate between successful and unsuccessful foragers are not mandatory, but it requires that dominance is associated with higher foraging success. The foraging benefits to subordinates may select for colonial behavior, but information transfer is not necessarily an important selective force. Colonial behavior is promoted mainly by the benefits arising at the colony to the dominants. 2 . Recruitment Center Model In the recruitment center model the successful forager benefits at the food patch by feeding in a larger group. The driving force for the selection of colonial behavior is not the benefit that unsuccessful foragers derive from information transfer, but the benefits that a successful forager can obtain at the food patch by feeding in the larger group that was recruited at the colony. 3. Cooperation Model True cooperation based on the benefits to the individual and therefore based on individual selection is likely only if two individuals have the occasion to interact over a prolonged period of time. This situation may occur between neighboring birds in a breeding colony that raise chicks over a few weeks or months. This is the only case in which the information transfer per se over time benefits both the donors and recipients of information, and therefore the only case in which for both interactors the benefits from information transfer select for colonial behavior. 4 . Kin Model In the kin model the successful forager that passes information to his unsuccessful relatives may derive direct or indirect genetic benefits; the unsuccessful ones derive direct benefits from the information gain regarding the location of food.

IV. A CRITICAL VIEW OF SOMEPREDICTIONS OF THE HYPOTHESIS

In order to evaluate the information center hypothesis many studies have tested explicit predictions of the hypothesis. Most of the predictions concern the evidence that information had been transferred from success-

22

HEINZ RICHNER AND PHILIPP HEEB

ful foragers to other individuals. A few problems associated with some of the predictions have already been pointed out by Mock et al. (1988). A. SYNCHRONY OF DEPARTURE

The simplest prediction is that information exchange at the colony will lead to synchronous departure of birds from the colony. Clumping of departures can, however, occur for many other reasons (see, e.g., Krebs, 1978; Evans, 1982b; Bayer, 1982; Mock et al. 1988) and is therefore only weak support for the hypothesis. For example, food availability and prey activity may show die1 or tidal patterns and promote departure synchrony. Further, the antipredator advantages of groups, the benefits arising from flying in formations, and the use of winds or thermals as means of transport may all favor clumped departures (Bayer, 1982). Social foraging by itself can provide superior ability to localize food patches (e.g., local enhancement, network foraging) and therefore also predicts departure synchrony. Further, related individuals may leave the colony simultaneously (Rabenold, 1987). Several mechanisms that provide direct benefits from group foraging as outlined in Section I1 will similarly favor synchronous departure. Some examples include the cases in which the most profitable prey can more efficiently be caught by hunting as a group, or if individuals enjoy higher feeding rates as a consequence of learning from others where food can be found within the patch and how it is exploited (Gochfeld and Burger, 1982; Pitcher, Magurran, and Winfield, 1982; Krebs et al., 1972). Furthermore, individuals in groups may reduce vigilance as group size increases and consequently be able to feed faster (Elgar, 1986). In contrast, the observation of single departures cannot be used to reject the hypothesis that information has been transferred, since animals may use a precise language at the colony to communicate food location (e.g., bees, humans) and then travel singly. Or else, if the food can occur only in a few compass directions (e.g., up or down a shoreline) it is sufficient for a bird to observe the direction of departure and fly singly at a later time (Bayer, 1982).

B.

SUCCESSFUL FORAGERS ARE MORELIKELY TO BE FOLLOWED THANUNSUCCESSFUL ONES

The prediction that successful foragers are followed by other birds more often than unsuccessful ones is widely accepted as a proof of information transfer at the colony. However, if high foraging success is associated with certain phenotypes (e.g., adults, dominants), then it is simply required

INFORMATION CENTERS AND THE ALTERNATIVES

23

that the unsuccessful foragers are able to identify these phenotypes in order to increase their chances of finding food. Transfer of information regarding these phenotypes’ previous foraging success is not necessary. This prediction is therefore of limited value, and should be used as a proof of information transfer at the colony only if it has been shown that all individuals have the same chance of finding food. This has only rarely been assessed (Brown, 1986; Wilkinson, 1992).

c.

NUMERICAL INCREASE OF ANIMALS AT THE FOODPATCH

The information center hypothesis holds that animals in a colony will find the food patch through the information provided at the colony by the discoverers of the patch. From this, some studies (e.g., Loman and Tamm, 1980; Andersson, Gotmark, and Wicklund, 1981; Fleming, 1981; Kiis and Mgller, 1986) derived the prediction that the number of birds at the patch will increase steeply after the discoverers return from an intervening visit to the colony. If the animals are not marked, as is the case in most studies so far, this prediction requires a few assumptions. First, it has to be assumed that most of the successful foragers also return to the previously discovered feeding site. Otherwise the newly informed birds may compensate for the ones not returning, and this would lead to no change in bird numbers, which could be wrongly used to disprove the hypothesis of information transfer. Second, it has to be assumed that there is no undetected turnover of birds at the food patch between the discovery and the subsequent revisit of the patch. If there is turnover, the number of birds at the food patch may stay constant, but many more birds know of the site than are seen at any given time. If, after a visit to the colony, all the informed birds return, it will create a steep increase in the number of birds compared to before even though no bird was informed at the colony. This effect has been demonstrated for carrion crows (Coruus corone corone) (Richner and Marclay, 1991). Unequivocal data to exclude turnover as the effect leading to an increase in numbers require that all birds that discovered the patch leave together, or else animals need to be individually marked to allow discrimination between the individuals that discovered the food patch by themselves and those that followed the discoverers from the information center. Furthermore, this prediction can be used to test information transfer at the colony only if local enhancement on the way from the colony to the food patch can be excluded. For example, the animals that leave the colony for a subsequent visit to the food patch may be recognized by their way of traveling, or by other cues that allow individuals that they pass to recognize that they must be going to a good food source.

24

HEINZ RICHNER AND PHlLlPP HEEB

D. INFORMATION CENTERS APPEARWHENFOODIS SCARCE Ward and Zahavi (1973) took the fact that roosts of many birds (e.g., starling, white wagtail [Motacilla alba], chaffinch [Fringilla coelebs], brambling [Fringillamontifringifla],red-winged blackbird [Agelaius phoeniceus], brown-headed cowbird [Molothrus ater], common grackle [Quiscalus quiscala], red-billed quelea [Quelea quelea]) (Wynne-Edwards, 1962; Moffat, 1931; Newton, 1972; Neff and Meanley 1957; Ward, 1965) become larger when food becomes seasonally scarce as one piece of evidence for the information center function of roosts. Moreover, they concluded that during these seasons the need for information is greatest, and the pool of information in the roost at its maximum. Many bird species change their diet from invertebrate or vertebrate prey in spring and summer to plant food (e.g., seeds, nuts, berries, grain, plant matter) in winter. The diet selection model (MacArthur and Pianka, 1966; hlliam, 1974; Krebs and Davies, 1993) predicts that of two or more food types the less profitable one (energy/handling time of one prey item) should be ignored provided the more profitable prey is sufficiently abundant. Thus

holds (where s is the search time for prey type 1 , and E is the energy content and h the handling time of prey types 1 and 2, respectively). Under this condition, prey type 2 can be superabundant and will still be ignored. If, throughout the summer and autumn, the abundance of prey type 1 decreases, the search time for this prey consequently increases until finally

holds. At this point a sudden switch from specializing on prey type 1 to eating both prey types will occur. Since prey type 2 can at this point be superabundant, Ward and Zahavi’s (1973) conclusion that during the season when birds use roosts they are in greatest need of information is not compelling. Neither is therefore the pool of information predicted to be at its maximum during that time. It may well be that prey type 2 (e.g., berries, seeds) but not prey type 1 (e.g., arthropods) has a patchy and ephemeral distribution that promotes a colony-based foraging strategy. Whether this is the case has to be assessed. The occurrence of roosts in itself cannot be taken as evidence that roosts are used as information centers.

INFORMATION CENTERS AND THE ALTERNATIVES

E.

25

LARGE COLONIES HOLDMOREINFORMATION

Ward and Zahavi (1973) predicted that the amount of information available in a colony increases with colony size, and it was therefore expected (Hoogland and Sherman, 1976) that unsuccessful individuals in large colonies could find a successful forager and potential leader faster than individuals in small colonies. As a consequence, individuals in large breeding colonies would enjoy a higher reproductive success than individuals in small colonies. However, Bayer (1982) argued that the positive correlation between offspring number and colony size obtained in observational studies may be unrelated to information exchange. Comparisons between colonies of different sizes are of limited value since colony size may be adapted to local conditions. Brown (1988) predicted that the amount of time spent looking for foraging associates may decrease with an increase in colony size of cliff swallows (Hirundo pyrrhonota), and that consequently foraging success of individuals should increase. Brown’s results showed that foraging benefits increase with colony size, and he suggested that this effect was due to a high information transfer in larger groups. This conclusion was criticized by Shields (1990) on the basis of Brown’s definition of colony size, methodological problems due to uncontrolled factors that may covary with colony size, statistical problems in the sampling method, and pseudoreplication in some analyses. Shields’ comments illustrate the difficulties involved in field experiments on the topic of information transfer in colonies, and point to the problems that should be considered when investigating the correlation between information exchange and colony size.

V.

EVALUATION OF THE EMPIRICAL EVIDENCE FOR INFORMATION TRANSFER AT THE COLONY

Our brief review of the literature is guided by the previous conclusions that: ( I ) most studies on the information center question have tried to provide evidence for information transfer at a colony, but have not tested whether the information center mechanism as proposed by Ward and Zahavi ( 1973) leads to colonial behavior; (2) the information center mechanism is only one among several mechanisms that equally predict information transfer at a colony; and (3) compared to the other mechanisms that involve information transfer at a colony, the information center mechanism is the least likely to explain the evolution of colonial behavior. We restrict this review to a simple evaluation of the evidence for information

26

HElNZ RICHNER AND PHILIPP HEEB

transfer at a colony, but it has to be born in mind that evidence for information transfer does not allow discrimination between the several hypotheses outlined above. Information transfer has been inferred from a variety of observations, such as departure synchrony when leaving the colony, following of successful foragers by unsuccessful ones when leaving the colony, and numerical increase of individuals at the patch after an intervening visit to the colony. Such evidence is highly insufficient and it has been pointed out a few times (e.g., Weatherhead, 1987; Mock et al., 1988; Richner and Marclay, 1991) that without the use of marked individuals these observations cannot provide evidence of information transfer at the colony. Detailed observations of leading and following using marked individuals in relation to their foraging success are necessary. Comparativeapproaches have also been used for evaluatingthe relationship between nesting type and feeding dispersion of birds in relation to their diet. Lack (1968) has pointed out that among seabirds, offshore feeders have larger colonies than inshore feeders. Krebs (1978) found an association between social feeding and colonial nesting in ciconiiformes and suggests that “it is an indication of the importance of food exploitation via an information center.” Erwin (1977, 1978) compared the colonial behavior of various species of terns and found a greater tendency for breeding in colonies in species feeding further away from the nesting sites. Gori (1988) pointed out that “there is a general association in vertebrates between sociality and patchily distributed unpredictable food resources.” In a recent review (Clode, 1993; but see Heeb and Richner, 1994) on the effects of predation and food resources on seabird coloniality, it was concluded that “the theory of social information centres fits the observed pattern of seabird aggregation.” We rather agree with Krebs’s (1978) conclusion that “post hoc interpretations of comparative evidence is not enough.”

A. OBSERVATIONAL EVIDENCE FOR INFORMATION TRANSFER

I . Mammals a. Bats. In evening bat (Nycticeius humeralis) nursery colonies consisting of females and their young, Wilkinson (1992) found that the unsuccessful foragers, as determined by their behavior and weight, gained more weight after following a successful forager than after leaving the colony singly. He concluded that unsuccessful bats could improve their foraging success by following previously successful foragers, thereby locating and exploiting richer prey patches. Bats apparently alternated between follow-

INFORMATION CENTERS AND THE ALTERNATIVES

27

ing and leading during the course of the summer. Which cues do unsuccessful foragers use to locate successful ones? One answer was provided by Barclay (1982). He performed playback experiments with little brown bats (Myotis lucifugus),using the echolocation calls of its own species but also the ones of the big brown bat (Eptesicusfuscus). Little brown bats approached the speaker in response to the calls of both species, and Barclay concluded that echolocation calls of bats at feeding sites (or roosting sites) could be used by other individuals as a cue for locating resources. This attraction to calls by bats is the acoustic equivalent to the visual attraction of birds to feeding flocks, and demonstrates the operation of local enhancement. Another answer, suggested by Wilkinson, is based on the observation that successful foragers tend to urinate more after returning from a foraging trip than do unsuccessful ones. Unsuccessful bats could then use the smell of fresh urine as a cue to an individual’s foraging success. It is not clear, however, how an unsuccessful forager locates a patch after leaving the vicinity of the roost. They may simply respond to the prey location calls of foraging bats, and in this case the return of successful foragers to the colony would merely indicate to the bats the temporal abundance of food. In Wilkinson’s study, bats did not show strong spatial or temporal association while foraging and, in addition, there is no evidence to suggest that the unsuccessful bats followed successful ones to their food patches. 2. Birds a. Ospreys. Greene (1987) investigated the information center question in a breeding colony of 11 pairs of ospreys (Pandion haliaerus) in Nova Scotia. Since ospreys carry the prey in their claws during flight, their foraging success and also the prey species brought back to the colony could easily be recorded. Greene showed that unsuccessful ospreys followed successful ones to the foraging patches. He observed that some ospreys performed conspicuous flight displays after having caught schooling fish, and this display recruited other ospreys to the food patch. Greene’s result seemed to show that ospreys not only benefited through information obtained at the colony enabling them to locate unpredictable food sources but that, surprisingly, successful foragers were indicating the food type to the other individuals in the colony! Does Greene’s study demonstrate information transfer at the colony? Fleming (1990), on a visit to Greene’s study area, noted that the foraging sites were visible from the colony. Furthermore, certain species of fish seemed to use predictable spawning sites. Fleming concluded that Greene’s result should be interpreted as a case of local enhancement rather than as evidence of information transfer at the colony. The question

28

HElNZ RICHNER AND PHILIPP HEEB

of the functional significance of the displays shown by successful ospreys remained unanswered. Results from a larger colony of ospreys (50-60 pairs) situated inland in coastal North Carolina (Hagan and Walters, 1990) did not support Greene’s result. Following of birds was independent of the leaders’ previous foraging success, and ospreys seem to have individually preferred foraging sites. In summary, information transfer at osprey colonies that could indicate to unsuccessful birds the location of distant food patches (Hagen and Walters, 1990; Poole, 1989; Fleming, 1990) has not been demonstrated. b. Swallows. Brown (1986) showed that unsuccessful cliff swallows followed successful ones when leaving the colony. This increased the follower’s foraging success. All individuals were equally likely to follow or be followed. The results suggest that, for cliff swallows, colony-based foraging is beneficial. In a further study, Brown, Brown, and Shaffer (1991) found that cliff swallows hunting swarming insects call to attract other swallows to the insect swarm. These specific food calls are particularly common during poor feeding conditions. Playback experiments demonstrated that these calls can function to recruit foragers to the patch. Stoddard (1988) described a call in cliff swallows that also appeared to recruit others away from the colony. Mock et al. (1988) claimed that some of the social foraging benefits observed by Brown might have arisen through local enhancement since some of the food patches could have been observable from the colonies. Studies on two closely related species, the barn swallow (Hirundo rustics) and the bank swallow, found no evidence for information transfer at the breeding colony. Barn swallows left the colony together but did not follow each other to the foraging patches (Hebblethwaite and Shields, 1990). Instead, they foraged singly in different foraging areas where aerial insects were abundant. Individual swallows cue in on insect-hunting conspecifics in order to locate food patches, and therefore aggregate at a food patch as a result of local enhancement rather than information transfer at the colony. In the bank swallow, both Hoogland and Sherman (1976) and Stuchbury (1988) did not observe following from the colony to food patches. Evidence of information transfer at the colony is lacking, and therefore social foraging does not seem important for the evolution of coloniality in barn swallows (Snapp, 1976; Mflller, 1987; Hebblethwaite and Shields, 1990) or bank swallows (Hoogland and Sherman, 1976; Stutchbury , 1988). c. Terns. Erwin (1978) suggested that breeding colonies of sandwich terns (Sterna sanduicensis) and common terns (Sterna hirundo) function as information centers, since terns forage in flocks over wide areas on

INFORMATION CENTERS AND THE ALTERNATIVES

29

seemingly unpredictable food supplies. Gotmark (1990) observed that sandwich terns headed off in the same direction when leaving the colony, but soon after diverged and flew to different feeding sites. Furthermore, unsuccessful foragers did not follow other birds and successful birds were not followed more often than any other birds. Even during two marked periods of food shortage, Gotmark could not observe information exchange within the colonies. He concluded that in sandwich terns feeding through local enhancement is of more importance. Gochfeld and Burger (1982) have shown for the nonbreeding season that foraging sandwich terns often find feeding sites through local enhancement. Waltz (1987) observed that common terns left the breeding colonies synchronously, and that the birds that left together went preferentially toward similar feeding areas. If the amount of time spent away from the colony was considered as an index of foraging success (successful birds return faster to the colony than unsuccessful ones), then unsuccessful birds were more likely to follow others than successful ones. Furthermore, the terns that arrived at the foraging areas in groups were more likely to catch a fish than were birds arriving alone. In Waltz’s study, however, the importance of local enhancement for the foraging behavior of unsuccessful foragers was not determined (Waltz, 1987). d. Ardeids. Overall, the results of studies on ardeid species provide little evidence of information transfer at colonies (for a review, see Mock et al., 1988). An observational study by Mock et af. (1988) on large mixed heronries of great blue herons and great egrets did not support the prediction that unsuccessful birds should follow successful foragers of ecologically similar species (Ward and Zahavi, 1973; Krebs, 1978).

B.

EXPERIMENTAL STUDIES

1 . Mammals

a . Rats. Galef and his collaborators have investigated whether Norway rats ( R a m s noruegicus) possess the behavioral and cognitive abilities required for food-related information transfer between individuals (see references in Galef, 1992, 1994). Norway rats are highly social animals that live in groups all year round and feed on a variety of food sources. Chemical cues found on resources (food type, nest sites) used by other rats provide a mechanism for social learning, which is used in a variety of contexts (Laland and Plotkin, 1991). Galef (1988) proposed that in most cases social learning in rats can be explained by local enhancement. Experiments with Norway rats showed that a naive rat can acquire sufficient information from a recently fed conspecific (the demonstrator)

30

HEINZ RICHNER AND PHlLlPP HEEB

to identify the food type this demonstrator ate before (Galef and Wigmore, 1983; Posadas-Andrews and Roper, 1983). Rats use olfactory cues consisting of two components: first, a diet-identifying component related to the odor of the food eaten, and second, a contextual component indicating that the food was safe to eat (Galef, Kennett, and Stein, 1985). Galef, Mason, Preti, and Bean (1988) found that carbonyl sulfide (COS) and carbon disulfide (CS,) from the breath of demonstrator rats enhanced the attractiveness of the food they ate to observer rats. The presence of CS2 in the food augments its attractiveness to rats (Mason, Bean, and Galef, 1989). Rats ate more than three times more food at baiting stations with food containing a solution of CS2 than they ate at unscented baiting stations. To test the transfer of information about the location of distant food, Galef, Mischinger, and Malenfant (1987) studied the behavior of rats following each other in a maze. They found that rats trained to follow conspecifics through a maze will follow rats that have eaten a “safe” food with a higher probability than that associated with rats that have eaten “unsafe” (poisonous) food. These results support the hypothesis that rat colonies function as information centers in which unsuccessful foragers could acquire information from the more successful colony mates about the locations of food sources, the types of food that can be safely eaten, and those that should be avoided (Galef, 1992, 1994). Galef (1991) also found that foragers mutually exchanged information about the food type they ate and this affected their later feeding behavior. Galef suggests that this information exchange among successful foragers could explain why they should return to the colony, even if they are not certain of reciprocation in the future. Recent experiments in a semi-natural enclosure suggest that rats use olfactory information transmitted by other “demonstrator” rats when choosing between two new foods (Berdoy, 1994). 2 . Birds a . Weauerbirds. Red-billed queleas, an agricultural pest in parts of central Africa, nest in colonies and may roost in groups larger than a million birds. In large aviaries, De Groot (1980) tested whether roosts serve as a center for food-related information exchange between birds. For the experiment, a communal roosting area was connected by four entrance funnels with four compartments. One group of birds was trained to find food in one of the compartments, and another group was trained separately to find water in another compartment. The evening before the experiment, the two groups were allowed to roost together. For the first test, they were deprived of food overnight. De Groot observed that the birds trained to find water followed the birds trained for food. In the

INFORMATION CENTERS AND THE ALTERNATIVES

31

second test the birds were deprived of water and it was then observed that the birds trained on food followed the birds previously trained on water. The results suggest not only that information transfer occurred between roosting birds, but also that birds could assess the resource type that the other birds had previously accessed. The results have been criticized (Mock et al., 1988) on the argument that the birds in the roosting area could probably tell which compartment held the resource simply by measuring how long birds stay in a chamber after having entered. The results would therefore demonstrate local enhancement rather than information transfer. This criticism is, however, invalid since birds entered the compartment on the side of the roosting area, but could only leave it at the other end, which was connected to an aviary that was invisible from the roosting area. A criticism of De Groot’s experiment should rather point to the fact that the birds in the roosting area could see the compartments where they were trained to find food, and could see from their roosting position where other birds entered. Hence there was no need to recognize the knowledgeable foragers at the roost, and De Groot’s observation thus cannot distinguish between information transfer at the roost and local enhancement from the roost. The compartments should have been visually blocked from the roosting area in order to test whether naive birds follow knowledgeable birds when leaving the roost. The same results would then unequivocally demonstrate information transfer at the roost. b. Crows. The method of creating rich food patches placed randomly and remote from the colony has commonly been used to test for information transfer at the colony. Information transfer is inferred if, after a visit of a few birds to the patch (All), many more return (NJ after an intervening trip to the colony (i.e., N2 > N J . Loman and Tamm (1980) created rich patches composed of dead pigs and chickens, and then counted the number of hooded crows (Corvus corone cornix)and ravens (Corvus corm) visiting these patches the first day (N,) and early the following morning (N&. In 13 out of 25 trials N 2 was actually smaller than N,. Weatherhead (1987) has pointed out that these 13 trials should not be interpreted as negative evidence of information transfer but rather as failed experiments, since not all the birds present the first day returned the day after. In 8 out of 11 trials where birds found the patch the first day, N 2 was larger than N,. These results could be interpreted as information transfer, but even Loman and Tamm (1980) do not exclude the possibility that the effect was due to local enhancement. Moreover, the birds were not individually marked and therefore the importance of turnover of birds at the food patch on the first day could not be assessed. If birds stay for only brief periods at the patch and if they are not individually marked,

32

HEINZ RICHNER AND PHILIPP HEEB

the number of birds that informed themselves at the food patch the first day remains unknown, and therefore a comparison of N2 with N , becomes meaningless (Richner and Marclay, 1991). Turnover at the food patch was demonstrated for carrion crows where almost 700 birds ( ~ 2 0 %of) a large population (Richner, 1989a, 1989b, 1992) were individually marked (Richner and Marclay, 1991).Because of turnover at the patch, the number of birds present at the patch at any one time during the first day was much lower than the number of birds that had acquired the information up to that time, and the results of Richner and Marclay’s (1991) study were more consistent with local enhancement than with information transfer at the roost. c. Gulls. In a breeding colony of black-headed gulls (Larus ridibundus), Anderson et al. (1981) tested the prediction that successful foragers should be followed back to their food patch by unsuccessful ones. After having successfully fed at an experimental patch, these foragers flew back to the breeding colony. Although they returned in 48 of 50 trials to the experimental patch, in none of the cases were they followed by another gull when leaving the colony. Directions of departure also indicated that other gulls left the colony independently of the experimental foragers. The result was the same when the feeding conditions in the area became very poor and following of successful foragers could have been expected. It shows that information transfer at the colony from successful to unsuccessful foragers was not relevant for food finding. Instead, some of the gulls had their preferred feeding sites, and others joined feeding gulls at their food patches. This identifies feeding by local enhancement as a prevalent mechanism for food finding. In colonies of black-billed gulls (Larus bulleri), Evans (1982a)observed that some leaders produced calls that attracted other birds. Calling leaders attracted more birds than silent leaders, and leaders called more than followers. Playback experiments confirmed the attractiveness of the calls. Evans suggests that the calls serve to recruit other birds to the callers’ foraging site, and that calling may be selected by benefits arising from group foraging. However, it is not clear from Evans’s study why in 60% of the cases the leaders did not call when leaving the colony, and it is also not known whether the calling or the silent leaders knew the location of profitable food patches. Evans argues that birds that know the location of a food source should call as much as birds that do not know the location of food, and calling could therefore not be evolutionarily stable. d . Wagtails. Fleming (1981) distributed a gallon of maggots over the winter foraging area of a pair of pied wagtails, which had been feeding at the food-supplemented site for a week prior to the experiment. The experiment was repeated five times involving five different (unmarked)

INFORMATION CENTERS AND THE ALTERNATIVES

33

pairs of birds. During the time of the experiments the ground was frozen and food was probably in short supply. In none of the cases was the target pair followed by other birds the following morning. Pied wagtails commonly join overnight roosts in the winter season. However, it is unknown whether the target birds were roosting with other birds or alone. e. Finches. Kiis and Mgller (1986) provided sunflower seeds, a preferred food of greenfinches (Carduelis chloris), at randomly chosen sites. In 7 out of 16 trials, no bird was attracted to the patch. In the other 9 trials birds were seen feeding at the baited sites, but in only one of the cases did a larger number of birds appear at the food patch the following morning. The 16 experiments suggest that the baited sites were too poor compared to other, natural food sources. Therefore, the results are of little use even as evidence against information transfer at a roost. f. Vultures. Rabenold (1987) tested information transfer at roosts in a partially marked population of black vultures (Coragyps atratus). The vultures discovered the experimental food supplement in 13 out of 30 trials, and also returned to the food in 7 of these 13 trials the following day. In the 7 successful trials, 73% of the marked birds returned to the food patch. Among the birds arriving the second day, these returning birds arrived earlier than other birds. The number of birds present increased from the first to the second day in all 7 trials, but was significantly higher than expected only in the 3 winter trials. The expected number of birds was calculated “as the number of birds present the first day plus the same proportion of naive birds in the roost as arrived on the first day when all roostmates were naive.” In another experiment Rabenold released 13 adult and 19 juvenile vultures that were held in captivity for 2 days, into a roost at nightfall. It was assumed that these birds did not know the current food distribution in the study area. The following morning, regardless of age, most experimental, “uninformed” birds left the roost later than the unhandled ones. Furthermore, uninformed birds were found at the rear of departing groups. Both observations correspond to predictions if information transfer at the roost has occurred. However, Rabenold could not exclude local enhancement as a mechanism that would also account for her results, and information transfer at the roost was not proven. She found that a demonstrably better class of food finders (adults) was routinely followed by another less capable class (juveniles)of birds. If there are persistently successful and unsuccessful foragers in a colony, there is no need for an unsuccessful forager to identify the previously successful ones. It is sufficient to follow consistently the better class in order to increase one’s foraging success. In other words, information transfer regarding previous foraging success is not required.

34

HEINZ RICHNER AND PHILIPP HEEB

There is evidence that in black vultures, parents and offspring maintain ties for several months past fledging, and observations show that parents continue to feed their young over prolonged periods of time (Jackson, 1975; McHargue, 1977; Rabenold, 1986). Outside the breeding season, family members participate in communal feeding and roosting behavior. Long-term associationsmay be mutually beneficial, and Rabenold suggests that “aid in feeding may be the largest single factor favoring retention of strong ties among members of black vulture families.” Roosts may therefore serve as meeting places for family members. If related birds can benefit from food-related information exchange at roosts, then the costs of returning to the roosts may be largely compensated for by the genetic benefits given by the higher fitness accrueing to kin. Prior and Weatherhead (1991a) investigated whether information transfer occurs at communal roosts of turkey vultures (Cuthartes aura). In a partially marked population they tested the prediction that more birds than expected (assuming independent discovery and/or local enhancement) should arrive at the novel patches the day following the discovery of the patches. In only 3 out of 13 successful trials did more birds arrive on the day after discovery than could be expected. Interactions of vultures at the food patch suggested high competition between birds, and also monopolization of food by socially dominant individuals. This competition may limit the benefits for the birds in greatest need of food, and would therefore also reduce the benefits that subordinates could derive from following (Prior and Weatherhead, 1991b). g. Yellow-Headed Blackbirds. Gori (1988)tested for information transfer at a breeding colony of individually marked yellow-headed blackbirds (Xunthocephalus xanthocephalus) by creating rich food patches of sunflower seeds and mealworms out of sight of the colony. Should information transfer at the colony occur, it was predicted that recruitment to food patches should be greater than estimated rates of recruitment by local enhancement only. This was confirmed. Gori estimated recruitment rates from local enhancement by counting the number of new birds flying in the direction where the food would be provided over a fixed period of time, and also by counting the number of birds landing and flying over the site between the time that the first bird located the site and his first repeat visit. Furthermore, the birds that returned to the food patch after provisioning their chicks at the colony were more likely to be accompanied by colony members than birds departing in other directions. In seven out of eight replicates, colony neighbors of the initial site discoverers had a significantly greater probability of being recruited to the sites than did nonneighbors. This recruitment pattern is predicted assuming that information about foraging success is more easily obtained from nearby individ-

INFORMATION CENTERS AND THE ALTERNATIVES

35

uals than from more distant ones (Krebs, 1974). Gori (1988) concluded that coloniality in yellowheads facilitates the location of good foraging areas, and that information transfer may be implicated in this process. He suggests that unsuccessful foragers may use the rate of food delivery at the nest as a cue of who to follow. C. SOCIAL HYMENOPTERANS In social hymenopterans cooperative behavior can evolve via kin selection (Hamilton, 1964; see also Section 111,B). The workers within a colony are the daughters of one or a few laying queens. By helping to raise their own siblings, workers increase their inclusive fitness (see review by Seger, 1991), and the costs faced by a leader returning to the colony may be offset by kin benefits. a. Ants. Social hymenopterans have efficient mechanisms for recruiting nestmates to food sources and to new nesting sites (Holldobler, 1977; Holldobler and Wilson, 1990). In many ant species individuals use chemical substances as a recruiting signal by depositing an odor trail between the resource and the nest site (Wilson, 1971; Holldobler, 1977). In the most efficient recruitment systems, as, for example, in Solenopsis spp., the presence of a scent trail is enough to recruit a large number of individuals (Wilson, 1971; Holldobler, 1977; Holldobler and Wilson, 1990). This sophisticated chemical communication system is a form of recruitment through local enhancement, where the recruited ants follow the odor trails. Moglich, Maschwitz, and Holldobler (1974) described the more “primitive” recruitment mechanism of Leptothorax aceruorum. In this ant species a successful forager returns to the colony and regurgitates some of the food to nest mates. She then raises the gaster with her sting exposed extruding a droplet of liquid containing pheromones. These pheromones attract nest mates, which touch the leader on the hind legs or gaster with their antennae and then tandem running starts (Moglich et al., 1974). During tandem running the leader runs in front while the follower keeps close antenna1 contact with the leader. The presence of a pheromone trail is not required. The two ants keep in close physical contact until they reach the food source. Tandem running is considered to be the most primitive recruitment mechanism in ants. The next step in complexity consists of “group recruitment” observed in Camponorus spp. (Holldobler, 1977; Holldobler and Wilson, 1990). In this mechanism, one ant recruits 5-30 nestmates at a time; the recruited ants follow closely behind the leader ant to the target area. In Camponofus socius, the scout leaves chemical cues around the newly discovered food source and lays a trail back to the colony. The pheromone trail alone does not have a recruitment

36

HElNZ RICHNER AND PHILIPP HEEB

effect and the presence of a leader is essential. Holldobler (1977) suggests that in ants, as the chemical recruitment system became more sophisticated, mechanical signals between leader and follower became less important. b. Stingless Bees. Bees are another group of social hymenopterans in which efficient recruitment to resources has been observed (Holldobler, 1977; Heinrich, 1978). In stingless bees, comparative studies have shown different levels of communication using recruitment techniques of varying complexity. In Trigona spp. recruiting signals given by returning foragers range from buzzing sounds and zigzag runs inside the hive to the laying of chemical signposts at certain intervals enabling the recruiting bee to return to the food (Holldobler, 1977). In certain Melipona species leaders do not lay odor trails, but the duration of the sounds produced by returning foragers appears to vary directly with the distance to the food source (Esch, 1967). Within this system, a guide bee is generally necessary to lead up to more than fifty nestmates to the food source. When leaving a nest, leaders apparently show the direction to the goal by a short zigzag guidance flight. Competition among certain neotropical bees at their food sources appears to be one of the most important factors in their foraging strategy (Hubbell and Johnson, 1978). The ability to recruit large numbers in a short period of time enables the bees to outcompete other foragers (Heinrich, 1978; Johnson and Hubbell, 1974; Hubbell and Johnson, 1978). Compared to the honeybee, experimental studies on the recruitment mechanisms in stingless bees remain scarce. Experiments are essential to understand the details of the recruitment dynamics and the physical or chemical cues used by the stingless bees. c . Honeybees: The Dance Language Controversy. Karl von Frisch proposed that the dances made by foraging bees from the genus Apis returning to the hive enabled them to recruit other bees in the hive to distant food sources (von Frisch, 1967; Lindauer, 1961). The “dance language hypothesis” proposed by von Frisch was widely accepted and became a famous paradigm in animal communication. Supporters of the dance language hypothesis claim that a honeybee colony is able to forage over a wide area because of the scouts who recruit other bees to their newly discovered flower patches by their dance in the hive. Experiments suggest that the dance effectively transmits information about the distance and direction of the food sources (von Frisch, 1967; Gould, 1975; Michelsen, Anderson, Storm, Kirchner, and Lindauer, 1992). Therefore, the honeybee colony appears to be a “classic” example of an information center with the honeybee dance as the mechanism by which the information is transferred between individuals (Seeley, 1985a, 1985b). This interpretation is not, however, accepted by supporters of the “odor search hypothe-

INFORMATION CENTERS AND THE ALTERNATIVES

37

sis,” who claim that food searching behavior in honeybees is much like that exhibited by other flying insects, that is, essentially based on odors (Wells and Wenner, 1973; Wenner, 1989). This alternative interpretation led to an ongoing controversy over the recruitment mechanism in honeybees (Wenner and Wells, 1990; Vadas, 1994). The odor search hypothesis is a more parsimonious mechanism for food finding in bees and does not require the exceptional cognitive capabilities assumed by the dance language hypothesis (Wells and Wenner, 1973; Wenner and Wells, 1990; Wenner, Meade and Friesen, 1991; Vadas, 1994). According to supporters of the odor hypothesis, the experiments carried out by the dance language proponents appeared to be flawed in various respects. They point out that the dance language researchers usually concentrated their attention on the bees effectively recruited by the dance, while the majority of bees foraged without apparently using it (Wenner and Wells, 1990; Wenner et al., 1991). Furthermore, they found that most bees searched for food, water, and new hives close to their home hive, where the information obtained through the dance would be of little use (Wenner et al., 1991; Vadas, 1994), and the time taken by recruits to find the food patches was greater than would be expected if the dance informed them of the location and distance of the food patches. As a way out of this controversy, Wenner et al. (1991) proposed that instead of concentrating on one single hypothesis, honeybee researchers need multiple working hypotheses in the context of a “strong inference” approach (Platt, 1964). In a review of the controversy, Vadas (1994) suggested that instead of assuming a priori the importance of the dance language, it has to be determined in what circumstances dances and/or odors are important for recruiting bees to food and other resources. In summary, both hypotheses appear to be relevant in honeybee recruitment and there is little doubt that the beehive functions as an information center. The challenging question is whether coloniality in bees originally evolved through the benefits of information exchange per se or through the benefits of information exchange between kin only.

D.

CONCLUSIONS FROM THE

EMPIRICAL EVIDENCE

Evidence that colony members exchange information concerning the location of distant food patches is scant. This is even more pronounced if one considers the large number of studies conducted over the past twenty years that were designed to test information transfer. Many studies were purely observational, most studies used unmarked animals requiring specific assumptions, and in much of the work the simpler mechanism of foraging by joining animals at their food patch (i.e., local enhancement)

38

HElNZ RICHNER AND PHILIPP HEEB

could not be excluded. Furthermore, other hypotheses that make the same predictions as the information center hypothesis were in most cases not even considered. The lack of empirical support makes it unlikely that the function of roosts and colonies as information centers is of much importance for the evolution of colonial behavior. VI. SUMMARY The evolution of coloniality in birds, mammals, insects, and other species is still a riddle. In contrast to the hypothesis of Lack (1968) that the antipredator function of bird roosts and colonies led to the evolution of colonial behavior, Ward and Zahavi (1973) reached the conclusion that communal roosting and breeding in birds has evolved for the exploitation of patchy food sources. This is now known as the information center hypothesis. The information center mechanism holds that individuals at the colony exchange information about the location of distant food patches, and that a foraging individual that is unsuccessful on one occasion can follow a successful individual from the colony; when it is successful it will be followed by unsuccessful ones when leaving the colony. The hypothesis rests therefore on the assumption that the individuals participating in an information center benefit and lose in turn, and by chance. The altruistic act of an individual is expected to be reciprocated by others in the future. The information center hypothesis further claims that this information transfer was the main selective force for the evolution of coloniality. The key problem of the functioning of colonies as information centers concerns the question of why a successful forager should return to the colony and thereby pay time and energy costs of food trips and parasitization by unsuccessful foragers. Without reciprocated cooperation, the originally proposed information center mechanism rests on group selection. Although the reciprocal cooperation model satisfies the evolutionary criterion that selection should be based on benefits to the individual, the stringent conditions posed by cooperation models and the high mobility of animals in colonies will most likely preclude stable cooperation between individuals in colonies. Several other models that will favor colonial behavior, and are based on individual selection, do not require reciprocal cooperation: the generalized two-handed strategy model (successful foragers return to the colony for the benefits they obtain by communal roosting or breeding), the recruitment center model (successful foragers benefit at the food patch by communal feeding and return to the colony for recruiting other foragers), and

INFORMATION CENTERS AND THE ALTERNATIVES

39

the kin model (successful foragers return to the colony to inform kin about the location of food patches and thereby benefit by increasing their inclusive fitness). Information transfer will occur in the recruitment center model and the kin model, and may occur but is not required in the generalized two-handed strategy model. However, in all three models it is not the information transfer to unsuccessful foragers at the colony that is at the origin of the evolution of colonial behavior, but either the benefits that the successful foragers derive at the colony (in the generalized twohanded strategy model) or at the food patch (in the recruitment center model), or the benefits to the successful forager from promoting its genes through helping kin (in the kin model). Nevertheless, in empirical studies the occurrence of information transfer between successful and unsuccessful foragers was thought to be sufficient evidence for the information center model, thus ignoring all other models where information transfer is also predicted. It is therefore not surprising that a review of many observational and experimental studies provides nearly no support for the information center model. Many predictions of the information center model cannot be upheld and, partly because much research was centered around a single hypothesis, research into the evolution of colonial behavior has stagnated. In our review, we (1) analyzed the costs and benefits of colony-based group foraging, (2) analyzed the levels and objects of selection of the various models that can lead to coloniality, (3) examined some common predictions of the information center model in the light of recent theory, and finally (4) evaluated the empirical evidence for information transfer at the colony. We propose that the original information center hypothesis for examining the evolution of coloniality should be abandoned in favor of the promising alternative hypotheses that are explicitly based on individual selection. Acknowledgments

We are grateful to Nick Davies. Manfred Milinski, Peter Slater, and Charles Snowdon for their comments on the manuscript. The work was supported by the Swiss National Science Foundation.

References

Alexander, R . D. (1974). The evolution of social behavior. Annu. Rev. Ecol. Syst. 5,325-383. Anderson, M., Gotrnark, F., and Wicklund, C. G. (1981). Food information in the blackheaded gull, Larus ridibundus. Behav. Ecol. Sociobiol. 9, 199-202. Barclay, R. M. R. (1982). Interindividual use of echolocation calls: Eavesdropping by bats. Behav. Ecol. Sociobiol. 10, 271-275.

40

HElNZ RICHNER AND PHILIPP HEEB

Barta, Z., and Szep, T. (1992). The role of information transfer under different food patterns: A simulation study. Behav. Ecol. 3, 318-324. Bayer, R. D. (1982). How important are bird colonies as information centers? Auk 99, 31-40.

Berdoy, M. (1994). Making decisions in the wild: Constraints, conflicts, and communication in foraging rats. I n “Behavioral aspects of feeding” (B. G. Galef Jr., M. Mainardi, and P. Valsecchi, eds.). Ettore majorana international life science series, Vol. 12, pp. 289-3 13. Harwood Academic Publishers, Chur, Switzerland. Birkhead, T. R., and Meller, A. P. (1992). “Sperm Competition in Birds: Evolutionary Causes and Consequences.” Academic Press, London. Brown, C. R. (1986). Cliff swallow colonies as information centers. Science 234, 83-85. Brown, C. R. (1988). Enhanced foraging efficiency through information centers: A benefit of coloniality in cliff swallows. Ecology 6Y, 602-613. Brown, C. R., and Brown, M. B. (1986). Ectoparasitism as a cost of coloniality in cliff swallows (Hirundo pyrrhonota). Ecology 67, 1206-1218. Brown, C. R., Stutchbury, B. J., and Walsh, P. D. (1990). Choice of colony size in birds. Trends Ecol. Evol. 5 , 398-403. Brown, C. R., Brown, M. B., and Shaffer, M. L. (1991). Food sharing signals among socially foraging cliff swallows. Anim. Behav. 42,551-564. Caccamise. D. F., and Morrison, D. W. (1986). Avian communal roosting: Implications of diurnal activity centers. A m . Nat. 128, 191-198. Caraco, T., Blanckenhorn, W. U., Gregory, G. M., Newman, J. A., Recer, G. M., and Zwicker, S. M. (1990). Risk-sensitivity: Ambient temperature affects foraging choice. Anim. Behav. 39, 338-345. Cartar, R. V., and Dill, L. M. (1990). Why are bumblebees risk-sensitive foragers? Behav. EcoI. Sociobiol. 26, 121-127. Clark, C. W., and Mangel, M. (1986). The evolutionary advantages of group foraging. Theor. POP. B i d . 30, 45-75. Clode, D. (1993). Colonially breeding seabirds: Predators or prey? Trends Ecol. Evol. 8, 336-338.

De Groot, P. (1980). Information transfer in a social roosting weaverbird (Quelea quelea, Ploceinae): An experimental study. Anim. Behav. 28, 1249-1254. Dugatkin, L. A., and Wilson, D. S. (1991). Rover: A strategy for exploiting cooperators in a patchy environment. A m . Nut. l38,687-701. Ekman, J., and Hake, M. (1988). Avian flocking reduces starvation risk: An experimental demonstration. Behav. Ecol. Sociobiol. 22, 91-94. Elgar, M. A. (1986). House sparrows establish foraging flocks by giving chirrup calls if resources are divisible. Anim. Behav. 34, 169-174. Elgar, M. A. (1989). Predator vigilance and group size in mammals and birds: A critical review of the empirical evidence. Biol. Rev. 64, 13-33. Emlen, S. T., and Demong, N. J. (1975). Adaptive significance of synchronized breeding in a colonial bud: A new hypothesis. Science 188, 1029-1031. Enquist, M., and Leimar, 0. (1993). The evolution of cooperation in mobile organisms. Anim. Behav. 45,747-757. Erwin, R. M. (1977). Foraging and breeding adaptations to different food regimes in three seabirds: The common tern, Sterna hirundo, royal term, Sterna maxima, and black skimmer, Rynchops niger. Ecology 58, 389-397. Erwin, R. M. (1978). Coloniality in terns: The role of social feeding. Condor 80, 21 1-215. Esch, H. (1967). Die Bedeutung der Lauterzeugung fur die Verstandigung der stachellosen Bienen. 2. vergl. Physiol. 56, 199-220.

INFORMATION CENTERS AND THE ALTERNATIVES

41

Evans, R . M. (1982a). Foraging flock recruitment at a blackbilled gull colony: Implications for the information center hypothesis. Auk 99, 24-30. Evans, R. M. (1982b). Flock size and formation in black-billed gulls (Larus bulleri). Can. J . ZOO/.60, 1806-1811. Fisher, J. (1952). “The Fulmar.” Collins, London. Fisher, J. (1954). Evolution and bird sociality. In “Evolution As a Process.” (J. S. Huxley, A. C. Hardy, and E. B. Ford, eds.), pp. 71-83. Allen and Unwin, London. Fleming, S. P. (1990). Making sense of information. Nature (London) 348, 291. Fleming, T. H. (1981). Winter roosting and feeding behaviour of pied wagtails Motacilla a/ba near Oxford, England. Ibis lZ3, 463-476. Frisch. K. von (1967). “The Dance Language and Orientation of Bees.” Harvard Univ. Press, Cambridge, Massachusetts. Galef, Jr., B. G. (1988). Imitation in animals: History, definitions and interpretation of the data from the psychological laboratory. In “Social Learning” ( T . Zentall and B. G. Galef, Jr., eds.), pp. 3-28. Lawrence Erlbaum, Hillsdale, New Jersey. Galef. Jr.. B. G. (1991). Information centres of Norway rats: Sites for information exchange and information parasitism. Anim. Behau. 41, 295-301. Galef, Jr., B. G. (1992). A primer of olfactory communication about distant foods in Norway rats. In “Chemical Signal in Vertebrates” (R.L. Doty and D. Miiller-Schwarze, eds.), Vol. 6, pp. 451-456. Plenum Press, London. Galef, Jr., B. G. (1994). Olfactory communications about foods among rats: A review of recent findings. In “Behavioral aspects of feeding” (B. G. Galef Jr., M. Mainardi, and P. Valsecchi, eds.) Ettore majorana international life science series, Vol. 12, pp. 83-101. Harwood Academic Publishers, Chur, Switzerland. Galef. Jr., B. G., Kennett, D. J., and Stein, M. (1985). Demonstrator influence on observer diet preference: Effects of simple exposure and the presence of a demonstrator. Animal Learning and Behavior 13,25-30. Galef, Jr., B. G., Mason, J. R., Preti, G., and Bean, N. J. (1988). Carbon disulfide: A semiochemical mediating socially induced diet choice in rats. Physiology and Behavior 42, 119-124. Galef, Jr., B. G., Mischinger, A., and Malenfant, S. A. (1987). Hungry rats’ following of conspecifics to food depends on the diets eaten by potential leaders. Anim. Behau. 35, 1234-1239. Galef, Jr.. B. G., and Wigmore, S. W. (1983). Transfer of information concerning distant foods: A laboratory investigation of the information centre hypothesis. Anim. Behau. 31,748-758. Gochfeld. M., and Burger, J. (1982). Feeding enhancement by social attraction in the sandwich tern. Behau. Ecol. Sociobiol. 10, 15-17. Gori, D. F. (1988). Colony-facilitated foraging in Yellow-headed blackbirds: Experimental evidence for information transfer. Ornis Scand. 19, 224-230. Goss-Custard, J. D. (1980). Competition for food and interference among waders. Ardea 68, 31-52. Gotmark. F. (1990). A test of the information-centre hypothesis in a colony of sandwich terns Sterna sanduicensis. Anim. Behau. 39, 487-495. Gould, J. L. (1975). Honey bee recruitment: The dance-language controversy. Science 189, 685-692. Greene, E. (1987). Individuals in an osprey colony discriminate between high and low quality information. Nature (London)329, 239-241. Greenwood, P. J. (1980). Mating systems, philopatry and dispersal in birds and mammals. Anim. Behau. 28, 1140-1 162.

42

HElNZ RICHNER AND PHILIPP HEEB

Hagan, J. M., and Walters. J. R. (1990). Foraging behavior, reproductive success, and colonial nesting in ospreys. Auk 107, 506-521. Hamilton, W. D. (1964). The genetical theory of social behaviour. I, 11. J . Theoret. Biol. 7, 1-52.

Hamilton, W. D. (1971). Geometry for the selfish herd. J . Theoret. Biol. 31, 295-311. Hebblethwaite, M. L., and Shields, W. M. (1990). Social influences on barn swallow foraging in the Adirondacks: A test of competing hypotheses. Anim. Behau. 39, 97-104. Heeb, P., and Richner. H. (1994). Seabird colonies and the appeal of the information center hypothesis. Trends Ecol. Evol. 9, 25. Heinrich, B. (1978). The economics of insect sociality. I n “Behaviour Ecology: An Evolutionary Approach, 1st ed. (J. R. Krebs and N. B. Davies, eds.), pp. 97-128. Blackwell, Oxford. Hinde, R. A. (1961). Behaviour. In “Biology and Comparative Physiology of Birds” (A. J. Marshall, ed.). Academic Press, London. Hinde, R. A., and Fisher, J. (1951). Further observations on the opening of milk bottles by birds. British Birds 44, 393-3%. Holldobler, B. (1977). Communication in social Hymenoptera. I n “How Animals Communicate” (T. A. Sebeok, ed.), pp. 418-471. Indiana University Press. Bloomington. Holldobler, B., and Wilson, E. 0. (1990). “The Ants.” Harvard University Press, London. Hoogland, J. L., and Sherman, P. W. (1976). Advantages and disadvantages ofbank swallow (Riparia riparia) coloniality. Ecol. Monogr. 46, 33-58. Horn, H. S. (I%@. The adaptive significance of colonial nesting in the Brewer’s blackbird (Euphagus cyanocephalus). ECOIORY 49,682-694. Houston, A. 1. (1993). Mobility limits cooperation. Trends Ecol. Euol. 8, 194-196. Hubbell, S. P., and Johnson, L. K. (1978). Comparative foraging behavior of six stingless bee species exploiting a standardized resource. Ecology 59, 1123-1 136. Jackson, J. A. (1975). Regurgitative feeding of young black vultures in December. Auk 92, 802-803.

Johnson, L. K., and Hubbell, S. P. (1974). Aggression and competition among stingless bees: Field studies. Ecology 55, 120-127. Kiis, A., and Moller, A. P. (1986). A field test of information transfer in communally roosting greenfinches (Carduelis chloris). Anim. Behav. 34, 1251-1255. Krebs, J. R. (1974). Colonial nesting and social feeding as strategies for exploiting food resources in the great blue heron (Ardea herodias). Behaviour 51, 99-134. Krebs, J. R. (1978). Colonial nesting in birds, with special reference to the ciconiiformes. I n “Wading Birds” (A. Sprunt, J. C. Ogden, and S. Winkler, eds.), pp. 299-314. Res. Rep. No. 7 of the National Audubon Society. Krebs. J. R., and Davies, N . B. (1993). “An Introduction to Behavioural Ecology,” 3rd ed. Blackwell, Oxford. Krebs, J. R., MacRoberts, M. H.. and Cullen, J. M. (1972). Flocking and feeding in the great tit-an experimental study. Ibis 114, 507-530. Kruuk, H. (1964). Predators and anti-predator behaviour of the black-headed gull (Larus ridibundus). Behau. Suppl. 11, 1-129. Lack, D. (1968). “Ecological Adaptations for Breeding in Birds.” Methuen, London. Laland, K. N., and Plotkin, H. C. (1991). Excretory deposits surrounding food sitesfacilitate social learning of food preferences in Norway rats. Anim. Behau. 41, 997-1005. Lindauer, M. (1961). “Communication among Social Bees.” Harvard University Press. Cambridge, Massachusetts. Loman, J., and Tamm. S. (1980). Do roosts serve as “information centers” for crows and ravens? Am. N a t . 115,285-289.

This Page Intentionally Left Blank

44

HEINZ RICHNER AND PHILIPP HEEB

Rabenold, P. P. (1987). Recruitment to food in Black Vultures: Evidence for following from communal roosts. Anim. Behau. 35, 1775-1785. Richner, H. (1989a). Habitat-specific growth and fitness in Carrion Crows ( C o r m s corone corone L). J . Anim. Ecol. 58,427-440. Richner, H. (1989b). Phenotypic correlates of dominance in carrion crows and their effects on access to food. Anim. Behau. 38,606-612. Richner, H. (1992). The effect of extra-food on fitness in breeding Carrion Crows. Ecology 73, 330-335. Richner, H., and Heeb, P. (in press). Communal life: Honest signalling and the recruitment centre hypothesis. Behau. Ecol. Richner, H., and Marclay, C. (1991). Evolution of avian roosting behaviour: A test of the information center hypothesis and of a critical assumption. Anim. Behau. 41,433-438. Seeley. T. B. (1985a). “Honeybee Ecology.” Princeton University Press, New Jersey. Seeley, T. B. (1985b).The information-centre strategy of honeybee foraging. In “Experimental Behavioral Ecology and Sociobiology” (B. Holldobler and M. Lindauer, eds.), pp. 75-90. G. Fischer Verlag, Stuttgart. Seger, J. (1991). Cooperation and conflict in social insects. In “Behavioural Ecology: An Evolutionary Approach,” 3rd ed. ( J . R. Krebs and N. B. Davies, eds.), pp. 338-373. Blackwell, Oxford. Shields, W. M. (1990). Information centers and coloniality in cliff swallows: Statistical design and analysis-a comment. Ecology 71, 401-405. Snapp, B. D. (1976). Colonial breeding in the barn swallow (Hirundo rusrica) and its adaptive significance. Condor 78,47 1-480. Stoddard, P. K. (1988). The ‘bugs’ call of the cliff swallow: A rare food signal in a colonially nesting birds species. Condor 90, 714-715. Stutchbury, B. J. (1988). Evidence that bank swallow colonies do not function as information centers. Condor 90, 953-955. Thorpe, W. H. (1956). “Learning and Instinct in Animals.’’ Methuen, London. Tristram, H. B. (1859). In On the Ornithology of Northern Africa. Ibis 1, 280. Trivers. R. L. (1971). The evolution of reciprocal altruism. Q. Rev. B i d . 46, 35-57. Trivers, R. L. (1985). “Social Evolution.” Benjamin/Cummings Press, Menlo Park, California. Vadas, Jr., R. L. (1994). The anatomy of an ecological controversy: Honey-bee searching behavior. Oikos 69, 158-166. Wagner, R. (1993). The pursuit of extra-pair copulations by female birds: A new hypothesis of colony formation. J. Theorer. Eiol. 163, 333-346. Waltz, E. C. (1982). Resource characteristics and the evolution of information centers. A m . Nat. 119, 73-90. Waltz, E. C. (1987). A test of the information centre hypothesis in two colonies of common terns, Sterna hirundo. Anim. Behau. 3548-59. Ward, P. (1965). Feeding ecology of the black-faced dioch Quelea quelea in Nigeria. Ibis 107, 173-214. Ward, P., and Zahavi, A. (1973).The importance of certain assemblages ofbirds as “information centres” for food finding. Ibis 115, 517-534. Weatherhead, P. (1983). Two principal strategies in avian communal roosts. A m . Nar. l21, 237-243. Weatherhead, P. J. (1987). Field tests of information transfer in communally roosting birds. Anim. Behau. 35,614-615. Wells, P. H., and Wenner, A. M. (1973). Do honeybees have a language? Nature (London) 241, 171-175.

INFORMATION CENTERS AND THE ALTERNATIVES

45

Wenner, A. M. (1989). Concept-centered versus organism-centered biology. Amer. Zool. 29, 1177-1197. Wenner, A. M., and Wells, P. H. (1990). “Anatomy of a Controversy: The Question of a ‘Language’ among Bees.” Columbia Univ. Press, New York. Wenner, A. M., Meade, D. E., and Friesen, L. J. (1991). Recruitment, search behavior, and flight ranges of honeybees. Am. Zool. 31,768-782. Wilkinson, G. S. (1984). Reciprocal food sharing in the vampire bat. Nature (London) 308, 18 I - 184. Wilkinson, G. S. (1992). Information transfer at evening bat colonies. Anim. Behau. 44, 501-5 18. Wilson, E. (1971). “The Insect Societies.” Harvard Univ. Press, Cambridge. Massachusetts. Wittenberger, J. F., and Hunt, Jr., G. L. (1985). The adaptive significance of coloniality in birds. I n “Avian Biology” (D. S. Farner, J. R. King, and K. C. Parkes, eds.), Vol. VIII, pp. 1-78. Academic Press, New York. Wynne-Edwards, V. C. (1962). “Animal Dispersion in Relation to Social Behaviour.” Oliver and Boyd, Edinburgh. Yom-Tov, Y., Imber, A., and Otterman, J. (1976). The microclimate of winter roosts of the starling Sturnus vulgaris. Ibis 119, 366-368. Zahavi, A. (1971). The function of pre-roost gatherings and communal roosts. Ibis 1l3, 106-109.

This Page Intentionally Left Blank

ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 24

Maternal Contributions to Mammalian Reproductive Development and the Divergence of Males and Females CELIAL. MOORE OF PSYCHOLOGY DEPARTMENT UNIVERSITY OF MASSACHUSETTS BOSTON BOSTON,MASSACHUSETTS

I. INTRODUCTION Reproduction lies at the heart of both evolution and development, and animal behavior is at its most flamboyant in the reproductive arena. Small wonder, then, that the study of reproductive behavior has captured so much attention. Of the various reproductive phenomena that have received particular scrutiny, those of sex differences are perhaps most favored. How, and why, do males and females within the same species become as different as they sometimes do? Of course, conspecific males and females are far more similar than different, but the fact of any phenotypic difference holds great interest for a developmentalist. How do reliable differences emerge during development when the genetic differences between males and females are so limited? The answer must lie in epigenetic processes within the developing individuals or in epigenetic interactions between the individual and its environment. The environment for a developing mammal during the most relevant stages is provided primarily by the mother. In this chapter, the contributions of maternally provided environments to reproductive development and to the developmental divergence of males and females are explored. Masculine sexual behavior in rodents is perhaps overrepresented in this review. That is because the current state of the literature is most complete in this research domain and because my own research has focused on this issue. 47

Copyri@I Q 1995 by Academic Press, Inc. All rights of reproduction in any form reserved.

48

CELlA L. MOORE

11. ADAPTATION AND

THE

NATURE OF SEXDIFFERENCES

It is commonly assumed that sex differences are adaptive, shaped to fit differential selective pressures on the two sexes. Differences in selective pressures for males and females intrigued Darwin and continue as central Darwinian themes (Campbell, 1972; Williams, 1975). However successful the adaptive arguments have been in accounting for the “why” of many differences, the fact of a sex difference is not sufficient for concluding that the difference is adaptive. Some traits are neutral with respect to extant selective pressures and, even when maladaptive, traits can be carried in populations for considerable periods of time, perhaps even for the life of the species (Hailman, 1982; Kirkpatrick and Lande, 1989). Males and females within a given species share many of the same selective pressures, but there are also differences in these pressures, the magnitude of which varies with the species. Thus, a character that is adaptive in males may be neutral or maladaptive in females, and vice versa. Both in speciation and sexual divergence, the problem faced is one of achieving divergence in characters to meet whatever differences in selective pressures are present, despite little in the way of genetic difference that might be used to separate either the two populations or the two sexes. The problem is even more acute for sexual divergence within a species because, with two diverging populations, the genetic differences can accumulate during long periods of reproductive isolation. By its very definition, sexual reproduction implies a continuing, high level of genetic mixing between the sexes within a species. This mixing is complete for all but the heteromorphic Y chromosome in mammals. Mammalian males have an XY and females an XX chromosome complement. This chromosomal difference might be thought to harbor a number of genetic differences, but this is not the case (reviewed by Bull, 1983). Several lines of evidence (Bull, 1983) indicate that the Y chromosome is essentially “degenerate”: despite some important functions, there are apparently few active genes on this chromosome. One fact in support of this conclusion is that a YO complement is lethal, whereas an XO is not. Another is that linkage studies in humans have found many X-linked effects, but only a handful of Y-linked effects. Furthermore, the Y is considerably smaller than the X, its cytological appearance suggests inactivity, and much of its DNA consists of short, repeated fragments. The fact that females have two X chromosomes and males have only one also has limited ability to explain divergent development. One X chromosome in each of the somatic cells of females is inactivated, which provides both males and females with only one functional X chromosome in each cell, thereby equalizing the dosage of genes carried on this chromosome across

MATERNAL CONTRIBUTIONS

49

the two sexes. In summary, male and female mammals have an equal number of freely recombining autosomes and one functional X chromosome that also freely recombines, allowing no genetic sex differences to accumulate on these chromosomes. There is a sex difference in one chromosome, but only a very small genetic difference: the Y chromosome, which is present only in males, has few functional genes, and the second X, which is present only in females, is not active. Whatever sex differences emerge during the development of mammals cannot be explained by hypothesizing the differential accumulation of genes in the course of evolution. This statement can be made even more strongly for some nonmammalian species, where two sexes having a variety of phenotypic differences can develop in the complete absence of opportunity to sequester genetic differences between the two sexes. Many fish and reptiles have environmental sex determination, where individuals with identical genotype will develop as males or females depending on such environmental variables as temperature during early embryonic stages (Bull, 1983). Therefore, the two sexes within a species may exhibit differences in a variety of behavioral, physiological, and anatomical phenotypes even though genetic differences are quite small or nonexistent. These differences emerge through regulatory and developmental processes. It is perhaps ironic that the same developmental principles that underlie the development of adaptive differences also set contraints that limit the emergence of differences when, in fact, these differences would be adaptive, or produce differences that may in fact be adaptively neutral or maladaptive (Cheverud, 1984; Gould, 1989). A.

DEFINING DIFFERENCES

1 . Descriptive Units Sex differences are typically approached initially by using functional categories of behavior. There are sex differences in such nonreproductive behavior as foraging, but most behavioral sex differences are found in some aspect of reproduction. Descriptions of sex differences usually draw attention to the fact that the sexes differ in likelihood of performing courtship, copulation, or parental behavior. One of the consequences of beginning with functional categories is that sex differences are cast in terms of functionally complementary behavior. For example, sex differences in rat copulation may be described by stating that males mount, whereas females perform lordosis. Or sex differences in the reproduction of song birds may be characterized by the observation

50

CELIA L. MOORE

that males patrol the territory while females attend the nest. These descriptions convey something about the complementarity of the behavior, either in terms of fitting together during an interaction (e.g., mount and lordosis) or carving out different specializations within the overall reproductive duties of a pair (e.g., patrolling and attending the nest). However, they juxtapose behavioral patterns that have only a functional similarity and induce one to conceive of a character in functional terms. Thus, when “copulation” is the character, lordosis and mounting become two alternative states of that character. Similarly, when “care of the young” is the character, patrolling and nest attendance become two alternative ways to provide care. The functional approach fits well with our ordinary conceptions of masculinity and femininity, two complementary suites of traits, the yin and yang of the social order. However, the concepts of masculinity and femininity map uneasily on actual males and females and pose many difficulties when used in causal explanations of behavior (Moore, 1985a). Although the behavior of a pair of nesting songbirds may reflect an efficient division of labor, there is no common ground in the causal mechanisms used in territory maintenance and incubation. This point is readily seen, but the analogous point for male and female copulatory behavior, mounting and lordosis, took many years and much effort to appreciate. Mounting was considered an index of masculinity and was thought to lie at one pole of a continuum, with lordosis, an index of femininity, at the other pole. The two behavioral patterns fit together as the copulatory organs do, and the analogy between genital anatomy and the neuroanatomical substrates of behavior was compelling. For some time, the neural mechanisms underlying masculine and feminine sexual behavior were thought to diverge from common origins in a manner similar to the development of male and female genitalia from the same embryonic tissue. However, mounting and lordosis are completely independent behavioral patterns: the ability to perform one has no bearing on the ability to perform the other. Given the appropriate contexts, the same animal may readily perform both behavioral patterns (Beach, 1971; Goldfoot and Wallen, 1978). Despite functional complementarity, there is no causal basis for placing the two behavioral patterns on the same continuum (reviewed in Moore, 1985a). It is possible to arrive at a more complete and more accurate description of sex differences when the behavioral patterns themselves are used to describe the differences, even though this approach often leads to more cumbersome descriptions than those based on functional consequences. For example, the statement that males mount whereas females perform lordosis glosses over some descriptive information. More completely:

MATERNAL CONTRIBUTIONS

51

given a particular set of internal and external conditions, males are more likely to mount than females; secondly, and quite separately, given another set of internal and external conditions, females are more likely to perform lordosis than males. This sets the explanatory problems so that they can be solved by further dissection of the behavior and by analytic studies of development and mechanism. 2 . Ability and Performance It is admittedly nonsensical from an evolutionary perspective for female mammals to engage in the motor patterns associated with mounting, intromission, and ejaculation. They have no sperm to deliver. Yet, given the right circumstances, females do perform masculine copulatory responses (Beach, 1971). This is less an evolutionaryconundrum than it might appear because in the normal course of female lives, the appropriate circumstances for masculine sexual behavior rarely present themselves. Among other things, circulating testosterone, which is critical for the behavior, is usually missing. Phenotypic expression is the relevant factor in evolutionary considerations, but developmentalists must also attend to ability. The ability to perform some behavioral pattern means that an individual has developed the underlying mechanisms. Because of the way that development works (Gould, 1977), it is probably the rule that males and females develop similar underlying mechanisms even when the probability of behavioral expression may differ markedly. This is not to say that there will be no differences in these mechanisms, Indeed, the number of known sex differences in neuroanatomy and neurophysiology with behavioral repercussions continues to grow (Goy and McEwen, 1980; Yahr, 1988; Breedlove, 1992). Nevertheless, there has been little success in relating differences in the nervous system to wholesale behavioral differences in males and females. Instead, differences at the behavioral level tend to be differences in probability, threshold, completeness, or stimulus context.

B. ARESEXDIFFERENCES ADAPTIVE?

The evolutionary function of mounting in males and lordosis in females is readily apparent. It is less apparent how mounting in females or lordosis in males should be characterized. Are these patterns maladaptive or adaptively neutral? The appropriate characterization becomes even more remote when the focus is placed on underlying ability rather than behavioral expression. Thus, a distinction should be drawn between examining the adaptiveness of behavior for a given sex and examining the adaptiveness

52

CELIA L. MOORE

of differences between the sexes. The necessity for keeping this distinction in mind becomes even more acute for nonreproductive behavior. If a functional advantage for spatial ability, let us say, is discovered for males in some species, it does not necessarily follow that it is adaptive for males and females to differ in this ability. Male meadow voles (Microtus pennsyluanicus) have larger home ranges than conspecific females, a sex difference that is related to the polygynous mating system in this species. In contrast to a monogamous congener (M. ochrogaster, prairie voles), M.pennsylvanicus males outperform females on spatial learning mazes (Gaulin and Fitzgerald, 1989). Gaulin and Fitzgerald hypothesize that the sex difference is an adaptive response to differential selection on the two sexes, with selection favoring males having greater spatial learning ability that can be put to use in establishing their breeding ranges. However, in a study of reproductive success as a function of space use, it was found that size of breeding range was unrelated to reproductive success among the adult male meadow voles in the study area (Ostfeld, Pugh, Seamon, and Tamarin, 1988). But the most reproductively successful females were those having smaller home ranges than other females in the study area. There are several conclusions and further questions that are generated by these findings. First, if understanding fitness is the goal, the relevant comparison is differential use of space among members of the same sex. Second, selective pressures on size of home range in M.pennsyluanicus may be acting onfemale home range size, rather than that of males. Third, there is a distinction between spatial ability and spatial use, although they are, of course, interrelated. The same ability may be applied differently in two contexts: males and females need not differ in ability to differ in use. Finally, engaging in spatial tasks increases spatial ability. Both male and female adolescent rats improve their performance on spatial learning tasks from the same kinds of experiences with spatially complex environments during development (Juraska, Henderson, and Muller, 1984). Because males and females share many developmental processes, it may be developmentally easier for both sexes to respond to selective pressure for some behavioral capacity rather than for only one sex to do so, even when differential reproductive success applies in only one sex (cf. Gould, 1977; Raffand Kaufman, 1983). Although any difference that is observed between the sexes may be functionally irrelevant, there are cases where the difference itself may have functional significance. For example, when males and females specialize on different food sources, this difference effectively increases the available food supply and reduces competition (Selander, 1972). Another difficulty in the way of addressing the adaptiveness of differences is posed by the developmental phenomenon of equifinality. Equifi-

MATERNAL CONTRIBUTIONS

53

nality refers to the process by which some developmental end point, performance of some particular behavioral pattern, for example, is achieved through more than one pathway. It is entirely possible for the two sexes to exhibit no difference when measurements are made at the behavioral level, but to have large phenotypic differences in the mechanisms underlying the behavioral performance. It is necessary to examine the various levels and components of a system to determine where sex differences lie and whether these differences have functional consequences. For example, there is a striking sexual dimorphism in the number of neurons in the superior cervical ganglion of rats, with males having 40% more neurons than females from two weeks of age. At two weeks of age, the mass of target tissue (submandibulargland and pineal gland) for norepinephrine, the neurotransmitter produced by neurons in the superior cervical ganglion, is not sexually dimorphic; in adults, these glands are larger in males, as is the entire body. However, functional equivalence between males and females is achieved at each age by adjusting the amount of neurotransmitter produced per neuron (Beaston-Wimmer and Smolen, 1991). Thus, one neural sex difference (number of neurons in the ganglion) is compensated by another (rate of neurotransmitter production per neuron) to achieve equifinality (match between amount of neurotransmitter and target size). The neural mechanisms underlying parental behavior in prairie voles (Microrus ochrogaster) provide another excellent example of equifinality (Bamshad, Novak, and DeVries, 1993). Prairie voles have a monogamous breeding system with fathers participating in parental care. Paternal mammals do not experience the hormonal changes of pregnancy, parturition, and lactation that are known to be important regulators of maternal behavior (Rosenblatt and Siegel, 1981) and must achieve parental behavior through some alternative means. There is now evidence that parental behavior in male and female prairie voles is controlled by two different neural systems. Male prairie voles exhibit a change in the vasopressin innervation of the lateral septum when they become parents. Neither female prairie voles nor male meadow voles (Microtus pennsyluanicus), a species that exhibits no paternal behavior, exhibit similar changes upon becoming parents (Bamshad et al., 1993). Experimental manipulation of vasopressin in the lateral septum of male prairie voles implicates this system in the onset of paternal behavior (Wang and DeVries, 1993; Wang, Ferris, and DeVries, 1993). These examples of similar functional outcomes achieved by different neural mechanisms complicate the question of whether sex differences are adaptive. Some neural differences are inconsequential, because they are offset by other compensatory differences. Other neural differences

54

CELlA L. MOORE

are alternative pathways to the same behavioral end. The question of difference must be addressed at multiple levels.

111. DEVELOPMENTAL PROCESSES

A. REGULATORY AGENTS Regulatory developmental mechanisms allow small differences to be magnified during the course of ontogeny (Gould, 1977; Edelman, 1988; Purves, 1988). A powerful example of this phenomenon is heterochrony, where small differences in developmentaltiming can account for the evolution of striking morphological novelties (Gould, 1977). Gonadal steroids are also powerful regulatory agents in development, with effects that cascade over many levels of organization and many organ systems within the body. Various small differences in either production (type of steroid and quantity or timing of secretion) or responsiveness (enzyme activity, receptor formation, developmental state of target) to gonadal hormones can occur during the course of development (e.g., Yahr, 1988; Breedlove, 1992). Through cascading effects, small hormonal differences can be magnified and elaborated into highly varied developmental outcomes. Thus, multiple phenotypes can occur even with identical structural genotypes through hormonally mediated means. The environmental milieu of the developing organism is another source of regulatory agents that can produce varied outcomes from the same beginnings. Variations in temperature, light, pH, and other dimensions of the physical environment can produce morphological variants from genetically identical origins (Lewontin, 1982; Gottlieb, 1992). Parents select or provide major aspects of the developmental environment in many animal species. Because of pregnancy and lactation, this environment is primarily a maternal one in mammals. B.

MATERNALEFFECTS

Geneticists have recognized that there are sometimes positive or negative correlations between an individual’s genotype and its maternal environment, because there are correlations between offspring and maternal genotype and between maternal genotype and the environment she provides to her offspring (Arnold, 1987; Kirkpatrick and Lande, 1989; Roubertoux, Nosten-Bertrand, and Carlier, 1990; Atchley and Hall, 1991). These indirect pathways can lead to some phenotypic correlations between parent and offspring. For example, there are negative correlations between

MATERNAL CONTRIBUTIONS

55

offspring weight and maternal weight in some animals. This is accounted for by the fact that larger dams have larger litters. Young from larger litters tend to be smaller, because of limits on the total mass that can be gestated (Arnold, 1987). Maternal effects refer to the fact that correlations between maternal and offspring phenotypes are greater than those between paternal and offspring phenotypes. The greater maternal correlation occurs because the mother provides so much of the developmental environment for the offspring. At the earliest stage, the cytoplasmic environment of the zygote comes primarily from the egg. In mammals, there are also the intrauterine environment, milk, and maternal behavior. Experimental studies have been able to tease out some of these maternal effects, demonstrating that some inherited phenotypes are due to the correlated maternal variable. Both prenatal and postnatal maternal effects have been documented for a range of behavioral and physiological outcomes (Roubertoux et al., 1990). Geneticists are interested in maternal effects because these effects set constraints on responses to artificial or natural selection, making change easier in some cases and difficult or impossible in others (Arnold, 1987; Kirkpatrick and Lande, 1989).The work on maternal effects within genetics is also highly significant for developmentalists. It is certainly important that variation in the maternal environment can produce different offspring phenotypes from identical origins (Roubertouxet al., 1990).Genetic experiments on maternal effects necessitate complex designs, and most work on the developmentalconsequences of the maternal environment has been done without genetic manipulations. Male and female siblings provide natural genetic variants, and the study of maternal contributions to the differential development of males and females provides an accessible means for a detailed analysis of the variety of ways that the maternal environment can regulate offspring development.

Iv. A.

MATERNALCONTRIBUTIONS TO T H E DEVELOPMENT OF REPRODUCTION IN MAMMALS

PRENATAL EFFECTS

The intrauterine environment offers rich opportunities for the mother to affect the development of mammalian fetuses (Alberts and Cramer, 1988). These maternal contributions may be produced by various agents that travel from the maternal to the fetal circulatory system, through the placental blood connection. Among the possible agents are nutrients,

56

CELIA L. MOORE

immune factors, hormones, viruses, and toxins. Furthermore, the behavioral and physiological processes of the mother generate vestibular, tactile, thermal, auditory, and chemosensory stimuli that are detectable by the fetus. Mothers exhibit circadian rhythms in activity, body temperature, and other physiological processes (Reppert, 1985); they locomote, groom themselves, and engage in a range of other behavioral patterns that can generate stimuli detectable by fetuses (Previc, 1991; Ronca, Lamkin, and Alberts, 1993);they ingest and digest food, altering both the nutritive and stimulative aspects of the intrauterine environment in the process; and maternal conditions affect the volume, viscosity, and chemical composition of the amniotic fluids, which can both constrain and stimulate fetal behavior (Smotherman and Robinson, 1988). Among the factors comprised by the prenatal maternal environment are many that have long-term consequences for reproductive development in the offspring. Some consequences are reached through indirect pathways, and not all are specific to reproduction. For example, the developing circadian rhythms in young rats owe their origin to the patterned transmission of maternal factors during late gestation (Reppert, 1985).These circadian rhythms eventually contribute to many physiological functions, including the synthesis and release of reproductive hormones (Morin, 1986). To take another example, rats, like many other animals, begin the process of identifying their species during the prenatal period. Initially, olfactory cues that are learned prenatally are used to direct the first nursing episode to the mother’s nipples (Hofer, Shair, and Singh, 1976; Teicher and Blass, 1977; Pedersen and Blass, 1982), but these early interactions serve as a foundation for further learning about the stimulus properties of conspecifics (Rosenblatt, 1976, 1983; Pedersen, Greer, and Shepherd, 1988), eventually exerting effects on the olfactory guidance of reproductive behavior (D’Udine and Alleva, 1983;Moore, 1985b; Fillion and Blass, 1986a, 1986b). These prenatal maternal influences are indeed important, but most of them are there for all offspring and cannot, therefore, contribute directly to sex differences or to within-sex individual differences. There is a necessary, underlying constancy to intrauterine environments within a species because there is a close relationship between uterine conditions and physiological processes that are necessary either for maternal life or for the maintenance of pregnancy. Despite the sharp limits to the degree to which the intrauterine environment can differ from one fetus to another, some differences are possible, and these may be consistently different for the two sexes. One such example is provided by the placenta, which is reliably larger for male fetuses in rats (Ward, Karp, and Aceto, 1977) and mice

MATERNAL CONTRIBUTIONS

57

(Bulman-Flemingand Wahlsten, 1991). It is possible that biased prenatal environments contribute to the developmental divergence of the sexes. This possibility is most obvious when single offspring are gestated, but can also occur, in a probabilistic fashion at least, when there are fetal litters with a skewed sex ratio. Under these circumstances, it is possible that the mother will adjust the intrauterine environment differently for males and females. For example, the pregnant animal might perform different amounts of particular activities as a function of the sex of the fetus or the sex composition of her fetal litter, which could have the consequence of altering the kind or amount of stimulation impinging on the young. No one has investigated this possibility directly, but the related question of whether fetal androgens can increase the performance of androgendependent behavior by pregnant females has been studied in primates and gerbils.

1 . Fetal Testosterone and Maternal Behavior It has been established for some time that testosterone is produced by the testes of fetal male primates. The level of testosterone measured from umbilical arteries is greater in male than in female rhesus monkeys from at least day 60 of gestation (Resko, Malley, Begley, and Hess, 1973). Thus, fetal androgens can elevate the level of androgens in maternal circulation by passing through the placental connection. As an indirect indicator that androgens of fetal origin may have biological significance to the pregnant female, Sackett (1981) reported a relationship between fetal sex and maternal agonism. Specifically, he examined a large number of colony records in pigtail macaques and found that females were more likely to be wounded during their pregnancy if they were pregnant with a male fetus than if they were pregnant with a female fetus. One possibility is that androgens originatingfrom the fetal males affected either the behavior or the stimulus properties of the mother so that fighting was more likely to occur. An alternative possibility is that there was something about the females that were more likely to fight that set a positive bias to the probability that they conceived males. The Sackett study, though provocative, was a retrospective study with no data to indicate the behavioral sources of the increased wounding of the females that were pregnant with male fetuses. Furthermore, a subsequent study with direct behavioral observations failed to substantiate an effect of fetal sex on aggressive interactions in pregnant females of another primate, stumptail macaques (Nieuwenhuijsen,Slob, and de Neef, 1988). However, recent work with Mongolian gerbils lends new support to the idea that fetal androgens can effect the behavior of the mother

58

CELIA L. MOORE

(Clark, Crews, and Galef, 1993). Gerbil litters are sufficiently large that both sexes are likely to be represented in each litter. However, the number of males and females varies across litters. It is possible, therefore, to examine the behavior of pregnant animals as a function of the number or proportion of males in their litter. This research strategy was followed by Clark et al. To control for possible maternal differences that might both bias the sex ratio and affect behavior in the pregnant animal, they compared the same females during two successive pregnancies. Scent-markingbehavior, an androgen-sensitive behavior that involves lowering the body and dragging the ventral scent gland across an object or the floor, was measured, as was circulating titers of testosterone. They found that the change in scent marking from one pregnancy to the next was positively correlated with the change in number of males in the litter. The number of males in a litter was also positively correlated with the level of testosterone measured in the dams. It is undoubtedly the case that when a pregnant rodent rubs its ventrum across surfaces, pressure and other stimulation will be provided to the fetuses gestating in the abdominally placed uterine horns. There are currently no available data to determine whether stimulation from this source is sufficiently great to be distinguished from background levels of similar stimulation from other maternal activities, or whether biased fetal sex ratios can generate biologically meaningful differences in the level of stimulation received by male and female fetuses. These possibilities, although difficult to investigate, are interesting. It is likely that the overwhelming majority of intrauterine factors are identical for males and females. Nevertheless, sex differences present in the fetuses at the time of exposure to some factor may result in different developmental consequences. The most unambiguous prenatal sex difference is that males have testes and females ovaries. Thus, the nature of gonadal response to some agent is likely to differ for the two sexes. Prairie voles (Microtus ochrogaster) provide an interesting example (Nelson, 1991). These voles are opportunistic breeders with a short life-span. They enter breeding condition in response to a dietary metabolite (6-MBOA) obtained from young shoots of vegetatively growing plants. The sons of dams fed a diet containing 6-MBOA during pregnancy had larger testes and higher sperm counts than controls had at 6 weeks of age. Although weaned females respond to 6-MBOA in their own diet with ovarian growth, there was no effect of maternal diet on reproductive development in female offspring (Nelson, 1991). There are differences in the hormones produced by testes and ovaries prenatally, which can lead to differences in the internal hormonal milieu of the two sexes. This, in turn, may lead them to respond differently to

MATERNAL CONTRIBUTIONS

59

the same maternally provided agent. It may well be, for example. that exposure to the same level of some odorous substance or the same tactile stimulus will have a different effect on two fetuses. Hormones can alter the time of onset or rate of progression of developmental changes in the brain or elsewhere (e.g., Yahr, 1988; Tobet and Fox, 1992).and this could interact with a particular stimulus level to produce divergent developmental outcomes.

2. Fetal Response to Maternal Stress The most thoroughly studied example of sex differences in the consequences of the same prenatal event is provided by offspring response to maternal stress during the gestation of laboratory rats and mice. The developmental repercussions of maternal stress during pregnancy are substantially different for male and female offspring because of sex differences in embryonic production of gonadal steroids. In rats, much of the reproductive development that is regulated by early gonadal steroids occurs during the neonatal period. Sex differences in the neonatal hormonal milieu are produced by a brief surge of testicular testosterone shortly after birth in males, followed by a steady production of this hormone at lower levels throughout the neonatal period (Baum, Brand, Ooms, Vreeburn, and Slob, 1988; Dohler and Wuttke, 1975). Despite the importance of the neonatal period, sexual development and sex differences in development begin prenatally in this species. This prenatal development is also regulated by steroids. Fetal testosterone levels are remarkably similar for male and female rats except for gestational days 18-19, when males produce a surge of testosterone from their testes (Weisz and Ward, 1980). This surge is primarily responsible for the sex differences in prenatal development. When pregnant rats of the Sprague-Dawley strain are stressed by repeated restraint and exposure to the heat and bright light of floodlamps, their male fetuses exhibit a surge in testosterone production on gestational day 17, followed by low levels of testosterone during days 18-19, which is the period when unstressed males produce high levels of this hormone (Ward and Weisz, 1984). In Ward’s (1972) original study, the male offspring of the stressed dams exhibited lower thresholds to perform feminine copulatory behavior and higher thresholds to perform masculine copulatory behavior that those of males from unstressed dams. Although there are substantial differences among studies from different laboratories that may be related to both the strain of the rat and the nature of the prenatal stressor (e.g., Dahlof, H k d , and Larsson, 1977; Chapman and Stern, 1978), under some circumstances prenatal stress can reduce sex differences in the performance of sexual behavior (Ward, 1992). The most

60

CELlA L. MOORE

consistently reported effect is that males from stressed dams are more likely to exhibit lordosis behavior when treated as adults with estrogen and progesterone and stimulated by a mounting male partner (i.e., they behave more like a female and less like a normally reared male with respect to this behavior). The effects of maternal stress during gestation are apparently due to elevated levels of corticosteroids of maternal origin that lead to a temporal displacement of the fetal testosterone surge to a period too early to affect those processes that result in the typical degree of divergence between the sexes (Ward and Weisz, 1984). Ward (1992) reviewed recent studies and concluded that the relevant difference between males from stressed and unstressed mothers is adequate levels of testosterone after day 17 of gestation. (Birth typically occurs on the 22nd day of gestation in SpragueDawley rats.) Indeed, when embryonic development occurs under conditions of prenatal stress, the degree of sexual dimorphism in the medial preoptic nucleus (Anderson, Rhees, and Fleming, 1985) and in the lumbar spinal cord (Grisham, Kerchner, and Ward, 1991) is reduced. Both of these structures are typically larger in males, a fact that has been traced to the testosterone that is available to males during early development (Gorski, Gordon, Shryne, and Southam, 1978; Breedlove and Arnold, 1983). Herrenkohl (1979, 1986) found that maternal stress during pregnancy can also disrupt some aspects of reproductive development in the female offspring. Specifically, the offspring of stressed dams were less likely to become pregnant or to produce viable pups. Whether these effects result from a direct action of maternal cortiocosteroids, a changed gonadal steroid environment, or some direct effect of the specific manipulations used to induce stress (heat-light-restraint) is unknown. Maternal stress during gestation can also alter offspring maternal behavior (Kinsley and Bridges, 1988),aggression (Harvey and Chevins, 1985),and other aspects of behavioral development, which may at best be only indirectly related to reproduction (e.g., exploration; performance on learning tasks) ( Joffe, 1969). The currently favored mechanism for many of the effects of maternal stress involves the action of maternal corticosteroids on the timing or amount of testosterone that is produced by male embryos during late gestation. 3. Uterine Position and Effects of Siblings' Hormones

Even under relatively stable, nonstressful environmental conditions, there is substantial within-sex variation in the level of intrauterine testosterone in some rodents. There are local variations in the amount of this hormone because it is produced by the developing fetuses, a phenomenon

MATERNAL CONTRIBUTIONS

61

first reported by Clemens (1974). The production of testosterone affects the internal milieu of the animals producing the testosterone and, because some of the hormone diffuses away from the producing fetus, it also affects the milieu of neighboring fetuses. This diffusion may occur across membranes or through the local circulatory system. Male fetuses produce more testosterone than females. Thus, the more males there are in a local region of the uterus, the higher will be the local testosterone titer. Female Norway rats (Clemens, 1974; Clemens, Gladue, and Coniglio, 1978; Meisel and Ward, 1981), house mice (Gandelman, vom Saal, and Reinisch, 1977), and Mongolian gerbils (Clark and Galef, 1988) that are gestated in positions with locally high levels of testosterone exhibit morphological, physiological, and behavioral evidence that elevated prenatal testosterone has affected their development. These effects are greater when the number of neighboring males is greater. There are species differences in whether effects are obtained from flanking, contiguous males or from males located “upstream” (i.e., in the more caudal part of the uterine horn). Although the effects of neighboring males are perhaps more obvious when measured in females, they can also be detected in males. Thus, males gestated near other males are exposed to more testosterone than are males gestated near females, and this can affect their reproductive development (Lephart, Fleming, and Rhees, 1989; vom Saal, 1989; Clark, Crews, and Galef, 1991; Clark, Tucker, and Galef, 1992). Furthermore, there is some evidence that neighboring female fetuses can reduce the amount of testosterone produced by male fetuses, perhaps by negative feedback effects of their hormones on gonadotropic releasing hormones in the males (Vomachka and Lisk, 1986). If a maternal rodent can affect either the number or the local concentration of males in her uterus, she may be able to alter the development of those offspring characteristics that are sensitive to the level of prenatal testosterone. There is recent evidence that such maternal effects do occur in Mongolian gerbils. Using a large sample of gerbils delivered by Cesarean section during the course of several experiments, Clark and Galef (1990) kept records of the sex distribution of fetuses in the left and right uterine horns. They found that males and females were distributed nonrandomly, such that more than half of the fetuses in the right horns were males and more than half of the fetuses in the left horns were females (Fig. 1A). This biased distribution makes it more likely that males will develop in the local environment of other males, exposed to higher testosterone levels, and that females will develop in the local environment of other females, sequestered from elevated testosterone levels. Similar biases were not found in mice (Clark, Galef, and vom Saal, 1991). There may other reasons for higher testosterone levels in the right horn,

62

CELIA L. MOORE 70

70 A.

Diswbution in homs

8. Composition of litter

60

f 20

10

Right

Uterine Horn

Males Females Dam's Flanking Siblings

FIG. 1. Departures from randomness in the intrauterine distribution of male and female fetuses in Mongolian gerbils. (A) Fetuses in the right uterine horn are significantly more likely to be male, whereas those in the left horn are more likely to be female. Data from Clark and Galef (1990). (B) Dams that were themselves gestated between two males have male-biased litters, whereas dams gestated between two females have female-biased litters. This relationship is correlated with age at reproductive maturity: females developing in ufero between flanking males both mature later and have male-biased litters. Data from Clark, Karpiuk, and Galef (1993).

however. Although there is no evidence for a nonrandom distribution of rat fetuses across the two horns, there is evidence of a greater androgenic effect in the right horn of rats: the anogenital distances (a testosteronesensitive morphological feature) of fetuses in the right horn were longer than the distances of those in the left horn (Lephart et al., 1989). This difference may reflect differences in blood flow to or within the two horns. In rats, there are also differences in local environments along the ovariancervical extent of a horn. Testosterone is more likely to affect neighboring females when males are located on the cervical end of the horn (Meisel and Ward, 1981),and rat fetuses gestated at midhorn locations are heavier than those gestated at the ends of the horn (Lephart et al., 1989). In some mouse strains, however, both fetuses and placentas occupying either cervical or ovarian ends of the horn are heavier than those in the middle location (Bulman-Fleming and Wahlsten, 1991). There are more fetuses in the right than in the left horn of mice (Bulman-Flemingand Wahlsten, 1991; Clark, Galef, et al., 1991), and placentas are heavier in the right horn (Bulman-Fleming and Wahlsten, 1991).

MATERNAL CONTRIBUTIONS

63

Microenvironmental differences as a function of uterine location have the consequence of increasing individual differences in developmental outcome (including degree of difference between a given male and female), but they can produce reliable divergence between the sexes only if there is a nonrandom distribution of fetuses within uterine horns or reliably biased sex ratios within particular litters. These biases are present under some circumstances. For example, rats that mate during their postpartum estrus or late in the day of their cycling estrous period have female-biased litters (Hedricks and McClintock, 1982, 1990). Two mechanisms that bias the sex ratio within the uterine environment have been reported for Mongolian gerbils. Not only are males and females distributed unequally in the left and right horns (Clark and Galef, 1990), but there is a bias to sex ratios within litters that differs for individual mothers. Female gerbils can be divided into two groups on the basis of age at reproductive maturity. Late-maturing females have litters with a greater percentage of males, and early-maturing females have a greater percentage of females. Perhaps for this reason, late-maturing dams (and their fetuses) have higher levels of circulating androgen during pregnancy than do early-maturing dams (Clark, Crews, et al., 1991). Female offspring born to late-maturing dams have a greater likelihood of developing near males in utero, which in turn increases the probability that they will become late-maturing dams and have male-biased litters (Fig. IB). Thus, there is a nongenetic inheritance of this reproductive pattern, which is mediated by intrauterine exposure to sibling testosterone (Clark, Karpiuk, and Galef, 1993). 4 . Intrauterine Environments and the Origin of Species Differences

Some species differences in the degree of sexual divergence can be traced to differences in the prenatal environments provided by mothers. Like other rodents, male and female hamsters share uterine environments, and it is possible for fetuses to share their gonadally produced hormones. But golden hamsters have a shorter (16 vs. 22 day) gestation than do rats and are more altricial at birth. Androgen levels are therefore much lower in fetal hamsters (Vomachka and Lisk, 1986). Because female hamsters are never exposed to significant amounts of intrauterine testosterone from their siblings, and sex difference in probability of mounting is consequently greater in hamsters than in rats (Beach, 1971). The rat-hamster difference illustrates that the early hormonal environment can be modified by a change in the timing of parturition relative to the extent of neuroendocrine development in the offspring. Heterochrony in these processes may explain the origin of some novel patterns among related species in the details of sexual differentiation and the degree of

64

CELIA L. MOORE

sexual divergence. In these instances, the role of the mother is to alter the possibility of exposure to sibling hormones by adjusting the relative time of birth. Other species differences in the degree of sexual divergence have arisen through more direct maternal effects on the prenatal hormonal milieu. Nowhere is this more evident than in spotted hyenas (Crocuta crocuta). Unlike other hyena species, female spotted hyenas have an external genital morphology that is similar to that of males-an erectile clitoris as large as the male’s penis and fused labia with a scrota1 appearance. The internal reproductive organs follow the typical mammalian female pattern, however, and females have no difficulty supporting a pregnancy. Mating occurs through an enlarged and elastic urogenital meatus, which opens at the tip of the clitoris. The erectile clitoris is tucked into the abdomen during copulation. Delivery occurs through this same route, which is not large enough to accommodate the fetus and routinely tears during the birth process (Glickman, Frank, Licht, Yalcinkaya, Siiteri, and Davidson, 1992).

The unusual genital morphology of female spotted hyenas can be traced developmentally to unusually high levels of testosterone in the intrauterine environment during the time of external genital development. Adult female hyenas produce high levels of androstenedionefrom ovarian tissue, particularly during pregnancy. Androstenedione can be aromatized to estrogen if the appropriate enzymes are present, and placental tissue in other mammals (e.g., humans) aromatizes androstenedione and other androgens of maternal origin to estrogen. However, the placenta of spotted hyenas does not have aromatizing capacity. Androstenedione is, therefore, metabolized to testosterone (which is the other possible metabolic pathway) by the action of 17/3-hydroxysteroiddehydrogenase. The testosterone passes freely through the placental circulation in both directions, yielding high levels of fetal exposure to this masculinizing hormone and high levels of circulating testosterone titers in the pregnant female (Glickman et al., 1992).

The spotted hyena is the only species of hyena with masculinized female genitalia. It is likely, therefore, that this developmental pattern resulted from the loss of aromatase in placental tissue in this species, with the resulting accumulation of high levels of testosterone converted from maternal ovarian androstenedione by the placenta and conveyed through the umbilical vein to the fetus. It is instructive in this regard that a female human infant has recently been reported with extensively masculinized genitalia as a result of aromatase deficiency in the placenta (Shozu, Akasofu, Harada, and Kubota, 1991).

MATERNAL CONTRIBUTIONS

65

B. POSTNATAL EFFECTS I.

General Effects of the Maternal Environment Geneticists who study maternal effects typically divide the postnatal maternal environment into two portions: the nutritive and the stimulative (Atchley and Hall, 1991; Roubertoux etal., 1990). Milk is the major, usually only, source of nutrients in young mammals, but milk is a complex substance that contains far more than nutrients, including hormones and immune factors. It is also a source of gustatory and olfactory stimuli. Both the production and composition of milk can be affected by stimuli and other factors (e.g., urinary water; Friedman, Bruno, and Alberts, 1981) from the young. It is difficult to maintain a clear separation between the stimulative and the nutritive aspects of the maternal environment within developmental studies, because the two are entwined for both the mother and the offspring (e.g., Alberts and Gubernick, 1983). Variation in milk composition can have significant reproductive consequences and may contribute to sex differences, a possibility that has received little attention. Hormones relevant to sexually differentiated behavior, for example, can be transmitted through milk. Medroxyprogesterone acetate (MPA),a progestin that has been reported to have both antiandrogenic and androgenic effects, that had been injected into lactating rats was subsequently measured in the milk, and it can have physiological and behavioral effects on offspring when transmitted through this route (Birke and Sadler, 1983, 1985; Holzhausen, Murphy, and Birke, 1984). Play behavior is reduced in juvenile offspring as a result of MPA injected into the mother during lactation and transmitted to the offspring through the milk (Birke and Sadler, 1983), and there are apparently changes in the stimulus characteristics of pups as a result of this treatment, which have the effect of increasing maternal licking of the affected pups (Birke and Sadler, 1985). Stimuli available to a young mammal from the maternally provided environment arise from the physical presence of the dam, her behavior when with the young, and some consequences of her behavior, such as a burrow or nest, or the siblings in the cluster of offspring she maintains. There is individual variability within the stimulative postnatal environment, but many aspects are ubiquitous within the species because conspecific dams share many behavioral and physical characteristics. The consistency within the maternal environment provides reliable foundations for species-appropriate development of social behavior, food preferences, predator-prey relations, and other behavior (West and King, 1987).

66

CELlA L. MOORE

a. Developing Attractions to Social Stimuli. Young animals become familiar with the stimulus properties of conspecifics through exposure to self-generated or socially provided stimuli. Stimulus familiarity can be used later in life to restrict the range of potential mates to those sharing the familiar stimuli. This is the well-studied phenomenon of sexual imprinting (Bateson, 1978, 1983). As Bateson (1983) has emphasized, a preference for a mate that is only slightly different from the familiar early companions is an effective means of achieving optimal outbreeding. Most of the work on sexual imprinting has been done in birds, but there are studies to suggest similar processes in mammals. Experimental rearing with parents having altered stimulus properties has been carried out in several rodent and sciurid species. This rearing procedure can change social preferences, including mate preferences (for reviews, see D’Udine and Alleva, 1983, and Holmes, 1988). Cross-fostering is the most commonly used experimental rearing procedure. Whether fostering is carried out between species or between strains, most studies report that adult males or females exhibit an increased likelihood of affiliating with companions that are similar to the foster parent and, conversely, a reduced likelihood of affiliating with individuals that are like the biological parent. In some, but not all, cross-fostering studies, these altered preferences have been associated with sexual behavior, and may be related to mate choice (D’Udine and Alleva, 1983). The stimuli guiding both affiliativebehavior and mate choice in mammals are likely to be olfactory (Cheal, 1975). In addition to cross-fostering, therefore, some investigators have applied artificial odors to parents or other companions and examined later responses to these odors in social contexts. In one such study, female mice from two inbred strains were reared with parents bearing the scent of a violet perfume (Violette de Parme). As adults, females from one strain (but not the other) spent more time near males scented with the perfume, whereas control females had no preference (Alleva, D’Udine, and Oliverio, 1981). Thus, rearing with artificially scented parents can shift social responsiveness toward conspecifics bearing the artificial scent, but this does not always occur. The bias set by the parental environment can be either reinforced or reversed by subsequent social experience, depending on whether postweaning companions are from the same or different strain or species (Albonetti and D’Udine, 1986). Familiarity can be achieved through interactions with parents and siblings and through exposure to self-generated stimuli (phenotype matching). Both phenotype matching and familiarity with parents and nestmates are effective means for distinguishing close kin from unrelated or distantly related individuals. Thus, preference for slightly unfamiliar olfactory stim-

MATERNAL CONTRIBUTIONS

67

uli can be used to achieve optimal outbreeding in mammals (Gilder and Slater, 1978; Holmes, 1988) as in birds (Bateson, 1983). There is evidence that female mice prefer the slightly unfamiliar over the very familiar or the very unfamiliar, although how these distinctions related to the kinship categories used in these studies (sibling, same strain, other strain) varies with the strain (D’Udine and Alleva, 1983; Gilder and Slater, 1978). There are, then, two steps involved in arriving at a choice of mating partner-an earlier step in which conspecifics assume attractiveness and a later step in which distinctions among conspecifics are made (Bateson, 1987). Initial attraction to conspecific olfactory cues develops gradually during the early preweaning period in rats. In the first two weeks after birth, thermotactile cues predominate in guiding approach responses to both the dam and siblings within the huddle; olfactory cues begin to predominate only on day 15 (Alberts and Brunjes, 1978). Indeed, neonatal rats find a warm plasfic tube as attractive for a huddling “partner” as a conspecific. The attraction to olfactory cues develops through an association between odor and warmth, such as that provided by the warm, odorous bodies of the dam and littermates or by the artificial pairing of foreign odors and contact with a warm inanimate surface (Alberts and May, 1984). Contact with a plastic tube filled with warm water was as effective as an active, lactating dam in conditioning approaches to novel odors during the first two weeks of life (Alberts and May, 1984). Although warmth is a sufficient reinforcer in rats, various other stimuli can also reinforce early filial attraction to novel odors. Attractions are formed to odors paired with tactile stimuli, such as that provided by maternal licking and handling, contact with the body of the moving dam, or artificial brush stimulation (Coopersmithand Leon, 1984;Pedersen, Williams, and Blass, 1982; Sullivan and Leon, 1986); to odors paired with suckling or milk ingestion (Fillion and Blass, 1986a; Johanson and Terry, 1988); or even, in 6-dayold or younger rats, to odors paired with mildly painful electric shock (Camp and Rudy, 1988). Presumably, with repeated exposure to one’s own olfactory profile or to those of particular conspecifics in the rearing environment, sufficient familiarity is eventually obtained to allow individuals to make distinctions among conspecifics. These distinctions allow particular social responses to be directed preferentially to particular individuals in the social surround. To date, most of the experimental work on this second stage has been addressed to issues of kin selection and optimal outbreeding (Bateson, 1983; Holmes, 1988; Holmes and Sherman, 1983). However, animals also face the tasks of distinguishingbetween male and female conspecifics and among the different stages of reproductive readiness within a sex. Although this area has received experimental attention, most of the research

68

CELIA L. MOORE

has approached the question within a hormonal context, exploring whether and how developmental or adult hormones might affect the expression of mate preference (Adkins-Regan, 1989). There are numerous hormone-dependent, sexually dimorphic characters that can serve as reliable indicators of sex and reproductive status. These range from the olfactory cues that are likely to be used by mammals to the vocalizations and visual displays that are more likely to characterize birds. Despite the prevalence of these dimorphic stimuli, it is unclear how animals become able to link particular stimuli with the distinctions that need to be made before sexual behavior is addressed to the appropriate partner. For example, the olfactory differences between male and female rodents are readily distinguished, but a preference for the odor of the opposite sex or for the odor of a sexually active partner may need to be acquired. In rats, at least some of this acquisition occurs during the course of initial sexual experiences. Carr, Loeb, and Dissinger (1965) reported that naive, adult, male rats spent equivalent amounts of time investigating screened compartments containing estrous and diestrous females in simultaneous choice tests within their home cages, whereas sexually experienced males spent more time investigating stimuli from estrous females. Furthermore, olfactory stimuli left by housing the females in the compartments for two hours were sufficient to mediate this preference. Using a T maze choice apparatus, Le Magnen (1952) had previously reported that sexually naive males were more likely to approach the odors of estrous females. It is important to take the particular method used to measure choice behavior into account when assessing the results of these studies. It is an open question whether either measure of choice behavior corresponds to the likelihood that a mating attempt will be directed preferentially toward the source of the “chosen” stimulus. Regardless of experimental procedure, however, neither juvenile (Le Magnen, 1952) nor castrated (Le Magnen, 1952; Carr et al., 1965) male rats exhibited a preference for estrous odors. Hamster females have a vaginal gland that produces large quantities of an odorous secretion that is attractive to both naive and sexually experienced males (Murphy, 1980). Although sexual experience does not affect approaches to vaginal secretion, it does affect the likelihood that stimuli (e.g., anesthetized male or female hamsters) bearing the odor will be mounted (Macrides, Johnson, and Schneider, 1977; Murphy, 1980). Female hamsters are aggressively territorial and live alone with their dependent young in dispersed burrows. They use a scent-marking behavior to deposit odors from the vagina onto the substrate, and their marking behavior increases substantially both shortly before estrus and during late lactation. This has the effect of attracting males to their vicinity in time for mating (Johnston, 1990).

MATERNAL CONTRIBUTIONS

69

Exposure to maternal odors can provide the developmental background for attraction to conspecific estrous females. The similarity of lactating and estrous female odor was noted by Macrides et al. (1977). Secretions taken from either estrous or lactating Syrian golden hamster females were equally attractive to adult males and equally effective in eliciting mounting behavior when placed on the genitals of anesthetized stimulus males. From about 10 to 16 days of age, hamster pups show great interest in vaginal deposits ( Johnston and Coplin, 1979).Furthermore, sexually inexperienced Syrian (Mesocricetus auratus) and Turkish (Mesocricetus brandti) hamster males exhibit a preference for conspecificestrous females when presented with a choice between the two species (Murphy, 1980). The species preference requires olfaction, but the vomeronasal system is not necessary. Vaginal secretions provide sufficient, although perhaps not the only, stimuli for the sexually naive males to express a preference. Experience with maternal odors provides a partial explanation for the species preference. In choice tests between odors from the conspecific and foster species, cross-fostered males from both species exhibited a significantlyreduced choice of conspecific odor as adults, but their preferences were not reversed (Murphy, 1980). As Murphy pointed out, these males were also exposed to their own odor and to the odor of conspecific siblings during development, and these odor sources may be important. Social responses are likely to involve complex mixtures of odors. Johnston (1990), for example, concludes that hamsters make distinctions among species by processing odor patterns, which is a function of the main olfactory system. There is also a hormonal consequence in male hamsters of exposure to vaginal secretions, but this seems to be mediated through a rather circumscribed vomeronasal pathway. Both sexually naive and sexually experienced adult males that are exposed to vaginal secretions or to a female in estrus will respond with an increase in androgen secretion, even when primary olfaction is eliminated by peripheral zinc sulfate. However, when the vomeronasal system is removed, only sexually experienced males will exhibit androgen secretion, and then only when interacting with the estrous female. This suggests that the functionalconnection between vaginal odor and androgen secretion is mediated by the vomeronasal system and has already been established for sexually naive males, perhaps through interactions with their mother. However, other stimuli, including nonolfactory stimuli, can acquire this function during sexual experience (Johnston, 1990). Fillion and BIass (1986a, 1986b)reported that young rats learn about the odor of estrous females during interactions with their dams. Anesthetized females in estrus were as effective as anesthetized, lactating females in eliciting probing responses from pups, but nonestrous adult females did not elicit probing (Fillionand Blass, 1986a).Probing is a behavioral pattern

70

CELIA L. MOORE

exhibited by pups as a preliminary to nipple attachment. This suggests an overlap in the odors of lactating and estrous females. Fillion and Blass (1986a, 1986b) hypothesized that the ubiquitous association between the suckling and ingestion of milk, on one hand, and the odors that are normally found on the ventral surface of a lactating female, on the other, will lead to an attraction to the maternal odors during the preweaning period of rat pups. Further, they speculated that the similarity of these odors to those produced by estrous females might affect the mating behavior of adult male offspring. As described above, dams provide reinforcers other than milk and the opportunity to suckle, namely warmth and tactile stimulation, that also contribute to the formation of attraction to maternal odors. Whether suckling or other reinforcers underlie the preferences, the perceived similarity to rat pups of estrous and lactating female odors suggests a means whereby a male might remember the natural odors of his dam and use this memory to direct his behavior selectively toward females in estrus. To test this idea, Fillion and Blass (1986b) applied an artificial odor (citral) either to the ventral or the dorsal surface of lactating dams from parturition to weaning. Although they did not provide the adult male offspring with a choice test, they did provide them with two mating opportunities, one with a citral-scented estrous female and one with an estrous female bearing no artificial scent. Sexual behavior was exhibited toward both partners, but some aspects of copulatory behavior occurred more rapidly with the citral-scented female for those males having early experience with that odor on the maternal ventrum; this was not the case for controls or for offspring of dams with citral applied to their backs. In an earlier study, Marr and Gardner (1965) used a similar rearing paradigm, adding the extraneous odor of rose cologne or methyl salicylate to both dams and their litters, or leaving them untreated. All male offspring were subsequently tested as adults with naturally scented estrous females. They found that the normally reared males performed more mounts than males from either of the other rearing conditions. These results are consistent with those of Fillion and Blass (1986b)but, because only normally scented stimulus females were used in the test of sexual behavior, Marr and Gardner’s findings may reflect a reduced propensity to mount regardless of the odor of the stimulus partner (Birke and Sadler, 1987; see Section B ,3,a, below). The rat studies indicate a lasting effect of the artificial maternal odor that can affect sexual behavior, but do not of themselves indicate that male rats use remembered maternal odors to guide choices among potential mates. We (Moore, Jordan, and Wong, unpublished) have recently tested the relative attractiveness of citral-scented or control ovariectomized fe-

MATERNAL CONTRIBUTIONS

71

males presented in a T maze to males that had been reared with either citral-scented or control dams. Males in both groups responded randomly when the females were not treated with hormones but, when treated with estrogen and progesterone, the citral-scented females were more likely to be chosen by males across both rearing conditions. Nevertheless, when citral-scented and unscented estrous females were presented in the same mating arena, they were equally likely to be mounted and to receive an ejaculation. These data suggest that a foreign odor when added to the odors of an estrous female can act as an attention-getting cue that can bias approach from a distance, while having no effect on the probability of mating with females in the immediate vicinity. A similar suggestion was made by Macrides et al. (1977) for dimethyl disulfide, a natural, volatile component of the hamster vaginal secretion that attracts males when presented alone but that, unlike the whole secretion, has no effect on the probability that an anesthetized male hamster will be mounted when scented with the substance. Dimethyl disulfide is found in various sources of odor in rodents, including saliva and amniotic fluid, and has been implicated in the first nursing episode of rat pups (Pedersen and Blass, 1981). In addition to distinguishing estrous from nonestrous females, males must also distinguish females from other males; the same is true for females. The development of attraction toward heterosexual stimuli has been studied primarily within the context of early hormonal effects. The typical pattern of attraction to stimuli produced by the opposite sex has been experimentally reversed by manipulation of the early gonadal hormone environment in pigs (Ford, 1983; Adkins-Regan, Orgeur, and Signoret, 1989), rat (de Jonge, Muntjewerff, Louwerse, and van de Poll, 1988), dogs (Beach, Johnson, Anisko, and Dunbar, 1977), and ferrets (Baum, Erskine, Kornberg, and Weaver, 1990). For example, male ferrets are more likely to approach and remain near a sexually active female when given a choice between this female and a sexually active male. Females exhibit the opposite choice. However, if females are treated with testosterone during their early (prenatal and neonatal) development, they will exhibit a male-typical choice (Baum et al., 1990). Animals acquire their particular odor profile as a result of various physiological processes, including the actions of gonadal steroids secreted during early development and at other times (Dunbar, Buehler, and Beach, 1980). Because of the hormonal effects, an individual’s odor is more similar to same-sex than to opposite-sex conspecifics, although all conspecifics share some odor characteristics, regardless of sex. If Bateson’s (1983) “slight unfamiliarity” principle is applied, a preference for the odors of the opposite sex would be predicted, because an individual becomes

72

CELIA L. MOORE

thoroughly familiar with its own odor by the time it reaches sexual maturity. This principle also accounts for why conditioned attraction to the mother’s odors in female rat pups (Fillion and Blass, 1986a)results eventually in an attraction to male odors. The odor of a developing female is like that of the mother, whereas that of the developing male differs in some ways. This leads to a sex difference in what is most familiar in the adult olfactory environments. The odor of an adult female is not as novel to another female as it is to a male, but that of an adult male is more novel. This may be the basis of the female’s attraction to male odors. This argument can be extended to account for the effects of developmental hormones on partner preference. The more an individual comes to smell like a member of the opposite sex, through the actions of heterotypical gonadal hormones at the appropriate developmental stages, the more its preference would be expected to shift toward members of the same sex. Although this hypothesis can account parsimoniously for the attraction of odors provided by members of the opposite sex and the reversal of this attraction with hormone treatments, it is currently unknown just how the early hormonal condition interacts with exposure to particular odors during development to shape partner preferences in sexual contexts. b. Restricted Rearing, There is a substantialliterature, gathered primarily during the 1950s through the 1970s, in which the effects of rearing in restricted social environments on the development of sexual behavior have been reported. These studies have been reviewed elsewhere (e.g., Larsson, 1978; Moore, 1985b); therefore, only the main findings will be summarized here. Environmental restriction is typically achieved by hand-rearing apart from conspecifics, or perhaps by rearing apart from some subset of the normal cadre of companions, such as the mother or siblings. In some studies, sensory restriction is employed, perhaps in combination with social restriction. And in other studies, social experience is manipulated in combination with hormonal manipulations. Rearing apart from the normal social environment has deleterious effects on sexual behavior in various mammalian species, but the nature and extent of the deficits vary with the species. Perhaps the most drastic effects occur in primates, as exemplified by Harlow’s (1965) work with rhesus monkeys. The difficulties that females have with maternal behavior and the difficulties that both males and females have in mating after rearing apart from other monkeys are profound, but cannot be separated from the difficulties they have in other aspects of social behavior (Fedigan, 1982). Indeed, the effects of extensive, long-term social deprivation are often so pervasive and multifaceted as to be uninterpretable.

MATERNAL CONTRIBUTIONS

73

Specifically sexual behaviors that have been examined in rhesus monkeys reared in social environments with different degrees of restriction include the foot-clasp mount, which is a component of masculine copulation, and the present posture, which is used in various contexts but serves as a component of feminine copulation. When reared in naturally complex social groups, monkeys of both sexes perform each of these behavioral patterns as young infants, with sex differences in probability of performance appearing later in the first year (Hanby, 1976). However, even in relatively intact laboratory environments (peers or mother plus peers), the foot-clasp mount is rarely seen in females and does not appear in males until several years of age. Furthermore, males may often perform mounts without the foot-clasp (Goy and Goldfoot, 1974). Apparently, the relevant social experience for the developmental pattern found in natural troops is provided not only by the mother and peers, but also by other adult females in the troop (Hanby, 1976). Males reared in laboratory groups of five mothers and same-aged peers developed the foot-clasp mount earlier and performed it more consistently than those reared with a single mother plus peers (Goy and Goldfoot, 1974). Primates normally develop in a rich social environment that includes the mother and a variable number of other individuals of varied ages and relatedness, some female and some male. These individuals also differ in their patterns of interaction and relationships with the mother and with the mother-infant dyad (Hinde and Stevenson-Hinde, 1976). In complex social groups, peers and older companions can influence the development of an individual both indirectly, by affecting the mother so that she treats the infant differently, and directly, by interacting with the infant. Some of these social factors have been linked to behavioral sex differences. In one such study, the social environment of developing rhesus monkeys was reduced to a set of peers. The monkeys were far more likely to exhibit heterotypical behavioral patterns when in an isosexual group than was the case in a mixed-sex group. For example,juvenile females rarely exhibited male-typical mounting behavior when the rearing group consisted of both males and females, but performed rather high levels of this behavior when in an all-female group (Goldfoot and Wallen, 1978). A similar effect of the companion’s sex on play behavior has been reported for cotton-top tamarins (Saguinus 0 . oedipus), a monogamous primate species that exhibits both twinning and extensive paternal contributions to care of young. There was no overall difference in the play behavior of young males and females. However, they both engaged in an equivalently high level of wrestling play when the play companion was male and performed wrestling play at reduced levels when the companion was female (Cleveland and Snowdon, 1984).

74

CELlA L. MOORE

The social environment also has effects on sexual development in other mammals, including rats (Gruendel and Arnold, 1969; Larsson, 1978), dogs (Beach, 1968), and guinea pigs (Gerall, 1963). This is most frequently revealed through social restriction studies. In rats, the earlier that social isolation is imposed (weaning, two weeks postpartum, or at birth), the greater are the consequences. Furthermore, there are progressively deleterious effects of removing the littermates of the opposite sex, all littermates, or the mother plus littermates (reviewed by Larsson, 1978). In fact, male rats reared artifically from birth never succeeded in performing a properly oriented mount of a female during four separate tests (Gruendel and Arnold, 1969). In many studies on these male laboratory mammals, the effects of rearing apart from companions have been at least partially reversed with repeated sexual experience in adulthood (Larsson, 1978; Moore, 1985b). This experience requires the joint action of circulating testosterone and exposure to sexually active companions. Apparently, nonsexual social experience provides the necessary background for an animal to use these internal and external conditions for the development of sexual behavior. Once sexual behavior has been initiated, the testosterone becomes less important (Rosenblatt and Aronson, 1958). Even when isolated from conspecifics, an animal has access to some self-generated stimuli. Self-generated auditory stimuli are important in the embryonic development of the ability of some birds, such as Peking ducks, to direct their social responses to conspecifics (Gottlieb, 1971). Selfproduced odors may function similarly in mammals. Evidence for this is provided by factorial studies combining sensory and social restriction. Both olfactory bulbectomy at 30 days and social isolation have deleterious effects on the development of masculine sexual behavior in rats, but their combination is particularly damaging (Wilhelmsson and Larsson, 1973). The behavior of bulbectomized adolescents reared subsequently with females was rather like that of males bulbectomized as adults, namely slower and less complete (Larsson, 1971), suggesting both that olfactory input is important for the expression of sexual behavior and that animals deprived of olfactory input can make use of other modalities if given social experience. Sexual behavior in rats in controlled by multiple modalities, such that no one modality is essential for its expression, at least in experienced rats (Beach, 1942; Stem, 1990). Although intact isolates are deprived of stimuli from conspecifics, they are not deprived of rat odors, because selfproduced odors are available. Perhaps this is why intact isolates are able to gain experience during repeated testing and eventually exhibit sexual behavior. Bulbectomized isolates are isolated both from self-produced olfactory stimuli and from other forms of conspecific stimuli, and these

MATERNAL CONTRIBUTIONS

75

animals fail to mount the female during multiple opportunities (Wilhelmsson and Larsson, 1973). Taken together, the social restriction studies demonstrate that the early social environment contributes importantly to the development of sexual behavior. Some of the relevant experience is gained through interactions with peers, but the mother also plays a significant role. She may provide the developmental foundation for the young animal to make appropriate use of the stimuli provided by peers and other social companions, during play, for example, and she may provide contributions that feed directly into developing components of reproductive behavior (Moore, 1985b, 1990, 1992). Of course, when young animals are removed from their mother, even if only for several hours, they are subjected to more than a restricted stimulative environment. The mother’s behavior, physical presence, and milk affect many physiological processes in the developing young, including thermoregulation, cardiovascular activity, adrenocorticalactivity, and growth hormone production (Hofer, 1983; Alberts and Cramer, 1988; Kuhn, Pauk, and Schanberg, 1990; Stanton and Levine, 1990). Because of the multiple and extensive effects of wholesale social restriction, the analysis of maternal influence on development has moved in recent years to more discrete, well-defined manipulations that have more limited and therefore more interpretable effects. 2. Sex-Related Biases in the Maternal Environment a . Differential Rearing of Males and Females. Most aspects of the early rearing environment are identical for male and female mammals, although differences in hormonal condition, rate of development, or other organismic factors can cause males and females to react differently to the same environmental condition. There is, however, accumulating evidence of some reliable biases in maternal behavior directed toward males and females. From the developing animals’ perspectives, these maternal biases constitute sex differences in the stimulative environments. The sex of a newborn is immediately important to human parents, but the extent of deliberately different rearing practices varies across cultures (Mead, 1935). Apart from these deliberate parental practices, there are more subtle differences in the way that parents treat males and females from earliest infancy (Moss, 1967). In describing the characteristics of their infants, parents make systematic errors that conform with cultural stereotypes, and adults describe the same newborns differently when cues (names; clothing)identifying their sex are systematically manipulated (Rubin, Provenzano, and Luria, 1974). Thus, some parental differences

76

CELIA L. MOORE

apparently stem from expectations, whereas others may arise from behavioral or physical differences in the infants. There has been considerable interest in determining the extent and nature of differential treatment of males and females by human parents. The literature reviews on sex-related parental socialization (Lytton and Romney, 1991; Maccoby and Jacklin, 1974; Siegal, 1987) indicate that parents do treat boys and girls differently, although perhaps not as differently as might be expected from the importance placed by the culture on sex differences. It is difficult to arrive at an accurate estimate of the differential treatment because of methodological variation from study to study. However, two findings emerge consistently from the reviews: parental behavior clearly supports culturally defined sex typing, which means that sex differences will be reinforced for some offspring behaviors; and fathers are more likely than mothers to treat sons and daughters differently. Lytton and Romney (1991) conclude from their meta-analytic study that differential parental treatment in various aspects of socialization other than sex typing (e.g., restrictiveness; encouragement of achievement) either does not occur, or is moderated by so many other factors that it cannot be detected consistently from study to study. They also report that the effect sizes of differential treatment, when it has been found, are generally quite small. It should be noted that the effect size measured in the meta-analysis is the size of differential socialization, not the effect of the differential socialization on offspring development. In the absence of this information, it is impossible to know the relationship between the size of the parental difference and the degree of effect on developmental outcome. Small differences in some input variable often multiply and cascade during development, leading to large outcome differences. Furthermore, behavioral sex differences may be fostered by differential parental treatment that has not been examined because it has not been categorized as socialization of the behavior. It is often the case that developmentally important variables have no obvious functional connection to the behavioral outcomes that they influence, which can be recognized in advance (Moore, 1990). In summary, there are some clear differences in parental socialization related to the sex of the child, but the differences that have been identified are rather few in number and their effects on behavioral development have yet to be determined. There are also scattered reports of differential treatment of male and female offspring among nonhuman primates. The genitalia of newborn primates are distinctly different as a consequence of prenatal testosterone in males, and adults devote a great deal of interest to inspecting the genitalia. In many species, males and females also grow at different rates and exhibit other morphological differences, such as canine size. These

MATERNAL CONTRIBUTIONS

77

dimorphic anatomical characters develop in response to the androgens that are secreted prenatally by males. Genetic females will develop to have masculinized genitalia and other morphological features, as well as masculinized play and sexual behavior, if they are treated prenatally with exogenous androgens (the “pseudohermaphroditic” female; Goy, 1970, 1978). Male rhesus monkeys also secrete testosterone during the first three postnatal months (Eider, Tannenbaum, Mann, and Wallen, 1993). Thus, concurrent as well as prenatal hormones are available that may produce detectable sex differences among neonatal primates. There have been no systematic studies in this area, but there are possibly olfactory as well as visual cues that can be used to distinguish the sexes in neonatal primates. Furthermore, there may be behavioral differences in the infants that arise from differences in their hormonal states. Studies of primates provide the earliest reports of differential rearing in nonhumans (reviewed in Fedigan, 1982). The early studies were driven by an interest in the development of independence in children, which affected the kind of data that was gathered. Jensen, Bobbitt, and Gordon (1967, 1968) were the first to report maternal discrimination of infant sex in primates. They found that pigtail macaque mothers rebuffed the attempts of male offspring to initiate contact from an earlier age than was the case for female offspring. This had the effect of reducing the amount of time that young males spent in contact with the mother, relative to females of the same age. In a study of captive rhesus macaques, Mitchell (1968) found that mothers more often held female infants in body contact, and more often actively rebuffed approaching male infants. These laboratory studies suggest sex differences in the amount or nature of maternal handling during infancy. Some field studies report similar data, but others do not (Fedigan, 1982). Thus, it is difficult to know whether there are reliable, systematic differences in the rearing environments of male and female primates under field conditions, or even how these environments should be characterized. Even under the best of conditions, it will require a substantial amount of data to identify sex differences in the rearing of primates, or other large animals giving birth to single offspring, particularly when maternal care occurs in the context of a social group. This is because differences in food availability, foraging time, the makeup of the group, the age and panty of the mother, the sex of prior offspring, and a host of other factors can also influence maternal behavior. Unless the bias is large, it will be difficult to disentangle it from other confounded factors with small samples of mothers and infants. In this respect, animals with mixed-sex litters offer advantages. On the other hand, animals with litters may be unable to exhibit sex-

78

CELIA L. MOORE

related differences that can be observed in mother-infant dyads. Most obviously, when there is only one offspring, a mother can alter the nature or amount of milk she provides, the time of weaning, or aspects of her nursing behavior as a function of sex of offspring. Red deer, for example, with male calves nurse more frequently and for longer bouts than is the case with female calves (Clutton-Brock, Guinness, and Albon, 1982). It is often the case that litters are nursed as units, which may preclude differential nursing of males and females. Under some circumstances, when young are sufficiently mobile to nurse outside the nestbound huddle, for example, differential nursing of males and females may be possible. When lactating wood rats are food restricted, growth and survival are lower in male offspring, perhaps because of differential maternal attention (McClure, 1981). Rat litters are sufficiently large that virtually all litters contain both males and females. The pups are clustered in the same nest, and nursing occurs when the dam assumes a crouched posture over the entire litter. The pups probe against the exposed ventrum of the dam and grasp the nipples without active maternal assistance. Shortly after the pups initiate sucking behavior, the dam enters a quiescent state during which milk ejections occur. Thus, when litters contain both sexes, there is not opportunity for differential delivery of milk to males and females, differential patterns of nest visits, or differences in the chemical, tactile, and thermal stimuli that are provided during nursing. However, between entering the nest and beginning to nurse, the dam exhibits bouts of extensive licking, mouthing, and other manipulations of the pups (Rosenblatt and Lehrman, 1963; Stem and Johnson, 1990). We have found a reliable and sustained difference in the amount of stimulation provided to males and females during these bouts of active maternal behavior in Long-Evans rats (Moore and Morelli, 1979; Fig. 2). In one study, undisturbed dams were observed during the early part of their daily dark cycle with litters culled to three males and three females. Unmarked pups were sexed by genital morphology when the dams turned them upside down to lick their anogenital regions. Both the number of bouts and the average bout duration of AGL (maternal anogenital licking) were greater for male pups than for their female siblings. This biased atttention to males was apparent at the earliest observation (the day after birth) and continued until the maternal pattern declined by the 18th postpartum day. Furthermore, when data from individual dams were considered separately, each of the nine dams exhibited a significant bias toward males as objects of AGL. Particular care was taken in this study to avoid differences in maternal attention that might be artifacts of the experimental procedure, such as

79

MATERNAL CONTRIBUTIONS

25

-I

I

0

/

2

'+ -

1

4

6

8

10

12

14

16

18

20

Days Postpartum

FIG. 2. Maternal rats exhibit reliable differences in anogenital licking (AGL) behavior directed toward the male and female offspring in their litters. In a sample of dams, each rearing 3 male and 3 female pups, male-directed AGL occurred more frequently throughout the postpartum period, and the durations of licking bouts were consistently longer when directed to males. Data are means and SEMs. Redrawn from Moore and Morelli (1979).

responses to sex differences in the reactions of pups to experimenter handling or marking, Each of these standard laboratory procedures does affect maternal licking (Smotherman and Bell, 1980). However, similar results were obtained when an individual male or female foster pup was placed into the cage, regardless of the sex composition of the home litter (Moore and Morelli, 1979). Furthermore, Richmond and Sachs (1984) observed the same male-biased maternal AGL when dams were moved from their home cage to a testing arena and presented with individually marked foster pups one at a time. Birke and Sadler (1985), working with Wistar rats, also reported more licking of male pups. Thus, the maternal discrimination is both robust and reliable. Maternal AGL is a specific form of licking that serves the immediate function of eliciting urination and, less frequently, defecation from pups. Young pups are unable to perform these functions without the assistance

80

CELIA L. MOORE

of their dams, which ingest the excreta. Maternal AGL is, then, the licking that is used during ingestion as well as during stimulation of elimination. This method of dealing with waste products from nestbound infants has the advantage of maintaining a sanitary nest. Furthermore, the urine of young animals is dilute, and it constitutes a significant source of both water and salt for dams, which have an increased need of these resources during lactation (Alberts and Gubernick, 1983). Both the actions of the dam and those of the pup contribute to an AGL bout (Moore and Chadwick-Dias, 1986). When the dam enters the nest, she begins to move about over the pups, sniffing and licking various parts of their bodies, and moving them from one part of the huddle to another, using her forepaws or mouth. During such episodes, she is likely to pick up an individual pup, turn it so that it is both supine and oriented with its head against her upper body, and hold it about the midriff so that its perineum is in close proximity to her mouth. Once settled with its anterior ventrum in contact with the dam’s body, the pup becomes quiescent with relaxed limbs. In the absence of anterior contact, the pup is likely to engage in righting responses. If the pup remains quiescent and the dam stimulates the perineal region with AGL, the pup will extend its hindlegs and become fairly rigid. Apart from the unusual orientation, this leg extension response is similar to the posterior part of the lordosis response shown by adult females. The pup’s behavior has the effect of creating a stationary target and a suitable surface for the accumulation of urine. Shortly after assuming the leg extension posture, the pup is likely to release urine. If the licking bout is prolonged, feces will also be released. This description applies to the majority of licking interactions, but they can occur in a variety of postures and orientations. We have observed no sex-related difference in the form of the behavioral pattern. Once an individual pup is voided and clean, the dam will turn her attention to a second pup, and so forth, until the probing and sucking activities of the litter stimulate her to assume a nursing posture and cease licking (Stem and Johnson, 1990). The majority of AGL occurs before nursing begins, during the initial part of nest visits, although pups may also be licked during breaks in nursing or just before the dam terminates a nest visit. The amount of licking each pup receives is in part a function of litter size. Larger litters elicit the crouch posture more readily (Stern and Johnson, 1990), thereby shortening the time available for AGL. Large litters also reduce the probability that individual pups will be selected, because there are more pups to choose among. In typical litters, some pups are likely to be skipped during a particular licking session. However, nest visits are initiated every hour or so throughout the day, providing multiple opportunities for licking to occur. Every pup is licked sufficiently

MATERNAL CONTRlBUTlONS

81

to accomplish elimination, but the proximate controls of licking behavior contribute to the variance among pups in amount of licking received. Favored pups can receive a disproportionate share of licking by being selected more frequently, and this is part of the reason why males receive more AGL than females (Moore and Morelli, 1979). When pups are about 17-18 days of age, they begin to move away from the dam’s initiation of AGL rather than to assume cooperative postures. AGL interactions are observed only infrequently by 3 weeks of age, although licking of other body regions continues (Moore and ChadwickDias, 1986). Licking of other parts of the body is variable in form, but is unlikely to be concentrated on a single location for a prolonged period. Under some observational conditions, it is impossible to distinguish body licking from contact sniffing (nosing), and we typically collapse these two behavioral categories. In one study, body licking and AGL were measured in 4-hour videotaped sessions for 12 dams, each rearing litters of 4 males and 4 females on days 3,6,9,12, and 15 postpartum. These two behavioral categories were correlated (positively)only on day 3. In general, individual dams sustain the same level of AGL until it declines in the third week, whereas they vary the amount of body licking as a function of changes in pups (e.g., growth of fur, which is groomed by the dam) (Moore, 1988). Although the separate behavioral patterns that constitute the rather heterogeneous category of body licking have not been subjected to close scrutiny, there is no evidence that body licking is delivered differentially to the two sexes. In rats, therefore, sex differences in the postnatal maternal environment are apparently restricted to differences in stimulation received from dams during the course of anogenital licking. This stimulation is typically delivered as a complex that includes concentrated tactile stimulation of the perineal region from the licking itself, tactile input to various skin regions from the forepaws and fur of the dam, thermal stimulation (heating from the dam’s paws and evaporative cooling from the licking),vestibular stimulation from the rotation to a supine position, olfactory stimulation from odors on the ventral region of the dam, and proprioceptive feedback from the leg extension and other postural adjustments. The differential maternal stimulation of males may be widespread in rodents. Clark, Bone, and Galef (1989) observed Mongolian gerbil dams with litters of 2 male and 2 female pups on days 5, 9, 13, 17, and 21 postpartum. They found that significantly more anogenital licking was directed to the males in the litter. They also found a significant interaction between pup sex and age, which was due to particularly high levels of licking directed to males on days 9 and 13. As was true for rats (Moore and Morelli, 1979), anogenital licking behavior declined during the third

82

CELIA L. MOORE

postpartum week. There is also some evidence that mice spend more time licking male than female offspring. Dams with all-male litters had higher licking scores than dams with mixed-sex or all-female litters, significantly so with older pups (Alleva, Caprioli, and Laviola, 1989). The measure of licking used in this study combined anogenital and other body licking, so the specificity of the differential maternal attention is not known for mice. b. Sources of Bias in Oflspring Stimuli and Maternal Sensory Processes. Removing the olfactory bulbs (Charton, Adrien, and Cosnier, 1971; Fleming and Rosenblatt, 1974), cutting the vomeronasal nerves (Fleming, Vaccarino, Tombosso, and Chee, 1979), and destroying the main olfactory receptors by intranasal application of zinc sulfate (Moore, Dou, and Juraska, 1992; Moore and Power, 1992) all reduce maternal licking. So does reducing pup odor by covering the perineum with collodion (Charton et al., 1971; Moore, 1981) and masking pup odors with perfume applied to their skin (Moore, 1981; Birke and Sadler, 1987). Thus, both the main olfactory system and the vomeronasal system are important for maternal licking behavior. Furthermore, odors emanating from the perineal region of pups play an important role in regulating maternal licking, particularly anogenital licking. There are various sources of odor in the perineal region. Of these sources, the preputial gland has been identified as important in several experiments. The preputial gland is a paired, subcutaneous gland found in both sexes. It has ducts that open alongside the urethral meatus. Thus, preputial secretions are found in the genital region and become mixed with urine on pup skin. This gland is a source of attractants that direct maternal licking to the perineum. In a series of experiments, BrouetteLahlou, Vernet-Maury, and Chanel (1991) reported that, in comparison with distilled water, perineal smears (excluding urine and feces) taken from 5-day-old pups attracted sniffing by adult males and adult females in various reproductive conditions. Furthermore, when these deposits were placed on the head of pups whose anogenital region had been cleaned, maternal licking was directed to the head region. This implies that anogenital licking is normally directed to the perineum by odors typically found on the perineal skin. Parallel results were obtained when extracts were taken directly from the preputial glands of pups: the extract attracted more sniffing than the solvent alone, and extract placed on the head of pups elicited licking of the head region by lactating dams. Dodecyl propionate was subsequently extracted from the preputial glands of pups, characterized chemically, and examined in behavioral tests. It is as attractive to dams in sniff tests as either the crude gland extract or perineal smears, and may be a sufficient cue to direct the dam to the perineum (BrouetteLahlou, Amouroux, et al., 1991). Furthermore, removing the vomeronasal

MATERNAL CONTRIBUTIONS

83

organ of dams significantly reduced the amount of licking directed to smears of dodecyl propionate placed on the heads of pups as well as the amount of sniffing directed to this odorant on filter paper (BrouetteLahlou, Vernet-Maury, Godinot, and Chanel, 1992). In addition to regulating maternal licking of pups of both sexes, perineal odors are also a major source of the differential licking directed to males. Covering the perineum with collodion reduced maternal anogenital licking directed toward 3-day-old male and female pups; it also eliminated the biased licking exhibited toward controls. Furthermore, female pups scented on the perineum with a drop of urine from a male pup received more anogenital licking than female siblings treated similarly with a drop of urine from a female pup (Moore, 1981; Fig. 3). Urine was collected from the perineal skin of pups after stroking the genitalia with a pipette. Thus, the urine contained glandular secretions available in the genital region. In a subsequent study, urine was collected in a similar manner from male and female pups every third day from day 2 to day 17 postpartum. The pups’ urine was presented to their dams in a simultaneous choice

FIG. 3. Differential maternal attention to male and female rat pups is based on differences in pup odor. Maternal anogenital licking observed after introducing two male and two female foster pups yielded a male bias when pups were untreated, but not after their perineal regions were coated with collodion or after a masking perfume odor was placed on the pups. In the urine condition, a drop of urine collected from either male or female pups was placed on the perineum of female pups before introducing them to the cage. Females bearing the odor of male pups received more licking. Data from Moore (1981).

84

CELIA L. MOORE

test as 20 pI drops on paper-towel-covered blocks presented outside the home cage. Dams spent more time sniffing the deposits collected from male pups throughout the study. When individuals were examined, each dam spend more time sniffing male urine, and these differences yielded significant individual preferences in 9 of the 15 subjects. The urine from male pups attracted more sniffing across pup ages. It also increased the probability of maternal anogenital licking. This was found by placing urine collected from 5- to 7-day-old donors on filter paper and dropping this paper into the nest region of an otherwise undisturbed dam and her mixedsex litter of six pups. Each dam was observed twice: once with a filter paper wet with 20 p1 of saline to establish a baseline, and immediately thereafter with a second filter paper wet with 20 pl of urine collected from either a male or a female. Male and female urine were equally effective in stimulating body licking, but only male urine increased anogenital licking above baseline levels. These data suggest that odors in the male urine can act as a signal to initiate maternal anogenital licking (Moore, 198%; Fig. 4). Taken together, the studies on pup odor and maternal behavior summarized thus far indicate that chemicals that are normally present on the perineum and that are carried in excreted urine can be used by dams to

I

+ Males + Females

I1

I

I

I

I

I

I

I

1

0

2

4

6

8

10

12

14

16

'd 18

Days Postpatturn

FIG. 4. Odor present in excreted pup urine is sufficient to direct maternal attention to males. Rat dams given simultaneous male-female choice tests every 3 days from day 2 through 17 postpartum (birth = day 1) with 20-4 drops of urine collected from their own pups spent significantly more time snifiing the urine deposits from male pups. Redrawn from Moore (19853.

MATERNAL CONTRIBUTIONS

85

direct their attention to the anogenital region, to increase the probability that licking of this region will occur, and to provide a greater amount of licking to males than to females. As a first step in identifying the sexbiasing factor(s), we explored the role of the preputial gland in a series of experiments that used sniffing of deposits by maternal rats as the dependent variable (Moore and Samonte, 1986). In one experiment, urine was collected directly from the bladder to avoid secretions from the preputial and other glands located “downstream” from the bladder. Dams responded equivalently to bladder urine collected from male and female pups. The preputial gland was implicated as the relevant source of odor because urine deposits from male and female pups with their preputial glands removed were also treated equivalently, whereas urine deposits from sham-operated littermates led to the expected male preference. Furthermore, deposits prepared from whole-gland homogenates of preputial glands taken from 10-day-old males elicited more sniffing from dams than deposits of equal volume prepared from female littermates. Thus, the preputial gland is sufficient to produce a male bias in maternal attention to pups. Whether the relevant difference is one of concentration of the same chemical or a qualitative difference in chemical production remains to be determined. It is known that adult preputial glands contribute to differences in the behavioral responses addressed to male and female rats, and that the male preputial odor depends on testicular hormones (Gawienowski, Orsulak, Stacewicz-Sapuntzakis, and Joseph, 1975; Gawienowski, 1977). There is testosterone secreted by neonatal male rats (Resko, Feder, and Goy, 1968; Baum et al., 1988) that may also underlie a sex difference in glandular activity. Under the test conditions in the Moore and Samonte (1986) study; the preputial glands seem to be necessary for maternal discrimination of sex. No differences in maternal attention to either bladder urine or urine from preputialectomized pups were found. Other, more sensitive measurements of maternal attention may reveal other olfactory cues, such as metabolites of testosterone in the urine, that can be used by dams to distinguish males from females, but differential activity of the preputial gland may turn out to be the only relevant olfactory difference. Although they were not concerned with the question of sex differences, Brouette-Lahlou, Vernet-Maury, et al. (1991) reported evidence consistent with the idea that preputial secretions may play a necessary role in directing licking to pups. In part of their argument, they reasoned that secretions from the preputial gland were used by the dam to direct licking to previously unlicked pups. Licking both removes preputial scent and stimulates elimination, so directing licking to pups with preputial secretions remaining on the perineum would have the effect of stimulating

86

CELlA L. MOORE

elimination from those pups with full bladders or bowels. This conclusion is supported by the facts that heads smeared with preputial secretions were far more likely to be licked than were cleaned perineal regions. The evidence in support of a necessary role for the preputial gland in directing maternal licking was collected by removing the preputial glands of all pups in a litter either on the day of birth or at 6 days of age. They found no difference in total amount of licking in operated litters when compared to either sham-operated or control litters, but they found a high level of mortality in the preputialectomized pups that was associated with distended intestines. They interpreted these results to mean that the dams persevered in licking some pups while ignoring others so that some eventually died from failure to eliminate waste products. They also found that removing the vomeronasal organ, which detects preputial secretions, increased anogenital licking of pups (Brouette-Lahlou et al., 1992). It is presently unclear how to explain the difference between these findings and other data that indicate a reduction in maternal licking after interference with either vomeronasal or primary olfactory functioning in the dam. Furthermore, Moore and Samonte (1986) observed no mortality and no difference in weight gain measured at 14 days of age in pups preputialectomized at 8 days, suggesting that even preputialectomized pups can receive adequate levels of anogenital licking to sustain normal growth. The reason for mortality after preputial gland removal in the Brouette-Lahlou, Vernet-Maury, et af. (1991) study and the absence of mortality in the Moore and Samonte (1986) study is not clear, but three differences in procedure deserve consideration. In the Moore and Samonte experiment the pups were slightly older (8 vs. 6 days) at the time of surgery, and sham controls were from the same litters as the preputialectomized pups, thereby providing a source of normal pup odors to the dams. The surgical procedure also differed. Moore and Samonte made small, bilateral incisions directly over the glands, whereas Brouette-Lahlou et af. accessed the glands through a single, larger incision 1 cm above the genitalia (BrouetteLahlou, Vernet-Maury, et al., 1991). Although this surgical route avoided wounds in the genital region, it necessitated lifting a flap of skin, which may have damaged the perineal innervation in some animals and compromised their ability to eliminate in response to tactile stimuli. Closer observations of mother-infant interactions with intact and preputialectomized pups are needed to clarify how odors from the preputial gland and other sources both regulate maternal behavior and bias attention to males in the litter. The weight of the evidence from studies of social interactions among adults supports the conclusion that rodents make use of multiple sources of odors and complex combinations of odors in their

MATERNAL CONTRIBUTIONS

87

social behavior (Brown, 1979; Johnston, 1990), and maternal animals can also be expected to make use of multiple odors emanating from their pups to regulate the amount of licking each pup receives. Olfaction is not the only regulator of maternal anogenital licking. Rat dams are attracted to the water and salt in urine, and the amount of licking they provide can be experimentally decreased by providing dietary salt or increased by temporarily withholding access to water (Alberts and Gubernick, 1983;Gubernick and Alberts, 1983). Rats from the F344 strain have an aversion to dietary salt at concentrations preferred by other strains of rats, such as Long-Evans (Bernstein, 1988; Midkiff, Fitts, Simpson, and Bernstein, 1985). Consistent with this strain difference, dams from the F344 strain exhibit lower levels of maternal anogenital licking than do Long-Evans dams, whether directed toward Long-Evans or F344 foster pups (Moore, Wong, and Daum, 1992). It is possible that olfactory cues from the perineum acquire attractiveness in part through their association with gustatory cues from the dilute urine produced by pups. Male pups in both rats (Clark and Galef, 1989) and Mongolian gerbils (Clark et al., 1989)produce more urine than their female siblings and take longer to release the urine after onset of artificial stimulation of the perineum. These differences add to the amount of licking time required to complete the task of voiding and cleaning a pup. As Clark and her colleagues argue, the differences in the amount and the schedule of reward (salty urine) provided by males and females to their dams may contribute to the bias of dams to lick males more than females. In addition to olfactory differences and differences in the rate and amount of urine production, there are behavioral differences that contribute toward the male bias. When rat dams position their pups and begin to stimulate the perineum, some time is required before the pups assume the leg extension posture that occurs shortly before urine release. The latency from onset of maternal anogenital licking to leg extension was measured in 12-day-old pups and was found to be significantly shorter in males (545 vs. 847 ms) (Moore and Chadwick-Dias, 1986). This response may well encourage further licking because it is accompanied by immobility, and the greater hesitation of females to become immobile and extend the legs may contribute to the greater number of short bouts of licking directed toward this sex. It is possible that some of the differences between male and female pups are consequences of differences in stimulation received on previous days. Changes in latency to respond to a stimulus may certainly arise from repeated stimulation. There is some evidence that urine production

88

CELIA L. MOORE

is increased by anogenital stimulation (Capek and Jelinek, 1956), and differences in such stimulation received by males and females may contribute to differences in urine volume of pups. c. Hormonal Effects on Pup Stimuli. The reliable differences in stimulus properties between the sexes suggest a role for gonadal steroids. This idea was tested by injecting female rats on the day of birth with testosterone, dihydrotestosterone, estrogen, or the oil vehicle. The testosterone treatment led to maternal anogenital licking that was equivalent to that delivered to untreated males both the next day and 9 days later. The other two hormones, which are metabolites of testosterone, also increased maternal licking above the levels addressed to oil-treated females (Moore, 1982; Fig. 5 ) . Birke and Sadler (1985) reported that a progestin, medroxyprogestrone acetate, perhaps after metabolism to an androgenic metabolite by pups, increases maternal anogenital licking. It is apparent that various steroids affect the stimulus properties of pups and, consequently, maternal responses, but the particular mechanism through which the hormones act and the degree of specificity in these effects have yet to be worked out. The common thread in all of the hormonal effects is a change in the licking levels delivered by dams. One likely mechanism is that the testosterone

FIG. 5. The sex bias in distribution of maternal anogenital licking (AGL) is reversible with neonatal hormone treatment of females. Untreated males received more AGL than untreated or oil-injected females on both days 2 and 10 postpartum, but not more than females injected on the day of birth (day 1) with testosterone propionate (TP) or its metabolites, estradiol benzoate (EB) and dihydrotestosterone propionate (DHTP). Redrawn from Moore (1982).

89

MATERNAL CONTRIBUTIONS

normally present in neonatal males increases the activity of the neonatal preputial gland, producing a different olfactory profile in male pups. Other exogenous steroids may also affect this gland. Of course, additional effects are possible, including direct olfactory consequences of hormones or their metabolites excreted in urine. There is now evidence that the two known mechanisms of producing natural variation in prenatal testosterone, maternal stress and intrauterine sibling effects, also produce variation in maternal anogenital licking. Dams that are stressed during their pregnancy exhibit some differences in their behavior toward pups. They are more likely to remain near their pups and nurse them (Muir, Pfister, and Ivinskis, 1985) but less likely to lick them (Moore and Power, 1986). Furthermore, the offspring of prenatally stressed dams, even when fostered to unstressed dams, elicit less licking (Moore and Power, 1986).This reduced licking is specific to the anogenital licking of males, as it is not observed in licking of female offspring or in the body licking of either sex offspring. The difference is apparently due to a reduction in the attractiveness of urine from male offspring of stressed dams (Power and Moore, 1986; Fig. 6). Clark et al. (1989) examined individual differences in the level of anogenital licking received by gerbil

B

T

-

0 Control Stressed

Control Stressed

FIG. 6. Male offspring of dams stressed during pregnancy are less effective elicitors of maternal attention. (A) When pups were gestated normally. foster dams directed more anogenital licking (AGL) to males, but there was no difference in their treatment of male and female offspring of stressed dams. The stressor was housing with five adult males during late pregnancy. Foster dams were not stressed. (B) Urine collected from male offspring of stressed dams elicited less investigatory sniffing from maternal rats than urine from normally gestated males. Urine deposits were presented in simultaneous choice tests to unstressed colony dams. Redrawn from Power and Moore (1986).

90

CELIA L. MOORE

pups as a function of the sex of their neighboring fetuses in utero. They found that the level of licking received by females was positively related to the number of neighboring males during gestation. Conversely, males that had gestated next to fewer females received higher levels of licking. In summary, the differential attention directed to males and females in rodents is based on differences in their gonadal hormones. Put another way, factors that alter gonadal hormones alter maternal behavior in predictable ways. However, it is possible that the stimulation that mothers provide to their offspring also affects their hormone secretion. Thus, the higher levels of stimulation that newborn males receive because of testosterone surges during late gestation and on the first postnatal day may initiate a reverberating system of continued testosterone and maternal stimulation that could last for the first 2-3 weeks of life. Both male and female adult rodents exhibit neuroendocrine responses to genital stimulation that can increase production of gonadal steroids (Allen and Adler, 1985). The relevant studies have not yet been done to determine whether neonatal rodents have similar responses. Clearly, hormones can affect the stimulus properties of the young, which in turn can produce sex-related differences in maternal behavior. However, the generality of these effects among mammals has yet to be established. It is often assumed that the early, developmental production of testosterone occurs during gestation but not postnatally in most male mammals. Of course, sex differences in offspring characteristics (such as primate genitalia; Goy, 1970, 1978) produced by these prenatal hormones may elicit different patterns of maternal care postnatally even when there are no neonatal sex differences in circulating hormones. However, the opportunities for these effects are greater when neonatal hormones are available that can, for example, produce differences in odor. Many of the hormonally mediated developmental processes that have been identified in fetal organisms can occur at later ages (Arnold and Breedlove, 198% and there is recent evidence of postnatal testosterone secretion in young males of various mammals other than rodents, including carnivores (ferrets; Baum et al., 1990), primates (rhesus monkeys; Eisler et al., 19931, and ungulates (pigs; Ford, 1983). Estrogen levels are very high for long developmental periods in juvenile male pigs, and this hormone is the active steroid for defeminizing the behavior of males. Because estrogen exerts its effects gradually during the first four months of life, there is ample time for extensive contributions from the social environment to act in concert with the hormone (Adkins-Regan, Orgeur, and Signoret, 1989). The opportunitiesfor reciprocal interactions of hormones and social stimulation during early mammalian development may be far greater than has previously been assumed.

MATERNAL CONTRIBUTIONS

91

3 . Consequences of Experimentally Manipulated Maternal Stimulation The fact that manipulating the endocrine condition of pups will alter their treatment by mothers (Moore, 1982) raises the possibility that maternal behavior may mediate some of the developmental effects of hormones. Maternal behavior is a known mediator of some early interventions, such as experimenter handling of rat pups (Smotherman and Bell, 1980).Indeed, maternal mediation of environmental perturbations can cross generations. This point has been made for diverse phenomena, including nutritional deficits (Birch, 1971), the stress and nutritional consequences of social crowding (Christian and Lemunyan, 1958), susceptibility to stress ulcers (Skolnick, Ackerman, Hofer, and Weiner, 1980), and early handling (Denenberg and Whimbey, 1963). Maternally mediated handling effects are particularly interesting in the present context, because mothers respond to handled offspring with increased levels of licking and handling of the offspring (Lee and Williams, 1974). This is also true for licking that is addressed specifically to the anogenital regions (Moore and Morelli, 1979), which is the type of maternal difference that has been linked to neonatal hormonal condition (Moore, 1982). In order to determine whether maternal stimulation of developing offspring has consequences for reproductive development, it is necessary to manipulate the amount of maternal stimulation by some means other than changing the hormonal condition of the offspring. The interest that rat dams exhibit in perineal pup odors and in the salty taste of pup urine provides two avenues for such manipulation. a . Juvenile Behavior. Prepubertal mammals engage in a substantial degree of locomotor and social play. The function of this play has been difficult to determine (Martin and Caro, 1985), but one persistent idea is that play provides a context for learning the social and perceptual-motor skills that are later used in reproductive behavior (Dolhinow and Bishop, 1970; Lancaster, 1971; Meaney, Stewart, and Beatty, 1985). It is certainly the case that play provides the opportunity for this experience. Potential play companions are one or two parents in all mammals; age-matched siblings among animals that have litters; both older and younger siblings among animals with extended family groups; and animals of all ages among animals that maintain larger social groups. Adults are less likely than immature animals to serve as play companions, but do so to varying extents across species. When sex differences in play are found, males are often reported to play more than females, particularly when vigorous social (“rough-andtumble”) play is the measure (Meaney etal., 1985). This sex difference is

92

CELIA L. MOORE

apparent in both rats and rhesus monkeys. Furthermore, it can be reversed by providing females with androgens early in development (Goy, 1970, 1978; Meaney and Stewart, 1981b). Unlike these species, female spotted hyenas play considerably more than males, which may be related to the high endogenous levels of androgen among females of this species (Glickman et al., 1992). The extent of play is highly dependent on contextual factors (Fedigan, 1982; Thor and Holloway, 1984; Martin and Caro, 1985). These contextual factors are such that sex differences are sometimes observed and sometimes not. This has been reported among both rats (Thor and Holloway, 1984) and primates (Fedigan, 1982), including rhesus monkeys. The incorporation of mounting behavior into the play of rhesus monkey females occurs at male-typical levels when the females are reared in all-female groups, but at lower levels when the rearing group includes males (Goldfoot and Wallen, 1987). Sex differences in play with younger or older animals also depend on contextual factors. Juvenile rhesus males are less likely to interact with younger animals when females are in the group (Gibber and Goy, 1985), and the level of wrestling play is greater among cotton-top tamarin females when playing with male companions (Cleveland and Snowdon, 1984). The experience gained during play with younger siblings or other companions in a social group is thought to contribute to the development of parental skills (Dolhinow and Bishop, 1970; Lancaster, 1971; Meaney et al., 1985; Brunelli, Shindledecker, and Hofer, 1989; Snowdon, 1990). Juvenile rats incorporate components of parental care into their play behavior with infant rats (e.g., “pounce” followed by “carry” or “retrieve”), and manipulations, such as social isolation, that affect the level of play also affect the level of parental behaviors exhibited by the juveniles (Brunelli et al., 1989). Furthermore, when early-weaned isolates are kept from playing as juveniles and then tested as adults with infants, they exhibit interaction patterns that are typical of play behavior (Brunelli et al., 1989). These results suggest that the elements elicited during play become integrated during experience, which allows a mature form of parental behavior to occur. A similar argument has been made for “playmothering” among primates. Primates with such experience are reported to be more competent caregivers (Lancaster, 1971). In many primates, the juvenile females do more interacting with infants than is true for juvenile males, but when the group structure permits such interaction, males also engage in extensive playful or caregiving interactions with younger members of the group. Such experience may be fundamental to the acquisition of paternal behavior among species, such as marmosets

MATERNAL CONTRIBUTIONS

93

and tamarins, where paternal care is an integral part of the reproductive pattern (Snowdon, 1990). There is also evidence that the skills acquired during play are important for masculine sexual behavior (reviewed in Moore, 1985b and Ward, 1992). For example, juvenile male rats isolated from play companions exhibit deficiencies in their copulatory behavior (Gerall, Ward, and Gerall, 1967). Furthermore, rats with prepubertal lesions of the medial preoptic area, a region that normally serves an essential role in the control of sexual behavior, will nevertheless develop the ability to copulate if reared in social groups during the recovery period, but will not if kept alone during this time (Leedy, Vela, Popolow, and Gerall, 1980). The opportunity to play is the most likely candidate for this difference. There are no sex differences in the form of play in rats, but males exhibit higher levels of social play in some contexts (Meaney and Stewart, 198I a). This sex difference can be reversed by neonatal androgen manipulation (Meaney and Stewart, 1981b)or by prenatal stress, which alters the prenatal hormonal environment (Ward, 1992).The play behavior includes chasing, pouncing, mounting, and other components that share features with copulatory behavior. Although males may play more than females, and play may be an important precursor to sexual behavior, there is no evidence that sex differences in amount of play contribute to sex differences in likelihood of performing either male or female copulatory behavior. The stimulation received from mothers affects the amount of play in juvenile male rats. Birke and Sadler (1987) manipulated the amount of stimulation received from maternal anogenital licking during the first nine days of life and measured play behavior during the postweaning period. The manipulation consisted of placing perfume on the pups, which masked their natural odors and, therefore, reduced maternal licking. Play was measured at 35 days of age in paired encounters between individuals of the same sex but from different litters. Males from the experimental group played significantly more than control males, but there was no group difference in the play behavior of females. Moore and Power (1992) found similar results using two different manipulations of maternal stimulation and different testing procedures. Dams were either treated intranasally with zinc sulfate, which reduced their ability to smell, or were given dilute saline instead of tap water to drink, which reduced their appetite for salty pup urine. Both manipulations reduced maternal anogenital licking throughout the period when it normally occurs, but had no other effects on maternal behavior or on the growth of pups. Play behavior was observed at 33-42 days of age in same-sex sibling groups of four. As in the study by

94

CELIA L. MOORE

Birke and Sadler (1987), the male (but not female) offspring of the less stimulating dams played more than controls. It is possible that the effect on play in males is part of a broader effect on general activity level. The offspring of treated dams also ambulated and reared more when tested alone in the open field (Moore and Power, 1992). The results of the studies on juvenile behavior demonstrate that play behavior is affected by maternal stimulation in rats, but do not support the idea that this maternal stimulation mediates sex differences in play. Males are typically stimulated more than females (Moore and Morelli, 1979) but, for reasons that are not yet understood, experimentally reduced anogenital stimulation of males increases their play behavior, thereby enlarging rather than reducing the degree of sex difference in this behavior. It is unlikely, therefore, that maternal effects on masculine copulatory performance are mediated through effects on play during the juvenile period. b. Reproductiue Behavior. Aspects of the performance of masculine sexual behavior have been linked to maternal anogenital stimulation. Moore (1984) lined either one or both nares of postparturient rats with polyethylene tubing so that odorous compounds would bypass the olfactory receptors (Ruddy, 1980). This procedure reduced maternal licking behavior in a graded manner, with bilaterally intubed dams having the lowest licking scores. The male and female offspring of these dams and untreated controls were gonadectomized as adults and implanted subcutaneously with equivalent amounts of testosterone. Although all the males copulated, the males reared by less stimulating dams took longer to ejaculate and had longer postejaculatory refractory periods before resuming copulatory behavior. Testosterone treatment in adulthood typically leads to some mounting behavior of females in the Long-Evans strain of rats, including mounts with intromission patterns, which are defined by a characteristic dismount pattern that males use when intromission has occurred. On their first test, 89% of the female offspring of control dams exhibited at least one intromission pattern, whereas less than 20% of the females from the experimental groups did so. Upon repeated testing, the number of experimental females exhibiting the pattern gradually increased to approach control levels. Throughout testing, the mounting behavior of experimental females was less likely to include the intromission pattern than was the case for normally stimulated controls. In both males and females, the interval between intromission patterns was significantly shorter in the normally stimulated controls than in animals reared with reduced stimulation (Moore, 1984). Birke and Sadler (1987) did not distinguish between mounts with and without the intromission pattern, but measured intervals between mounts. They found that these intervals were signifi-

MATERNAL CONTRIBUTIONS

95

cantly longer in those male offspring that had been reared in a perfumed maternal condition. This maternal treatment reduces maternal anogenital licking. The studies by Moore (1984) and Birke and Sadler (1987) are consistent in finding that copulatory behavior occurs at a slower pace in offspring that receive less maternal stimulation of the perineum as neonates. The probability of observing an intromission pattern during a mount is decreased in males that received reduced amounts of anogenital stimulation as neonates (Moore, 1992). Whether maternal anogenital licking was reduced by peripheral, intranasal applicationsof zinc sulfate (Moore, 1992; Fig. 7) or by providing the dams with salty drinking water (C. Moore, unpublished data), the proportion of mounts with an intromission pattern was significantly reduced from control levels in intact, adult male offspring. This effect was maintained with repeated testing (Moore, 1992) and with multiple ejaculatory series during the same copulatory session (C. Moore, unpublished data). The reduced probability of intromission patterns during the mounts of females that were reared by less stimulating (nasally intubed) dams (Moore, 1984) is consistent with the findings for males. Furthermore, the number of intromission patterns can be increased in females by providing them with extra anogenital stimulation (Moore, 1985b; Fig. 8). Female -~

c

.i.- 90

Reduced MaternalAGL

c

0 Test 1

Test 2

FIG. 7. The percentage of mounts that include an intromission pattern was significantly greater in control males than in males reared by dams that provided reduced stimulation from maternal anogenital licking (AGL). AGL was reduced by rendering the dams hyposmic with intranasal zinc sulfate treatments. Intromission patterns were identified by a characteristic dismount. The two tests occurred at least one week apart. Redrawn from Moore (1992).

96

CELIA L. MOORE

FIG.8. The number of intromission patterns performed during a sexual behavior test was greater in female rats that had been provided with additional perineal stimulation during the first two weeks of life. Although most mounts did not include the intromission pattern (identified by a characteristic dismount) for females in both groups, intromission patterns were more likely to occur during mounting in females receiving extra neonatal stimulation in the perineal region. Controls were female siblings receiving identical amounts of stimulation directed to the shoulder region. Data from Moore (1985b).

offspring from untreated dams were provided with extra stimulation applied with a lubricated brush during the first two weeks of life. Some females were stimulated in the perineal region to mimic aspects of maternal anogenital licking, whereas others from the same litter were stimulated on the shoulder region. The adult female offspring were ovariectomized, treated with testosterone, and tested with a female partner in hormoneinduced estrus. Those females stimulated in the perineal region performed more intromission patterns, and they took fewer mounts before exhibiting an intromission pattern than was the case for shoulder-stimulated siblings (Moore, 1985b). In summary, specific aspects of maternally provided stimulation have been linked to specific aspects of masculine copulatory behavior in rats. When no isolation is imposed but the degree of maternally provided perineal stimulation is manipulated by intervening in the sensory control of anogenital licking, both the timing and the quality of the behavior are changed. Within the range of variation in stimulation that has been imposed, there is no evidence of a changed motivation to copulate. However, when stimulation is reduced, the mounts and intromission patterns occur more slowly, and mounts are less likely to culminate in an intromission pattern. Although males are far more likely to exhibit intromission patterns

MATERNAL CONTRIBUTIONS

97

than females, parallel changes in masculine copulatory behavior are found for male and female offspring. We have found no evidence of an effect of maternal anogenital stimulation on feminine copulatory behavior (C. Moore, unpublished data). c. Neural Mechanisms. The nature of the behavioral consequences of differential maternal anogenital licking indicates an effect on those neural mechanisms that underlie penile function. Erection and other actions of the penis in male rats are controlled by both autonomic and somatic mechanisms. There are several small, striated muscles associated with the penis, and these muscles contribute importantly to intromission, penile movements within the vagina, and ejaculation. There are now several studies in which a combination of techniques, including muscle excision, reflex measurement, and measurement of ejaculatory plug placement within the vagina, have been used to tease apart the several contributions of these muscles (Sachs, 1982, 1983; Hart and Melese-d’Hospita1, 1983; Wallach and Hart, 1983). The ischiocavernosus muscles attach laterally to the corpus cavernosus of the penis, and they function in penile “flips,” which are reflexes that underlie the dorsiflexion of the penis prior to insertion and that contribute to the displacement of previously placed sperm plugs from the cervix. The bulbospongiosus muscles surround the bulb of the penis, and their contraction contributes to full erection, which includes the formation of a penile cup, and to the formation and placement of an ejaculatory plug. The cup-shaped tip of the penis is associated with a well-formed plug that adheres firmly to the cervix without being withdrawn along with the penis. Intromission, plug formation and placement, and the ability to remove the plugs of competitors are compromised by excising either the ischiocavernosus or the bulbospongiosus muscles (Sachs, 1982; Hart and Melese-d’Hospital, 1983). The motor nuclei innervating the penile muscles of rats are located in the fifth and sixth segments of the lumbar spinal cord, with a dorsomedial nucleus innervating the bulbospongiosus muscles and a dorsolateral nucleus (DLN) innervating the ischiocavernosus muscles (Breedlove and Arnold, 1981, 1983; McKenna and Nadelhaft, 1986; Schrqbder, 1980). The dorsomedial nucleus has become known as the SNB, the spinal nucleus of the bulbocavernosus. Both the SNB and the DLN are sexually dimorphic as a consequence of the secretion of testosterone by males during the late prenatal and early neonatal period (Breedlove and Arnold, 1981, 1983; Jordan, Breedlove, and Arnold, 1982). Testosterone leads to a greater number of motor neurons in males primarily through its direct effect on the survival and further development of the penile muscles (Cihik, Gutman, and Hanzlkovi, 1970; Rand and Breedlove, 1988; Breedlove, 1992). These muscles involute during the

98

CELlA L. MOORE

neonatal period in females (CihAk et al., 1970). The loss of target tissue, in turn, leads to a greater death of motor neurons in females than that in males during this early developmental period and, consequently, to fewer neurons in the dimorphic lumbar nuclei of adults (Sengelaub and Arnold, 1986).

Although interactions with target muscles play a primary role in determining the number of motor neurons that survive the early period of selective cell death (Hamburger and Oppenheim, 1982; Oppenheim, 1981), there are also important contributions to this process from afferents (Okado and Oppenhelm, 1984). Unlike most motor nuclei (Oppenheim, 1986), cell death in the dimorphic lumbar nuclei extends well into the neonatal period (Sengelaub, Nordeen, Nordeen, and Arnold, 1989). Therefore, there is an opportunity for afferent input from maternal stimulation to affect the process. This possibility was examined by treating dams with intranasal zinc sulfate to render them anosmic and to reduce the amount of anogenital stimulation they provided to their pups. Tactile stimulation of the genitalia and surrounding perineum enters the spinal cord at the lumbar-sacral transition (Pfaff, 1980), in exactly those segments where dimorphic motor nuclei are found. Motor neurons were counted in the spinal nuclei of the bulbocavernosus of the adult male and female offspring of treated and control dams. The animals with reduced neonatal stimulation had 11% fewer motor neurons than controls, an effect that was focused in the rostra1 extent of the nucleus (Moore, Dou, et al., 1992; Fig. 9). Because the effect was regionally specific and was not accompanied by changes in either body size or gross size of the spinal cord, it is unlikely that some general effect, such as growth hormone secretion, underlies the phenomenon. It is possible that the maternal stimulation affected testosterone secretion in male pups, with higher levels in controls than in the offspring of anosmic dams. However, it is difficult to extend this particular explanation to the female offspring, which also show an effect of the stimulation. Furthermore, Mills and Sengelaub (1992) removed the dorsal roots serving one side of the dimorphic lumbar-sacral portion of the spinal cord, thereby reducing afferent input on the operated side relative to that on the intact side. This resulted in a reduction of motor neurons on the operated side, but not on the intact side. A reduction in testosterone secretion as a result of the lowered afferent input following dorsal rhizotomy would be expected to affect both sides of the spinal cord equally. There is a third sexually dimorphic nucleus in the lumbar spinal cord of rats that is known to be regulated independently of testosterone. This is the cremaster nucleus, which innervates the muscles that retract the

MATERNAL CONTRIBUTIONS

99

FIG.9. When rats were reared with dams that provided reduced anogenital licking (AGL) as a result of intranasal zinc sulfate treatments, there was an 11% reduction in the number of motor neurons in a sexually dimorphic nucleus located in the dorsomedial region of the lumbar spinal cord. The motor neurons innervating the bulbospongiosus penile muscles are located in this nucleus. The reduction in neuron number was apparent in both males and females and was specific to the rostra1 region of the nucleus. After Moore, Dou, and Juraska (1992).

testes into the abdomen. The survival of motor neurons in the cremaster nucleus is regulated by the physical presence of the testes, independent of any hormonal contribution (Newton and Hamill, 1989). Afferent input resulting from the testicular mass may be an important factor. Afferent input may affect motor neuron survival by local increases in the production or transport of trophic substances to motor neurons. This might occur directly, through the provision of trophic substances from the afferents themselves, or indirectly, by stimulating muscles to change their interactions with motor neurons. There are sexual dimorphisms in afferent as well as motor systems in the lumbar region of rats. Afferent projections to the dimorphic lumbar motor nuclei are sexually dimorphic, with a substantially greater number of somatostatin fibers in males than in females (Newton, Unger, and Hamill, 1990). Males have more neurons in the dorsal root ganglia of the sixth lumbar and first sacral segments from about the fourth postnatal day (Mills and Sengelaub, 1993). These ganglia include a mixed population of inputs, including both cutaneous and muscle afferents from both dimorphic

100

CELIA L. MOORE

and nondimorphic structures. There is, however, a specific dimorphism among adult rats in the number of axons in the sensory branch of the pudendal nerve, with males having more axons than females have (Moore and White, 1994). This branch serves the perineal region and includes only axons of cutaneous receptors (McKenna and Nadelhaft, 1986). The age at which this input becomes dimorphic has not yet been determined. There are, therefore, feasible physiological mechanisms for local trophic effects on motor neurons from afferent inputs. Developing males may experience more afferent input from the perineum for two reasons: the greater number of afferent neurons they have in this region and the greater stimulation of these afferents by the behavior of the mothers. The effect of maternal stimulation on neuron number in lumbar motor nuclei is intriguing because it suggests that maternal behavior can contribute to neural development in highly specific ways that have specific consequences for reproductive behavior. No causal connection between the neural and behavioral effects has yet been established, but it is possible that the effect on motor neuron number underlies the effects on masculine copulatory behavior. However, penile reflexes normally occur in association with the gross motor patterns of copulatory behavior, and the integration of the spinal mechanisms controlling penile reflexes with the brain mechanisms underlying copulatory behavior is achieved by supraspinal inhibitory mechanisms (Sachs and Meisel, 1989; Meisel and Sachs, 1980). There is some evidence to suggest that the development of the brain as well as the spinal regions that are involved in the control of copulatory behavior can be affected by afferent input during the neonatal period. Olfactory input is critical to the performance of sexual behavior in rodents, and the early development of the olfactory system, including learning of specific odorants, is affected significantly by tactile stimulation of a kind that is typically provided by the dam (Pedersen and Blass, 1981; Pedersen et al., 1982; Coopersmith and Leon, 1984; Sullivan and Leon, 1986; Sullivan, Wilson, and Leon, 1989). The fact that males receive more tactile stimulation from the licking and handling behavior of their dams sets the stage for differences in effective olfactory input, olfactory learning, or integration of the olfactory systems with other developing neural systems. Other evidence in support of afferent contributions to dimorphic brain development is provided by pharmacological and other procedures that block afferent input to the developing preoptic-anterior hypothalamic area. These blocks interfere with the usual effects of neonatal androgen on the functional development of these regions (Beyer and Feder, 1987). Furthermore, afferent connections within and to the hypothalamus, as measured electrophysiologically, undergo dynamic developmental changes in neonates (Almli and Fisher, 1985). The medial preoptic area

MATERNAL CONTRIBUTIONS

101

does not assume its sexually dimorphic appearance until the neonatal period in rats (Gorski, 1984; Hammer and Jacobson, 1984; Murakami and Arai, 19891, allowing the opportunity for differential stimulation of males and females to play a role. As with behavior, there is a correlation between the effect that hormones have on the sexual dimorphism of brain regions (Gorksi, 1984) and the effect that they have on maternal behavior (Moore, 1982),but whether the two are casually related is currently unknown. Furthermore, the necessary experiments have yet to be done to determine whether the developmental effects of maternal stimulation on masculine copulatory behavior reside in brain mechanisms, spinal mechanisms, or both. d. Implications for Sex and Species Differences. The finding that specific aspects of maternal behavior can alter both behavioral and morphological aspects of offspring development, and do so differently for males and females, raises the interesting probability that some sex and species differences have arisen indirectly, through changes in the maternal environment. In the normal course of events, males in rats and at least some other rodents receive more stimulation of a particular kind than their female siblings receive, and the studies reviewed in the previous sections demonstrate that these differences account for some of the divergence of the two sexes in the development of sexual behavior and associated neuroanatomical structures. Differences between related species may also arise from maternal sources. Of course, some shifts in maternal behavior from one species to another may be elicited by changes in the cues provided by offspring. The use of olfactory differences between males and females is an example that points in that direction. However, the maternal environment may undergo change even though there are no changes in the offspring. The strain difference in maternal behavior between Fischer-344 (F344) and Long-Evans rats provides an insight into how some species differences might occur. The F344 strain of rats is known to have an aversion to sodium chloride in saline concentrations that are highly preferred to plain water by other strains of rats (Bernstein, 1988) and that spans the range of saline in the urine of young pups (Gubernick and Alberts, 1983). This gustatory difference may underlie the fact that F344 rats seem less interested in pup urine and may, therefore, affect their maternal behavior. Whether observed with Long-Evans or F344 pups, the F344 dams perform significantly less anogenital licking than Long-Evans dams (Moore, Wong, et al., 1992; Fig. IOB). The effect of this maternal difference on offspring behavior has yet to be determined, but it is predicted to underlie observed strain differences in masculin sexual behavior (Moore and Wong, 1992).

102

CELIA L. MOORE

FIG. 10. Correlations in reproductive success and maternal stimulation in two strains of rats. (A) Fischer-344 (F344) males sire fewer offspring than do Long-Evans (L-E) males when each male is allowed one ejaculation with the same female. There is a second-male advantage apparent for both strains, but this effect is much greater for L-E males. Redrawn from Moore and Wong (1992). (B)Whether interacting with F344 or L-E pups, L-E dams provide significantly greater amounts of maternal stimulation of the perineum in the form of anogenital licking (AGL). This maternal difference was predicted from differences in the reproductive performance of males in the two strains. Data from Moore, Wong, and Daum (1992).

MATERNAL CONTRIBUTIONS

103

V. MATERNAL CONTRIBUTIONS TO OFFSPRING REPRODUCTIVE SUCCESS Trivers and Willard (1973) formulated the prediction that when reproductive success is more variable in one sex than in the other, parents should invest more heavily in the more variable sex. Trivers (1972) defined parental investment in terms of competition between the parent and a given offspring: parental investment is provision of resources that increase the offspring’s reproductive success, while decreasing parental ability to direct resources to other offspring. The two areas that have received the greatest attention as a result of the Trivers-Willard hypothesis are differences in sex ratio and differential provision of nutritive resources (Charnov, 1982; Clutton-Brock and Albon, 1982; Clutton-Brock et al., 1982; Hrdy, 1987). The ability of mammals to vary the sex ratio is limited, although there are reports of departure from randomness (Altmann, Hausfater, and Altmann, 1988; Clark, Karpiuk, et al., 1993). Mammals commonly exhibit polygynous mating patterns. As a consequence of competition for mates, male mammals are expected to have greater variance in reproductive success than females have. They may also receive more parental investment as defined by Trivers (1972). There are some studies that confirm these predictions. The suckling frequency is higher for male calves in red deer, and mothers are more likely to be barren in the following year if their calf is male (Clutton-Brock et al., 1982). Similarly, Mongolian gerbil dams with all-male litters nurse more frequently and take longer to reenter reproductive condition (Clark and Galef, 1992). It is possible for mothers to influence the reproductive success of their offspring in ways that do not readily fall within the predictions of the Trivers-Willard hypothesis. The licking behavior that maternal rats provide so avidly to male offspring and less generously to female offspring contributes importantly to masculine reproductive development (Moore, 1984, 1990, 1992; Moore, Dou, er al., 1992), and the lesser stimulation that females receive apparently exacts no reproductivecosts. Furthermore, the behavior takes what is probably a negligible amount of the energy and daily time budget of the dam. The behavior is performed as a prelude to nursing, so it requires no special trips to the nest. The nature of the maternal effects on masculine copulatory behavior suggests that they may be consequential for sperm competition among males. Male rats do compete for estrous females with aggressive interactions (Thor and Flannelly, 1979), but males that are well acquainted with one another copulate successively with the same female without fighting (McClintock, Anisko, and Adler, 1982). Even when males are behaviorally cooperative, there is sperm competition.

104

CELlA L. MOORE

An individual male performs a series of ejaculations during each copulatory session, with each ejaculation preceded by several intromissions. Sperm supplies are depleted after about six ejaculations (Dewsbury, 1984). In order for sperm transport to occur, a sperm plug must be firmly attached to the female’s cervix for about 5 min (Adler and Toner, 1986).A postejaculatory refractory period prevents a male from dislodging his own plug before transport can occur, but an intromitting male can dislodge the plugs of competitors. Two intromissions are sufficient to displace a plug and interfere with sperm transport (Wallach and Hart, 1983). Maternal anogenital stimulationdecreases the interintromission interval (Moore, 1984) and increases the probability that a mount will include an intromission (Moore, 1985b, 1992). It also increases the number of motor neurons (Moore, Dou, et al., 1992)in the nuclei that innervate the perineal muscles used to set and dislodge sperm plugs (Sachs, 1982; Hart and Melese-d’Hospita1, 1983; Wallach and Hart,1983). These results suggest that maternal stimulation can affect aspects of copulatory performance that in turn can affect the reproductive success of sons. This possibility is supported by a recently observed correlation between reproductive success and maternal stimulation. F344 and Long-Evans rats have equivalent fertilizing capacity in noncompetitive matings, but Long-Evans males father about 75% of pups in matings that involve sperm competition (Dewsbury and Hartung, 1980; Moore and Wong, 1992). Long-Evans males were also more successful when each male was allowed one ejaculation. There is a second-male advantage for each strain, but this advantage is greater in Long-Evans males even when there is no strain difference in intromission behavior following the first male’s ejaculation (Moore and Wong, 1992). This difference suggests a strain difference in the ability to form, place, or remove plugs. F344 mothers perform lower levels of maternal anogenital licking than Long-Evans mothers, regardless of strain of pup (Moore, Wong, et al., 1992; Fig. 10). The pattern of findings suggests a maternal effect on offspring reproductive success, but experimental verification is needed. Experiments are currently under way to test the prediction that high levels of maternal stimulation will provide male offspring with an advantage over other males during competitive copulation.

VI. SUMMARY AND CONCLUSIONS The early developmental milieu in mammals is structured primarily by the physiology, physical characteristics, and behavior of mothers. The regulatory role of the maternally provided milieu is an important source of developmental order, helping to ensure species-typical developmental

MATERNAL CONTRIBUTIONS

105

outcomes in a range of offspring characteristics. Analytic studies of the consequences of specific maternal factors indicate that there are multiple, independent contributions to reproductive development of components present in both prenatal and neonatal environments. Some of these factors are present for offspring of both sexes, which accounts, in part, for similarities in developmental outcome for males and females. There are sexrelated biases in other factors, such that males and females are sometimes exposed to reliably different environments. These differences account for some sex differences in developmental outcomes. The characteristics of infants that give rise to sex-related biases in the rearing environment originate from differences in gonadal steroids. These hormones, when present in infants, affect maternal behavior by changing the olfactory, behavioral, and other properties of the young. Thus, some effects of gonadal hormones on sexual differentiation and the divergence of the sexes are exerted through indirect routes, mediated by the stimulation provided by the mother’s behavior. These same processes account for some of the individual differences within males or females that can be traced to differences in hormonal condition during early development. Differences in the amount or timing of testosterone have been linked to maternal stress during pregnancy and to the sex of neighboring fetuses in utero, both of which also alter maternal behavior directed toward the affected offspring. At this stage, support for these summary statements comes almost entirely from a series of experiments in rats. These experiments demonstrate that there are differences in a specific form of maternal stimulation that occur because of hormonally mediated olfactory differences in pups. There is evidence that the maternal stimulation contributes toward the development of masculine reproductive behavior as well as spinal mechanisms that underlie penile actions. Further work is needed to determine the extent and functional significance of these effects in rats and to determine the generality of these phenomena among mammals. Those individual differences in the development of reproduction that can be traced to differences in maternal behavior can, in principle, underlie differences in reproductive success of offspring. There is circumstantial evidence that variation in the maternal stimulation provided by maternal rats to their male offspring produce variation in the success of sperm competition when the males become adult copulators. This possibility needs direct experimental testing. If direct maternal contributions to offspring reproductive success can be established, the extensive information that has been gathered on the neural, endocrine, and behavioral mechanisms of rat reproduction holds forth the promise that a reasonably complete account of the developmental and causal foundations for this functional outcome can also be established.

106

CELIA L. MOORE

Acknowledgments Preparation of this chapter was supported by Grant IBN-9121238from the National Science Foundation. The work from my laboratory reported in the chapter was also supported by this and previous grants from the National Science Foundation. I am very grateful for this continuing support. I am also grateful to the editors of this volume and their reviewers for helpful comments that have improved an earlier draft.

References Adkins-Regan, E. (1989). Sex hormones and sexual orientation in animals. Psychobiol. 16, 335-347. Adkins-Regan, E., Orgeur, P., and Signoret, J. P. (1989). Sexual differentiation of reproductive behavior in pigs: Defeminizing effects of prepubertal estradiol. Horm. Behav. 23, 290-303. Adler, N. T., and Toner, Jr., J. P. (1986). The effects of copulatory behavior on sperm transport and fertility in rats. Ann. N . Y. Acad. Sci. 474, 21-32. Alberts, J. R., and Brunjes, P. C. (1978). Ontogeny of thermal and olfactory determinants of huddling in the rat. J . Comp. Physiol. Psychol. 92, 897-906. Alberts, J. R.. and Cramer, C. P. (1988). Ecology and experience: Sources of means and meaning of developmental change. In “Handbook of Behavioral Neurobiology” (E. M. Blass, ed.), Vol. 9, pp. 1-39. Plenum, New York. Alberts, J. R., and Gubernick, D. J. (1983). Reciprocity and resource exchange: A symbiotic model of parent-offspring relations. In “Symbiosis in Parent-Offspring Interactions” (L. A. Rosenblum and H. Moltz,eds.), pp. 7-44. Plenum, New York. Alberts, J. R., and May, B. (1984). Non-nutritive, thermotactile induction of filial huddling in rat pups. Dev. Psychobiol. 17, 161-181. Albonetti, M. E., and D’Udine, B. (1986). Social experience occurring during adult life: Its effects on socio-sexual olfactory preferences in inbred mice, Mus musculus. Anim. Behav. 34, 1844-1847. Allen, T. O., and Adler, N. T. (1985). Neuroendocrine consequences of sexual behavior. In “Handbook of Behavioral Neurobiology” (N. Adler, D. Pfaff, and R. W. Goy, eds.), Vol. 7, pp. 725-766. Plenum Press, New York. Alleva, E., Caprioli, A., and Laviola, G. (1989). Litter-gender composition affects maternal behavior of the primiparous mouse dam (Mus musculus). J . Comp. Psychol. 103,83-87. Alleva, E., D’Udine, B., and Oliverio, A., (1981). Effect d’une experience olfactive precoce sur les preferences sexuelles de neux souches de souris consanguines. Biol. Behau. 6, 73-78. Almli, C. R., and Fisher, R. S. (1985). Postnatal development of sensory influences on neurons in the ventromedial hypothalamic nucleus of the rat. Deuel. Brain Res. 18, 13-26. Altmann, J., Hausfater, G., and Altmann, S. A. (1988). Determinants of reproductive success in savannah baboons, Papio cynocephalus. In “Reproductive Success” (T. H. CluttonBrock, ed.), pp. 403-418. Univ. Chicago Press, Chicago. Anderson, D. K.,Rhees, R. W., and Fleming, D. E. (1985). Effects of prenatal stress on differentiation of the sexually dimorphic nucleus of the preoptic area (SDN-POA) of the rat brain. Brain Research 332, 113-118. Arnold, A. P., and Breedlove, S. M. (1985). Organizational and activational effects of sex steroids on brain and behavior: A reanalysis. Horm. Behav. 19, 469-498.

MATERNAL CONTRIBUTIONS

107

Arnold, S . J. (1987). Genetic correlation and the evolution of physiology. I n “New Directions in Ecological Physiology” (M. E. Feder, A. F. Bennett, W. W. Burggren, and R. B. Huey, eds.), pp. 189-215. Cambridge University Press, Cambridge. Atchley, W. R., and Hall, B. K. (1991). A model for development and evolution of complex morphological structures. Eiol. Rev. 66, 101-157. Bamshad, M., Novak, M. A., and DeVries, G. J. (1993). Sex and species differences in the vasopressin innervation of sexually naive and parental prairie voles, Microtus ochrogaster and meadow voles, Microtus pennsylvanicus. J . Neuroendocrinol. 5,247-255. Bateson, P. P. G. (1978). Early experience and sexual preferences. I n “Biological Determinants of Sexual Behaviour” (J. B. Hutchison, ed.), pp. 29-53. Wiley, Chichester. Bateson, P. (1983). Optimal outbreeding. I n “Mate Choice” (P. Bateson, ed.), pp. 257-277. Cambridge University Press, Cambridge. Bateson, P. (1987). Imprinting as a process of competitive exclusion. In “Imprinting and Cortical Plasticity” (J. P. Rauschecker and P. Marler, eds.), pp. 151-168. Wiley, New York. Baum, M. J., Brand, T., Ooms, M., Vreeburg, J. T. M.. and Slob, A. K. (1988). Immediate postnatal rise in whole body androgen content in male rats: Correlation with increased testicular content and reduced body clearance of testosterone. Biol. Reprod. 38, 980-986.

Baum. M. J., Erskine, M. S., Kornberg, E., and Weaver, C. E. (1990). Prenatal and neonatal testosterone exposure interact to affect differentiation of sexual behavior and partner preference in female ferrets. Eehau. Neurosci. 104, 183-198. Beach, F. A. (1942). Analysis of the stimuli adequate to elicit mating behavior in the sexually inexperienced male rat. J. Comp. Psychol. 33, 163-207. Beach, F. A. (1968). Coital behavior in dogs. 111. Effects of early isolation on mating in males. Eehauiour 30,218-238. Beach, F. A. (1971). Hormonal factors controlling the differentiation, development, and display of copulatory behavior in the ramstergig and releated species. I n “The Biopsychology of Development” (E. Tobach, L. R. Aronson, and E. Shaw, eds.), pp. 249-2%. Academic Press, New York. Beach, F. A., Johnson, A. I., Anisko, J. J., and Dunbar, I. F. (1977). Hormonal control of sexual attraction in pseudohermaphroditic female dogs. J. Comp. Physiol. Psychol. 91, 711-715.

Beaston-Wimmer, P., and Smolen, A. J. (1991). Gender differences in neurotransmitter expression in the rat superior cervical ganglion. Dev. Brain Res. 58, 123-128. Bernstein, I. L. (1988). Development of salt aversion in the Fischer-344 rat. Deu. Psychobiol. 21,663-670.

Beyer, C . , and Feder, H. H. (1987). Sex steroids and afferent input: Their roles in brain sexual differentiation. Annu. Reu. Neurosci. 49, 349-364. Birch, H. G. (1971). Levels, categories, and methodological assumptions in the study of behavioral development. I n “The Biopsychology of Development” (E. Tobach, L. R. Aronson, and E. Shaw, eds.), pp. 503-513. Academic Press, New York. Birke, L. I. A., and Sadler, D. (1983). Progestin-induced changes in play behavior of the prepubertal rat. Physiol. Eehau. 30, 341-347. Birke, L. I. A., and Sadler, D. (1985). Maternal behavior in rats and the effects of neonatal progestins given to the pups. Dev. Psychobiol. 18, 467-475. Birke, L. I. A., and Sadler, D. (1987). Differences in maternal behavior of rats and the sociosexual development of the offspring. Deu. Psychobiol. u),85-99. Breedlove, S . M. (1992). Sexual dimorphism in the vertebrate nervous system. J. Neurosci. 12,4133-4142.

108

CELIA L. MOORE

Breedlove, S. M., and Arnold, A. P. (1981). Sexually dimorphic motor nucleus in the rat lumbar spinal cord: Response to adult hormone manipulation, absence in androgeninsensitive rats. Brain Research 225, 297-307. Breedlove, S. M., and Arnold, A. P. (1983). Hormonal control of adeveloping neuromuscular system. 11. Sensitive periods for the androgen-induced masculinization of the rat spinal nucleus of the bulbocavernosus. J . Neurosci. 3, 424-432. Brouette-Lahlou, I., Amouroux, R., Chastrette, F., Cosnier, J.. Stoffelsma, J., and VernetMaury, E. (1991). Dodecyl propionate, a regulatory factor from rat pup, characterization and identification. J . Chem. Ecol. 17, 1343-1354. Brouette-Lahlou, I., Vernet-Maury, E., and Chanel, J. (1991). Is rat-dam licking behavior regulated by pups’ preputial gland secretion? Anim. Learn. Behau. 19, 177-184. Brouette-Lahlou, I., Vernet-Maury, E., Godinot, F., and Chanel. J. (1992). Vomeronasal organ sustains pups’ anogenital licking in primiparous rats. I n “Chemical Signals in Vertebrates VI” (R. L. Doty, and D. Muller-Schwarze. eds.), pp. 551-555. Plenum, New York. Brown, R. E. (1979). Mammalian social odors: A critical review. I n “Advances in the Study of Behavior” (J. S. Rosenblatt, R. A. Hinde, C. Beer, and M.-C. Busnel, eds.), Vol. 10, pp. 104-162. Academic Press, New York. Brunelli, S. A., Shindledecker. R. D., and Hofer. M. A. (1989). Early experience and maternal behavior in rats. Deu. Psychobiol. 22, 295-314. Bull, J. J. (1983). “Evolution of Sex DeterminingMechanisms.” Benjamin-Cummings,Menlo Park, California. Bulman-Fleming, B., and Wahlsten, D. (1991). The effects of intrauterine position on the degree of corpus callosum deficiency in two substrains of BALB/c mice. Deu. Psychobiol. 24, 395-412. Camp, L. L., and Rudy, J. W. (1988). Changes in the categorization of appetitive and aversive events during postnatal development of the rat. Deu. Psychobiol. 21, 25-42. Campbell, B. (Ed.). (1972). “Sexual Selection and the Descent of Man: 1871-1971.” Aldine, Chicago. Capek, K., and Jelinek. J. (1956). The development of the control of water metabolism 1. The excretion of urine by young rats. Physiol. Bohemoslou. 5, 91-96. Carr, W. J., Loeb, L. S., and Dissinger, M. L. (1965). Responses of rats to sex odors. J . Comp. Physiol. Psychol. 59, 370-377. Chapman, R. H., and Stem, J. M. (1978). Maternal stress and pituitary-adrenal manipulations during pregnancy in rats: Effects on morphology and sexual behavior of male offspring. J . Comp. Physiol. Psychol. 92, 1074-1083. Charnov, E. L. (1982). “The Theory of Sex Allocation.” Princeton University Press, Princeton, New Jersey. Charton, D., Adrien, J., and Cosnier, J. (1971). Declencheurs chimiques du comportement de lechage des petits par la ratte parturiente. Rev. Comport. Animal 5, 89-94. Cheal, M. (1975). Social olfaction: A review of the ontogeny of olfactory influences on vertebrate behavior. Behau. Biol. 15, 1-25. Cheverud, J. M. (1984). Quantitative genetics and developmental constraints on evolution by selection. J . Theoret. Biol. 110, 155-171. Christian, J. J., and LeMunyan, C. D. (1958). Adverse effects of crowding on lactation and reproduction of mice and two generations of their progeny. Endocrinol. 63,517-529. CihAk, R., Gutman, E., and Hanzfkovft, V. (1970). Involution and hormone-induced persistence of the M . sphincter (leuator) ani in female rats. J . Anat. 106, 93-1 10. Clark, M. M., Bone, S., and Galef, Jr., B. G. (1989). Uterine positions and schedules of urination: Correlates of differential maternal anogenital stimulation. Deu. Psychobiol. 22,389-400.

MATERNAL CONTRIBUTIONS

109

Clark, M. M., Crews, D., andGalef, Jr., B. G. (1991).Concentrations ofsex steroid hormones in pregnant and fetal Mongolian gerbils. Physiol. Behau. 49, 239-243. Clark, M. M., Crews, D., and Galef, Jr., B. G. (1993). Androgen mediated effects of male fetuses on the behavior of dams late in pregnancy. Deu. Psychobiol. 26, 25-35. Clark, M. M., and Galef, Jr., B. G. (1988). Effects of uterine position on rate of sexual development in female Mongolian gerbils. Physiol. Behau. 42, 15-18. Clark, M., and Galef, Jr., B. G. (1989). Male rat pups are more hesitant to urinate in response to anogenital stimulation than are their female sibs. Deu. Psychobiol. 22, 81-85. Clark, M. M., and Galef, Jr., B. G. (1990). Sexual segregation in the left and right horns of the gerbil uterus: “The male embryo is usually on the right, the female on the left” (Hippocrates). Dev. Psychobiol. 23, 29-37. Clark, M. M., and Galef, Jr., B. G. (1992). Unconfounded evidence of sex-biased, postnatal maternal effort by Mongolian gerbil dams. Deu. Psychobiol. 24, 539-546. Clark, M. M., Galef, Jr., B. G., and vom Saal, F. S. (1991). Nonrandom sex composition of gerbil, mouse, and hamster litters before and after birth. Deu. Psychobiol. 24, 81-90.

Clark, M. M., Karpiuk, P., and Galef, Jr., B. G. (1993). Hormonally mediated inheritance of acquired characteristics in Mongolian gerbils. Nature (London) 364, 712. Clark, M. M.. Tucker, L., and Galef, Jr., B. G. (1992). Stud males and dud males: Intrauterine position effects on the reproductive success of male gerbils. Anim. Behau. 43, 2 15-221. Clemens. L. G. (1974). Neurohormonal control of male sexual behavior. I n “Reproductive Behavior” (W. Montagna and W. A. Sadler, eds.), pp. 23-53. Plenum, New York. Clemens, L. G., Gladue, B. A., and Coniglio, L. P. (1978). Prenatal endogenous androgenic influences on masculine sexual behavior and genital morphology in male and female rats. Horm. Behav. 10,40-53. Cleveland, J., and Snowdon, C. T. (1984). Social development during the first twenty weeks in the cotton-top tamarin (Saguinus 0. oedipus). Anim. Behav. 32, 432-444. Clutton-Brock, T. H., and Albon, S. D. (1982). Parental investment in male and female offspring in mammals. I n “Current Problems in Sociobiology” (King’s College Sociobiology Group, eds.), pp. 223-247. Cambridge Univ. Press, Cambridge. Clutton-Brock, T. H., Guinness, F. E., and Albon, S. D. (1982). “Red Deer: Behavior and Ecology of Two Sexes.” University of Chicago Press, Chicago. Coopersmith. R.. and Leon, M. (1984). Enhanced neural response to familiar olfactory cues. Science 225,849-85 1. Dahlof, L.-G., HBrd, E., and Larsson, K. (1977). Influence of maternal stress on offspring sexual behaviour. Anim. Behav. 25,958-963. de Jonge, F. H., Muntjewerff, J.-W., Louwerse, A. L., and van de Poll, N. E. (1988). Sexual behavior and sexual orientation of the female rat after hormonal treatment during various stages of development. Horm. Behau. 22, 100-115. Denenberg, V. H., and Whimbey, A. E. (1%3). Behavior of adult rats is modified by the experiences their mothers had as infants. Science 142, 1192-1193. Dewsbury, D. A. (1984). Sperm competition in muroid rodents. I n “Sperm Competition and the Evolution of Animal Mating Systems” (R. L. Smith, ed.), pp. 547-571. Academic Press, Orlando. Dewsbury, D. A.. and Hartung, T. G. (1980). Copulatory behavior and differential reproduction of laboratory rats in a two-male, one-female competitive situation. Anim. Behau. 28,95-102. Dohler, K. D., and Wuttge, W. (1975). Changes with age in levels of serum gonadotropins, prolactin, and gonadal steroids in prepubertal male and female rats. Endocrinol. 97, 898-907.

110

CELlA L. MOORE

Dolhinow, P. J., and Bishop, N. (1970). The development of motor skills and social relationships among primates through play. Minnesota Symp. Child Psychol. 4, 141-201. D’Udine, B., and Alleva, E. (1983). Early experience and sexual preference in rodents. I n “Mate Choice” (P. Bateson, ed.), pp. 31 1-327. Cambridge University Press, Cambridge. Dunbar, I., Buehler, M., and Beach, F. A. (1980). Developmental and activational effects of sex hormones on the attractiveness of dog urine. Physiol. Behau. 24,201-204. Edelman, G . (1988). “Topobiology.” Basic Books, New York. Eisler, J. A., Tannenbaum, P. L., Mann, D. R., and Wallen, K. (1993). Neonatal testicular suppression with a GnRH agonist in rhesus monkeys: Effects on adult endocrine function and behavior. Horm. Behau. 27,551-567. Fedigan, L. M. (1982). “Primate Paradigms: Sex Roles and Social Bonds.” Eden Press, Montreal. Fillion. T. J., and Blass, E. M. (1986a). Infantile behavioural reactivity to oestrous chemostimuli in Norway rats. Anim. Behav. 34, 123-133. Fillion, T. J., and Blass, E. M. (1986b). Infantile experience with suckling odors determines adult sexual behavior in male rats. Science 231, 729-731. Fleming, A., and Rosenblatt, J. S. (1974). Olfactory regulation of maternal behavior in rats: 1. Effects of olfactory bulb removal in experienced and inexperienced lactating and cycling females. J . Comp. Physiol. Psychol. 86, 221-232. Fleming, A., Vaccarino, F., Tombosso, L., and Chee, P. (1979). Vomeronasal and olfactory system modulation of maternal behavior in the rat. Science 203, 372-374. Ford, J. J. (1983). Postnatal differentiation of sexual preference in male pigs. Horm. Behau. 17, 152-162.

Friedman, M. I., Bruno, J. P., and Alberts, J. R. (1981). Physiological and behavioral consequences in rats of water recycling during lactation. J . Comp. Physiol. Psychol. 95, 26-35.

Gandelman, R. F., vom Saal, F. S., and Reinisch, J. (1977). Contiguity to male fetuses affects morphology and behavior of female rats. iVature (London) 266, 722-724. Gaulin. S. J. C., and Fitzgerald, R. W. (1989). Sexual selection for spatial-learning ability. Anim. Behau. 37, 322-331. Gawienowski, A. M. (1977). Chemical attractants of the rat preputial gland. I n “Chemical Signals in Vertebrates” (D. Muller-Schwarze and M. M. Mozell. eds.), pp. 45-60. Plenum, New York. Gawienowski, A. M., Orsulak, P. J., Stacewicz-Sapuntzakis. M., and Joseph, B. (1975). Presence of sex pheromone in preputial gland of male rats. J. Endocrinol. 67,283-288. Gerall, A. A. (1963). An exploratory study of the effects of social isolation on the sexual behaviour of guinea pigs. Anim. Behau. 11, 274-282. Gerall, H. D., Ward, I. L., and Gerall, A. A. (1967). Disruption of the male rat’s sexual behavior induced by social isolation. Anim. Behau. 15, 54-58. Gibber, J. R., and Goy. R. W. (1985). Infant directed behavior in young rhesus monkeys: Sex differences and the effects of prenatal androgens. Am. J . Prirnatol. 8, 225-237. Gilder, P M., and Slater, P. J. B. (1978). Interest of mice in conspecific male odours is influenced by degree of kinship. Nafure (London) 274, 364-365. Glickman, S. E., Frank, L. G., Licht, P., Yalcinkaya, T., Siiteri, P. K., and Davidson, J. (1992). Sexual differentiation of the female spotted hyena: One of nature’s experiments. Ann. New York Acad. Sci. 662, 135-159. Goldfoot, D. A., and Wallen, K. (1978). Development of gender role behaviors in heterosexual and isosexual groups of infant rhesus monkeys. Recent Adu. Primarol. 1, 155-159. Gorski, R. A. (1984). Critical role for the medial preoptic area in the sexual differentiation of the brain. Prog. Brain Research 61, pp. 129-146.

MATERNAL CONTRIBUTIONS

11 I

Gorski, R. A., Gordon, J. H., Shryne, J. R., and Southam, A. M. (1978). Evidence for a morphological sex difference within the medial preoptic area of the rat brain. Brain Research 148, 333-346. Gottlieb, G. (1971). “Development of Species Identification in Birds.” University of Chicago Press, Chicago. Gottlieb. G. (1992). “Individual Development and Evolution.” Oxford Univ.Press, Oxford. Gould, S . J. (1977). “Ontogeny and Phylogeny.” Harvard Univ. Press, Cambridge, Massachusetts. Gould. S. J. (1989). A developmental constraint in Cerion, with comments on the definition and interpretation of constraint in evolution. Evolution 43, 516-539. Goy, R. W. (1970). Early hormonal influences on the development of sexual and sex-related behavior. In “The Neurosciences: Second Study Program” (F. 0. Schmitt, ed.), pp. 196-207. Rockefeller University Press, New York. Goy, R. W. (1978). Development of play and mounting behavior in female rhesus virilized prenatally with esters of testosterone or dihydrostestosterone. I n “Recent Advances in Primatology: Vol. I , Behaviour” (D. C. Chivers, ed.), pp. 449-462. Academic Press, London. Goy, R. W., and Goldfoot, D. A. (1974). Experiential and hormonal factors influencing development of sexual behavior in the male rhesus monkey. I n “The Neurosciences: Third Study Program” (F. 0. Schmitt and F. G. Worden, eds.), pp. 571-581. The M.I.T. Press, Cambridge, Massachusetts. Goy, R. W., and McEwen, B. S. (1980). “Sexual Differentiation of the Brain.” The M.I.T. Press, Cambridge, Massachusetts. Grisham, W., Kerchner, M., and Ward, I. L. (1991). Prenatal stress alters sexually dimorphic nuclei in the spinal cord of male rats. Brain Research 551, 126-131. Gruendel, A. D., and Arnold, W. J. (1969).Effects ofearly social deprivation on reproductive behavior of male rats. J . Comp. Physiol. Psychol. 67, 123-128. Gubernick, D. J., and Alberts, J. R. (1983). Maternal licking of young: Resource exchange and proximate controls. Physiol. Behau. 31, 593-601. Hailman, J. P. (1982). Evolution and behavior: An iconoclastic view. In “Learning, Development, and Culture” (H. C. Plotkin, ed.), pp. 205-253. Wiley, New York. Hamburger, V., and Oppenheim, R. W. (1982). Naturally-occumng cell death in vertebrates. Neurosci. Comment 1, 39-55. Hammer, R. P., and Jacobson, C. D. (1984). Sex difference in dendritic development of the sexually dimorphic nucleus of the preoptic area in the rat. Inf. J . Deu. Neurosci. 2, 77-85.

Hanby, J. (1976). Sociosexual development in primates. I n “Perspectives in Ethology” (P. P. G. Bateson and P. H. Klopfer, eds.), Vol. 2, pp. 1-67. Plenum, New

York. Harlow, H. F. (1965). Sexual behavior in rhesus monkeys. I n “Sex and Behavior” (F. A. Beach, ed.), pp. 234-265. Wiley, New York. Hart, B. L.. and Melese-d’Hospital. P. Y. (1983). Penile mechanisms and the role of striated penile muscles in penile reflexes. Physiol. Behau. 31, 807-813. Harvey, P. W., and Chevins, P. F. D. (1985). Crowding pregnant mice affects attack and threat behavior of male offspring. Horm. Behau. l9,86-97. Hedricks, C., and McClintock, M. K. (1982). “Regulation of mating and sex ratios during postpartum estrus.” Conference on Reproductive Behavior (June). East Lansing. Michigan. Hedricks, C., and McClintock, M. K. (1990). Timing of insemination is correlated with the secondary sex ratio of Norway rats. Physiol. Behau. 48, 625-632.

112

CELIA L. MOORE

Herrenkohl, L. R. (1979). Prenatal stress reduces fertility and fecundity in female offspring. Science 206, 1097-1099. Herrenkohl, L. R. (1986). Prenatal stress disrupts reproductive behavior and physiology in offspring. Ann. New York Acad. Sci. 474, 120-128. Hinde, R. A., and Stevenson-Hinde, J. (1976). Towards understanding relationships: Dynamic stability. In “Growing Points in Ethology” (P. P. G. Bateson and R. A. Hinde, eds.), pp. 451-479. Cambridge University Press, Cambridge. Hofer, M. (1983). The mother-infant interaction as a regulator of infant physiology and behavior. I n “Symbiosis in Parent-Offspring Interactions” (L. A. Rosenblum and H. Moltz, eds.), pp. 61-75. Plenum, New York. Hofer, M. A., Shair, H., and Singh, P. (1976). Evidence that maternal ventral skin substances promote suckling in infant rats. Physiol. Behau. 17, 131-136. Holmes, W. G . (1988). Kinship and the development of social preferences. I n “Handbook of Behavioral Neurobiology” (E. M. Blass, ed.), Vol. 9, pp. 389-413. Plenum, New York. Holmes, W. G., and Sherman, P. W. (1983). Kin recognition in animals. Am. Scienrisr 71, 46-55.

Holzhausen, C., Murphy, S., and Birke, L. 1. A. (1984). Neonatal exposure to a progestin via milk alters subsequent LH cyclicity in the female rat. J . Endocrinol. 100, 149-154. Hrdy, S. B. (1987). Sex-biased parental investment among primate and other mammals: A critical evaluation of the Trivers-Willard hypothesis. I n “Child Abuse and Neglect: Biosocial Dimensions” (R.Gelles and J. Lancaster, eds.), pp. 97-147. Aldine, New York. Jensen, G. D., Bobbitt, R. A., and Gordon, B. N. (1967). Sex differences in social interaction between infant monkeys and their mothers. I n “Recent Advances in Biological Psychiatry” (J. Wortis, ed.), Vol 9, pp. 283-293. Plenum, New York. Jensen, G. D., Bobbitt, R. A., and Gordon, B. N. (1968). Sex differences in the development of independence of infant monkeys. Behaviour 30, 1-14. Joffe, J. M. (1%9). “Prenatal Determinants of Behaviour.” Pergamon Press, Oxford. Johanson, 1. B., and Terry, L. M. (1988). Learning in infancy. I n “Developmental Psychobiology and Behavioral Ecology. Handbook of Behavioral Neurobiology” (E. M. Blass, ed.), Vol. 9, pp. 245-281. Plenum, New York. Johnston, R. E. (1990). Chemical communication in golden hamsters: From behavior to molecules and neural mechanisms. I n “Contemporary Issues in Comparative Psychology” (D. A. Dewsbury, ed.), pp. 381-409. Sinauer, Sunderland, Massachusetts. Johnston, R. E., and Coplin. 9. (1979). Development of responses to vaginal secretion and other substances in golden hamsters. Behau. Neural. B i d . 42, 158-168. Jordan, C. L., Breedlove, S. M., and Arnold, A. P. (1982). Sexual dimorphism and the influence of neonatal androgen in the dorsolateral motor nucleus of the rat lumbar spinal cord. Brain Research 249, 309-314. Juraska, J. M., Henderson, C., and Miiller, J. (1984). Differential rearing experience, gender, and radial maze performance. Dev. Psychobiol. 17, 209-215. Kinsley, C. H., and Bridges, R. S. (1988). Prenatal stress and maternal behavior in intact virgin rats. Response latencies are decreased in males and increased in females. Horm. Behau. 22,76-89. Kirkpatrick, M., and Lande, R. (1989). The evolution of maternal characters. Evolution 43, 485-503.

Kuhn, C. M., Pauk, J., and Schanberg, S. M. (1990). Endocrine responses to mother-infant separation in developing rats. Deu. Psychobiol. 23, 395-410.

MATERNAL CONTRIBUTIONS

113

Lancaster, J. B. (1971). Play mothering: The relations between juvenile females and young infants among free-ranging vervet monkeys (Cercopithecus aethiops). Folia Primatol. 15, 161-182. Larsson, K. (1971). Impaired mating performances in male rats after anosmia induced peripherally or centrally. Brain Behau. Evol. 4, 463-471, Larsson, K. (1978). Experiential factors in the development of sexual behaviour. In “Biological Determinants of Sexual Behaviour” ( J . B. Hutchinson, ed.), pp. 54-86. Wiley, New York. Lee, M. H. S., and Williams, D. I. (1974). Changes in licking behaviour following handling of young. Anim. Behav. 22,679-681. Leedy, M. G., Vela, E. A., Popolow, H. B., and Gerall, A. A. (1980). Effect of prepubertal medial preoptic area lesions on male rat sexual behavior. Physiol. Behav. 24,341-346. Le Magnen, J. (1952). Les phenomenes olfacto-sexuels chez le rat blanc. Arch. Sciences Physiol. 6 , 295-332. Lephart, E. D., Fleming, D. E., and Rhees, R. W. (1989). Fetal male masculinization in control and prenatally stressed rats. Dev.Psychobiol. 22, 707-716. Lewontin, R. C. (1982). Organism and environment. In “Learning, Development. and Culture” (H. C. Plotkin, ed.), pp. 151-172. Wiley, New York. Lytton, H., and Romney, D. M. (1991). Parents’ differential socialization of boys and girls: A meta-analysis. Psych. Bull. 109, 267-2%. Maccoby, E. E., and Jacklin, C. N. (1974). “The Psychology of Sex Differences.” Stanford University Press, Stanford, California. Macrides, F., Johnson, P. A., and Schneider. S. P. (1977). Responses of the male golden hamster to vaginal secretion and dimethyl disulfide: Attraction versus sexual behavior. Behav. Biol. 20, 377-386. Marr, J. N., and Gardner, Jr., L. E. (1%5). Early olfactory experience and later social behavior in the rat: Preference, sexual responsiveness, and care of young. J . Genet. Psychol. lCn, 167-174. Martin, P., and Caro, T. M. (1985). On the functions of play and its role in behavioral development. I n “Advances in the Study of Behavior” (J. S. Rosenblatt, C. Beer, M.-C. Busnel, and P. J. B. Slater, eds.), Vol. 15, pp. 59-103. Academic Press, New York. McClintock, M. K., Anisko, J. J., and Adler, N. T. (1982). Group mating among Norway rats. 11. The social dynamics of copulation: Competition. cooperation, and mate choice. Anim. Behav. 30, 410-425. McClure, P. A. (1981). Sex-biased litter reduction in food-restricted wood rats (Neotoma floridana). Science 211, 1058-1060. McKenna, K. E., and Nadelhaft, I. (1986). The organization of the pudendal nerve in the male and female rat. J . Comp. Neurol. 248, 532-549. Mead, M. (1935). “Sex and Temperament in Three Primitive Societies.” William Morrow, New York. Meaney, M. J., and Stewart, J. (1981a). A descriptive study of social development in the rat ( R a m s norvegicus). Anim. Behau. 29, 34-45. Meaney, M. J . , and Stewart, J. (1981b). Neonatal androgens influence the social play of prepubescent male and female rats. Horm. Behau. 15, 197-213. Meaney, M. J., Stewart, J., and Beatty, W. W. (1985). Sex differences in social play: The socialization of sex roles. In “Advances in the Study of Behavior” (J. S. Rosenblatt, C. Beer, M.-C. Busnel, and P. J. B. Slater, eds.), Vol. 15, pp. 1-58. Academic Press, New York.

114

CELlA L. MOORE

Meisel, R. L., and Sachs, B. D. (1980). Spinal transection accelerates the developmental expression of penile reflexes in male rats. Physiol. Behav. 24, 289-292. Meisel, R. L., and Ward, I. L. (1981). Female rats are masculinized by male littermates located caudally in the uterus. Science 2l3, 239-242. Midkiff, E. E., Fitts, D. A.. Simpson, J. B., and Bernstein, I. L. (1985). Absence of sodium chloride preference in Fischer-344 rats. Am. J . Physiol. (Regulatory, Integrative, Comp. Physiol.) 249, R438-R442. Mills, A., and Sengelaub, D. R. (1992). Effect of unilateral dorsal rhizotomy on motoneuron number in a sexually dimorphic rat spinal nucleus. SOC.Neurosci. Absrr. 18(1), 233. Mills, A., and Sengelaub, D. R. (1993). Sexually dimorphic neuron number in lumbosacral dorsal root ganglia of the rat: Development and steroid regulation. J . Neurobiol. 24, 1543- 1553. Mitchell, G. D. (I%@. Attachment differences in male and female infant monkeys. Child Dev. 39,611-620. Moore, C. L. (1981). An olfactory basis for maternal discrimination of sex of offspring in rats (Rattus norvegicus). Anim. Behav. 22, 679-681. Moore, C. L. (1982). Maternal behavior in rats is affected by hormonal condition of pups. J . Comp. Physiol. Psychol. 96, 123-129. Moore, C. L. (1984). Maternal contributions to the development of masculine sexual behavior in laboratory rats. Dev. Psychobiol. 17, 346-356. Moore, C. L. (1985a). Another psychobiological view of sexual differentiation. Dev. Rev. 5, 18-55. Moore, C. L. (1985b). Development of mammalian sexual behavior. I n “The Comparative Development of Adaptive Skills” (E. S. Gollin, ed.), pp. 19-56. Lawrence Erlbaum, Hillsdale, New Jersey. Moore, C. L. (188%). Sex differences in urinary odors produced by young laboratory rats (Rattus norvegicus). J . Comp. Psychol. 99, 76-80. Moore, C. L. (1988). Maternal and self-grooming in Norway rats: Mechanisms and consequences of dissociable components. Ann. N. Y. Acad. Sci. 525,425-427. Moore, C. L. (1990). Comparative development of vertebrate sexual behavior: Levels, cascades, and webs. I n “Contemporary Issues in Comparative Psychology” (D. A. Dewsbury, ed.), pp. 278-299, Sinauer, Sunderland, Massachusetts. Moore, C. L. (1992). The role of maternal stimulation in the development of sexual behavior and its neural basis. Ann. N. Y. Acad. Sci. 662, 160-177. Moore, C. L., and Chadwick-Dias, A.-M. (1986). Behavioral responses of infant rats to maternal licking: Variations with age and sex. Dev. Psychobiol. 19, 427-438. Moore, C. L., Dou, H., and Juraska, J. M. (1992). Maternal stimulation affects the number of motor neurons in a sexually dimorphic nucleus of the lumbar spinal cord. Brain Research 572,52-56. Moore, C. L., Jordan, L., and Wong, L. (1995). Novelty, olfactory experience, and choice of sexual partner in male Norway rats. (submitted) Moore, C. L., and Morelli, G. A. (1979). Mother rats interact differently with male and female offspring. J . Comp. Physiol. Psychol. 93,677-684. Moore, C. L., and Power, K. L. (1986). Prenatal stress affects mother-infant interaction in Norway rats. Dev. Psychobiol. 19, 235-245. Moore, C. L., and Power, K. L. (1992). Variation in maternal care and individual differences in play, exploration, and grooming of juvenile Norway rat offspring. Dev. Psychobiol. 25, 165-182. Moore, C. L.,and Samonte, B. (1986). Preputial glands of infant rats (Rattus norvegicus) provide chemosignals for maternal discrimination of sex. J . Comp. Psychol. LOO, 76-80.

MATERNAL CONTRIBUTIONS

115

Moore, C. L., and White, R. H. (1994). Sex differences in the sensory branch of the pudendal nerve of rats. SOC. Neurosci. Absrr. 20(1), 591. Moore, C. L., and Wong, L. (1992). Copulatory behavior, reproduction, and sperm competition in two strains of male rats. Physiol. Behav. 51, 569-573. Moore, C. L., Wong, L., and Daum, M. (1992). “Strain differences in maternal behavior predicted by strain differences in masculine sexual behavior.” Conference of International Society for Developmental Psychobiology (November), New Orleans. Morin, L. P. (1986). A concept of physiological time: Rhythms in behavior and reproductive physiology. Ann. N . Y . Acad. Sci. 474, 331-351. Moss, H. A. (1967). Sex, age, and state as determinants of mother-infant interaction. Merrill Palmer Quart. 13, 19-36. Muir, J. L., Pfister, H. P., and Ivinskis, A. (1985). Effects of prepartum stress and postpartum enrichment on mother-infant interaction and offspring problem-solving ability in Rarrus norvegicus. J . Comp. Psychol. 99, 468-478. Murakami, S., and Arai, Y. (1989). Neuronal death in the developing sexually dimorphic periventricular nucleus of the preoptic area in the female rat: Effect of neonatal androgen treatment. Neurosci. Lett. 102, 185-190. Murphy, M. R. (1980). Sexual preferences of male hamsters: Importance of preweaning and adult experience, vaginal secretion, and olfactory or vomeronasal sensation. Behau. Neural Biol. 30, 323-340. Nelson, R . J . (1991). Maternal diet influences reproductive development in male prairie vole offspring. Physiol. Behav. 50, 1063-1066. Newton, B. W., and Hamill, R. W. (1989). Target regulation of the serotonin and substance P innervation of the sexually dimorphic cremaster nucleus. Brain Research 485,149-156. Newton, B. W.. Unger, J.. and Hamill, R. W. (1990). Calcitonin gene-related peptide and somatostatin immunoreactivities in the rat lumbar spinal cord: Sexually dimorphic aspects. Neurosci. 37, 471-489. Nieuwenhuijsen, K . , Slob, A. K., and de Neef, K. J. (1988). Fetal gender and aggression in stumptail monkeys (Macaca arcroides). Dev. Psychobiol. 21, 277-282. Okado, N., and Oppenheim, R. W. (1984). Cell death of motoneurons in the chick embryo spinal cord. 1X. The loss of motoneurons following removal of afferent inputs. J . Neurosci. 4, 1639-1652. Oppenheim, R. W. (1981). Neuronal cell death and some related regressive phenomena during neurogenesis: A selective historical review and progress report. I n “Studies in Developmental Neurobiology: Essays in Honor of Viktor Hamburger” (W.M. Cowan, ed.), pp. 74-133. Oxford Univ. Press, London. Oppenheim, R. W. (1986). The absence of significant postnatal motoneuron death in the brachial and lumbar spinal cord of the rat. J . Cornp. Neurol. 246, 281-286. Ostfeld, R. S., Pugh, S. R., Seamon, J. 0..and Tamarin, R. H. (1988). Space use and reproductive success in a population of meadow voles. J . Anim. Ecol. 57, 385-394. Pedersen, P. E., and Blass, E. M. (1981). Olfactory control over suckling in albino rats. I n “The Development of Perception: Psychobiological Perspectives” (R.N. A s h . J. R. Alberts, and M. R. Petersen, eds.), Vol. 1, pp. 359-387, Academic Press, New York. Pedersen, P. E., and Blass, E. M. (1982). Prenatal and postnatal determinants of the first suckling episode in the albino rat. Deu. Psychobiol. 15, 349-356. Pedersen, P. E., Greer, C. A., and Shepherd, G.M. (1988). Early development of olfactory function. I n “Handbook of Behavioral Neurobiology” (E. M. Blass, ed.), Vol. 8, pp. 163-203. Plenum, New York. Pedersen. P. E., Williams, C. L., and Blass, E. M. (1982). Activationand odor conditioning

116

CELIA L. MOORE

of suckling behavior in 3-day-old albino rats. J . Exp. Psychol.: Anim. Behav. Proc. 8, 329-341. Ffaf€,D. W. (1980). “Estrogens and Brain Function.” Springer, New York. Power, K. L., and Moore, C. L. (1986). Prenatal stress eliminates differential maternal attention to male offspring in Norway rats. Physiol. Behav. 38, 667-671. Previc, F. H. (1991). A general theory concerning the prenatal origins of cerebral lateralization in humans. Psych. Reu. 98,299-334. Purves, D. (1988). “Body and Brain: A Trophic Theory of Neural Connections.” Harvard Univ. Press, Cambridge, Massachusetts. Raf€, R. A., and Kaufman, T. C. (1983). “Embryos, Genes, and Evolution.” Macmillan, New York. Rand, M. N., and Breedlove, S. M. (1988). Progress report on a hormonally sensitive neuromuscular system. Psychobiol. 16, 398-405. Reppert, S. M. (1985). Maternal entrainment of the developing circadian system. Ann. N . Y. Acad. Sci. 453, 162-169. Resko, J. A., Feder, H. H., and Goy, R. W. (I%@. Androgen concentration in plasma and testis of developing rats. J . Endocrinol. 40, 485-491. Resko, J. A., Malley, A., Begley, D., and Hess, D. L. (1973). Radioimmunoassay oftestosterone during fetal development of the rhesus monkey. Endocrinol. 93, 156-161. Richmond, G., and Sachs, B. D. (1984). Maternal discrimination of pup sex by rats. Dev. Psychobiol. 17, 87-89. Ronca, A. E., Lamkin, C. A., and Alberts, J. R. (1993). Maternal contributions to sensory experience in the fetal and newborn rat (Rattus noruegicus). J. Comp. Psycho/. 107, 1-14. Rosenblatt, J. S. (1976). Stages in the early behavioral development of altricial young of selected species of non-primate mammals. In “Growing Points in Ethology” (P. P. G. Bateson and R. A. Hinde, eds.), pp. 345-383. Cambridge Univ. Press, Cambridge. Rosenblatt, J. S. (1983). Olfaction mediates developmental transition in the altricial newborn of selected species of mammals. Deu. Psychobiol. 16, 347-375. Rosenblatt, J. S., and Aronson, L. R. (1958). The influence of experience on the behavioural effects of androgen in prepuberally castrated male cats. Anim. Behau. 6, 171-182. Rosenblatt, J. S., and Lehrman, D. S. (1%3). Maternal behavior of the laboratory rat. In “Maternal Behavior in Mammals” (H. L. Rheingold, ed.), pp. 8-57. Wiley, New York. Rosenblatt, J. S.. and Siegel, H. I. (1981). Factors governing the onset and maintenance of maternal behavior among nonprimate mammals: The role of hormonal and nonhormonal factors. In “Parental Care in Mammals’’ (D. J. Gubernick and P. H. Klopfer. eds.), pp. 13-76. Plenum, New York. Roubertoux, P. L., Nosten-Bertrand, M., and Carlier, M. (1990). Additive and interactive effects of genotype and maternal environment. In “Advances in the Study of Behavior” (P. J. B. Slater, J. S. Rosenblatt, and C. Beer, eds.), Vol. 19, pp. 205-247. Academic Press, New York. Rubin, J. Z., Provenzano, F. J., and Luria, Z. (1974). The eye of the beholder: Parents’ views on sex of newborns. Am. J . Orthopsychiatry 44, 512-519. Ruddy, L. L. (1980). Nasal intubation: A minimally obtrusive anosmia technique applied to rats. Physiol. Behau. 24, 881-886. Sachs, B. D. (1982). Role of the rat’s striated penile muscles in penile reflexes, copulation and induction of pregnancy. J. Reprod. Fertil. 66, 433-443. Sachs, B. D. (1983). Potency and fertility: Hormonal and mechanical causes and effects of

hl \ I I I(NAI ( ‘ O N I K I I I I I I I O N S

pciiilc

. i l . ( i o i i h 111 i.;ifs. I r i

mi.!. I:.

I’rovc.

;inti

117

“I loriiioiics ;ind I4ch;iviour in Higher Vertebrates” ( J . Baltha-

K. Gillcs, cds.). pp. Xh-l 10. Springer-Verlag, Berlin.

; i d Mciscl. K. I.. (1979).Puhertaldevelopmentofpenilereflexesandcopulation in male rats. f ’ . s ~ c ~ / r o r r i ~ r i r o i ~ t r ~4,~ ~287-296. ~~ritri~l. Sackett. G. P. (1981). Receiving severe aggression correlates with fetal gender in pregnant pigtailed monkeys. Dev. Psychobiol. 14, 277-282. Schrflder. H. D. (1980). Organization of the motoneurons innervating the pelvic muscles of the male rat. J. Comp. Neurol. 192, 567-587. Selander. R. K. (1972). Sexual selection and dimorphism in birds. In “Sexual Selection and the Descent of Man: 1871-1971.” (B. Campbell, ed.), pp. 180-230. Aldine, Chicago. Sengelaub, D. R., and Arnold, A. P. (1986). Development and loss of early projections in a sexually dimorphic rat spinal nucleus. J . Neurosci. 6 , 1613-1620. Sengelaub, D. R., Nordeen, E. J., Nordeen, K. W., and Arnold, A. P. (1989). Hormonal control of neuron number in sexually dimorphic spinal nuclei of the rat: 111. Differential effects of the androgen dihydrotestosterone. J . Comp. Neurol. 280, 637-644. Shozu, M., Akasofu, K.. Harada. T.. and Kubota, Y. (1991). A new cause of female pseudohermaphroditism: Placental aromatase deficiency. J. Clin. Endocrinol. Merab. 72,560-566. Siegal. M. (1987). Are sons and daughters treated more differently by fathers than by mothers? Dev. Rev. 7, 183-209. Skolnick, N. J., Ackerman, S. H., Hofer, M. A., and Weiner, H. (1980). Vertical transmission of acquired ulcer susceptibility in the rat. Science 208, 1161-1162. Smotherman, W. P., and Bell, R. W. (1980). Maternal mediation of early experience. In “Maternal Influences and Early Behavior” (R. W. Bell and W. P. Smotherman, eds.), pp. 201-210. Spectrum Publications, Jamaica, New York. Smotherman, W. P.. and Robinson, S. R. (1988). The uterus as environment. I n Handbook of Behavioral Neurobiology” (E. M. Blass, ed.), Vol. 9, pp. 149-196. Plenum Press, New York. Snowdon. C. T. (1990). Mechanisms maintaining monogamy in monkeys. In “Contemporary Issues in Comparative Psychology” (D. A. Dewsbury, ed.), pp. 225-251. Sinauer, Sunderland, Massachusetts. Stanton, M. E., and Levine, S. (1990). Inhibition of infant glucocorticoid stress response: Specific role of maternal cues. Deu. Psychobiol. 23, 41 1-426. Stern, J. M. (1990). Multisensory regulation of maternal behavior and masculine sexual behavior: A revised view. Neurosci. Eiobehav. Revs. 14, 183-200. Stern. J. M.. and Johnson. S. K. (1990). Ventral somatosensory determinants of nursing behavior in Norway rats. I. Effects of variations in the quality and quantity of pup stimuli. Phvsiol. Eehav. 47, 993-101 I . Sullivan, R. M.. and Leon, M. (1986). Early olfactory learning induces an enhanced olfactory bulb response in young rats. Deu. Brain Res. 27, 278-282. Sullivan, R. M.. Wilson. D. A,, and Leon, M. (1989). Norepinephrine and learning-induced plasticity in infant rat olfactory system. J . Neurosci. 9, 3998-4006. Teicher, M . H . . and Blass, E. M. (1977). The role of olfaction and amniotic fluid in the first suckling response of newborn albino rats. Science 198, 635-636. Thor, D. H.. and Flannelly. K. J. (1979). Copulation and intermale aggression in rats. J. Comp. Phvsiol. Psvchol. 93, 223-229. Thor. D. H.. and W. R. Holloway. (1984). Social play in juvenile rats: A decade of methodological and experimental research. Neurosci. Eiobehav. Revs. 8, 455-464. Tobet, S. A., and Fox. T. 0. (1992). Sex differences in neuronal morphology influenced

S;ichs. H . I)..

118

CELIA L. MOORE

hormonally throughout life. In “Handbook of Behavioral Neurobiology” (A. A. Gerall, H. Moltz, and 1. L. Ward, eds.), Vol. I I , pp. 41-83. Plenum, New York. Trivers, R. L. (1972). Parental investment and sexual selection. In “Sexual Selection and the Descent of Man: 1871-1971” (B. Campbell, ed.), pp. 136-179. Aldine, Chicago. Trivers, R. L., and Willard, D. E. (1973). Natural selection of parental ability to vary the sex ratio of offspring. Science 179,90-92. vom Saal, F. S. (1989). Sexual differentiation in litter-bearing mammals: Influence of sex of adjacent fetuses in utero. J. Anim. Sci. 67, 1824-1840. Vomachka, A. J., and Lisk, R. D. (1986). Androgen and estradiol levels in plasma and amniotic fluid of late gestational male and female hamsters: Uterine position effects. Horm. Behau. 20, 181-193. Wallach, S. J. R., and Hart, B. L. (1983). The role of the striated penile muscles of the male rat in seminal plug dislodgement and deposition. Physiol. Behau. 31, 815-821. Wang, Z.-X., and DeVries, G. J. (1993). Testosterone effects on parental behavior and vasopressin immunoreactive projections in prairie voles (Microtus ochrogasfer).Brain Research 631, 156-160. Wang, Z.-X., Ferris, C. F., and DeVries, G. J. (1994). The role of septa1 vasopressin innervation in paternal behavior in prairie voles (Microfus ochrogaster). Proc. N a f l . Acad. Sci., U.S.A. 91,400-404. Ward, I. L. (1972). Prenatal stress feminizes and demasculinizes the behavior of males. Science 175, 82-84. Ward, I. L. (1992). Sexual behavior: The product of perinatal hormonal and prepubertal social factors. I n “Handbook of Behavioral Neurobiology” (A. A. Gerall, H. Moltz, and I. L. Ward, eds.), Vol. 11, pp. 157-180. Plenum, New York. Ward, I. L., and Weisz, J. (1984). Differential effects of maternal stress on circulating levels of corticosterone, progesterone, and testosterone in male and female rat fetuses and their mothers. Endocrinol. 114, 1635-1644. Ward, W. F., Karp, C. H.. and Aceto, Jr., H. (1977). Developmental effects of the uterine environment: Dependence on fetal sex in rats. J . Reprod. Ferril. 50, 269-274. Weisz, J., and Ward, 1. L. (1980). Plasma testosterone and progesterone titers in pregnant rats, their male and female fetuses and neonatal offspring. Endocrinol. 106, 306-3 16. West, M. J., and King, A. P. (1987). Settling nature and nurture into an ontogenetic niche. Dew. Psychobiol. 20, 549-562. Wilhelmsson, M., and Larsson, K.(1973). The development of sexual behavior in anosmic male rats reared under various social conditions. Physiol. Behau. 11, 227-232. Williams, G . C. (1975). “Sex and Evolution.” Princeton Univ. Press, Princeton, New Jersey. Yahr, P. (1988). Sexual differentiation of behavior in the context of developmental psychobiology. I n “Handbook of Behavioral Neurobiology” (E. M. Blass, ed.), Vol 9, pp. 197-243. Plenum Press, New York.

ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 24

Cultural Transmission in the Black Rat: Pine Cone Feeding JOSEPH TERKEL DEPARTMENT OF ZOOLOGY TEL AVIV UNIVERSITY RAMAT AVIV, ISRAEL

I. INTRODUCTION

Although the majority of natural oak and pine forests in Israel were cut down at the end of the nineteenth century and beginning of the twentieth, over the last fifty years reforestation with pine has taken place in the northern and coastal areas of the country. About 14 years ago, Aisner (1981) noticed bare pine cone shafts piled up under some of the trees in the northern pine forests. When he brought some of these cones to my laboratory to discuss the phenomenon, we observed that they had been gnawed off the branches, systematically stripped of all their scales, and the underlying seeds had been removed. Cones found in similar condition in other parts of the world could have been the work of squirrels (Layne, 1954; Smith and Balda, 1979), but as there are no squirrels in Israel, they were clearly not responsible in this case. The crossbill (Loxiu curuirostru) also feeds on pine seeds, but employs a different method of extraction. It anchors the cones in cracks in the tree trunks and using its asymmetric bill it pries off the scales, exposing the seeds; moreover, there are only a relatively small number of crossbills in Israel in comparison to the large amount of discarded pine cones that had been observed. Once we had discounted this possibility, we were left with no idea whatsoever as to the identity of the pine cone stripping creature. Our preliminary efforts were directed at determining whether our mysterious “stripper” was a nocturnal or diurnal animal; but first we had to capture it. Aisner, therefore, laid out plastic sheeting beneath several of the trees where we had observed large quantities of discarded cone shafts. The result of counting the number of cones and scales dropped on the plastic sheeting at different times of the day showed that the main activity I19

Copyright 0 1995 by Academic h e s s . Inc. All rights of reproduction in any form reserved.

120

JOSEPH TERKEL

took place after dark. Rodent feces were also found among the pine cone shafts on the sheeting, providing evidence that the animal we were seeking was probably a nocturnal rodent. Aisner then placed traps on the same pine tree branches under which we had found the cone shafts-and our first black rats (Raftus raftus) were caught. At this stage we still had no confirmation that it was these black rats that were the pine cone openers. We brought the wild black rats back to the animal house at Tel Aviv University where they were given a short period of acclimation in standard rat cages before being presented with pine cones. The rats turned out to be extremely timid in the presence of humans, which made it difficult to observe their interaction with the cones. However, it was quite evident from the bare cone shafts found in the cages each morning that these rats had indeed stripped the scales from the shafts and eaten the nutritious seeds beneath. These shafts were identical to those we had found in the wild. Despite the lack of direct observation, this was nonetheless our first evidence that the black rats do strip the scales from the cones and feed off the exposed seeds beneath (Aisner, 1981). We next turned to solving the problem of actual observation of the rats at work, which we had failed to do so far in both the forest and the laboratory. Aisner constructed a large cylindrical metal container (2.5 m high X 1.5 m diameter) with a large Perspex window. Through this window we were able to observe and photograph the animals without disturbing them. We placed a large section of pine tree trunk complete with branches and cones in the container. In their new “home” the rats were able to move freely on the branches and demonstrate their natural feeding behavior. Following an acclimation period to their new captive conditions, we were able to watch them as they left their nest boxes each evening and set off to feed on the pine cones. They would begin by warily climbing over the branches, stopping from time to time to inspect a cone. Once the choice of a particular cone had been made, the rat would gnaw it free from the branch (Fig. I), a maneuver that demands formidable skill in order to prevent the cone from falling to the ground. The rat would then carry off the cone to a preferred branch, which is used as a feeding station. There, as we observed for the first time, it would proceed to strip the scales from the cone in a systematic and stereotyped fashion, starting from the base (Fig. 2A) and continuing around the shaft in a spiral fashion (Fig. 2B,C). The rat removes each individual scale by prising the upper distal free edge of the scale from the cone’s surface with its teeth, and it then gnaws off the proximal end (Fig. 3). Such a technique is quite energy efficient, as no more of the scale is gnawed than is necessary. When all the scales have been removed in this fashion, only the bare shaft remains

121

CULTURAL TRANSMISSION IN THE BLACK RAT

FIG. 1. A black rat on a branch of a Jerusalem pine, starting to detach cone from branch.

(Fig. 2D; Fig. 4), which the rat then drops to the ground below. This was the reason for the piles of accumulated bare shafts that Aisner had originally observed beneath the pine tree feeding sites (Fig. 5) (Aisner, 1981; Aisner and Terkel, 1985, 1992a, 1992b).

A

B

C

D

FIG. 2. Pine cones in different stages of being stripped. (A) Intact cone; arrow indicates where rat would begin stripping. (B) Cone with 3 rows of scales removed. (C) Half-stripped cone. (D) Bare cone shaft after all scales have been removed.

122

JOSEPH TERKEL

FIG. 3. A black rat actively stripping a pine cone. Note the use of the rat’s lower incisors in prising up the scales.

FIG. 4. Cones detached by rat from pine branch before and after stripping.

CULTURAL TRANSMISSION IN THE BLACK RAT

123

FIG. 5. Accumulation of bare cone shafts beneath pine tree, indicating black rat activity.

Black rats are known to nest in a variety of trees. This opportunistic behavior of the black rat, together with its flexible feeding techniques and generally extremely adaptable behavior mechanisms (Rozin, 1976), have combined to enable this rodent to invade new habitats with great success and to achieve a worldwide distribution (Canby, 1977). Nonetheless, finding black rats inhabiting the pine trees was quite surprising, as this was the first time that there had been any sign of the presence of these rodents in the pine forests in Israel. Their occurrence as camp followers of man, however, in roofs of buildings, for example, is well known, as is their ability to obtain food from a wide variety of geographical areas and from such sources as food factories and warehouses (Canby, 1977; Lore and Flannelly, 1977).However, our discovery that they were living and feeding in the pine forests led us to consider the possibility that the practice of food selection and a feeding technique was passed on from one generation to the next. Due to the relatively recent planting of these pine forests, it seemed to be a case of opportunistic exploitation. A persistent preference for feeding on pine seeds appears to have been retained by certain black rat communities for several years, although there is also edible fruit nearby. The next questions we asked ourselves, therefore, related to the ontogeny of this food preference and development of the feeding technique necessary to take advantage of it. We have been involved for the past few years in seeking answers to these questions.

124 A.

JOSEPH TERKEL

FIELDOBSERVATIONS

We carried out our observations in the field in two forests near the agricultural settlement of Kibbutz Ramot Menashe in northern Israel. Both of the forests are about 15 ha in size and are located at least 2 km from any human population. The pine trees there were planted about 30-50 years ago, at high density (ca. 300 trees/ha) and currently stand at about 15-20 m high. Because of this high density most pines in the middle of the forest have not developed well, for primarily only those at the edge of the forest are able to get enough light and rainfall to allow a welldeveloped root system and rich crop of cones (see Fig. 6). It is among these sturdier trees that we found most of the black rat activity taking place. In other parts of the world this particular niche is often occupied by tree squirrels, wood mice, and other animals. In our pine forests, however, it is the rats that build their nests in these trees (see Fig. 7). They skillfully interweave thin pine twigs to construct nests varying in size from 25 x

FIG. 6. Typical pine tree in study area, showing abundant crop of cones.

CULTURAL TRANSMISSION IN THE BLACK RAT

125

FIG. 7. A black rat maternal nest on a pine tree. The nest is constructed of pine twigs attached to a main vertical branch, with an opening at the nest bottom.

30 x 35 cm to 40 x 50 x 100 cm. The nests are situated on the branches about 5-10 m above the ground and are almost entirely closed off, with just one opening 4-5 cm in diameter. The rats use thin twigs to attach the nest to branches of the trees that are quite distant from the trunk, possibly to impede the approach of predators. Inside the nests we found the remains of food, droppings, small twigs, and both intact and stripped cones (Aisner and Terkel, 1991, 1992b). In the Israeli pine forests there are no other competitors for this treetop habitat and the rats are thus the only creatures to exploit the pine seeds that provide their main, and possibly only, source of nourishment. Their water requirements are provided by the dew that accumulates on the pine needles during the night from the moisture-laden breezes that blow throughout the year. We ascertained that the pine seeds do indeed provide a sufficient and complete diet for the rats by providing the animals that we had trapped and brought to the laboratory with a diet consisting of only pine seeds. The rats remained healthy, gave birth, and raised their

126

JOSEPH TERKEL

litters through to weaning. We believe, however, that this is the first evidence that Rattus can be included among those animals known to feed and survive exclusively on pine seeds. Our efforts to capture black rats by placing traps on the forest floor were unsuccessful, confirming our view that these rodents seldom if ever leave their arboreal environment. The cones from which they obtain the seeds are found on the upper branches of the trees, and intact cones are never found on the ground, except following accidental breakage of a branch due to storms and the like. Another possible way in which the rats might obtain the seeds other than by opening the cones is when the cone scales react to extreme heat, such as fire, by lifting and releasing their seeds. During exceptionally hot summer days this process can occur; however, the wind blows the seeds to the ground where the rats do not collect them. B. DIRECTION OF RESEARCH

The pine forest is characterized by a poor fauna and a limited undercanopy flora. The resin content of the pine needles that cover the ground lowers the soil pH and virtually sterilizes it, thus leaving the pine cone seeds as the main, if not the only, food source. We therefore began to consider that the clue to the black rat’s successful colonization of this new habitat, which appeared to offer little in the way of food resources, might lie in the efficient and energy-saving technique the animals had developed to open the cones and obtain the highly nourishing seeds. We set out next to carry out experiments aimed at determining the factors involved in this pine cone opening behavior. To examine these, we used both male and female black rats from two habitats: experienced “strippers,” trapped in the pine forests (i.e., rats that already possessed the ability to open pine cones by stripping the scales); and “naive” rats, unfamiliar with pine cones, that we had trapped in urban dwellings or agricultural warehouses. 11. THEABILITYOF NAWEADULTRATSTO LEARNTO OPENCONES

A. TRIALAND

ERROR

1. Rationale and Method

In our very first experiment we examined the naive black rat’s ability to learn to open the pine cones through a process of trial and error (Aisner and Terkel, 1992a). The naive rats were placed either individually or in

CULTURAL TRANSMISSION IN THE BLACK RAT

127

pairs in cages and supplied with pine cones either detached or still on a branch. Their normal food intake of rat chow was restricted to 85% so that they would be constantly slightly hungry. We monitored their physical well-being throughout the experiment. The wild rats are naturally timid and only rarely open pine cones in the presence of an observer. We thus had to determine their ability to open and strip the cones indirectly, by examining both the state of the cones and the physical condition of the tested animals, as clearly reflected in the appearance of their dark fur, which becomes clumped and matted after two days of food deprivation. Naive rats generally show little interest in pine cones unless they are extremely hungry, so once a week we withheld their rat chow for 48 hours and gave them instead only fresh pine cones at the same stage of ripeness as those fed upon in nature by adult rats. Starting from the first day that fresh cones were provided, once a day, for one hour, we observed the interaction of the rats with the pine cones and noted any attempts made to strip and open them. As the rats were quite hungry by this time, their hunger prevailed over their natural timidity in our presence and the majority of them began to inspect the cones at once. For those rats that did not manipulate the cones in our presence, we indirectly established their stripping ability by examining the state of the cones every hour for six hours, and again once the following day. The appearance of the organized structure of the scales around the cones made it easy to tell immediately whether a rat had learned the opening technique (see Fig. 4). 2. Results After three months of experimentation, none of the 32 naive animals tested had learned to open the cones efficiently by trial and error. Because of their extreme hunger, the rats had begun gnawing at the pine cones in a haphazard manner (Fig. 8) and had actually managed to get at some of the seeds. However, their technique was so inefficient that if we had not provided them with a supplementary diet they would have starved. Although in this experiment none of the rats learned to open the cones efficiently because of the complex structure of the cone, we did observe several cases in which the rats began to remove scales from the base of the cone. The spiral structure of the cone led the rat to continue gnawing around the cones following the spiral direction in a corkscrew pattern, so that the scales were removed from the base but with some rows being skipped (the term row is used to denote one complete turn of the spiral). The schematic presentation of Fig. 9 demonstrates the extent to which the spiral structure of the cone determines the method of opening it. Although the rat did not use the most efficient technique for opening the

128

JOSEPH TERKEL

FIG. 8. Comparison of cones that have been stripped by an experienced rat (A) and two others randomly gnawed by naive black rats (B and C).

cone (i.e., to begin at the base and systematically remove row after row following the spiral pattern), the cone’s structure nevertheless led the animal to gnaw its way spirally around the cone. €3.

OBSERVATION OF EXPERIENCED ADULT“STRIPPER”RATS BY NAIVEADULTRATS

1 . Rationale and Method After we had determined that naive adult black rats cannot learn to open the pine cones successfully by trial and error, we decided to provide them with a model from which to learn. The experiment was designed to determine whether naive adult rats are able to learn the pine cone opening technique by observing experienced stripper rats (Aisner and Terkel, 1992a). We caged each individual naive adult rat (N = 15) with a stripper rat of the same sex. Each pair of animals was supplied with pine cones throughout the entire experiment. The naive rat, which had free access to the cones at all times, was able to observe the stripper rat at work opening the cones and feeding on the seeds. The naive rats were removed from the cage once each day and given a supplement of 85% of their normal intake of rat chow in order to ensure their physical well-being. To determine whether the naive rats had learned to open the cones and

CULTURAL TRANSMISSION IN THE BLACK RAT

129

FIG. 9. Pine cone from which a naive rat has stripped the scales in a corkscrew pattern, leaving intact rows of scales in the process.

extract the seeds, we examined the cones as in the procedure described previously. 2. Results Even following three months of exposure to experienced stripper rats, none of the 15 naive rats had improved its technique and learned to open the cones efficiently. They simply continued to gnaw at them haphazardly in a manner similar to that of the naive rats housed individually and provided with pine cones.

111. THECONTRIBUTION OF GENETICS AND/OR EARLY EXPERIENCE TO THE ABILITY OF YOUNGRATS TO LEARN TO OPENCONES I N MOTHERS’MILK A. PINESEEDFLAVOR

1. Rationale and Method

We now knew that naive adult rats are unable to learn the pine cone opening technique in the laboratory either through trial and error or obser-

130

JOSEPH TERKEL

vation of an experienced rat. In the forest, however, this technique appears to be in constant use from one generation to the next. It might be possible that the pups’ later choice of diet is influenced or restricted by their early experience of the flavor of the mother’s milk, as has been shown previously (Galef and Henderson, 1972; Galef and Clark, 1972; Le Magnen and Tallon, 1968). Could it be possible, therefore, that the stripper mothers were passing on the flavor of the pine seeds they ate through their milk to the young, and that such a flavor cue, provided by the mothers in this way, might influence their offsprings’ preference for pine seeds upon weaning, and thus stimulate them to learn the requisite stripping technique? In this experiment we provided seven lactating naive black rat mothers with a diet only of pine seeds, that we had extracted manually from heated cones. We then examined their pups’ ability to learn to open the cones (Aisner and Terkel, 1992a). After the pups were weaned we gave them pine cones as their main source of nourishment, together with a supplement of rat chow, and then observed their ability to extract the seeds from the cones. 2 . Results Despite random attempts by the pups to gnaw at the cones, none of them (N = 77) successfully acquired the stripping technique and they were thus unable to open the cones and obtain sufficient nourishment from the seeds. As they became weaker and their physical condition deteriorated, we provided them with supplementary food to prevent their deaths.

B. PUPS RAISEDBY PINE CONE STRIPPING MOTHERS 1 . Rationale and Method At this stage we were still much in the dark regarding how the black rats were acquiring the pine cone opening technique. Our next step, therefore, was to follow nature and examine in the laboratory whether and how pups born and raised by experienced “stripper” mothers were able to learn the cone opening technique that was observed in generation after generation of black rats in the wild (Aisner and Terkel, 1992a). In this experiment we took five experienced stripper mothers and housed each of them separately with their own offspring. They were provided with pine cones ad lib., from which they efficiently extracted and ate the seeds. The mothers interacted normally with their pups throughout the entire experiment, which lasted three months. Throughout this time the pups were able to be close to the mother while she was occupied with

CULTURAL TRANSMISSION IN THE BLACK RAT

131

stripping the cones and feeding on the seeds. At the end of this period, we tested the pups for their “stripping” ability.

2 . Results Of the 33 pups born to the stripper mothers, 31 were able to strip the cones efficiently at the age of three months. Only two of them had failed to learn to strip scales and were still gnawing away haphazardly. C. RECIPROCAL FOSTERING 1. Rationale and Method At this stage it was clear that pups of stripper mothers were able to learn the technique of cone opening, so we next set out to ascertain whether the transmission of the stripping technique from mother to offspring is acquired through a process of social interaction (Aisner and Terkel, 1992a). In this experiment we cross-fostered pups born to naive mothers on stripper mothers and vice versa as follows: we took eight pups from two naive mothers on day 5 postpartum and exchanged them with seven pups of the same age from two stripper mothers. The two naive mothers thus raised a combined total litter of six natural and seven foster pups; and the stripper mothers suckled six natural and eight foster pups. As the naive mothers were, of course, unable to open the pine cones provided to all the animals ad lib. throughout the course of the 3-month experiment, we supplemented their diet with rat chow. The stripper mothers, on the other hand, fed exclusively on the pine seeds, which they extracted from the constant supply of intact pine cones. 2. Results At three months of age, when the developing pups were tested for their ability to strip the cones, we found that none of the pups that had either been born to and raised by naive mothers, or born to stripper mothers but raised by naive ones, had learned to open the cones and obtain the seeds. In contrast, all the pups born to and raised by their stripper mothers, as well as those pups born to naive mothers but fostered on the stripper mothers, had acquired the cone stripping technique and were competently prising open the scales and obtaining the seeds (Table I). The above reciprocal fostering experiment clearly indicated to us for the first time that this ability to strip pine cones efficiently is acquired by the pups through their social proximity to an experienced stripper mother.

132

JOSEPH TERKEL

TABLE I DOESCROSS-FOSTERING OF Pups TO “STRIPPER”OR NAYVEMOTHERS INFLUENCE PINE CONE OPENING BEHAVIOR OF PUPS? Conditions

Number of pups

Number of pups acquiring technique

Naive mother raising own and pups from “stripper” mother. Pups exposed to intact cones. “Stripper” mother raising own and foster pups from naive mother. Pups exposed to mother opening cones and to intact cones.

13

0

14

14

a

Modified from Aisner and Terkel(1992a).

IV. ALTERATION OF THE CONESTO MOTIVATEAND/OR ASSISTTHE RATSTO LEARN TO OPENTHEM

A. EXPOSING PUPSTO PARTIALLY OPENEDCONES I. Rationale and Method A pine cone that is completely closed (intact) provides no clues to the inexperienced pup regarding the nutritious meal it contains in the form of the hidden pine seeds. However, the partially stripped and open cones manipulated by the mother do provide such a clue. As the mother prises up the scales one by one, the seeds are disclosed and exposed to the pups that generally gather around her mouth at this time. Was it possible, therefore, that this exposure to partially open cones contributed first to the pups’ motivation to open them and later on also to their ability to do so? In this experiment we heated pine cones in an oven to a temperature of 50°C at which point their scales lift in response to the heat and the seeds become exposed (Aisner and Terkel, 1991). Such cones present no difficulty to naive rats, which are able to extract the seeds easily. We gave three naive pregnant rats these already-opened pine cones, starting at halfway through their pregnancy. After their pups were born and had reached the stage of eating solid food, they, too, were able to feed on the seeds from these cones. Having confirmed that the pups were able to obtain the seeds from the cones with the lifted scales, we then tested them for their ability to open completely closed cones (N = 17). 2 . Results Despite the pups’ ability to remove and feed on the exposed seeds of open cones, this early introduction to the nutritional source contained

(‘Ill IllKAl

I’KANSMISSION

IN ‘I’Hti BLACK RAT

133

within the coiies did not lead lo an improvement in their learning to strip the closed cones competently in the stereotyped spiral fashion that we had observed in the stripper rats. OF NAIVEADULTRATS TO PARTIALLY B. EXPOSURE STRIPPED CONES

Neither naive adults nor their pups, it thus appeared, were able to acquire the pine cone stripping technique even when provided with the impetus of cones with exposed seeds. On the other hand, we had seen that pups born to naive mothers but raised by stripper mothers, did acquire this technique. Our next decision, therefore, was to concentrate on any possible clues that the pups might be getting from the cones being stripped by the mother, and on determining how these cones differed from the opened cones with the exposed seeds that we had been providing. We thus set out to examine the pups’ ability to learn the pine cone stripping technique in relation to the geometric structure of the cones. The spiral structure of the cones is such that the first row of scales at the bottom of the cone (the widest part), partially overlaps the next row up. In order for the rat to extract the seeds efficiently, therefore, it needs to strip off the scales in a systematic fashion, starting at the base and completing a complete spiral before going on to the second row, and so on up to the top. At this stage of our experiment we were visited by B. G. Galef, who suggested that supplying our rat pups with already-started cones might provide them with a clue as to how to continue. Following this line of thought, and based on earlier studies, as detailed below, our aim now was to establish whether partially opened cones (i.e., cones from which we had stripped the first few rows of scales in the spiral manner used by the stripper rats) would provide the black rat pups with the necessary clue for them to commence learning the efficient spiral stripping technique. 1 . Rationale and Method Together with Daphna Yanai, we had previously initiated a pilot study that had indicated that when naive adult rats were provided with cones from which the first four rows of scales had been removed most of the animals would continue to strip the cones proficiently and feed on the seeds inside them. However, when we provided these same rats later on with intact cones, none of them were able to begin the opening process and they reverted to their former random gnawing (Yanai, 1989). We decided at this point to “assist” the rats by providing them with a series of graded clues in the form of cones with a consecutively decreasing number of rows from which the scales had been stripped, beginning with

134

JOSEPH TERKEL

four rows and decreasing to zero (an intact cone). After their exposure to a cone with four stripped rows, the rats were observed for their ability to continue stripping the cone in the spiral order. Those animals that succeeded in doing so were then given cones with only three rows of scales removed, and so on down to zero. For each new goal they were given one week of interaction with the cones; if they failed to open the new level of cone they were graded successful from the previous level of cone only (Yanai, 1989). This experiment was aimed at determining the need for and effect of partially opened cones as a stimulus in learning to open intact cones. We provided naive adult rats with pine cones from which we had either manually removed the first few rows of scales or cones that had been “professionally” partially stripped by stripper rats. In both cases the naive animals were given cones with the spiral direction in which to continue opening already indicated (see Fig. 10). The experiment was divided into two parts. In the first part we gave pine cones from which we had stripped four rows of scales to 5 1 individually housed naive adult rats. The rats were maintained in a state ofconstant hunger through the restriction of their normal food intake to 85%. Following our daily examination of the cones, the rats were considered to have acquired the stripping ability only if they succeeded in competently removing the scales in a systematic fashion until only the bare shaft was left. Rats that did not succeed in opening the cones in this way after a period of four weeks were considered as having failed to acquire the technique. In the second part of the experiment, we took 20 rats from among those that had learned to continue stripping the pine cones in the first part of the experiment. At weekly intervals, for a period of four weeks, we pro-

FIG. 10. Pine cones in different stages of opening, with the number of rows of previously stripped scales decreasing consecutively from right to left.

CULTURAL TRANSMISSION IN THE BLACK RAT

135

vided them with additional pine cones from which we had stripped a consecutively decreasing number of rows of scales, as described above. 2. Results Of the 51 rats that were provided with spirally opened cones in this experiment, 35 learned to continue to open the cones and extract the seeds. Of the rest of the rats (N = 16), some simply continued to gnaw away at them haphazardly, while others failed to relate to them as a source of nourishment in any way at all. Of the rats that had previously learned to continue stripping the cones with four rows of scales removed, 90% continued to remove systematically the scales from these new cones. They eventually demonstrated an ability to open intact cones followingtheir exposure to cones with a consecutively decreasing number of rows of previously removed scales. Of the 20 rats in the above experiment, only two remained unable to open intact cones, although they did succeed in continuing to open cones with 4, 3, 2, and I rows removed; however, by the end of the 4-week period the rest of the rats were all able to deal competently with intact cones, stripping them “professionally” and obtaining the seeds. V. SHAVING VERSUS SPIRAL METHODOF OPENINGPINECONES

So far, we have mentioned only one method used by the black rat to obtain the nourishing seeds from the pine cones, the spiral method. The rat first gnaws the cone free of the branch and then carries it to a “feeding site,” where it removes the scales in a systematic fashion, beginning at the base and following the spiral design of the overlapping scales around the cone to the apex. In nature, however, there are circumstances in which the cones are stripped while still attached to the branch: we believe (despite the lack of conclusive evidence) that when the rat finds itself in a less secure setting, such as when invading new and unfamiliar territory, it may prefer to save the valuable time otherwise expended in detaching the cones, and instead open it in situ. On certain of the more isolated and prominent trees in the pine forests we had observed the bare shafts of pine cones that had been totally stripped of all their scales but which still remained attached to the branches. Among these particular cones we noticed a new phenomenon: some of the cones had had their scales removed only along the axis of one half of the cone. In the forest this situation was encountered when the cones were not attached to the branch in the usual upright fashion, but the shaft of the cone had instead grown parallel to the branch (Fig. 11).

136

JOSEPH TERKEL

FIG. 11. Cones stripped while still attached to the branch. The rat has removed scales from one side of the cone only, using the “shaving” method.

In such a case, the rat obviously cannot reach the scales on that half of the cone that is pressed close to the branch; it therefore removes only those scales it can reach on the far side of the cone, beginning at the base. After it has completed its feeding on one half of such a cone, the rat then abandons the cone, which remains attached to the branch, appearing very much as if it has been “shaved” along the entire length of that side (see Fig. 11). Because of this appearance of the cone, we termed this method of obtaining the seeds the “shaving” method, as opposed to the “spiral” method more generally encountered.

Rationale and Method Unlike the piles of discarded “stripped” cones that we continued to find throughout the entire period of our investigation of the phenomenon of the black rats’ pine cone opening behavior, we never found any cones that had been completely detached from the branch by the rats and then

CULTURAL TRANSMISSION IN THE BLACK RAT

137

“shaved” open. All cones gnawed off the branch by the rats in the forest were stripped using the spiral technique. While continuing our observations in the laboratory, we noticed that some rat pups, however, were using both stripping and shaving methods while learning to open the pine cones. We believe that the age at which the pups first come into contact with the cones may affect the way they continue to open them. As adults, they then continue to employ the method that they acquired as pups. Opher Zohar suggested the possibility that since there are no seeds beneath the first few rows of scales, at the cone base, the young pups may lack the motivation as well as the physical strength to continue to strip the cones spirally around the complete circumference without any immediate reward for their efforts. If, on the other hand, they shave the cones, they are almost immediately reinforced by obtaining seeds. In the laboratory we observed that a rat that employs the shaving method, having finished shaving one side of the cone, often rotates it and shaves off the scales from the other side. When a rat employs this method, it encounters more overlap on the scales, and must therefore expend more energy gnawing away at scale parts than if it had used the spiral stripping method. In nature, however, except when the cones were still attached to the branches (as detailed above), we found no other evidence of the shaving method, which led us to consider the efficacy of the two methods and the possible energy saved by the stripping technique. The study of Yanai (1989) confirmed the significance of providing partially opened cones to the black rat pups as a clue for their learning to continue to open cones (Yanai, 1989). She provided naive rats with cones from which she had stripped four rows of scales only half way around the circumference of the base (termed “started to 180””) in contrast to cones from which the four rows of scales had been completely removed all the way around (3609, as in the earlier experiments. A. EXPOSURE OF NAYVEADULTRATS TO HALF-OPENED CONES

In this experiment 24 naive adult rats were housed individually and provided each day with only two rat chow pellets and two pine cones “started to M O O . ” Once again their permanent state of slight hunger motivated them to seek additional nourishment. Their diet was supplemented for 36 hours once a week with an ad lib. provision of rat chow. Every morning the previous day’s cones were removed from the cage and the rat’s progress in opening them was determined before the new cones were provided.

138

JOSEPH TERKEL

Results Although the experiment lasted four weeks, most of the rats that learned the cone opening technique managed to do so within five days of first receiving the cones and by the end of three weeks they had gained complete mastery of the technique. Of the 24 animals that received the "started to 180"" cones, 22 continued to open the cones in a similar manner, that is, using the shaving technique from base to apex along one side of the cone only (Table 11). Most rats that had received pine cones started around the entire circumference of the base and that had learned to strip them using the spiral technique continued to open intact cones in the same manner. In contrast, however, rats that had been given the cones with scales removed from only half the circumference of several rows developed an opening technique of first shaving one side of the cone totally before turning it and continuing on the other side.

B. EXPOSURE OF PUPSTO PINECONESIN VARIOUS STAGES OF

OPENING

I . Rationale and Method Now that we knew that most of the naive adult rats were able to learn to strip pine cones if provided with the correct clue (i.e., cones already started at the base), we next turned to examine the role of the started cone in the pups' learning process (Zohar, 1987; Zohar and Terkel, 1991).

TABLE I1 DOESEXPOSURE OF NAYVEADULTBLACKRATS TO PARTIALLY OPEN CONES TO EITHER 360" OR 180"ENHANCE THEIR ABILITY TO CONTINUE STRIPPING THE CONES?' Conditions Naive adults. Solitary, supplied with cones stripped to 360". Naive adults. Solitary, supplied with cones stripped to 180". From Yanai (1989).

Number of animals tested

Number of animals opening cones

51

35 (68.6%)

Spiral

24

22 (91.7%)

Shaving

Style of opening

CULTURAL TRANSMISSION IN THE BLACK RAT

139

In nature, a pup has a much greater chance of finding a started or partially open cone near its mother, than does an adult rat of finding such a cone. The following experiment, therefore, attempted to stimulate the natural conditions under which the cone stripping technique might be acquired and later passed on to the next generation. In this experiment we provided the pups with partially open cones at different stages, thus simulating more closely the conditions that the pups might be expected to encounter in nature, rather than the less natural process of consecutively decreasing rows of stripped scales, as in the previous experiment (Zohar and Terkel, 1991). We housed five naive lactating females, each with five pups, in individual glass terraria measuring 30 x 50 x 35 cm with wire mesh lids. Each terrarium was divided in half by a wire mesh barrier with a 2 x 2 cm opening, which permitted the pups, but not the mother, free access to both sides. The mother and pups were provided with rat chow powder for 3 hours daily and water was supplied ad lib. The mother and her pups were housed in one half of the terrarium and a new batch of eight cones was placed in the other half each day, after the previous batch had been removed. We provided two cones at each of four different stages of opening: two closed; two stripped of scales at the proximal end only; two halfstripped of scales; and two bare shafts with all scales removed but a few seeds still remaining. As all the cones had been stripped to the various stages by stripper rats from our black rat colony, it is possible that it was the residual odor of the conspecific rats that attracted the pups to the cones (Galef, 1982; Galef and Stein, 1985; Galef and Whiskin, 1992; Valsecchi, Moles, and Mainardi, 1993). Because the pups were receiving a reduced diet, their state of permanent slight hunger stimulated them to search for an alternative source of food and they began to show interest in the pine cones we had placed in the other half of the cage. When the pups were 80 days old we took the mother out of the cage and tested the pups for their ability to strip the cones. 2. Results Six out of 25 (24%) of the pups had independently learned to strip the cones, without the benefit of observing this feeding behavior performed by an experienced stripper rat. Having confirmed the performance of two different methods of opening the cones, Yanai, Ar, and Terkel(l991) carried out a comparison of the efficiency of the two methods.

140

JOSEPH TERKEL

VI. THEENERGETICS OF PINECONEOPENING OF STRIPPING OR SHAVING THE CONES COMPARATIVE ENERGETICS

1 . Rationale and Method

In this experiment we assessed the energetic requirements of opening a pine cone by measuring both the energy invested and the duration of the opening process. We used equipment specially constructed for this purpose, based on a metabolic chamber, that measured the oxygen consumption of the rats while they were engaged in opening cones (Fig. 12). The metabolic chamber (2-liter volume) was placed in the experimental setup and maintained under a constant temperature of 28 1°C controlled by heated water flowing through the chamber’s double wall (Fig. 12). Dry air at room temperature was pumped into the chamber at a constant rate, measured by a flow meter. We measured the amount of oxygen leaving the chamber and compared it to the amount of oxygen entering the setup, using an oxygen analyzer installed as part of the setup. Measurements were monitored on a chart recorder (Fig. 13). The rat’s oxygen consumption during both resting and active periods was determined from the curves obtained from the chart recorder. Oxygen consumption was measured in units of ml O,/g body weight/hr, and converted from ml oxygen to Joules (standard energy unit). The calculation

*

METABOLIC CHAMBER

PUMP

OXYGEN ANALYZER

CHART RECORDER

FIG. 12. Oxygen consumption measuring system.

CULTURAL TRANSMISSION IN THE BLACK RAT

w + a

CONE OFFERED

14 1

END OF CONE OPENING

SPECIFIC RATE OF INCREASE

TIME (min)

FIG. 13. Time course record of oxygen consumption during cone opening.

was made by deducting the rat’s oxygen consumption during a rest period from its oxygen consumption during pine cone opening activity. The total net oxygen consumption of a rat engaged in opening a complete cone was calculated by multiplying the increase in oxygen consumption during opening by the time taken to open the cone. The rats were given an acclimation period of several hours a day for several weeks in the metabolic chamber until they became accustomed to opening cones in the experimental setup. We then assessed the oxygen consumption for each rat under three conditions: (1) at rest; (2) while opening a cone; and (3) followingcone opening. To ensure equal conditions for all the rats, they were given cones of standard size. We carried out measurements for both the shaving and spiral stripping techniques of cone opening. 2. Results Table 111 summarizes the comparison of cone opening methods. As can be seen, the spiral stripping method is demonstrably more efficient: (1) the average time taken to strip a standard pine cone spirally is only one third of that required to open a cone using the shaving method; (2) the rate of increase in oxygen consumption is significantly lower in rats using the spiral strippingmethod than in rats using the shaving method.

142

JOSEPH TERKEL

TABLE 111 DURATION OF CONEOPENING, INCREASEI N OXYGEN CONSUMPTION DURING OPENING, AND NET OXYGEN CONSUMPTION FOR SPIRAL AND SHAVING OPENING TECHNIQUES” Net oxygen consumption Style of opening Spiral method

Duration per cone (min) increase [ m l k x hr)l 46.2

Specific rate (ml)

* 7.3

1.35

Wb

(4) *+d

*C

Shaving method

118.9 f 40.9

1.49

(8) ~~

* 0.31

* 0.30 (4)

Total per cone 201.65

k

44.18

(4) ***e

476.89 f 53.10 (4)

~

From Yanai, Ar and Terkel (1991). Number of animals in parentheses. * p < 0.01, t test for independent measurements. ** p < 0.05, f test for repeated measurements. *** p < 0.01, t test for repeated measurements.

A rat using spiral stripping thus consumes less oxygen per unit time and body weight than it would require if it shaved the cones; (3) the data on specific rate of increase in oxygen consumption and the duration of opening provided us with the total differences in energy cost between the two methods of opening the cones. Thus a rat employing the spiral method is able to open the cone more quickly and expend less energy in doing so. Why the rat should choose one method over the other became even more obvious when we calculated the relative benefit of the two techniques of opening the cones, that is, the amount of energy gained from feeding on the seeds less the amount of energy expended in obtaining them. A rat using the spiral method to open the cones expends 9.5% of the energy that it obtains from the cone; on the other hand, a rat that employs the shaving method must expend 22.6% of the energy that it obtains from the cone. The spiral opening method is thus approximately 2.5 times more efficient than the shaving one and a rat using the spiral method will expend only 42% of the energy expended by a conspecific using the shaving method (Table IV). Despite the lesser efficiency of the shaving method, a rat that employs this method does gain some benefit, although not to the same degree as that of a rat using the more “professional” technique of spiral stripping. In nature, however, it is unlikely that rats can afford to waste energy in obtaining their food, and they thus use the spiral method whenever possible.

143

CULTURAL TRANSMISSION IN THE BLACK RAT

TABLE IV

ENERGYINVESTED

BY A RAT IN OPENING A PINE CONE BY SPIRAL A N D SHAVING TECHNIQUES I N RELATIONTO THE CONE’SENERGETIC VALUE*

Style of opening Spiral

Net energy expended/ cone (J) (M f SD) 3960

-t

Available energy/ cone ( J ) (M f SD)

751

Relative cost of opening 9.55

41442 f 143ob Shaving Ratio

9366

f 903

22.6 0.42

From Yanai, Ar, and Terkel (1991). After Aisner (1984).

The structure of the pine cone, as we have already mentioned, consists of rows of scales that overlap one another in a spiral arrangement around the central shaft. Two seeds lie hidden beneath each scale, except for the first few rows of seedless scales at the cone base. We have termed the seedless part of the cone the “sterile section.” When a rat begins to open the cones from the sterile section, therefore, using the spiral technique, it does not immediately come upon any seeds. It must first gnaw off the sterile rows of scales before it can reach the nutritious seeds that lie beneath the scales from the fifth row of the cone up. Although this sterile section of the cone accounts for only about one eighth of its total length, the black rat, as Aisner discovered, nonetheless invests 25% of the time required to open the entire cone on this small section alone, and without any reward. This relatively long period is due to the extremely tight structure and relative thickness of the base scales, so that the rat has to expend valuable energy gnawing most of the scale away rather than simply being able to prise it up as it can with the scales higher up on the cone. Using six measurements taken in the oxygen consumption study, and by monitoring the rat’s behavior during cone opening, Yanai was able to determine the precise point at which the animal finished stripping the sterile section and continued with the rest of the cone (Yanai et al., 1991). We were able to calculate and compare the rat’s rate of oxygen consumption while opening the sterile section of the cone to its oxygen consumption while stripping the rest of the cone by noting on the chart recorder the time at which the rat moved on from the sterile section to the seed-bearing part of the cone. We found that oxygen consumption was significantly higher when the rat was at work on the sterile section (1.441 0.297 ml O,/g/hr) in comparison to the oxygen requirement for opening the rest of the cone (1.253 ? 0.330 ml O,/g/hr).

*

144

JOSEPH TERKEL

The above finding was of particular interest because in the forests we also found pine cone shafts from which the black rats had started stripping the scales from above the sterile section only, leaving the lower four rows of seedless scales in place (Fig. 14). The rats that opened the cones in this way completely bypassed the sterile section and thus conserved both energy and 25% of the time that would otherwise have been spent in opening this nonrewarding part of the cone. It can be easily calculated that a rat opening the cone in this way saves 27.4% of the energy expended by a rat that begins at the base and opens the entire cone. OF Two TYPESOF CONESIN SYMPATRIC POPULATIONS VII. OPENING OF BLACKRATS

SUBPOPULATIONS OF BLACKRATSFEEDING ON CONIFEROUS SEEDS 1. Rationale and Method In areas of Israel in which coniferous forests have been planted, we observed two subpopulationsof black rats, distinguishable from the overall population of R. rattus by their specific diet preferences and food handling techniques. These two subpopulations are sympatric with the general population, possibly because the habitat contains a variety of tree species and other food sources in adjacent areas. One subpopulation exploits the pine seeds of the Jerusalem pine (Pinus r

A

B

FIG. 14. Two pine cones stripped by different blackrats. (A) All scales removed beginning at the base. (B)Scales stripped from only the portion above the sterile section.

CULTURAL TRANSMISSION IN THE BLACK RAT

145

halepensis), while the other exploits the seeds of the cypress (Cupressus serniperuirens). Like the pine forests, the cypresses were also planted 30-50 years ago to form windbreaks around the pines. We discovered that black rats were feeding on the cypress cones in much the same way that they feed on pine cones: under certain cypress trees adjacent to the pine forests we found the remains of gnawed cypress cones. We trapped rats on these trees and took them back to the laboratory where we presented them with cypress cones. The animals gnawed on these cones in a manner that left the cones identical to those found in the field. The technique of obtaining cypress cone seeds is relatively simple, because of the simple arrangement of the cone scales (Fig. 15). The rat gnaws away the sides of the mature cone until the small seeds are exposed and can be eaten. Because of the proximity of the cypress to the pine trees and the fact that these black rats are able to obtain and feed on conifer cones, the question then arose as to whether their ability is specific to a given species of tree or more generalized. We tested whether a rat trapped on a pine tree and able to open pine cones was capable of feeding on cypress cones, and vice versa. Rats of both subpopulations were trapped and we observed their cone opening behavior. The different motor patterns of individuals of both subpopulations were studied and we tested them for their ability to obtain nourishment from the two types of cones (Aisner and Terkel, 1991).

2 . Results All 17 of the rats trapped on pine trees were successful in stripping pine cones, but only five of them were able to obtain the seeds from cypress cones. All the rats that we had captured on cypress trees were able to get the seeds from the cypress cones, but only two of them were able to

FIG. IS. Cypress cones at various stages of ripeness; (a) least ripe (closed); (b) partially open, suitable source of seeds; (c) overripe, seeds already dispersing.

146

JOSEPH TERKEL

obtain some seeds from closed pine cones, while the rest (N = 10) completely failed to open the cones. Naive, nonforest rats trapped while feeding at human settlement garbage sites, were generally unable to open either pine or cypress cones. Only seven out of 58 such rats, were able to extract the seeds when provided with cypress cones and only two rats out of 49 obtained all the seeds when presented with closed pine cones, while five animals obtained some of the seeds. It would appear that prior manipulation of one variety of cone does not influence the rats’ ability to open the other, previously unencountered variety, and that the behavioral pattern employed in obtaining the familiar food is extremely specific. AND VIII. DISCUSSION

CONCLUSIONS

Research into socially transmitted feeding behavior has on the whole been approached from two separate angles: the performance of experimental studies in the laboratory to analyze the mechanisms involved in the particular behavior; and observations in the field of the new behaviors that are believed to be socially transmitted. However, many of the reported field observations on novel methods of seeking and handling food in the animal world are either anecdotal or conjectural. A comprehensive list of such studies has been compiled by Lefebvre and Palameta (1988). The laboratory studies, on the other hand, have often dealt with experimental aspects of behavior that do not always apply to the animal’s ecology or behavior in the field (e.g. Gardner and Engel, 1971; Bullock and Neuringer, 19771, although there have been some studies that have proved the exception (e.g., Krebs, MacRoberts, and Cullen, 1972; Palameta and Lefebvre, 1985; Sherry and Galef, 1984; Lefebvre, 1986). Our own studies on the social factors involved in the black rats’ acquisition of the pine cone stripping technique are significant in that they combine laboratory experiments with field observations. We have thus been able to isolate and test each separate factor for its possible involvement in the acquisition of this feeding behavior in nature. In the Israeli pine forests the acidic soil and extremely dense planting of trees have turned the habitat into an almost completely sterile monoculture. For the black rats, therefore, the pine cones, with their nutritious seeds, provide the sole source of nourishment. It is this exclusivity of available food that contributes to the uniqueness of our study. In order for the young rats born into this habitat to be able to survive, it is essential that they learn the necessary technique to open the cones competently and obtain the seeds, for no alternative source of food is available to them. After our experiments with the naive adult black rats had shown that

CULTURAL TRANSMISSION IN THE BLACK RAT

147

they cannot learn this new feeding behavior, we turned to studying how the behavior was acquired by the pups. When we had determined that young black rats raised by an experienced stripper mother (whether they were her own pups or fostered pups from a naive mother) were indeed learning to strip the cones in a highly competent manner, we knew that the cone opening technique was not being transmitted from one generation to the next by genetic transfer. The mechanism involved appeared to be a case of complex social transmission. Opening nuts, in contrast, is a far simpler task and it is known that squirrels, for example, can begin learning to open nuts by trial and error (Eibl-Eibesfeldt, 1956) and then arrive at an improved technique through observation (Wiegl and Hanson, 1980). The fact that a pup’s preferred choice of diet can be influenced by the flavor of its mother’s milk had been shown previously (Galef, 1977; Posadas-Andrews and Roper, 1983; Roper, 1986; Valsecchi and Galef, 1989; Valsecchi, Moles, and Mainardi, 1993). In the case of our black rat pups, however, we were able to show that their motivation to strip the pine cones and eat the seeds was not related to whether the mother’s diet had included pine seeds while they were still suckling. Another possible factor that can influence food preference is that of social mediation through the presence of adults at a feeding site, stimulating social feeding. Galef (1977) has shown that pups may even be attracted to a particular site by the mere presence of adult odors there. The presence of an experienced stripper mother can also facilitate the learning process. It has been noted by Ewer (1971) that new behaviors appear at the time when the weaning rat pups start to leave the nest and try solid food for the first time: such behaviors include licking and sniffing the mother’s mouth. In our own black rats, we observed that the young did indeed gather around their mother’s mouth while she was occupied with stripping the cone of its scales and extracting the exposed seeds. The pups attempted at this time, not always successfully, to snatch some seeds. As they grew, we witnessed their attempts to snatch partially stripped cones from the mother while she was actively opening them. The successful pups would keep the cone and continue the opening process on their own (Fig. 16). A similar behavior was described by Kemble (1984), in regard to the facilitation of learning cricket predation by northern grasshopper mouse pups and in wild mice (Valsecchi et al., 1993). The exposure to cones in various stages of being stripped was shown to be critical to the learning process of stripping cones for both pups and adults. Encountering cones with a gradually decreasing number of previously stripped rows of scales proved to be the crucial factor that facilitated the rat’s development of an efficient opening technique for dealing with an intact cone. None of our naive rats learned to open cones

148

JOSEPH TERKEL

FIG. 16. Rat pup feeding next to its mother on a partially opened cone, which it has stolen from her.

when given only intact ones, but about 70% of them did continue to open and completely strip cones when provided wit,h the right stimulus to do so, that is, cones in which the first rows had been partially opened. In addition, of the naive adult rats that reached the stage of continuing to strip already started cones (first four rows of scales removed), 90% of them were eventually able to strip completely intact cones after they had been led through the steps of dealing with cones with a consecutively decreasing number of previously removed rows of scales. Although the method we adopted to “teach” the rats to strip cones through gradually exposing them to cones with fewer and fewer stripped rows had proved to be highly successful, such a deliberate step-by-step introduction to the required feeding behavior does not occur in nature. In our experiment in which we simulated natural conditions more closely and provided the rat pups with cones at various stages of opening simultaneously, bypassing the step-by-step procedure, only 24% of them learned the opening technique. However, when the element of social influence was added (the presence of a stripping mother), the number of pups that learned to open the cones increased significantly to almost 100%. As the pups remain with the mother for a limited period of time only, it would seem that the learning process cannot be delayed much beyond weaning, as the maternal influence is the only chance that they have to acquire this complex feeding technique that is so vital to their survival in the pine forest. We have been carrying out this research for over 10 years now, and in all this time we have never come across partially opened cones discarded on the pine forest floor, other than the odd cone that had been accidentally

CULTURAL TRANSMISSION IN THE BLACK RAT

149

dropped to the ground by a rat in the process of stripping it. This fact lends credence to our assumption that the black rat pups acquire the pine cone stripping technique by snatching cones that the mother has already partly opened: if they did not do so, they would have no other opportunity to encounter started cones and thus would be unable to learn to open them. It is possible that for some pups the presence of the mother is not an absolute prerequisite for the learning process, as has been shown by those pups that did manage to learn to open cones, when provided with partially opened ones, without the mother’s presence. Nonetheless, the partially opened cones, encountered in the close proximity of the experienced stripper mother and “stolen” from her, provide the key factor in the pups’ learning process and enable this fascinating feeding behavior to be culturally transmitted from one generation to the next. The importance of the mother chimpanzee for her offspring’s acquisition of a new feeding technique, using hammer and anvil to crack nuts, has been observed and documented in the tropical rainforest of Tai National Park, Ivory Coast (Boesch and Boesch, 1984; Boesch, 1991). Because the technique of this nut-cracking behavior is similar in many respects to that of the pine cone opening technique in the black rat, I believe that the following comparison of the two cases might be instructive. According to Boesch (1991) the chimpanzee mother may contribute to the development of her offspring’s nut-cracking technique, for example, by providing the infant with a tool or facilitating its learning by leaving the hammer or nuts near the anvil; the infant is thus given its first chance to use the hammer in the “correct” context. Chimps without infants were never observed to leave hammers. In other cases the mother may also actively guide the young in the correct way to hold the hammer and use it to open the nut. Boesch’s observations fit the criteria of Car0 and Hauser (1992:152), thus putting this phenomenon within the category of teaching: An individual actor A can be said to teach if it modifies its behavior only in the presence of a naive observer, B, at some cost or at least without obtaining an immediate benefit for itself. A’s behavior thereby encourages or punishes B’s behavior, or provides B with experience, or sets an example for B. As a result, B acquires knowledge or learns a skill earlier in life or more rapidly or efficiently than it might otherwise do, or that it would not learn at all.

In black rats, the mother pays an energetic cost when she provides her offspring with partially open cones because, as we have already shown, the cost of stripping the first rows of tough, seedless scales is about 25% of the total energetic expenditure in opening the entire cone. When the mother supplies her young with the stimulus of partially opened cones, she therefore pays the energetic cost with no immediate benefit to herself.

150

JOSEPH TERKEL

However, her “tolerant” behavior in allowing her young to snatch the started cones from her does not appear to occur in other social interactions. The similarities in the acquisition of the feeding behavior in the two species would appear to put the development of the pine cone opening technique also within the teaching category. In spite of this comparison, however, it should be remembered that in the rain forest the hard-shelled nuts are probably only a supplement to a diet based on a variety of fruit. If the young chimp fails to learn the cracking technique, it can still survive on these other food resources. This is in strong contrast with our black rats, whose primary and possibly sole source of food in their pine forest habitat is the pine cones: if they do not acquire the stripping technique they will not survive. Despite the availability of vast numbers of cones, the black rat population in the pine forest has remained stable, and throughout our entire period of observation we did not note any significant rise in the rat population in any given area. We believe that this zero growth level stems from the limitation in the number of pups able to learn to open cones while still at the age when the mother is willing to strip the cones and supply her offspring with them. Our laboratory observations suggest that the pups, too, must invest some effort in obtaining the partially stripped cones from their mother. In the laboratory our black rat pups competed among themselves for a relatively limited number of partially opened cones that the mother was able to supply, and which were vital for the development of the stripping skill. It is therefore possible that in the wild only the strongest pups are able to obtain opened cones and it is only these pups that can acquire the opening technique, while the weaker pups, lacking the opportunity to practice on started cones and with no other source of food available, cannot survive. Another intriguing feeding behavior was originally reported by Fisher and Hinde (1949), who studied the spread of the milk-bottle opening phenomenon in tits. They explained it as a case of observational learning by naive birds watching birds experienced in milk-bottle opening (Hinde and Fisher, 1972). They believed both social and nonsocial factors to be involved in the spread of this behavior, but did not elucidate the precise mechanism. Later on Sherry and Galef (1984, 1990) were able to show that the provision of previously opened milk-bottle tops was sufficient to stimulate and establish the bottle opening behavior and that social factors are not necessarily required for its transmission. Our own study, however, differs from the above example, as can be seen from the significantly greater number of pups that were able to learn to strip the cones when exposed to both environmental and social

CULTURAL TRANSMISSION IN THE BLACK RAT

151

stimulation, in comparison with the pups that received only the nonsocial stimulus of partially stripped cones, without the interaction with the stripper mother. There are important ecological implications regarding the black rats’ flexible adaptation to a new feeding strategy, as shown through their ability to open the cones and extract the nutritious seeds, together with the cultural transmission of this vital behavior from one generation to the next. Cultural rather than genetic transmission of an acquired adaptive behavior is advantageous in that it permits the rapid spread of the behavior through the population, so that its members can more easily invade and exploit new environments. This is plainly demonstrated by the black rats, whose adaptive ability to open the pine cones, feed on the seeds, and transmit this behavior from one generation to the next, has enabled their successful invasion and adaptation to this new habitat, which has a potentially high capacity for a dense rat population.

IX. AFTERWORD Having got this far, the question, as in all studies of cultural transmission, is: How did the phenomenon get started in the first place? At the present point in time we still do not have an answer to this intriguing mystery. There have been many descriptive studies on the transmission of feeding behavior from generation to generation in other mammals. However, the methodology of our case study differs from others, since it is based on field observations combined with a complete parametric laboratory analysis. Together these two approaches have produced a seminal analysis of the mechanisms that permit this behavior to be perpetuated in the wild.

X. SUMMARY Black rats ( R a m s r a m s ) have taken over a new habitat in recent years, the Jerusalem pine forests. Their only source of food in these otherwise sterile forests is the pine seeds, which they obtain from the cones by a complex feeding technique. We studied the rats’ acquisition of this feeding behavior and its transmission from one generation to the next. Nalve adult black rats were unable to acquire the technique through trial and error or observation of experienced rats, but only gnawed at the cones randomly in an energy-wasteful manner. Pups, however, were able to learn the technique through observation. Flavor of the mother’s milk was eliminated

152

JOSEPH TERKEL

as a possible cue. Cross-fostering pups born to “naive mothers on experienced mothers and vice versa revealed that the pine cone stripping technique is socially rather than genetically transmitted. When we exposed naive adult rats to pine cones from which the rows of scales at the base of the cone had been previously stripped, the majority of the naive adult rats did learn to strip the cones, although the number was less than that of pups learning from their mothers. Of the two methods observed to be employed by the rats to obtain the seeds (stripping and shaving), stripping was the preferred method and shown to be more energy efficient. The young rats must acquire this feeding behavior while they are still in the vicinity of the mother as this is their only opportunity to learn the stripping technique from her; without this skill they could not survive in the pine forest habitat. The present study exemplifies the significance of the ontogeny of a specific feeding behavior in examining the energy budget and transference of knowledge from one generation to the next by means of cultural transmission. Learning the most efficient way to obtain food is essential and often critical for survival. It is possibly the most important factor involved in the widespread distribution of the species in a natural habitat.

Acknowledgments

I am grateful to Naomi Paz for preparing and editing the manuscript and to Amelia Terkel for reading and improving earlier versions; to Amikam Shoob for the photography and Walter Ferguson for the drawings. I sincerely thank my graduate students, Daphna Yanai, Ran Aisner. and Ofer Zohar, without whose dedication this work could not have been accomplished. The research was partially supported by the National Geographic Society Grant No. 3684-85.

References

Aisner, R. (1981). . . . of rats and pine cones. Israel-Land and Nature 6, 94-91. Aisner, R.(1984).Aspects of learning of pine cone’s opening by the black rat (Rattus raftus) and the spread of this behauiour in Israel. Unpublished master’s thesis. Tel Aviv University. Aisner, R.. and Terkel, J. (1985). Habitat exploitation through cultural transmission: Pine cone feeding behavior in black rats (Rattus rattus). Proc. 19th Int. Ethol. Conf., Toulouse, France. Aisner, R.,and Terkel, J. (1991). Sympatric black rats (Rattus rattus) populations. Different food handling techniques. Mammalia 55,301-3 10. Aisner, R., and Terkel, J. (1992a). Ontogeny of pine cone opening behavior in the black rat, Rattus rattus. Anim. Behau. 44,321-336.

CULTURAL TRANSMISSION IN THE BLACK RAT

153

Aisner, R., and Terkel, J. (1992b). Pine cone opening behavior in the black rat (Rattus ruttus):Ontogeny and mechanisms. In “The Rodent and its Environment” (M. le Berre and L. le Guelte, ed.). Chabaud, Paris (in press). Boesch, C. (1991). Teaching among wild chimpanzees. Anim. Behau. 41, 530-532. Boesch, C., and Boesch, H. (1984). Possible causes of sex differences in the use of natural hammers by wild chimpanzees. J . Hum. Euol. W, 415-440. Bullock, D., and Neuringer, A. (1977). Social learning by following: An analysis. J. Exp. Anal. Behau. 25, 103-117. Canby, T. Y. (1977). The rat-lapdog of the devil. Nut. Geog. 152,60-87. Caro, T. M., and Hauser, M. D. (1992). Is there teaching in nonhuman animals? Q. Rev. B i d . 67, 151-174. Eibl-Eibesfeldt, I. (1956). Uber Ontogenetische Entwiklung der Technik des Nusseoffnens von Eichhornche (Sciurus uulgaris L.) Zeit. Saugertier. 21, 132-134. Ewer, R. F. (1971). The biology and behavior of a free-livingpopulation of black rats (Ruttus rattus). Anim. Behau. Mono. 4, 127-174. Fisher, J., and Hinde, R. A. (1949). The opening of milk bottles by birds. British Birds 42, 347-357. Galef, Jr., B. G. (1977). Mechanisms for the social transmission of acquired food preferences from adult to weaning rats. In “Mechanisms in Food Selection” (L. M. Barker et al., eds.), pp. 123-150. Baylor University Press, Austin, Texas. Galef, Jr., B. G. (1982). Studies of social learning in norway rat: A brief review. Deu. Psychobiol. 15, 279-295. Galef, Jr., B. G., and Clark, N. N. (1972). Mother’s milk and adult presence: Two factors determining initial dietary selection by weanling rats. J. Comp. Physiol. Psychol. 78, 220-225. Galef, Jr., B. G., and Henderson, P. W. (1972). Mother’s milk: A determinant of feeding preferences of weaning pups. J. Comp. Physiol. Psychol. 78, 213-219. Galef, Jr., B. G., and Stein, M. (1985). Demonstrator influence on observer diet preference: Analyses of critical social interactions and olfactory signals. Anim. Learn. Behau. 13, 31-38. Galef, Jr., B. G., and Whiskin, E. E. (1992). Social transmission of information about multiflavored foods. Anim. Learn. Behau. 20, 56-62. Gardner, E. L., and Engel, D. R. (1971). Imitational and social facilitatory aspects of observational learning in the laboratory rat. Psychon. Sci 25, 5-6. Hinde, R. A,, and Fisher, J. (1972). Some comments on the republication of two papers on the opening of milk bottles by birds. In “Function and Evolution of Behavior” (P. H. Klopfer and J. P. Hailman, eds.), pp. 377-378. Addison-Wesley, Reading, Massachusetts. Kemble, D. E. (1984). Effects of preweaning predatory or consummatory experience and litter size on cricket predation in northern grasshopper mice (Onychomys leucogaster). Aggress. Behau. 10, 55-58. Krebs, J. R., MacRoberts, M. H., and Cullen, I. M. (1972). Flocking and feeding in the Great Tit (Parus major): An experimental study. Ibis 114, 507-530. Layne, J. N. (1954). The biology of the red squirrel Tamiasciurus hudsonicus loquax (Bangs) in central New York.Ecol. Mono. 24, 227-267. Le Magnen, J., and Tallon, S. (1968). Preference alimentaire du jeune rat induite par I’allaitement maternel. Societe de Biologie Seance du 24 Fevrier 1%8. Lefebvre, L. (1986). Cultural diffusion of a novel food-finding behavior in urban pigeons: An experimental field test. Ethology 71, 295-304.

154

JOSEPH TERKEL

Lefebvre, L.,and Palameta B. (1988). Mechanisms, ecology and population diffusion of socially-learned, food-findingbehavior in feral pigeons. In “Social Learning” (T. Zentall and B. G. Balef, Jr., eds.), pp. 141-164. Lawrence Erlbaum, New Jersey. Lore, R., and Flannelly, K. (1977). Rat societies. Sci. Amer. 56, 106-116. Palameta, B., and Lefebvre, L. (1985). The social transmission of a food-finding technique in pigeons: What is learned? Anim. Behav. 33,892-896. Posadas-Andrews, A., and Roper, T. J. (1983). Social transmission of food preferences in adult rats. Anim. Behav. 31, 265-271. Roper, T. J. (1986). Cultural evolution of feeding behavior in animals. Sci. Prog. 70,571-583. Rozin, P. (1976). The selection of food by rats, humans and other animals. In “Advances in the Study of Behavior” (J. Rosenblatt, R. A. Hinde, E. Shaw, and C. Beer. eds.), Vol. 6, pp. 21-76. Academic Press, New York. Sherry, D. F., and Galef, Jr., B. G. (1984). Cultural transmission without imitation: Milk bottle opening by birds. Anim. Behav. 32, 937-938. Sherry, D. F., and Galef, Jr., B. G. (1990). Social learning without imitation: More about the milk bottle opening by birds. Anim. Behav. 40, 987-989. Smith, C. C., and Balda. R. P. (1979). Competition among insects, birds and mammals for conifer seeds. Amer. Zool. 19, 1065-1083. Valsecchi, P., and Galef, Jr., B. G. (1989). Social influences on the food preferences of house mice (Musmusculus). I n f . J. Comp. Psych. 2, 245-256. Valsecchi, P., Moles, A., and Mainardi, M. (1993). Does mother’s diet affect food selection of weanling wild mice? Anim. Behav. 46,827-828. Weigl, P. D., and Hanson, E. F. (1980). Observational learning and the feeding behavior of the red squirrel (Tamiasciurus hudsonicus): The ontogeny of optimization. Ecology 61,213-218.

Yanai, D. (1989). A comparison of different modes of pine cone opening (Pinus halepensis) by black rafs (Rattus rattus). Unpublished master’s thesis, Zoology Department, Tel Aviv University, Tel Aviv, Israel. Yanai, D., Ar, A., and Terkel, J. (1991). Energetic efficiency of two methods of pine cone opening behavior by the black rat. VI Int. Coll. Ecol. Taxon. Small African Mammals. Mitzpe Ramon, Israel. Yanai, D., and Terkel, J. (1991). Aspects of the learning ability of wild and laboratory mice encountering partially opened pine cones. Ann. Meet. Zool. SOC.Israel, pp. 40-42. Zohar, 0. (1987). The influence of social and environmental factors on pine cone opening learning behavior of black rats. Unpublished master’s thesis, Zoology Department, Tel Aviv University, Tel Aviv, Israel. Zohar, O., and Terkel, J. (1991). Acquisition of pine cone stripping behavior in black rats (Rattus rattus). Int. J . Comp. Psycho/. 5 , 1-6.

ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 24

The Behavioral Diversity and Evolution of Guppy, Poecilia reticdata, Populations in Trinidad A. E. MAGURRAN SCHOOL OF BIOLOGICAL AND MEDICAL SCIENCES UNIVERSITY OF ST ANDREWS ST ANDREWS, UNITED KINGDOM

B. H. SEGHERS DEPARTMENT OF ZOOLOGY UNIVERSITY OF OXFORD OXFORD, UNITED KINGDOM

P. W. SHAWAND G. R. CARVALHO SCHOOL OF BIOLOGICAL SCIENCES UNIVERSITY OF WALES, SWANSEA SWANSEA, UNITED KINGDOM

I. INTRODUCTION

In June 1992 over 150 nations met in Rio de Janeiro to hold an “earth summit” on biological diversity. The convention that was signed there promised to support the development of strategies for the “conservation and sustainable use of biological diversity” and underlined the global concern about the rapid erosion of biodiversity. Indeed, it seems that many species will become extinct, as a result of human activity, before they have even been described by scientists (May, 1990a, 1990b; KapoorVijay and White, 1992). Our ignorance of biodiversity extends far beyond the absence of a complete catalog of species; there are still considerable gaps in our understanding of how diversity is generated through the process of speciation (Rice and Hostert, 1993) and how it is maintained in ecological communities (Schluter and Ricklefs, 1993). The resolution of these problems is a significant intellectual challenge in its own right. However, it is more than that because effective conservation strategies demand fundamental knowledge of the ecology and evolution of biodiversity. 155

Copyright Q 1995 by Academic Press, Inc. All rights of reproduction in any form reserved.

156

A.

E. MAGURRAN ET AL.

When Niko Tinbergen pioneered the study of ethology he observed that there are four types of behavioral question (Tinbergen, 1963). One of these concerns the functional basis of behavior and asks why particular behaviors have evolved. In answering a functional question the ethologist’s aim is to determine how a behavior contributes to an animal’s survival and ultimately to its reproductive success. The rapid growth and popularity of the discipline of behavioral ecology (Krebs and Davies, 1993) demonstrates the significance of the functional approach. Although considerable effort (and several journals) have been devoted to studying the evolution of behavior, rather few biologists have considered how behavior might, in turn, influence evolution. Yet it is becoming clear that behavior can play a role in generating and sustaining diversity. For example, mating behavior and sexual selection contribute to reproductive isolation (Endler, 1989). Lande (1981) has shown how the Fisher “runaway” process of sexual selection can lead to rapid speciation. In addition, populations of a single species can be differentiated according to their behavioral adaptations to varying predation regimes or habitats (Bell and Foster, 1994a,b). This differentiation may be the precursor of speciation. The biodiversity convention signed in Rio recognizes that biodiversity “includes diversity within species.” Questions about the evolution of diversity through behavior can potentially be addressed through investigations of many different groups of animals. However, increasing numbers of studies are documenting marked variation in the behavior of fish, both at the individual and the population level (Wilson, 1989; Magurran, 1993; Huntingford, Wright and Tierney, 1994). Natural populations of the guppy, Poecilia reticulata, in Trinidad, provide a particularly good model system for exploring the evolution of biodiversity (Endler, 1995). For example, we have recently been able to confirm that genetic and behavioral diversity in the guppy is widespread and that it reflects both stochastic and deterministic processes. Our work complements the findings of other investigators and demonstrates that, while colonization events, founder effects, and genetic drift contribute to population differentiation,natural selection (particularly selection exerted by predators) and sexual selection are also responsible for initiating and maintaining variation. Most intriguing of all, we have discovered that interactions between selection and the evolutionary history of a population can enhance diversity. Indeed, population differentiation in the Trinidadian guppy appears to represent a case of incipient speciation. Our aim in this review, then, is to explore the behavioral diversification of guppy populations in Trinidad from an evolutionary perspective. In doing this we hope to demonstrate how behavioral investigations can throw light on the evolution of biodiversity.

GUPPY POPULATIONS IN TRINIDAD

157

11. BACKGROUND

A. THEGUPPY:BIOLOGYAND DISTRIBUTION The guppy is one of the best known members of the family Poeciliidae (Rosen and Bailey, 1963; see also Parenti, 1981; Parenti and Rauchenberger, 1989),a remarkable group of fish. Like other poeciliids, the guppy has internal fertilization and gives birth to live young (Wourms, 1981). It is an ovoviviparous species (Thibault and Schultz, 1978), that is, one where the female provides no additional nourishment to her offspring once her eggs have been fertilized. In the wild, females typically produce broods of between two and eight young once a month (Reznick and Endler, 1982; Reznick and Miles, 1989). Parental care ceases at birth and newborn guppies can have well-developed antipredator responses (Magurran and Seghers, 1990a). For example, guppies from the Aripo River in Trinidad display, from birth, well-integrated schooling behavior, effective predator inspection behavior, and have the ability to execute complex antipredator maneuvers such as the flash expansion. This sophisticated antipredator behavior is present in guppies from both the Lower and the Upper Aripo populations even though the latter is traditionally classified as a low predation site (see, e.g., Magurran and Seghers, 1994a). It may be that the antipredator skills of newborn guppies have evolved to thwart cannibalistic attacks. Ovulation follows parturition and coincides with an increase in female receptivity to male courtship (Liley, 1966).Male guppies (and other poeciliids: Constanz, 1989) have a modified anal fin, the gonopodium, which they use as an intromittent organ. Female guppies can store sperm, and need only a single mating in order to produce broods over many months (Winge, 1937). There is considerable sexual dimorphism in the species. Males are not only the smaller sex (Liley and Seghers, 1975) but also the more brightly colored one (Haskins, Haskins, McLaughlin, and Hewitt, 1961). Male growth, unlike that of females, is determinate and virtually ceases at sexual maturity. Thus, in Trinidadian populations, mean male standard lengths are often in the range of 14-17 mm, while mean female standard lengths usually exceed 20 mm, sometimes by considerable margins (Liley and Seghers, 1975). Females have a cryptic, beige coloration, presumably as a defence against predation. The conspicuouscolor patterns of males, on the other hand, are implicated in sexual selection (Haskins and Haskins, 1950). Guppies are endemic to northeastern South America and occur in the coastal rivers of Venezuela, Guyana, Surinam, and several of the Lesser Antilles including Trinidad and Tobago (Rosen and Bailey, 1963). Al-

158

A. E. MAGURRAN

ET AL.

though the species was first collected and described by W. Peters in Venezuela in 1859, it was an amateur Trinidad naturalist of the same period, R. J. Lechmere Guppy, who bequeathed the fish its popular name (Winer and Boos, 1991). In recent decades the range of the guppy has been greatly extended due to human influences. Guppies are popular with aquarists and a whole industry has been devoted to generating fish with spectacular finnage and coloration. Inevitably these exotic fish have found their way into natural drainage systems around the world. In addition, the use of the guppy in mosquito control has facilitated the spread of this little fish across the globe (Courtney and Meffe, 1989). The species is no longer confined to tropical regions. There are reports of thriving populations of guppies in warmwater effluent in such unlikely places as St. Helens, Lancashire in England (Maitland and Campbell, 1992) and Moscow (Zhuikov, 1993).

B. TRINIDAD Despite its ever-increasing range it is in Trinidad that the guppy has been most thoroughly investigated. Trinidad is a small island (4,828 km2) lying a mere 16 km from the coast of Venezuela. Trinidad was joined to mainland South America during the Pleistocene glacial period (ca. 10,000 B.P.). Evidence from oceanic corals suggests that a land bridge may have persisted until 1,000 years ago (Kenny, 1989). Guppies are widely distributed across Trinidad and occur in all but the most marginal freshwater habitats (Boeseman, 1960). They can be found in small, clear oligotrophic headstreams, in the large, turbid eutrophic rivers of the flood plains, and even in the rainwater puddles that collect on the surface of Pitch Lake, a 36-ha deposit of natural asphalt. In some instances guppies are members of diverse fish communities where they can potentially interact with 20 or more species (Boeseman, 1960; Liley and Seghers, 1975; Magurran and Seghers, 1991). At the other extreme, guppies may be part of an impoverished fauna consisting of two or three species of fish. This discontinuity between the distribution of guppies and other species is particularly marked in the river systems that drain the Northern Range Mountains. In a number of cases the presence of barrier waterfalls on rivers has prevented the upstream migration of one or more guppy predators. As a consequence, guppies can be studied in habitats that differ in predation risk but are otherwise virtually identical in ecological terms. The unique potential of rivers in the Northern Range was first recognized in 1936 by the American biologist Caryl P. Haskins. Haskins subsequently uncovered the link between variation in a morphological trait (male coloration) and the presence of predators (Haskins and Haskins, 1951 ;Haskins et al., 1961).

GUPPY POPULATIONS IN TRINIDAD

159

C. INTRASPECIFICVARIATION Guppy populations in Trinidad have two key features that appeal to evolutionary biologists. First, populations vary markedly in a whole host of morphological traits. Second, this variation can, to a large extent, be attributed to a single ecological factor, predation pressure. The clear link between natural selection and adaptation provides an unrivaled opportunity for investigating evolution in the wild. In addition, under favorable conditions, generation time is remarkably short: the period between her own birth and the production of a female's first brood can be as little as 68 days (B. H. Seghers, unpublished data). As such, the consequences of natural selection can be examined not merely from a human perspective of time but also within the life-span of a single research grant. The ease with which guppies can be bred and observed in the laboratory is a further reason for their popularity. AS A DIRECT CONSEQUENCE OF RISK 111. ADAPTIVEVARIATION

A. PREDATION REGIMES The Northern Range is dissected by a large number of rivers almost all of which have been colonized by guppies (Fig. 1). Two major river systems drain the southern slopes of this mountain range. The Caroni River flows westward into the Gulf of Paria, while the Oropuche system drains into the Atlantic Ocean to the east of Trinidad. The fish communities in these rivers are typically South American. Cichlids (including Crenicichla alta and Aequidens pulcher) and characins (such as Hoplias malabaricus and ~

'F

CARIBBEAN SEA

N 10 bn

GULF of PARlA

~aroni

NORTHERN

TRINIDAD

FIG. I . Map of northern Trinidad showing the location of rivers mentioned in the text.

160

A. E. MAGURRAN ET AL.

Astyanux bimaculatus) predominate in the downstream portions but are often replaced upstream, or above barrier waterfalls, by the cyprinodontid Riuulus hartii. In the northern flowing rivers, by contrast, the fauna has a distinctly Antillean flavor (Boeseman, 1960). The lower sections, close to the Caribbean Sea, are dominated by gobies (includingEleotris pisonis and Gobiomorus dormitor) and/or mullet (Agonostomusmonticola). High densities of freshwater prawns (Macrobruchium spp.) are a characteristic feature of the upstream portions of these northerly flowing streams. Macrobrachium is usually found in association with Riuulus, and the small goby, Sycidium sp., may also be abundant. Although it is indisputable that predation levels vary from population to population, assessments of risk are almost invariably qualitative. Most investigators equate high risk with the presence of likely predators such as the pike cichlid Crenicichla alta. Conversely, low-risk sites are defined as those locations where guppies occur with less obviously predacious species such as Riuulus hartii. Such classifications take no account of the abundance of a particular predator at a site, nor of its actual foraging preferences. For example, although Crenicichla is often described as a voracious predator, and perhaps even a specialized guppy predator (Haskins et al., 1961), there is remarkably little information on its predatory behavior in the wild. Seghers (1973) undertook a diet analysis of a range of potential predators and found that over 50% of the Crenicichla examined had fish remains in their stomachs. Yet, his Crenicichla were also frequent consumers of mollusks and insects. It is known that Crenicichla is an efficient predator in the laboratory (Liley and Seghers, 1975; Mattingley and Butler, 1994) and it can be observed attacking guppies in the wild (Endler, 1987).We also know that guppies derived from Crenicichlu localities are more wary in their reactions to predators (see below). Despite this, the exact relationship between risk and the density and diversity of predators is unclear. Similarly, although Riuulus is conventionally described as a minor predator that targets immature guppies and presents little threat to adult fish (Haskins et al., 1961; Liley and Seghers, 1975; Seghers, 1978), knowledge of its predatory activities in the wild is decidedly scant. Recent work on Trinidadian Riuulus by Doug Fraser, Jim Gilliam, and their colleagues is, however, beginning to reveal more about its ecology and behavior (Fraser and Gilliam, 1992;Gilliam, Fraser, and Alkins-Koo, 1993).Since predators are credited with playing a pivotal role in the evolution of morphological and behavioral traits in the guppy, it is perhaps surprising that our understanding of the ecology of the predatory/prey system is still so incomplete. For the purposes of this chapter we divide the Trinidad predation regimes into two main categories: high-risk sites, which have a range of piscivores including Crenicichla, and low-risk sites where guppies are

GUPPY POPULATIONS IN TRINIDAD

161

found with Riuulus. In certain cases we will also discuss sites where freshwater prawns from the genus Macrobrachiurn are abundant. We recognize that this classification oversimplifies the complex mosaic of predation regimes to which Trinidadian guppies are exposed.

B. ANTIPREDATOR RESPONSES 1 . Schooling

Schoolingbehavior is a highly effective defense against a range of piscivorous predators (Shaw, 1970; Neil1 and Cullen, 1974; Magurran, 1990a; Pitcher and Parrish, 1993). Large groups of fish are more vigilant than are small ones (Magurran, Oulton, and Pitcher, 1985) and predators prefer to attack solitary individuals rather than schools. If attacked, the confusion effect decreases a predator’s likelihood of success, while the dilution effect diminishes the probability of any given individual becoming the victim. Group tactics, such as the fountain maneuver (where the school divides, streams past the predator, and reforms behind it) and the flash expansion (during which a compact school “explodes” away from the predator) are additional antipredator benefits of schooling. Seghers (1974) was the first person to explore the relationship between predation risk and schooling behavior. He collected fish from five Trinidadian populations, three from Riuulus sites and thus at relatively low risk, and the remaining two from high-risk Crenicichla localities. The guppies were bred and raised under standard conditions in the laboratory for 3-4 generations and had no experience of predator attack. Schooling tendency (measured using an index of cohesion) was greater in those populations that had originated from high-risk sites (Fig. 2). The fact that the behavioral differences persisted in laboratory stocks provides support for the idea that schooling tendency is inherited. Population variation in schooling behavior is particularly evident when fish are observed in the wild. Females from Crenicichla localities spend a large proportion of their time in cohesive schools (Fig. 3). It is important to note that a low schoolingtendency does not necessarily imply inadequate antipredator behavior. Guppies in the Paria population, located in an upstream section of the north-flowing Paria River, coexist not just with Riuulus but also with high densities of Macrobrachiurn crenulaturn. This freshwater prawn is a guppy predator (Chace and Hobbs 1969; Endler, 1978; Magurran and Seghers, 1990b). However, the prawn does not only rely on vision to locate its prey; it also hunts using olfactory and tactile cues. In such circumstances schooling confers no advantages. Indeed, schooling may well be a handicap because it will alert the predator (which is not susceptible to the confusion effect) to a rich food source.

162

A. E. MAGURRAN ET AL.

T

T

FIG. 2. Mean index of cohesion (+ 95% confidence limits) for five laboratory stocks of guppies. Groups of 10 fish, 5 of each sex, were placed in an aquarium (48 x I10 cm with water 3 cm deep). The tank floor was covered by a grid of 10 squares each measuring 528 cm2. Fish were allowed to explore the tank. Their schooling behavior was recorded after 5 hours. The index of cohesion was calculated by scoring the maximum density of fish (in a 1-min period) for any of the 10 squares. Scores were averaged over the 30 observations. The index has a theoretical minimum of 1 and a maximum of 10. Five populations were compared. There were 10 replicates for each of these five stocks. Black columns indicate those (high-risk) sites where the pike cichlid, Crenicichla, also occurs. Low-risk (from fish predators) populations are denoted by open columns. Redrawn from Seghers (1974).

2. Inspection Behavior When an attack is imminent, information about the identity and intentions of the assailant is a valuable asset. It is for this reason that guppies (Seghers, 1973; Dugatkin, 1988), and other fish including minnows, Phoxinus phoxinus (Magurran, 1986; Murphy and Pitcher, 1991;Pitcher, 1993), and three-spined sticklebacks, Gasterosteus aculeatus (Milinski, 1987), approach potential predators in order to inspect them (George, 1960;Licht, 1989). Several studies have shown that fish from high-predation localities are particularly wary during inspection; they form larger groups and maintain a greater distance between themselves and the predator (Magurran, 1990a,b; Huntingford et al., 1994). Guppies are no exception. Field observations reveal that fish from Crenicichla sites are more cautious when inspecting a realistic model of a fish predator (Fig. 4).

163

GUPPY POPULATIONS IN TRINIDAD

T A

.-E s

r

I

75-

CI

Y

.--0

en

C

50-

0

r 0

a

25-

FIG. 3. The percentage of time (+ 95% confidence limits) that female guppies from nine Trinidadian guppy populations devote to schooling in the wild. Black columns indicate those (high-risk) sites where the pike cichlid, Crenicichla, also occurs. See Magurran and Seghers (1994b) for details.

Inspection is inevitably a risky endeavor (Dugatkin and Godin, 1992). However, this risk can be reduced if fish avoid the dangerous “attack cone” around the jaws of the predator (George, 1960). Guppies from a Crenicichla population are reluctant to approach the mouth region of a fish predator, preferring instead to inspect the eye and body region behind the attack cone (Magurran and Seghers, 1990b).Such caution is abandoned when the fish are confronted by a novel predator, such as a Mucrobruchium prawn. In such circumstances the otherwise wary guppies court danger by approaching the prawn’s head and claws. By contrast, Paria guppies (which co-occur with Mucrobrachium) show clear attack cone avoidance when inspecting a prawn but swim close to the mouth of a fish predator (Magurran and Seghers, 1990b).

C. COURTSHIP A N D RISK Behavioral responses to predation are not merely confined to defensive tactics. Other activities, such as courtship, are profoundly influenced by risk. The ardor of male guppies is impressive. In certain populations,

164

A. E. MAGURRAN

ET AL.

0 8-

.-8cn

3cn C

.9 U

:

0 0

0

6-

0

4-

A A

2-

A 0

4

8

12

16

closest approach distance (cm)

FIG. 4. The relationship between the mean closest approach distance and the median group size of fish during inspection of a realistic predator model in the wild. The populations are the same as those shown in Fig. 3. High-risk (Crenicichla) sites are indicated by an open circle and low-risk (Riuulus) ones by a triangle. It is evident that the guppies in the Crenicichla populations reduce the risk of inspection by approaching in larger groups and maintaining a greater distance between themselves and the predator. See Magurran and Seghers (1994b) for details.

paradoxically those where predators are abundant, males spend at least half of their time pursuing females (Magurran and Seghers, 1994a). Male guppies have two main methods of achieving a successful copulation (Liley, 1966). They can display, using the characteristic sigmoid posture, which may induce a receptive female to mate with them. However, since females are sexually receptive only as virgins or for short periods following the birth of a brood, such advances are rarely successful (Liley, 1966). Alternatively, males may attempt a sneaky mating by thrusting their gonopodium toward the female’s genital pore. All males adopt both tactics, though there may be a tendency to specializein one or the other (Reynolds, Gross, and Coombs, 1993). In contrast to the typical poeciliid pattern, smaller males do not specialize in gonopodial thrusting (Constanz, 1975, 1989). Instead, as Reynolds et al. (1993) showed, high levels of thrusting are correlated with greater gonopodium length. It is not unusual for fish in the wild to perform at least one display and/ or thrust per minute (Luyten and Liley, 1985).Such behavior may increase a male’s vulnerability. After all, displays that have evolved to catch the

GUPPY POPULATIONS IN TRINIDAD

165

eye of a female are also likely to attract the attention of a predator. However, males in a dangerous habitat decrease their probability of attack by reducing conspicuous displays in favor of sneaky mating attempts in the presence of a predator (Magurran and Seghers, 1990c, and see Fig. 5 ) . Males are also less likely to display at high light levels when they may be particularly at risk of predation (Endler, 1987; Reynolds er al., 1993). Investigations by Liley (1966) and Luyten and Liley (1991) indicate that a mating between a displaying male and a receptive female carries a much greater chance of success than its sneaky counterpart. Responsivefemales do not mate randomly. Rather, they weigh up potential suitors before selecting the most appropriate partner.

D. MALECOLORATION A N D FEMALE CHOICE It is particularly fitting that one of the local names for the guppy in Trinidad is "seven colors." Vivid splashes of iridescent colors, such as blue and silver, and structural colors, such as red and black, cover the body (Endler, 1980). As we noted in the introduction, it is only the males that are brightly colored. During the sigmoid display a male's color markings are shown to their best advantage. Male coloration plays an important role in female choice (Endler, 1983; Kodric-Brown, 1985; Houde, 1987), although other characters such as body size may have an influence (Reynolds and Gross, 1992). Natural populations of guppies are so polymorphic that no two males

Lower Aripo

Upper Aripo

predator absent

FIG. 5. Courtship behavior of male Lower Aripo and Upper Aripo guppies in the absence and presence of predators. The Lower Aripo is a high-risk site (Crenicichla present), whereas the Upper Aripo is low risk. The pie diagrams show the proportion of sigmoid displays (black) and sneaky matings (white). See Magurran and Seghers (1990~)for details.

166

A. E. MAGURRAN ET AL.

are identical (Endler, 1983). However, intrapopulation polymorphism pales into insignificance when variation in color patterns across populations is considered. Male guppies in low-risk habitats, such as Riuulus sites, tend to be most flamboyant. As predation intensity rises, males become more cryptic and the number and size of color spots decrease (Endler, 1978). Females generally prefer brightly colored males (Endler, 1980, 1983; Kodric-Brown, 1985) and thus the population variation in coloration is largely due to the balance between female preference for and predator selection against conspicuousness. Yet, in certain cases this conflict has been sidestepped by use of a channel of communication not readily accessible to the predator (Endler, 1991a,b, 1992). For instance, in populations where there are high densities of the Macrobrachiurn prawn (such as Paria), females are particularly attracted to males with conspicuous orange coloration (Long and Houde, 1989; Houde and Endler, 1990). Endler (1978, 1991a, 1991b) has pointed out that crustaceans are insensitive to red. Males can thus signal their freedom from parasites (Houde and Torio, 1992) and/or their foraging capabilities (Kodric-Brown, 1989) to females without making themselves more obvious to the predator. It has also been suggested that females from high-risk (Crenicichla) populations favor more cryptically marked males (Breden and Stoner, 1987; Stoner and Breden, 1988, but see Endler, 1988). Male coloration and female choice in the guppy have been an important focus of investigation over recent years for John Endler and Anne Houde and others in this field. Indeed, Houde (1994) has recently found that selection for increased orange in males from the Paria population resulted in a shift in the mating preferences of females (see also Pomiankowski and Sheridan, 1994). E. LIFEHISTORYTACTICS Predation risk not only can explain variation in behavior and morphology; it can also account for population differences in life history tactics. Guppies that occur in Crenicichla sites increase their investment in reproduction relative to fish in other populations. For example, females mature at an earlier age and smaller size than their counterparts in Riuulus sites. In addition, females from the Crenicichla populations produce larger broods of smaller young at shorter intervals and devote proportionally more of their body weight to reproduction (Reznick, 1982; Reznick and Endler, 1982; Reznick, 1989). David Reznick and his colleagues have worked extensively on this topic (which is beyond the scope of this review).

GUPPY POPULATIONS IN TRINIDAD

167

IV. INDIRECT CONSEQUENCES OF PREDATION RISK A.

BEHAVIORAL TRADE-OFFS ARISING FROM IMPROVED ANTIPREDATORTACTICS

Predators exert influence that extends far beyond any direct encounters with their prey. Effective predator defense, for instance, has its own attendant costs. As in human society, there is a social price to pay for investment in the arms race. Schooling is a highly effective means of countering predators (see above). Yet schooling, by its very nature, relies on uniformity. Any individuals that appear or behave differently are particularly vulnerable (Ohguchi, 1981; Theodarkis, 1989). The necessity for coordinated behavior (Pitcher and Parrish, 1993) also limits scope for resource defense. Thus we might predict that guppies in high-risk environments are less likely to defend food patches or compete for mates. A laboratory investigation of feeding aggression revealed that food patches are guarded less assiduously by fish from sites with many predators, and high levels of schooling (Magurran and Seghers, 1991). The aggression consists of nipping and chasing, circling, and tail beating. At times fish defend food patches against all competitors and chase away any individual who attempts to consume a morsel of food. A time-budget investigation of female guppies in seven wild populations confirmed that these behaviors occur in the natural environment as well as in the laboratory. In certain populations, such as the low-risk Upper Tunapuna, large females guard algal patches and viciously repel any competitors. Once again, there is an inverse correlation between schooling and aggression (see Fig. 6). These results (and see also Huntingford, 1982) point toward a trade-off between antipredator behavior and resource defense and illustrate how selection for defensive skills has costs as well as benefits. B. BIASEDSEX RATIOS

Haskins et al. (1961) drew attention to the fact that certain wild populations of guppies can be female biased. Seghers (1973) confirmed this observation in an extensive survey of populations in Trinidad, and related variation in sex ratio to predation risk. He found that Riuulus habitats have an excess of females. This contrasts with Crenicichla sites where the sexes tend to be equally represented (Fig. 7). Rodd (1994) similarly noted that sex ratios are more female biased in Riuulus localities than in Crenicichla ones. Why might this be? There is no evidence for differential mortality of the sexes during gestation (Farr, 1981). Nor can the biased sex ratios be attributed to an overproduction of female offspring. When

168

A.

E. MAGURRAN ET AL.

1

2

3

4

5

6

7

schooling low

high

FIG. 6. The relationship between schooling and aggression in the wild. The populations are ranked from one to seven, with seven representing the greatest degree of schooling or aggression. All populations in Fig. 3 (with the exception of L. Tacarigua and Paria) are shown. High-risk (Crenicichla) populations are indicated by an open circle and low-risk (Riuulus)ones by a triangle. Aggression was measured by recording the frequency of aggressive acts that females either delivered or received over a 3-min period. See Fig. 3 for details of schooling behavior.

guppies from female-biasedpopulations are bred in the laboratory, the sex ratio of the young approaches unity (Seghers, 1973). The most appealing explanation invokes differential predation on males (Liley and Seghers, 1975).We know that males have a lower schooling tendency than females (Magurran, Seghers, Carvalho, and Shaw, 1992),employ fewer antipredator tactics in the wild (Magurran and Seghers, 1994a), and are less adept at evading capture when attacked (Seghers, 1973).Conspicuous coloration and courtship behavior seem likely to magnify male vulnerability further. Paradoxically, male and female abundances are most closely matched in the high-risk Crenicichlu sites, so heightened male susceptibility to predation cannot be the sole cause of the population variation in sex ratio. The observed differences in sex ratio must arise through greater vulnerability of males in Rivulus habitats. It could be that males, being smaller, are particularly at risk of capture from a size-limited predator such as Rivulus. As we noted earlier, Rivulus is indeed known to consume

GUPPY POPULATIONS IN TRINIDAD

T 1

50-

40

-

169

0

T 1

L\

30

4n L"

-

I

I

low risk

high risk

FIG. 7. Males as a percentage of adult guppies (+ 95% confidence limits) in nine lowrisk (Riuulus)and ten high-risk (Crenicichla)populationsin Trinidad.The data were collected in 1967. The figure excludes sites such as Yarra and Paria where Macrobrachiurn prawns are abundant. Data from Seghers (1973).

small guppies (Seghers, 1978). However, the species is not a specialized piscivore and guppies of both sexes from Riuulus habitats have poorly developed antipredator responses. Differential risk could nonetheless interact with other factors to accelerate male mortality in such apparently low-risk localities. Riuulus habitats are less productive than the downstream Crenicichla sites (D. N. Reznick, personal communication). Males devote a great deal of energy and time to courtship and spend relatively short periods foraging (Magurran and Seghers, 1994a). Thus, when food is in short supply (e.g., during the wet season; Winemiller, 1993),males will have fewer reserves to draw on and could be more likely to succumb to starvation, disease, or predation. Thus, although Riuulus might not be a specialized guppy predator it could pick off fish that are diseased or in poor condition. Male guppies may also be prone to a variety of physiological stresses (Snelson, 1989). It has been shown that males of another poeciliid species, the mosquitofish, Garnbusiu afinis, are more susceptible than females to extremes of temperature, shortage of food, and overcrowding (Krumholz, 1948). In fact, as Snelson (1989) points out, wild populations of many different poeciliid species have female-biased sex ratios.

170

A. E. MAGURRAN ET AL.

C. SEXUAL HARASSMENT Biased sex ratios, combined with a differential perception of risk by males and females, shift the balance of power in the battle of the sexes (Dawkins, 1989). The goal of both sexes is to maximize reproductive success. However, the routes that males and females follow in order to achieve this objective are quite different. A female’s fecundity is dictated by her body size (Reznick, Bryga, and Endler, 1990), which is in turn a product of her foraging ability and age. By avoiding predation, and mating with a suitable male, her reproductive output is ensured. Because females are only occasionally receptive to male advances, it is not surprising that their time budgets are largely devoted to foraging. A male’s priorities differ. Because of female choice (Houde, 1987, 1988a,b), and possibly also as a result of male-male competition (Kodric-Brown, 1992), mating opportunities are distributed unequally among males. It is therefore inevitable that some individuals will father few or no offspring irrespective of their longevity. Males must maximize their mating opportunities in order to ensure fitness. It is for this reason that they engage in continuous sexual activity (Luyten and Liley, 1985)and that females are subjected to frequent sneaky mating attempts. This asymmetry can also explain the male guppy’s apparent disregard of predation risk. Male disregard of predation risk could alternatively be attributed to the size-selective hunting behavior of guppy predators. It is often assumed that Crenicichlu prey predominantly on large, mature guppies (see, e.g., Reznick and Endler, 1982). Indeed, size-selective predation has long been invoked as the driving force behind life history evolution in Trinidadian guppy populations. If Crenicichlu do specialize on large fish, then females should be especially vulnerable to predation. Males, of course, being smaller, could then afford to take more risks. However, a recent experimental study of Crenicichlu foraging (Mattingley and Butler, 1994) cast doubt on the earlier certainties. Mattingley and Butler found that although larger Crenicichlu consume larger prey, small prey form a substantial part of the diets of predators of all sizes. It thus seems unlikely that small size protects males from predation. Sexual harassment can be costly for a female. The persistent attention may force her to leave a profitable foraging site (Magurran and Seghers 1994~).Furthermore, sneaky matings can thwart her choice of a particular male (Luyten and Liley, 1991). Females in high-risk Crenicichlu sites are exposed to the greatest degree of harassment and suffer approximately one sneaky mating attempt per minute (Fig. 8; Magurran and Seghers, 1994a). This is partly due to the higher incidence of males in such habitats but is also a consequence of the male’s exploitation of female antipredator

GUPPY POPULATIONS IN TRINIDAD

=

171

I

FIG. 8. The mean (+ 95% confidence limits) frequency of sneaky matings received by (n =30) females per minute in Trinidadian guppy populations in the wild. Black columns indicate those (high-risk) sites where the pike cichlid, Crenicichlu, also occurs. Low-risk (from fish predators) populations are denoted by open columns.

behavior. When females detect predators they inspect them. Males swim behind the inspecting females, and take advantage of their preoccupation with the predator in order to launch a sneaky mating attempt (Magurran and Nowak, 1991). An alternative viewpoint sees male harassment not as a cost, but as a genetic benefit to females. One possibility (M. Milinski, personal communication) is that females receive the good genes of males who are able to successfully mate and simultaneously approach a predator. Other advantages might accrue to females who “accept” the matings of the most vigorous males (Farr, 1989). Beardmore and Shami (1979), for example, discovered a link between heterozygosity and fecundity in guppies. Farr (1989) contends that in most species of poeciliids female choice in itself is unlikely to maximize offspring heterozygosity. As there is little point in choosing, a female might just as well accept the outcome of male-male competition. Yet we know that female choice does occur in guppies from high-predation locations (Houde, 1988b) and that the offspring fathered by such means benefit as a result, for example, from faster growth rates (Reynolds and Gross, 1992). The costs, and evolutionary consequences,

172

A. E. MAGURRAN ET AL.

of sexual harassment clearly need to be resolved (see also Smuts and Smuts, 1993). There is evidence that male sexual vigor in poeciliids is linked to high genetic heterozygosity (Farr, 1983, 1989). This observation could explain why male mating behavior is especially frenetic in Crenicichlu sites, which are also the most heterozygous populations of guppies (Shaw, Carvalho, Magurran, and Seghers, 1994) (see Section VI,C below).

V. EVOLUTION IN THE WILD The dovetailing of morphological variation and predation risk provides compelling evidence for evolution in the wild. Yet, although persuasive, observations based solely on the geographical association of predator communities and prey populations are of limited value because they depend on correlation. They offer no direct test of the hypothesis that a shift in selection regime results in an adaptive change in the trait in question. This shortcoming has prompted biologists to seek experimental support for their conclusions. For example, Endler (1980) undertook a series of experiments in which he monitored shifts in male body coloration in greenhouse ponds with different predatory communities. However, some of the most convincing support for predator-induced change has been provided by transplants of wild populations. Although guppies are widely distributed in Trinidad, it is still possible to find streams where the species is absent. These are usually, but not invariably, remote locations with few potential predators. Guppies collected from high-risk sites and introduced into such safer habitats provide an indication of what might happen when predation pressure is relaxed. Conversely, if predators are placed in habitats that they have hitherto been unable to colonize, guppies will be forced to adapt to an increase in predation risk. Though they are powerful tests of evolutionary theory such transplants bring their own ethical dilemmas. Trinidadian river systems are already threatened by environmental degradation (Seghers, 1992) and the number of pristine sites diminishes at a depressing rate. Moreover, once a hardy and fecund species such as the guppy is introduced into a site it is almost certainly there in perpetuity. The dangers, and value, of such transplants are illustrated graphically in the next example.

INTRODUCTION A. HASKINS’S As in so many other areas of guppy biology in Trinidad, the transplant experiment was pioneered by C. P. Haskins. In 1957, Haskins collected

GUPPY POPULATIONS IN TRINIDAD

173

200 guppies from a river (probably the Arima) in the Caroni drainage and moved them to previously guppy-free and (with the exception of Riuulus hartii)predator-free headwater of the Turure River in the Oropuche drainage system. Prior to their collection, the transplanted guppies were exposed to predation from Crenicichla and other piscivores. Descendants of the introduced guppies soon colonized the upper portion of the Turure River. The experiment was never published and might have remained undetected. However, by good fortune, our investigation of the genetic differentiation of Trinidadian guppy populations included a sample from the Lower Turure (Shaw, Carvalho, Magurran, and Seghers, 1991). This sample was initially puzzling because it did not appear to belong to the Oropuche system. In fact, it bore all the hallmarks of a Caroni population (guppies from the two drainages show marked genetic divergence [Carvalho, Shaw, Magurran, and Seghers, 1991; Fajen and Breden, 19921 and see below for a fuller discussion). Additional samples and correspondence with C. P. Haskins led us to conclude that the Upper Turure fish were descendants of the introduced guppies. From the genetic evidence we were also able to deduce that the transplanted stock had moved downstream below the barrier waterfall and had replaced the indigenous guppy population in the Lower Turure River (Shaw, Carvalho, Seghers, and Magurran, 1992). In addition to being an intriguing detective puzzle, the Turure introduction offered a unique opportunity of linking long-term changes in predation risk with behavioral evolution. As noted above, the guppies were moved from a risky habitat to a safe one. Indeed, as Riuulus is scarce in the Upper Turure, the site must be one of the least threatening (from a guppy’s perspective) in Trinidad. We know that predator defense can be costly (Magurran and Seghers, 1991; Dugatkin and Godin, 1992). Did the behavior of the Upper Turure guppies thus evolve following the reduction of risk? In order to answer this question fish were collected from the Upper and Lower Turure sites and from two control sites (Upper and Lower Aripo) in the Caroni drainage. These fish were housed and bred in the laboratory and their offspring raised with no experience of predation. When schooling and inspection behavior were measured it emerged that the behavior of the Turure guppies varied with risk. Thus, the Upper Turure fish (just like the Upper Aripo ones) had a reduced schooling tendency and moved close to a predator during inspection, while the Lower Turure (and Lower Aripo) guppies spent much time schooling and were cautious when inspecting (see Fig. 9 and Magurran et al., 1992). The 34 years between the transplant and our 1991 investigation represents at least 100 generations. Is there any evidence that behavior can evolve more swiftly? We turned to a more recent transplant in an attempt to answer the question.

A. E. MAGURRAN

ET AL.

B 26 A

-E

a 20 0 C

(D

c

cn

16

c 0

:10 P 0 U

8

-

6

0 0

0

i T -

.-a8

A

-I-

&

't

a

3

FIG. 9. The mean (+ 95% confidence limits) time spent schooling, out of a total of 300 s (A) and the closest approach to a realistic predator model during inspection (B) by female guppies in a laboratory test. All fish had been bred and raised in controlled conditions in the laboratory. Open columns denote those populations where the current predation regime is low. The Upper Turure females are descendants of fish transplanted from a highrisk Caroni site in 1957. See Magurran et al. (1992) for details.

B. ARIPOINTRODUCTION Endler, like Haskins before him, decided that an introduction was the most effective way of manipulating the evolution of male coloration. In 1976 he (Endler, 1980; Reznick et al., 1990) collected 200 guppies from the Lower Aripo (a high-risk site) and placed them in an upstream tributary of the same river. Only Riuulus previously inhabited this site; a waterfall had proved an insurmountable barrier to guppies and other fish species. Once again the transplant was successful and the descendants of the introduced fish flourished. The change in male coloration was remarkable. Endler (1980) monitored guppies in the introduction site over the next two years and observed that male descendants of the transplants became more conspicuous as the sizes and frequencies of their color spots increased. Changes in the introduced fish were not confined to male color patterns. Marked evolution of life history traits was also observed. Phenotypic shifts in offspring size

GUPPY POPULATIONS IN TRINIDAD

175

and the proportion of female body mass devoted to reproduction were apparent within two years of the transplant (Reznick and Endler, 1982). The heritability of such differences was confirmed two years later when guppies were collected from the wild and held in a common environment in the laboratory for two generations (Reznick and Bryga, 1987). The populations continued to diverge over the 11 years that the transplant was monitored (Reznick et ul., 1990). Field analyses of behavior in 1992 suggested that it, too, had evolved in the 16 years following the introduction. Females from the introduction site spent only a small fraction of their time schooling, just like their sisters in the low-risk Upper Aripo site (see Fig. 10). The descendants of the introduced fish also showed little caution when inspecting a model of a Crenicichlu. Thus, in terms of antipredator behavior observed in the wild, the guppies in the introduction site had diverged from the ancestral Lower Aripo stock and had come to resemble fish from a low-risk habitat. However, when fish were collected from the wild, and their offspring were raised and studied in laboratory conditions, a rather different picture began to emerge. There were highly significant differences in the schooling

Upper Aripo Rivulus

Aripo introduction site 200 guppies transplanted from Lower Aripo in 1976

W-

\ Middle Arlpo Crenicichfe introduced 1980 banier

I e-

Lower Aripo Crenicicha, characins etc

FIG. 10. Summary of recent transplants of fish in the Aripo system showing behavior of the wild guppies. The black segments of the pie diagrams represent the percentage of time that females devote to schooling. Mean closest approach distance to a realistic model predator during inspection is indicated by the arrowed lines. The greatest (mean) approach distance of 10 cm was observed for the Lower Aripo fish. Other approach distances are drawn to scale. See Magurran and Seghers (1994b) for details.

I

176

A. E. MAGURRAN

ET AL.

behavior of guppies from the three populations (one-way ANOVA: females F2.50 = 15.45, p < 0.001; males F2.50 = 6.98, p = 0.002; Fig. ll). (These data have not been reported elsewhere, so details of the statistical analysis are provided.) For both sexes, schooling tendencies differed significantly between the Aripo (I) and Upper Aripo guppies ( p < 0.05, Scheffd F test), but not between the Aripo (I) and Lower Aripo fish. During inspection (Fig. l l ) , the closest distance between fish and the predator model also varied significantly among the three populations (females F2,40 = 15.15, p < 0.001; males F2,39= 7.07, p < 0.002). Once again, for both males and females, there were significant differences between the Aripo (I) and Upper Aripo populations ( p < 0.05). The observation that behavioral change has lagged behind the evolution of morphological and life history patterns in the Aripo (I) site is puzzling. At least three (not necessarily mutually exclusive) explanations can be advanced to resolve the paradox. First of all, it might simply be that insufficient time has elapsed in order for the behavior to evolve. We know

B

-t a e

25

1

20

0

4

15

c 0

g

10

0

-Q %

5

0

0

8

8. K a a

FIG. 11. The mean (+ 95% confidence limits) time spent schooling out of a total of 300 s (A) and the closest approach to a realistic predator model during inspection (B) by guppies in a laboratory test. All fish had been bred and raised in controlled conditions in the laboratory. Females are denoted by open columns, males by shaded ones. Aripo (I) guppies are the descendants of fish transplanted from the Lower Aripo in 1976. The sample sizes (number of different individuals) in the tests were as follows: female schooling, Lower Aripo 24, Upper Anpo 15, Aripo (I) 12; female inspection. Lower Aripo 15, Upper Aripo 14, Anpo (I) 12; male schooling, Lower Aripo 24, Upper Aripo 15, Aripo (I) 12; male inspection, Lower Aripo 14, Upper Aripo 14, Aripo (I) 12. Behaviors were measured by the methods described in Magurran et al. (1992).

GUPPY POPULATIONS IN TRINIDAD

177

that behavioral changes have occurred in the guppies transplanted from a high-risk site to the Upper Turure River. This would suggest that it takes more than about 30-40 generations for a behavioral shift to become apparent in these fish. Studies of other fish have nonetheless indicated that heritable behavioral change can be rapid. For instance, Ruzzante and Doyle (1991, 1993) found that the schooling and agonistic behavior of medaka (Orytius lutipes) shifted after two generations of selection on growth. Bakker (1985, 1986) obtained a significant change in the aggressiveness of juvenile sticklebacks, Gusterosteus aculeatus, after only a single generation of selection. A second, related explanation is that antipredator behavior is phenotypically plastic. Guppies in the Aripo introduction site are clearly flexible in their behavior; their antipredator response is contingent upon the habitat in which they are raised. Phenotypic plasticity can be advantageous in a variable environment (West-Eberhard, 1989; Thompson, 1991;Via, 1993). After all, there is little point in developing sophisticated predator evasion tactics if early experience suggests that they are unlikely to be needed. Once a flexible response to predation risk has been established in a population, behavior could be resistant to further rapid evolution. This is particularly likely to be the case when predation has been relaxed. Although there are costs to antipredator behavior (Curio, 1993), these would be expected to be small relative to the degree of selection imposed by an increase in predation threat. Several studies indicate that antipredator responses are retained in populations no longer at risk from predation. Arctic ground squirrels, Spermophilus spp., diverged from Californian populations (now in the subgenus Otospermophilus) during the Miocene some 5 million years ago. Coss and Owings (1985)estimate that the Arctic ground squirrels have not been exposed to snake predation for at least 3 million years. Although less competent at avoiding snakes, the Arctic ground squirrels have retained the same motor patterns as their snakesympatric Californian cousins. European minnows, Phoxinus phoxinus, from a population isolated from pike (Esox lucius) for several thousand years likewise adopt the same evasive maneuvers as minnows under continuous heavy predation from pike (Magurran and Pitcher, 1987; Magurran, 1990b). The behavioral differences between the two minnow populations are quantitative rather than qualitative: fish from the low-risk site integrate their antipredator tactics less effectively. Behavioral differences in the antipredator responses of stickleback populations follow a similar pattern (Giles and Huntingford, 1984; Huntingford et al., 1994). Unfortunately for this argument, phenotypic plasticity is also evident in the life histories of guppies (Reznick and Bryga, 1987) and other poeciliids (Trexler, 1989a, 1989b, 1989c) and, as we know, life histories did evolve soon after the transplant.

178

A. E. MAGURRAN ET AL.

Finally, there may have been insufficient genetic variance for behavior relative to the other traits already assessed. Although we know that 200 fish were introduced into the Aripo tributary in 1976, the effective population size could have been much lower than that (because of high mortality) or indeed much greater as a consequence of sperm storage and multiply inseminated broods. Subsequent analysis of allozyme frequencies in the introduced population and the ancestral population below Haskins’s Fails (Aripo (I) and Aripo 6, respectively, in Fig. 12) has shown there to be almost no differences (P.Shaw, unpublished data). This suggests that very little genetic variation was lost during the founding of the new population, probably due to rapid initial increase in population size. There are fewer opportunities to investigate an increase in predation risk. However, Crenicichla was absent from the Middle Aripo until 1980 when it was introduced by D. N. Reznick (personal communication). Although a small breeding population of Crenicichla has persisted, we found only low densities of this predator when we visited the site in April 1992. Our field investigations show that the schooling behavior and inspection behavior of guppies in the Middle Aripo fall between the levels observed in the Upper and Lower Aripo (Fig. 10). Some behavioral modiGENETIC IDENTITY 0.90 0.91 I

0.92 0.93 0.94 0.95 0.96 0.97 0.98 0.99 1.00

I

I

I

7 RIVERS (31

1

U.TURURE

h

UPLAND

I

ARlMA

I

LOWLAND RIVERS (4) PRIORITY BUSROUTE ARlPO I 0

2

ARlPO 6 U. ARlPO

ri I I

I

I PARlA

I OROPUCHE DRAINAGE

KENNY’S POND YARRA

FIG. 12. Dendrogram of mean genetic identities (Nei 1972) between samples of guppies from N. Trinidad. Calculations (UPGMA)are based on frequencies of 37 alleles across 25 gene loci. Sample locations are shown in Fig. 1, or described in the text.

GUPPY POPULATIONS IN TRINIDAD

179

fication has therefore occurred, though we cannot confirm at this stage whether the changes are heritable. VI. GENETIC DIVERGENCE OF POPULATIONS A. THEPROCESS OF GENETIC DIFFERENTIATION Guppies, in common with most freshwater fishes, display a discontinuous distribution, determined by interactions between historical processes (geography and colonization), the availability of suitable habitats, and various abiotic and biotic factors. Populations are therefore isolated to varying degrees and, depending on the exchange of migrants and influence of deterministic forces (selection) and stochastic forces (drift and founder effects), will display a corresponding range of genetic differentiation. Anadromous fishes tend to exhibit the most marked genetic divergence among fish species (Gyllensten, 1985; Ward, Woodwark, and Skibinski, 1994), though even those without natal homing may still diverge, depending on local factors such as mating patterns, effective population size, and selective regimes (Carvalho, 1993). Molecular markers provided by allozymes and mitochondria1 and nuclear DNA sequence variation (Hoelzel, 1992) provide a universally comparative approach to quantify genetic divergence across taxa. Molecular genetic data can provide information on the levels and distribution of genetic variability within and between populations, influence of mating patterns, relative roles of microevolutionary forces, and patterns of colonization and time since divergence. Genetic data are most usefully interpreted in conjunction with background information on the environment and biology of the species in question. A comprehensive knowledge has accumulated in recent years for P. reticulata populations in Trinidad (e.g., Seghers, 1974; Endler, 1978; Carvalho et al., 1991; Fajen and Breden, 1992; Magurran et al., 1993; Shaw et al., 1994), and this now provides a model system for examiningthe interplay of evolutionary forces underlying population differentiation. In addition, P. reticulata is rare in occupying a wide range of contrasting habitats differing particularly in degree of physical isolation and intensity and nature of predation. Such broad distribution facilitates studies on the relationship between environmental variation and genetic population structure.

B. PATTERNS OF GENETIC DIVERGENCE Extensive allozyme data are available on the distribution of genetic variability in Trinidadian guppies, especially among rivers from the North-

180

A. E. MAGURRAN ET AL.

ern Range (Carvalho et al., 1991; Shaw et al., 1991; Shaw et al., 1992; Shaw et al., 1994). Underlying marked population variability in courtship behavior, antipredator behavior, and morphology (Endler, 1980) is marked genetic differentiation (Carvalho et al., 1991; Shaw et al., 1992), both between rivers in the same drainages, and between drainage basins (Fig. 12). Several mean genetic identities (I), a measure of genetic similarity based on a comparison of allele frequencies from different loci (Nei, 1972), fell below that normally encountered for conspecificpopulations (I > 0.90; Thorpe 1983), with occasional fixed allelic differences, indicative of reproductive isolation. Of six polymorphic loci, four showed major differences between populations from different river basins, while two showed differentiation within and between different drainages. The proportion of genetic diversity attributable to differences among populations, GST(Nei, 1973), is indeed one of the highest disclosed among freshwater fishes (Ward et al., 1994), providing further evidence for the action of isolating and/or diversifying evolutionary forces. The extent of genetic divergence fitted the predicted continuum based on degree of physical isolation: highest genetic similaritieswere observed between samples from the same river (with no obvious barrier to dispersal), followed by subpopulations separated by only a waterfall barrier, to finally the most distinct populations deriving from separate drainage basins. The genetic distinctness of two major drainages, the Oropuche and Caroni systems, was particularly marked. The pattern of allozyme divergence among populations was supported by an independent study using polymerase chain reaction (PCR)-based analysis of sequence variation in the control region of the mitochondria1 genome (Fajen and Breden, 1992). The level of sequence variation among populations was high: 30 variable sites were detected within a 363-basepair (bp) section of the control region (excluding the proline tRNA gene). This differentiation contrasts, for example, with that observed in 363 bp of the same area of the genome in African cichlids in Lake Victoria (Meyer, Kocher, Basasibwaki, and Wilson, 1990), where only 15 variable sites were found among 14 species. Whereas approximately 4% sequence divergence was detected between two distinct lineages of haplochromine species, comparable levels were found in P.reticufata among samples taken from geographically proximate river basins. Such findings illustrate the scale of the genetic divergence in Trinidadian guppies, and indicate extremely effective geographical isolation and/or high rates of divergence.

FORCES C. ROLEOF EVOLUTIONARY In much the same way that behavioral diversity can afford opportunities to explore selection in the wild, patterns in the distribution and levels of

181

GUPPY POPULATIONS IN TRINIDAD

genetic diversity can provide a means to examine microevolutionary forces in natural populations. Allozyme electrophoresis has facilitated descriptions of genetic differentiation among fish populations, but only rarely have the underlying causes of genetic differentiation been examined (reviewed by Gyllensten, 1985; Waples, 1987; Carvalho, 1993; Ward et al., 1994).

Among the various approaches for studying evolutionary processes in the wild (Endler, 1986), two are particularly informative: first, the examination of correlations between genetic and environmental variation, and second, the study of populations that have been manipulated, enabling the genetic consequences of the perturbation to be investigated. Both approaches have provided valuable data on the relative roles of stochastic and deterministic forces in Trinidadian guppies, through examining the spatial distribution of genetic diversity (Shaw et al., 1991, 1994) and evolutionary consequences of artificial introductions (Reznick et al., 1990; Shaw et al., 1992; Magurran et al., 1992; Carvalho, Shaw, Hauser, Seghers, and Magurran, in press). In addition to the broad correspondence of genetic variability in P . reticulata with geography, patterns have been detected in the distribution of genetic diversity in relation to river order (Shaw et al., 1992, 1994). A comparison of 18 sites revealed markedly higher levels of genetic diversity (mean observed heterozygosity, H,) in lowland ( H , = 0.0382 0.006 SE, n = 9), compared with upland populations ( H , = 0.0112 2 0.0034 SE, n = 9). Several explanations could account for such a correlation, including the geographical isolation and probable smaller effective population sizes of fragmented upland sites, together with unidirectional gene flow bolstering levels of variation in downstream sites. Increased heterogeneity of habitats in lowland rivers could further promote genetic diversity (Shaw et al., 1991). Finally, differences in mating tactics between upland and lowland sites could contribute to the observed differences in heterozygosity. The presence of predators in lowland sites may allow increased levels of sneaky mating (Luyten and Liley, 1985; Magurran and Nowak, 1991; Magurran and Seghers, 1991), circumventing the strong female mate choice and associated strong sexual selection seen in guppies (Houde and Endler, 1990). Compared to the more randomly mating lowland populations, therefore, strong mate choice and sexual selection in low-predation upland rivers may lead to increased inbreeding and decrease of local effective population size, both of which facilitate the loss of genetic diversity. In a study designed to disentangle effects due to geographic isolation and contrasting predation regimens (Shaw et al., 1994), heterozygosity was estimated at sites encompassing guppy habitats along two complete river courses. The two rivers differed in predation regimes in that the

*

182

A. E. MAGURRAN

ET AL.

upland reaches of one (the Aripo) were devoid of major predators, whereas the other (the Tacarigua) had no impassable waterfalls, allowing predators almost to the upper limits of the guppy distribution. Behavioral observations were also conducted at two sites on each river to assess mating tactics. Although genetic diversity increased progressively in samples taken downstream (e.g., for the Tacarigua river, Po,95,the number of polymorphic loci, and H,, mean observed heterozygosity, over 8 loci: upland sitesP = 1-3, H , = 0.06-0.10;lowlandsitesP = 3-8,H0 = 0.11-0.211, there was no direct correspondence with predation regime, and mating tactics were not as divergent between high- and low-predation sites as laboratory observations had suggested. It appears therefore that the observed spatial patterns of genetic variability arise primarily as a consequence of stochastic forces leading to the chance loss of alleles (founder effects, drift) in upland sites arising from physical isolation and small effective population sizes. The effects of stochastic forces in promoting genetic divergence among populations have been well illustrated by genetic studies on artificial introductions of guppies (Shaw et al., 1992; G. Carvalho et al., in press). In the case of Haskins’s introduction of Caroni fish into the upper Turure river (see Section V,A), allozyme analysis of gene frequencies in the introduced population 34 years after the introduction revealed a reduction in genetic diversity compared to the ancestral population, as predicted by theory, because of sampling effects in the initial colonizers and drift during the ensuing inbreeding. Levels of genetic identity between the proposed ancestral population and the introduced populations are as low (mean 1 = 0.988) as those previously recorded between guppy populations from separate rivers within the same drainage (Fig. 12; Carvalho et al., 1991; Shaw et al., 1991). For such a genetic distance to have accumulated between two populations derived from the same source in just 34 years (approximately 120 generations) is again indicative of founder events or population bottlenecks, which are known to accelerate microevolutionary rates (Chakraborty and Nei, 1977). Since an inseminated female guppy can store sperm, it can produce offspring over several brood cycles, and so it is feasible that a single female may initiate a population. Direct evidence for this was provided by an artificial introduction, in Trinidad by J. Kenny, of a single pregnant female into a garden pond (“Kenny’s Pond”; 0.5 m diameter, 0.5 m depth) in 1981 (G. Carvalho et al., in press). The female founded a population still in existence, providing an opportunity for comparing the introduced fish with their ancestral population in a drainage culvert near the Priority Bus Route (PBR site). Allozyme analysis of fish collected in 1991, at least

GUPPY POPULATIONS IN TRINIDAD

183

30 generations after the introduction, detected an extreme founder effect, with low genetic similarity between the two populations (I = 0.938; Fig. 12), and a marked reduction in genetic diversity (Kenny’s Pond, P0.95 = 0.08, H , = 0.019; PBR, Po.95 = 0.28, H, = 0.081). The study

demonstrates not only that extreme founder events are possible in P . reticulata, but that such events may greatly promote genetic divergence, perhaps accounting for some genetically divergent natural populations (e.g., Yarra River; Fig. 12). Although this is only one example of a population founded by a single female it does confirm that such events are possible in the wild. In addition to stochastic forces, deterministic forces arising from contrasting predation pressures undoubtedly play a major role in promoting genetic divergence. Broad geographic patterns of differentiation can be related to probable past geological and historical events and degree of physical isolation, though the existence of divergence without obvious physical barriers (Carvalho et al., 1991) indicates the importance of extant evolutionary forces. Genetic divergence in Trinidadian guppies may in part be driven by their complex color polymorphism, which is subject to intense selection pressures (sexual selection and predation), as well as the stochastic influences of small effective population sizes due to founder effects and their apparent restricted vagility.

D. PHYLOGEOGRAPHIC IMPLICATIONS A N D SPECIATION Available molecular genetic evidence (Carvalho et al., 1991; Fajen and Breden, 1992) indicates that guppies from the Oropuche and Caroni drainages are derived from distinct ancestral stocks (Fig. 12), providing support for the “two arcs” hypothesis of colonization of Trinidad originally proposed to explain lizard distributions (Boos, 1984). Indeed, a striking case of genetic divergence was found between the Aripo River samples (Caroni drainage) and the Oropuche samples (Carvalho et al., 1991), which is surprising because the two watersheds are separated by only 1 km of savannah. The relevance of the two arcs hypothesis to fish movements has also been given credence by the pronounced genetic differentiation of characin fishes from the Caroni and Oropuche river basins (Ali, 1989). Molecular data can be employed to estimate times of divergence (Thorpe, 1983) and hence provide indications of speciation rates in different lineages (e.g., Meyer et al., 1990). The translation of genetic distance measures into evolutionary time must, however, be exercised with caution, especially since different molecules evolve at variable rates, and genetic divergence itself may be greatly magnified in specific circumstances. For example, using Nei’s (1975) estimate of the relationship of

184

A. E. MAGURRAN ET AL.

genetic distance ( D ) to time since separation (1.0 D = 5 million years), the D value for Kenny’s Pond (0.064) would indicate a separation time of 320,000 years; direct evidence indicates this not to be the case. Here, founder effect has greatly inflated the apparent rate of evolutionary divergence. Bearing such limitations in mind, it is still worthwhile examining probable times of divergence between guppy stocks, especially where the same populations have been examined using independent molecular tools. It is reassuring that similar times of divergence have been arrived at for some populations using both mitochondria1 DNA (mtDNA) sequence data (Fajen and Breden, 1992) and allozymes (Carvalho et al., 1991; Shaw et al., 1991, 1992). mtDNA and allozyme comparisons between north coast and Caroni drainage sites (mtDNA = 200,000 yr; allozymes = 170,000 yr), and Oropuche and Caroni sites (mtDNA = 500,000 yr; allozymes = 330,000 yr), revealed values of similar magnitude. Such data indicate that the genetic divergence among guppy populations occurred prior to separation from South America (Kenny, 1989). Trinidadian guppies therefore display high levels of genetic and phenotypic differentiation,some of which may be approaching levels of incipient speciation. Speciation does, however, depend on the intensity of isolating mechanisms, and is not merely a function of evolutionary time since separation. For example, the extraordinary proliferation of cichlid species in Lake Victoria has occurred within only 500,000-750,000 years (Fryer and Iles, 1972), although, in contrast to guppies, it has been characterized by extensive phenotypic differentiation among species underlain by high genetic similarity (Meyer et al., 1990). In guppies, marked genetic divergence has proceeded in parallel with phenotypic differentiation, though the evolutionary fate of disparate stocks depends ultimately on opportunities for adaptive radiation and effectiveness of isolating mechanisms. In conclusion, molecular data have revealed marked genetic divergence among populations, and have provided a means of examining the mechanisms determining such divergence. Patterns observed arise from the interplay between historical events and intense selection pressures, which have been magnified by stochastic forces arising from chance fluctuations in effective population size.

AND STOCHASTIC INFLUENCES ON BEHAVIOR VII. HISTORICAL

The previous sections of this chapter have sought adaptive explanations for variation in behavior. Yet, as we have just seen, stochastic factors do contribute to the genetic diversification of populations. We therefore now

GUPPY POPULATIONS IN TRINIDAD

185

ask whether there are cases where population variation in behavior of Trinidadian guppies cannot readily be attributed to the adaptive model. As we know, there is a strong correlation between levels of schooling behavior and the presence of predators, especially the pike cichlid, Crenicichla. The conclusion that selection (exerted by predators) molds behavior (and other traits) seems unassailable (Endler, 1980; Reznick et al., 1990). Nonetheless, our investigations reveal that there can be major differences in the schooling behavior of populations within the same predation regime. Take the case of the Oropuche and Caroni river systems, which drain the southern slopes of the Northern Range. At their closest point the two systems are separated by less than 1 km of savannah. The drainages are similar ecologically and support almost identical fish communities. In both systems fish diversity increases in the downstream portion of streams. Crenicichla is widespread, especially at lower elevations. Our example focuses on four rivers, the Aripo and Guanapo Rivers from the Caroni drainage and the Oropuche and Quare Rivers from the Oropuche drainage. The populations under consideration are from the lower, high-predation sections of the rivers. In the past five years we have consistently observed high densities of adult and juvenile Crenicichla in the Lower Aripo and Oropuche and have also confirmed the predator’s presence in the Quare and Guanapo. J. A. Endler (personal communication) recorded that Crenicichla was abundant in the Oropuche in 1976(see also Douglas and Endler, 1982), while one of us (B. H. Seghers) observed the species in the Quare in 1977. It is likely that Crenicichlu has been present at the four locations for a long period before guppy biologists became interested in the species. Guppies were collected from the four sites in Trinidad. Behavioral tests on their descendants (which had been bred and raised in the laboratory) two to three generations later, revealed that both populations from the Oropuche drainage had a lower schooling tendency than their compatriots from the Caroni drainage (Fig. 13). Because all guppies were equally wary when inspecting a model Crenicichla (Fig. 13), it seems unlikely that this behavioral difference was symptomatic of an impoverished antipredator response in Oropuche guppies. How might we explain the differences in the schooling behavior of guppies from populations that appear to be under the same predation regime? There are two main possibilities. It may be that Oropuche (drainage) guppies are trading off schooling tendency against some other behavior. For example, as we observed earlier, schooling has associated costs, particularly in the context of competition for food. It is interesting that Oropuche (River) females defend a food patch more assiduously than do Lower Aripo females (Magurran and Seghers, 1991).This was shown in a laboratory test where female guppies (in groups of eight all drawn from the same population) were offered food

186

A

A. E. MAGURRAN ET AL.

300

T

:i'1l Q)

--E

200

-c

8r

100

0

.......... :.:.:.:.:. i:iiiig .......... ..... ..... ..... ..... .......... ..... ..... ..... .......... ..... ..... .......... ..... ..... ..... iiiiiiiii; .......... ..... ..... ..... .......... ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... .'".

.......... ..... ..... :*.*.*.*.'. . :.:.:.:. .......... :...... :.:.:.:. ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... .*.*.*.*.. ..... .......... ..... ..... .......... ..... ..... ..... ..... ..... ..... ..... .......... ..... ..... ..... .......... ..... ..... ..... .......... ..... *........... *.*.*.*. .......... ..... ..... .......... ..........

FIG. 13. The mean (+ 95% confidence limits) time spent schooling, out of a total of 300 s (A) and the closest approach to a realistic predator model during inspection (B) by female guppies in a laboratory test. All fish had been bred and raised in controlled conditions in the laboratory. Stippled columns denote those populations (Guanapo and Lower Aripo) that occur in the Caroni drainage, while shaded ones (Quare and Oropuche) belong to the Oropuche drainage. See Magurran ef a/. (1992) for details.

in a single concentrated patch (a pebble dipped in gelatine and then in Promin'" fish food). Oropuche females were quicker to commence aggression and had a higher overall frequency of aggressive interactions. Unlike Lower Aripo females, Oropuche females competed for control of the patch. In the wild guppies will monopolize patches of benthic food (Magurran and Seghers, 1994c), so this experiment was a realistic test of the natural situation. We do not know whether the different behavior patterns are equally successful (in terms of reproductive fitness), though this would be an interesting topic for future study. Alternatively, suboptimal strategies could be retained in some populations as a genetic legacy from the founding fish. This hypothesis is exceedingly difficult to test, although it is notable that the schooling differences in the two drainages are paralleled by marked genetic divergence (see above). The most extreme case of genetic population differentiation at the allozyme level also occurs in a population (the Upper Yarra, northern coast of Trinidad) where the schooling behavior is inappropriate for the

GUPPY POPULATIONS IN TRINIDAD

187

extant predation regime. As in other north coast rivers, Upper Yarra River guppies are exposed to predation from the prawn Macrobrachiurn crenulatum. However, unlike the population in the nearby Paria River, which displays the low levels of schooling predicted for this predation regime (see Section III,B), the Upper Yarra population, in an apparently identical environment, has a high schooling tendency (Magurran et al., 1993). The Yarra River guppies are genetically the most distinct natural population surveyed to date in Trinidad, lying even outside the “extreme founder event” population in Kenny’s Pond in the Trinidad population group (Fig. 12) as a result of being fixed for one allele at every allozyme locus (Carvalho et al., 1991). The Upper Yarra population has almost certainly gone through an extreme bottleneck in population size in the past, most probably the result of a chance colonization event. Two previous studies, one on male coloration, the other on life history tactics, have rejected the hypothesis that stochastic factors (such as historical events) contribute to observed patterns of variation. Douglas and Endler (1982) undertook a multivariate analysis of 41 guppy sites in Trinidad. They assessed male color patterns in relation to predation regime, altitude, and watershed. Four alternative evolutionary models were tested. The first two models, which proposed that color patterns varied clinally with altitude and patchily as a consequence of predation risk, could not be rejected. By contrast, Douglas and Endler found no effect of watershed (taken as a measure of historical factors) nor of geographic isolation (which should detect effects due to genetic drift). They concluded that male color patterns are a consequence of selection imposed by predators, rather than due to stochastic processes. Strauss (1990) reanalyzed Reznick and Endler’s (1982) data on life history variation among guppy populations located in the Oropuche and Caroni drainages. He demonstrated that predation regime accounts for all but 17% of the variation in life history traits and attributed this residual variation to environmental heterogeneity. Molecular markers used in our studies (Carvalho et al., 1991; Shaw et al., 1991, 1992, 1994) and those of Fajen and Breden (1992) have indicated that stochastic events such as colonization events or random genetic drift do have significant effects on guppy populations in Trinidad, at least at the level of these supposedly neutral genetic markers. In upland rivers guppy populations display progressive loss of genetic variability as the upper limits of the species’ distribution are approached, with populations in different headstreams being fixed for different allelic forms of particular enzyme coding genes. We have concluded that this effect is more likely to result from founder effects and bottlenecks in population size due to physical substructuring of the population, rather than inbreeding caused

188

A.

E. MAGURRAN ET AL.

by predation regime-mediated sexual selection (Shaw et al., 1994). Our genetic analyses of artificial introductions of guppies have demonstrated the dramatic changes in gene frequencies that may result from colonization events. None of the three introduction examples displays a link between changes at allozyme loci and observed shifts in heritable morphological, life history and behavioral characters, but they do illustrate the potential for rapid genetic differentiation of populations following random historical events. Accelerated microevolution and drastic genetic reorganization following founder events and consequent bottlenecks are thought to be important factors in intraspecificdifferentiation and subsequent speciation processes (Mayr, 1963; Nei, Maruyama, and Chakraborty, 1975). The circumstantial correlations between genetic and behavioral divergence in the Caroni and Oropuche drainages, and the Paria and Upper Yarra rivers are very suggestive of such historical influences on guppy population evolution.

VIII. LINKSBETWEEN BEHAVIOR AND DIVERGENCE How do the behavioral differences we have documented contribute to the genetic divergence that hasjust been described? Behaviorally mediated reproductive isolation can arise in two ways. First of all, groups of individuals can be physically separated (because of habitat preferences, schooling tendency, and so on) and have no possibility of interbreeding. Alternatively, particular individuals may have the opportunity to mate but choose not to. A. FISHDISTRIBUTION, BEHAVIOR, AND GENE FLOW Any behaviors that decrease migration or cause a patchy distribution of individuals should help limit gene flow. Morphologically, guppies are poorly adapted to life in fast current and are not normally found in water flowing above 0.3 m/s (Seghers, 1973). Preference for benthic food (Dussault and Kramer, 1981) also means that guppies are most likely to be found in shallow sections of streams or at river margins. However, habitat preferences can be further influenced by the need to avoid predators, or even competition from other species. Portions of rivers where predators are abundant are particularly inhospitable. Seghers (1973) found that guppies in the Guayamare were unwilling to venture into sections of the river inhabited by the predatory characin Hoplias, even though the habitat was physically suitable in other respects. He further demonstrated that guppies

GUPPY POPULATIONS IN TRINIDAD

189

from the high-risk sites had an inherited tendency to remain near the surface of the water column and close to the riverbank. Schooling behavior is potentially capable of inducing a patchy distribution of fish. This would occur, for example, if individuals are faithful to particular schools and if these schools are dispersed in the habitat. There is evidence that associations within schools are nonrandom. We (Magurran, Seghers, Shaw, and Carvalho, 1994) have recently shown that, in the laboratory, females prefer to school with familiar individuals. In this experiment females were housed in groups for a period of two months. They were then, as individuals, given the opportunity to school with either a familiar female (from their own tank and population) or an unfamiliar female (of the same population but from a different tank). Females spent significantly more time schooling with familiar individuals. Since all fish were originally bred in mass cultures in the laboratory, it is unlikely that familiar females were more closely related than unfamiliar ones. Familiarity could nevertheless account for the weak kin associations that we have observed in the wild. Guppy schools are not random genetic groupings of individuals. Using multilocus genetic fingerprinting with a human minisatellite probe (Jeffreys, Wilson, and Thein, 1985), we have demonstrated that individuals within the majority of schools caught in the wild are more closely related than the average for the population (unpublished data). The level of relatedness (see Lynch, 1988) within schools was low, in the range of 4th-order relatives (mean r = 0.077 to 0.099), but significantly higher than expected for unrelated individuals (mean r = 0.023 in the test population). So, whether individuals recognize kin and choose to school with them, or whether they associate with familiar individuals who from birth happen to be related because of viviparous brood production and well-developed site fidelity, schooling appears to maintain a subtle level of local population substructuring. Although guppies in larger rivers are not forced into close proximity with one another, they have the option to associate over long periods of time if they choose to do so. Indeed, when we undertook a mark-recapture survey in a section of the Quare River (at which the river was approximately 8 m wide and 40 cm deep), we discovered that female guppies show strong site fidelity. Nearly 70% of marked fish were recaptured within a few meters of the release site five days later (Magurran et al., 1994). More extensive mark-recapture studies by D. N. Reznick and his colleagues (personal communication) likewise show little short-term movement by female guppies. This degree of association in the wild provides an opportunity to learn the identity of other individuals in the locality. It is now well established that fish have the “cognitive” skills to discriminate individuals and to modify their behavioral responses on the

190

A. E. MAGURRAN ET AL.

basis of past interaction (Milinski, Kulling, and Kettler, 1990a; Milinski, Huger, Kulling, and Kettler, 1990b; Dugatkin and Alfieri, 1991). Does restricted vagility resulting from habitat choice, site specificity, and schooling enhance genetic substructuring of populations? Guppy populations in Trinidadian rivers certainly display extensive microgeographic heterogeneity in gene frequences at all levels, from differences between adjacent rivers (Carvalho et al., 1991)down to significant allele frequency changes between adjacent pools in upland streams. Repeated sampling of allozyme frequencies along the complete course of two rivers (the Tacarigua and the Aripo) demonstrated that substructuring of populations could be related, not only to the presence of physical barriers to fish movement, but also to distance between sites (Shaw et al., 1994). Temporal studies have also shown that gene frequencies within sites may remain relatively stable over several years (Shaw et al., 1991). Substructuring of the population through “isolation by distance” (Slatkin, 1993) that is relatively stable over time would be consistent with restricted effective migration by individuals, so it is not improbable that site specificity and some degree of behaviorally mediated social cohesion in guppy populations will promote small-scale genetic differentiation. DIVERSITY B. MATINGBEHAVIORA N D GENETIC Female guppies show strong preference for particular males (Houde, 1987; Houde and Endler, 1990). To what extent does this preference facilitate population differentiation? Lande (1981) showed how Fisher’s “runaway process” can amplify differences in sexually selected characters and lead to rapid speciation through reproductive isolation. Endler (1989) argues that sexual selection in the guppy can have the effect of generating or exaggerating premating isolation. Because levels of sexual harassment are reduced in low-risk habitats, it is here that females ought to have most freedom to exercise choice. Low-risk sites are often physically isolated, and more likely to have been subject to founder effects. Female choice may thus exaggerate genetic differences that have arisen by stochastic means. On the other hand, the high incidence of sneaky mating attempts in more dangerous habitats could potentially undermine the role of female choice, and maintain random outcrossing. Remarkably little is known about the relative success of sneaky matings. Laboratory studies suggest that males may only infrequently father offspring by this means (Luyten and Liley, 1991; Kodric-Brown, 1992). Males, however, would seem most unlikely to indulge in such frenetic activity if it did not bring some reproductive rewards. Establishing the extent of those rewards remains a major challenge for the future.

GUPPY POPULATIONS IN TRINIDAD

191

The importance of sneaky mating has been amply demonstrated in some other species of fish. Unlike guppies, coho salmon (Oncorhynchus h u t c h ) and bluegill sunfish (Lepornis rnacrochirus) males specialize in different strategies (Dominey, 1980; Gross, 1982, 1985). Frequency-dependent selection determines the percentage of sneaky males in the population. If there are too many sneaky males, then their reproductive success declines; if there is an abundance of conventional males, then the opportunities for sneaking increase. Thus (in theory at any rate), there should be a stable polymorphism of the two strategies, with each, on average, achieving the same lifetime reproductive success. We know that females from populations isolated by physical distance, or barriers, use different criteria by which to judge their partners. As a consequence, males who stray into a strange population could well be shunned by the resident females. This behavioral effect would buffer populations against migrants and reduce gene flow. However, in contrast to other observers, Farr (1977) has suggested that females actually prefer rare males. The role of female choice in population differentiation thus remains a contentious issue. Nonetheless, Liley (1966) demonstrated clear behavioral differences in the reproductive behavior of four sympatric species of poeciliid fishes, and found that females responded only to males of their own species. His studies included guppies and their closest relative P. picta, and provides strong circumstantial support for the assertion that female choice plays a crucial role in population differentiation and speciation.

DIVERSIFICATION AND SPECIATION C. POPULATION The guppy is only one of a number of species that exhibit marked population variation. An equally impressive example is the three-spined stickleback (Bell and Foster, 1994b). Like guppies, stickleback populations differ in their antipredator responses (Huntingford et al., 1994), morphology (Reimchen, 1994), reproductive behavior (Foster, 1994), life history tactics (Baker, 1994), and allozymes (Buth and Haglund, 1994). The morphological, behavioral, and life history divergence appears to be driven by natural selection (McPhail, 1994). There can even be different forms in the same habitat. McPhail(l994) describes three types of divergent sympatricor parapatric populations that occur in the Strait of Georgia region of British Columbia. The first contrast is between stream-resident and anadromous forms. The second pair of divergent populations are the lake- and the stream-dwelling sticklebacks. Limnetic (plankton-feeding) and benthic sticklebacks represent the third set of populations. Since the Strait of Georgia region was glaciated during the last ice age, the maximum age of these populations is in the region of 12,000 years. Reproductive

192

A. E. MAGURRAN ET AL.

isolating mechanisms have evolved in at least one of the divergent pairs, the benthic and limnetic sticklebacks. Females choose to mate with males of their own type, which they can distinguish on the basis of nuptial coloration and courtship behavior (Ridgway and McPhail, 1984). Although the forms mate at the same time of year, and may occupy adjacent tenitories, the two types of males choose to nest in different microhabitats. McPhail rejects the idea that the divergent forms arose sympatrically as a consequence of feeding specializations. Instead, he attributes their reproductive isolation to separate colonizations by the marine form. Irrespective of its origin, it is clear that behavior is vital in maintaining the differentiation of the benthic and timnetic sticklebacks. Speciation might seem the inevitable consequence of population differentiation. Yet, although the stickleback populations are obviously capable of rapid and adaptive variation, the stickleback genus Gasterosteus is, paradoxically, a species-poor one (McPhail, 1994). The lineage has been present in North America since the Miocene (Bell, 1977; Bell and Foster, 1994a),but is represented by only two well-defined species. McPhail(l994) argues that the species is primarily marine and that isolated populations are vulnerable to extinction or recolonization. Foster and Bell (1994) suggest that low potential for divergent exploitation of limited resources prevents the coexistence of many stickleback species and that the ephemeral nature of freshwater habitats limits the amount of time available for the evolution of reproductive isolation. Superficially,the poeciliids appear more speciose than the sticklebacks. However, the subgenus Lebistes consists of only six species and estimates from mtDNA and allozymes put the separation of Trinidad guppy populations at many hundreds of thousands of years (Fajen and Breden 1992; this chapter). Why are there not more species of “guppy”? Solving the relationship between population differentiation and speciation is one of the greatest problems facing evolutionary biologists today.

IX. FUTUREDIRECTIONS Ever since the landmark paper by Haskins et al. (1961) the interest in Trinidadian guppies has increased exponentially. By the end of 1994 over 100 papers were published on these populations, twice the number published between 1961 and 1990 (Seghers, 1992). Trinidad guppies are firmly established as a model system for testing many hypotheses in evolutionary biology, especially in the domain of predator-prey interactions, life history evolution, and sexual selection. Few vertebrates exhibit such a high degree of phenotypic and genotypic variation over such a small geographic area and are so amenable to field and laboratory observation.

GUPPY POPULATIONS IN TRINIDAD

193

We raise two concerns about recent trends in research on these fish: first, the relevance of laboratory studies to what is actually occurring in the wild, and second, the future well-being of the populations themselves. Because guppies are relatively easy to study in small aquaria they have become popular subjects in behavioral ecology. Much recent work has been on mate choice and sexual selection. Food is usually provided ad lib. and females normally “choose” between two or more males (often behind glass) in a small, predator-free environment. We acknowledge that such work is an important first step in identifying potential selection pressures in the real and highly complex natural world where females spend a large proportion of their time acquiring food and maintaining vigilance against predators and sneaky mating attempts by males. Females in the wild seem to have little time to assess, associate with, or mate with specific males from an array of perhaps dozens or hundreds of potential suitors that may occur in a single pool. We are not suggesting that highly controlled laboratory experiments have no relevance to the situation in nature, but we do urge workers to devote more time and effort to validating laboratory results with field observations or experiments. With this species we have the good fortune of relatively easy access to natural populations. Ideally, investigators should combine both a field and a laboratory element. A recent series of experiments (in the laboratory) has indicated that female guppies seem to copy the mate choice of others (Dugatkin, 1992; Dugatkin and Godin, 1993). If this observation is to radically change our way of viewing female choice, it is essential that copying be shown to occur in the wild. Likewise, we have found that harassment of females by males is widespread in the wild. In order to ascertain the extent to which it undermines female choice, we will need to resort to appropriate laboratory experiments. If the Trinidad guppy is to remain a model for studies on factors promoting or retarding population differentiation, that is, the origin of biodiversity, it is imperative that we give urgent attention to conservation measures. As we have noted elsewhere in this review, field transplants of guppies or predators have provided convincing evidence for the rapid action of natural selection. Nonetheless, this information has been acquired at the cost of irreversible changes in the impacted river systems. Our work on the Turure River has underlined the confusing picture that may arise for future workers on introduced or mixed populations, especially if molecular evolution is of primary interest. We thus urge future workers to use restraint when manipulating or harvesting the remaining natural populations. It would also be tragic if those of us who seek to understand the evolution of biodiversity contributed to its loss. For this reason it is vital that ethical considerations, as well as scientific ones,

194

A. E. MAGURRAN ET AL.

play a central role in any investigation. In addition we should continue to develop local and international awareness of the importance of these tiny colorful fish to the world’s scientific community. There is still much to learn about the behavioral and genetic diversity of guppy populations in Trinidad. Our hope is that future workers will be able to profit from the unique opportunities provided by these fish and test new hypotheses to explain their remarkable diversity.

X. SUMMARY Trinidadian guppy populations display impressive behavioral diversity and marked genetic differentiation. Much of the behavioral variation observed in Trinidadian guppies can be attributed to deterministic processes. Of these the most important is selection exerted by predators. Predation risk directly affects evasion responses such as inspection behavior and schooling tendency but has further indirect consequences. For instance, mating decisions can be compromised by risk. Behavioral interactions between males and females are also influenced by the sex ratio, which is itself indirectly related to risk. Improved antipredator responses bring attendant costs. For instance, guppies with a greater schooling tendency, which necessitates uniform behavior, are less adept at competing for limited resources. Transplant experiments, where guppies are moved to sites with a different predation regime, confirm that behavior can evolve in the wild. It seems that antipredator defenses are lost slowly following a relaxation of predation risk, whereas coloration and life history tactics evolve rapidly. Although most population variation in behavior can be attributed to variation in predation risk, there are behavioral differences that do not seem to have an adaptive explanation. Historical and stochastic events may thus also mold behavior. Analysis of neutral molecular genetic markers has revealed a complex pattern of genetic variability and differentiation in guppy populations. Substantial evidence exists for the importance of stochastic processes such as colonization history, founder effects, and population bottlenecks in the patterns observed. Stochastic effects are most notable in isolated populations, but conceivably have influenced the diversity of the genome available for selection by deterministic processes in many Trinidadian populations. Acknowledgments

It is a pleasure to thank J. S. Kenny and his colleagues in the University of the West Indies, St. Augustine, Trinidad, for their help and hospitality. J. A. Endler, D. F. Fraser,

GUPPY POPULATIONS IN TRINIDAD

195

C. P. Haskins, D. N. Reznick, and H. F. Rodd kindly provided details of unpublished work. The editors made helpful comments on the paper. Funding from N.E.R.C. and the Royal Society (London) is gratefully acknowledged.

References

Ali, N. (1989). Affinities of some freshwater fishes from Trinidad. Unpublished masters thesis, University of the West Indies, Trinidad. Baker, J. A. (1994). Life history variation in female threespine stickleback. In “The Evolutionary Biology of the Threespine Stickleback” (M. A. Bell and S. A. Foster eds.), pp. 144-187. Oxford University Press, Oxford. Bakker. T. C. M. (1985). Two-way selection for aggression in juvenile, female and male sticklebacks (Gasterosteus aculearus L.), with some notes on hormonal factors. Behauiour 93, 69-8 I . Bakker, T. C. M. (1986). Aggressiveness in sticklebacks (Gasterosteus aculeatus L.): A behavior-genetic study. Behauiour 98, 1-144. Beardmore, J. A., and Shami, S. A. (1979). Heterozygosity and the optimum phenotype under stabilising selection. Aquilo Series Zoologica 20, 100-1 10. Bell, M. A. (1977). A late Miocene marine threespine stickleback, Gasterosteus aculeatus aculeatus, and its zoogeographic and evolutionary significance. Copeia 1977,277-282. Bell, M. A., and Foster, S. A. (Eds.). (1994a). “The Evolutionary Biology of the Threespine Stickleback.” Oxford University Press, Oxford. Bell, M. A., and Foster, S. A. (1994b). Introduction to the evolutionary biology of the threespine stickleback. In “The Evolutionary Biology of the Threespine Stickleback” (M. A. Bell and S. A. Foster, eds.), pp. 1-27. Oxford University Press, Oxford. Boeseman, M. (1960). The fresh-water fishes of the island of Trinidad. Studies ofthe Fauna of Curacao and other Caribbean Islands 48, 72-153. Boos, H. E. A. (1984). A consideration of the terrestrial reptile fawna on some offshore islands northwest of Trinidad. Living World 1983/1984, 19-26. Breden, F., and Stoner, G. (1987). Male predation risk determines female preference in the Trinidad guppy. Nature (London) 329, 83 1-833. Buth, D. G., and Haglund, T. R. (1994). Allozyme variation in the Gasterosteus aculeatus complex. In “The Evolutionary Biology of the Threespine Stickleback” (M. A. Bell and S. A. Foster, eds.), pp. 61-84. Oxford University Press, Oxford. Carvalho, G. R. (1993). Evolutionary aspects of fish distributions: Genetic variability and adaptation. J. Fijh Biol. 43 (Supplement A), 53-73. Carvalho, G. R., Shaw, P. W., Magurran, A. E., and Seghers, B. H. (1991). Marked genetic divergence revealed by allozymes among populations of the guppy. Poecilia reticulafa (Poeciliidae) in Trinidad. Biol. J . Linn. SOC.42, 389-405. Carvalho, G. R., Shaw, P. W., Hauser, L., Seghers, B. H., and Magurran, A. E. (1995). Artificial introductions, evolutionary change, and population differentiation in Trinidadian guppies (Poecilia reticulafa:Poeciliidae). Biol. J . Linn. SOC.(in press). Chace, F. A., and Hobbs, H. H. (1%9). The freshwater and terrestrial decapod crustaceans of the West Indies with special reference to Dominica. Bulletin: United States National Museum 292, 1-258. Chakraborty, R., and Nei, M. (1977). Bottleneck effects on average heterozygosity and genetic distance with the stepwise mutation model. Evolution 31, 347-356. Constanz, G. D. (1975). Behavioral ecology of mating in the male gila topminnow, Poeciliopsis occidentalis (Cyprinodontiformes: Poeciliidae). Ecology 56, 966-973.

196

A.

E. MAGURRAN ET AL.

Constanz, G. D. (1989). Reproductive biology of Poeciliid fishes. I n “Ecology and Evolution of Livebearing Fishes (Poeciliidae)” (G. K. Meffe and F. F. Snelson, eds.), pp. 33-50. Prentice Hall, New Jersey. Coss, R. G., and Owings, D. H. (1985). Restraints on ground squirrel antipredator behavior: Adjustments over multiple time scales. I n “Issues in the Ecological Study of Learning” (T. D. Johnston and A. T. Pietrewicz, eds.), pp. 167-200. Lawrence Erlbaum, Hillsdale, New Jersey. Courtney, W. R., and Meffe, G . K. (1989). Small fishes in strange places: A review of introduced poeciliids. I n “Ecology and Evolution of Livebearing Fishes (Poeciliidae)” (G. K. Meffe and F. F. Snelson, eds.), pp. 319-331. Prentice Hall, Englewood Cliffs, New Jersey. Curio, E. (1993). Proximate and developmental aspects of antipredator behavior. Adu. Study Beh. 22, 135-238. Dawkins, R. (1989). “The Selfish Gene.” OUP, Oxford. Dominey, W. J. (1980). Female mimicry in male blue gill sunfish-a genetic polymorphism? Nafure (London) 320,62-63. Douglas, M. E., and Endler, J. A. (1982). Quantitative matrix comparisons in ecological and evolutionary investigations. J. Theoret. Biol. 99, 777-795. Dugatkin, L. (1988). Do guppies play Tit for Tat during predator inspection visits? Beh. Ecol. Sociobiol. 23, 395-399. Dugatkin, L. A. (1992). Sexual selection and imitation: Females copy the mate choice of others. Am. Nat. 139, 1384-1389. Dugatkin, L. A., and Alfieri, M. (1991). Guppies and the Tit for Tat strategy: Preference based on past interaction. Eeh. Ecol. Sociobiol. 28, 243-246. Dugatkin, L. A., and Godin. J.-G. J. (1992). Prey approaching predators: A cost-benefit perspective. Ann. Zool. Fennici 29, 233-252. Dugatkin, L. A., and Godin, J.-G. J. (1993). Female mate copying in the guppy (Poecilia reticulafa):Age-dependent effects. Beh. Ecol. 4,289-292. Dussault, G. V., and Kramer, D. L. (1981). Food and feeding behavior of the guppy Poecilia reticulafa (Pisces: Poeciliidae). Can. J. Zool. 59, 684-701. Endler, J. A. (1978). A predator’s view of animal colour patterns. Evol. Biol. 11, 319-364. Endler, J. A. (1980). Natural selection on color patterns in Poecilia reticulafa. Evolution 34,76-91. Endler, J. A. (1983). Natural and sexual selection on color patterns in poeciliid fishes. Environ. Biol. Fish. 9, 173-190. Endler, J. A. (1986). “Natural Selection in the Wild” Princeton University Press, Princeton, New Jersey. Endler, J. A. (1987). Predation, light intensity and courtship behavior in Poecilia reticulafa (Pisces: Poeciliidae). Anim. Behau. 35, 1376-1385. Endler, J. A. (1988). Sexual selection and predation risk in guppies. Nafure (London) 332, 593-594. Endler, J. A. (1989). Conceptual and other problems in speciation. I n “Speciation and Its Consequences” (D. Otte and J. A. Endler, eds.), pp. 625-648. Sinauer, Cambridge, Massachusetts. Endler, J. A. (1991a). Interactions between predators and prey. I n “Behavioural Ecology: An Evolutionary Approach” (J. R. Krebs and N. B. Davies, eds.), pp. 169-196. Blackwell, Oxford. Endler, J. A. (1991b). Variation in the appearance of guppy color patterns to guppies and their predators under different visual conditions. Vision Res. 31, 587-608. Endler, J. A. (1992). Signals, signal condition and the direction of evolution. Am. Nat. 139, s125-s 153.

GUPPY POPULATIONS IN TRINIDAD

197

Endler, J. A. (1995). Multiple-trait evolution and environmental gradients in guppies. Trends E d . E v o ~ .10, 22-29. Fajen, A., and Breden, F. (1992). Mitochondria1 DNA sequence variation among natural populations of the Trinidad guppy, Poecilia reticulata. Evolution 46, 1457-1465. Farr, J. A. (1977). Male rarity or novelty, female choice behavior, and sexual selection in the guppy, Poecilia reticulata Peters (Pisces: Poeciliidae). Evolution 31, 162-168. Farr, J. A. (1981). Biased sex ratios in laboratory strains of guppies, Poecilia reticulata. Heredity 47, 237-248. Farr, J. A. (1983). The inheritance of quantitative fitness traits in guppies. Poecilia reticulata (Pisces: Poeciliidae). Evolution 37, 1193-1209. Farr, J. A. (1989). Sexual selection and secondary sexual differentiation in poeciliids: Determinants of male mating success and the evolution of female choice. In “Ecology and Evolution of Livebearing Fishes (Poeciliidae)” (G. K. Meffe and F. F. Snelson, eds.), pp. 91-123. Prentice Hall, Englewood Cliffs, New Jersey. Foster, S. A. (1994). Evolution of the reproductive behaviour of threespine stickleback. In “The Evolutionary Biology of the Threespine Stickleback” (M. A. Bell and S. A. Foster, eds.), pp. 381-398. Oxford University Press, Oxford. Foster, S. A,, and Bell, M. A. (1994). Evolutionary inference: The value of viewing evolution through stickleback-tinted glasses. In “The Evolutionary Biology of the Threespine Stickleback” (M. A. Bell and S. A. Foster, eds.), pp. 472-486. Oxford University Press, Oxford. Fraser, D. F., and Gilliam, J. F. (1992). Nonlethal impacts of predator invasion: Facultative suppression of growth and reproduction. Ecology 73, 959-970. Fryer, G., and Iles, T. D. (1972). “The Cichlid Fishes of the Great Lakes of Africa.” Oliver and Boyd, Edinburgh. George, C. J. W. (1960). Behavioral interaction of the pickerel (Esox niger and Esox americanus) and the mosquitofish (Garnbusia patruelis). Unpublished doctoral dissertation, Harvard University, Cambridge, Massachusetts. Giles, N., and Huntingford, F. A. (1984). Predation risk and interpopulation variation in anti-predator behaviour in the three-spined stickleback, Gasterosteus aculeatus L. Anim. Behav. 32, 264-275. Gilliam, J. F., Fraser, D. F., and Alkins-Koo, M. (1993). Structure of a tropical stream community: A role for biotic interactions. Ecology 74, 1856-1870. Gross, M. R. (1982). Sneakers, satellites and parentals: Polymorphic mating strategies in North American sunfishes. Z. Tierpsychol. 60, 1-26. Gross, M. R. (1985). Disruptive selection for alternative life hsitories in salmon. Nature (London) 313,47-48. Gyllensten, U. (1985). The genetic structure of fish: Differences in the intraspecific distribution of biochemical genetic variation between marine, anadromous and freshwater species. J . Fish Biol. 26, 691-699. Haskins, C. P., and Haskins, E. F. (1950). Factors affecting sexual selection as an isolating mechanism in the poeciliid fish, Lehistes reticularus. Proc. Natl. Acad. Sci. U.S.A. 36, 464-476. Haskins, C. P., and Haskins, E. F. (1951). The inheritance of certain color patterns in wild populations of Lebistes reticulatus in Trinidad. Evolution 5 , 216-225. Haskins, C. P., Haskins, E. F., McLaughlin,J. J. A., and Hewitt, R.E. (1961). Polymorphism and population structure in Lebistes reticularus, an ecological study. In “Vertebrate Speciation” (W.F. Blair, ed.), pp. 320-395. University of Texas Press, Austin. Hoelzel, A. R. (1992). “Molecular Genetic Analysis of Populations”. IRL Press, Oxford. Houde, A. E. (1987). Mate choice based upon naturally occurring color pattern variation in a guppy population. Evolution 41, 1-10.

198

A. E. MAGURRAN

ET AL.

Houde, A. E. (1988a). The effects of female choice and male-male competition on the mating success of male guppies. Anim. Behau. 36, 888-896. Houde, A. E. (1988b). Genetic difference in female choice between two guppy populations. Anim. Behau. 36,510-516. Houde, A. E. (1994). Effect of artificial selection on male colour patterns on mating preferences of female guppies. Proc. R. Soc. Lond. B . 256, 125-130. Houde, A. E., and Endler, J. A. (1990). Correlated evolution of female mating preferences and male color pattern in the guppy, Poecilia reticulata. Science 248, 1405-1408. Houde, A. E., and Torio, A. J. (1992). Effect of parasitic infection on male color pattern and female choice in guppies. Behau. Ecol. 3, 346-351. Huntingford, F. A. (1982). Do inter- and intra-specificaggression vary in relation to predation pressure in sticklebacks? Anim. Behav. 30, 909-916. Huntingford, F. A., Wright, P. J., andTierney, J. F. (1994). Adaptivevariationinantipredator behaviour in threespine stickleback. In “The Evolutionary Biology of the Threespine Stickleback” (M. A. Bell and S. A. Foster, eds.), pp. 277-296. Oxford University Press, Oxford. Jeffreys, A. J., Wilson, V.,and Thein, S. L. (1985). Individual-specific “fingerprints” of human DNA. Nature (London) 316,76-79. Kapoor-Bijay, P., and White, J. (1992). “Conservation Biology.” Commonwealth Secretariat, London. Kenny, J. S. (1989). Hermatypic scleractinian corals of Trinidad. Studies ofthe Fauna of CuraGao and other Caribbean Islands U3,83-100. Kodric-Brown, A. (1985). Female preference and sexual selection for male coloration in the guppy (Poecilia reticulata). Beh. Ecol. Sociobiol. 17, 199-206. Kodric-Brown, A. ( 1989). Dietary carotenoids and male mating success: An environmental component to female choice. Beh. Ecol. Sociobiol. 25, 393-401. Kodric-Brown, A. (1992). Male dominance can enhance mating success in guppies. Anim. Behau. 44, 165-167. Krebs, J. R., and Davies, N. B. (1993). “An Introduction to Behavioural Ecology.” Blackwell, Oxford. Krumholz, L. A. (1948). Reproduction in the western mosquitofish, Gamhusia affinis affinis (Baird and Girard) and its use in mosquito control. Ecol. Monogr. 18, 1-43. Lande, R. (1981). Models of speciation by sexual selection on polygenic traits. Proc. Nafl. Acad. Sci.. U.S . A . 78, 3721-3725. Licht, T. (1989). Discriminating between hungry and satiated predators: The response of guppies (Poecilia reticulafa) from high and low predation sites. Ethology 82, 238-242.

Liley, N. R. (1966). Ethological isolating mechanisms in four sympatric species of Poeciliid fishes. Behaviour Suppl. 13, 1-197. Liley, N. R., and Seghers, B. H. (1975). Factors affecting the morphology and behaviour of guppies in Trinidad. In “Function and Evolution in Behaviour” ( G . P. Baerends, C. Beer, and A. Manning, eds.), pp. 92-1 18. Clarendon Press, Oxford. Long, K. D., and Houde, A. E. (1989). Orange spots as a visual cue for female mate choice in the guppy (Poecilia reticulata). Ethology 82, 316-324. Luyten, P. H., and Liley, N. R. (1985). Geographic variation in the sexual behaviour of the guppy, Poecilia reticulara (Peters). Behaviour 95, 164-179. Luyten, P. H., and Liley, N. R. (1991). Sexual selection and competitive mating success of male guppies (Poecilia reticulata) from four Trinidad populations. Beh. Ecol. Sociobiol. 28, 329-336. Lynch, M. (1988). Estimation of relatedness by DNA fingerprinting. Mol. Biol. Euol. 5, 584-599.

GUPPY POPULATIONS IN TRINIDAD

199

Magurran, A. E. (1986). Predator inspection behaviour in minnow shoals: Differences between population and individuals. Beh. Ecol. Sociobiol. 19, 267-273. Magurran, A. E. (1990a). The adaptive significance of schooling as an anti-predator defence in fish. Ann. Zool. Fennici 27, 51-66. Magurran, A. E. (1990b). The inheritance and development of minnow anti-predator behaviour. Anim. Behav. 39,834-842. Magurran, A. E. (1993). Individual differences and alternative behaviours. In “Behaviour of Teleost Fishes” (T. J. Pitcher, ed.), pp. 441-477. Chapman and Hall, London. Magurran, A. E., and Nowak, M. N. (1991). Another battle of the sexes: The consequences of sexual asymmetry in mating costs and predation risk in the guppy, Poecilia reticulata. Proc. R . SOC.Lond. B . 246, 31-38. Magurran, A. E., Oulton, W. J., and Pitcher, T. J. (1985). Vigilant behaviour and shoal size in minnows. Z. Tierpsychol. 67, 167-178. Magurran, A. E., and Pitcher, T. J. (1987). Provenance, shoal size and the sociobiology of predator evasion behaviour in minnow shoals. Proc. R. SOC.Lond. B . 229, 439-465. Magurran, A. E., and Seghers, B. H. (1990a). Population differences in the schooling behaviour of newborn guppies, Poecilia reticulata. Ethology 84, 334-342. Magurran, A. E., and Seghers, B. H. (1990b).Population differences in predator recognition and attack cone avoidance in the guppy, Poecilia reticulata. Anim. Behau. 40,443-452. Magurran, A. E., and Seghers, B. H. (1990~).Risk sensitive courtship in the guppy (Poecilia reticulata). Behaviour 1l2, 194-201. Magurran, A. E., and Seghers, B. H. (1991). Variation in schooling and aggression amongst guppy (Poecilia reticulata) populations in Trinidad. Behaviour 118, 2 14-234. Magurran, A. E., and Seghers, B. H. (1994a). Sexual conflict as a consequence of ecology: Evidence from guppy, Poecilia reticulata, populations in Trinidad. Proc. R. SOC.Lond. B 255, 31-36. Magurran, A. E., and Seghers, B. H. (1994b). Predator inspection behaviour covaries with schooling tendency amongst wild guppy, Poecilia reticulata, populations in Trinidad. Behaviour 128, 121-134. Magurran, A. E., and Seghers, B. H. (1994~).A cost of sexual harassment in the guppy, Poecilia reticulata. Proc. R. SOC.Lond. B. 258, 89-92. Magurran, A. E., Seghers, B. H., Carvalho, G. R., and Shaw, P. W. (1992). Behavioural consequences of an artificial transplant of guppies, Poecilia reticulata in Trinidad: Evidence for the evolution of anti-predator behaviour in the wild. Proc. R. SOC.Lond. B. 248, 117-122. Magurran, A. E., Seghers, B. H., Carvalho, G. R., and Shaw, P. W. (1993). Evolution of adaptive variation in antipredator behaviour Mar. Behav. Physiol., 23, 29-44. Magurran, A. E., Seghers, B. H., Shaw, P. W., and Carvalho, G. R. (1994). Schooling Fish . Biol. 45,401-406. preferences for familiar fish in the guppy, Poecilia reticulata. .I Maitland, P. S., and Campbell, R. N. (1992). “Freshwater Fishes.” Harper Collins, London. Mattingley, H. T., and Butler, M. J. (1994). Laboratory predation on the Trinidadian guppy: Implications for the size-selective predation hypothesis and guppy life history evolution. Oikos 69, 54-64. May, R. M. (l99Oa). How many species? Phil. Trans. R. SOC.Lond. B. 330,293-304. May, R. M. (1990b). Taxonomy as destiny. Nature (London) 347, 129-130. Mayr, E. (1963). “Animal Species and Evolution.” Harvard Univ. Press, Cambridge, Massachusetts. McPhail, J. D. (1994). Speciation and the evolution of reproductive isolation in the sticklebacks (Gasterosteus)of south-western British Columbia. In “The Evolutionary Biology of the Threespine Stickleback” (M. A. Bell and S. A. Foster, eds.). pp. 399-437. Oxford University Press, Oxford.

200

A. E. MAGURRAN ET AL.

Meyer, A., Kocher, T. D., Basasibwaki, P., and Wilson, A. C. (1990). Monophyletic origin of Lake Victoria cichlid fishes suggested by mitochondria1 DNA sequences. Nature (London) 347, 550-553. Milinski, M. (1987). Tit for Tat in sticklebacks and the evolution of cooperation. Nature (London) 325,433-435. Milinski, M., Kiilling, D., and Kettler, R. (1990a). Tit for Tat: Sticklebacks trusting a cooperating partner. Behav. Ecol. 1, 7-10. Milinski, M., Pfluger, D., Kiilling, D.. and Kettler, R. (1990b). Do sticklebacks cooperate repeatedly in reciprocal pairs? Beh. Ecol. Sociobiol. 27, 17-21. Murphy, K. E., and Pitcher, T. J. (1991). Individual behavioural strategies associated with predator inspection in minnow shoals. Ethology 88, 307-319. Nei, M. (1972). Genetic distance between populations. Am. Nut. 106, 283-292. Nei, M. (1973). Analysis of gene diversity in subdivided populations. Proc. Natl. Acad. Sci., U.S.A. 70,3321-3323. Nei, M. (1975). “Molecular Population Genetics and Evolution.” North Holland, Amsterdam. Nei, M., Maruyama, T., and Chakraborty, R. (1975). The bottleneck effect and genetic variability in populations. Evolution 29, 1-10. Neill, S. R. S. J., and Cullen, M. (1974). Experiments on whether schooling by their prey affects the hunting behaviour of cephalopod and fish predators. J . Zoo/. London. 172, 549-569. Ohguchi, 0 . (1981). Prey density and the selection against oddity by three-spined sticklebacks. Z . Tierpsychol. Supplement 23, 1-78. Parenti, L. (1981). A phylogenetic and biogeographic analysis of cyprinodontiform fishes (Teleostei, Antherinomorpha). Bull. Amer. Mus.Natur. Hisr. 168, 335-557. Parenti, L., and Rauchenberger, M. (1989). Systematic overview of the Poeciliines. In “Ecology and Evolution of Livebearing Fishes (Poeciliidae)” ( G . K. Meffe and F. F. Snelson, eds.), pp. 3-12. Prentice Hall, Englewood Cliffs, New Jersey. Pitcher, T. J. (1993). Who dares, wins: The function and evolution of predator inspection behaviour in shoaling fish. Netherlands Journal of Zoology 42, 371-391. Pitcher, T. J., and Parrish, J. K. (1993). Functions of shoaling behaviour in teleosts. I n “Behaviour of Teleost Fishes” (T. J. Pitcher, ed.), pp. 363-439. Chapman and Hall, London. Pomiankowski, A., and Sheridan, L. (1994). Linked sexiness and choosiness. Trends Ecol. E v o ~ .9, 242-244. Reimchen, T. E. (1994). Predators and morphological evolution in threespine stickleback. I n “The Evolutionary Biology of the Threespine Stickleback” (M. A. Bell and S. A. Foster, eds.), pp. 240-276. Oxford University Press, Oxford. Reynolds, J. D., and Gross, M. R. (1992). Female mate preference enhances offspring growth and reproduction in a fish, Poecilia reticulata. Proc. R . Lond. B. 250, 57-62. Reynolds, J. D., Gross, M. R., and Coombs, M. J. (1993). Environmental conditions and male morphology determine alternative mating behaviour in Trinidadian guppies. Anim. Behav. 45, 145-152. Reznick, D. N. (1982). The impact of predation on life-history evolution in Trinidadian guppies: Genetic basis of observed life history patterns. Evolution 36, 1236-1259. Reznick, D. N. (1989). Life-history evolution inguppies: 2. Repeatability of field observations and the effects of season on life histories. Evolution 43, 1285-1297. Reznick, D. N., and Bryga, H. (1987). Life-history evolution in guppies (Poecilia reticulafa): 1. Phenotypic and genetic changes in an introduction experiment. Evolution 41, 1370- 1385.

GUPPY POPULATIONS IN TRINIDAD

20 1

Reznick, D. N., Bryga, H., and Endler, J. A. (1990). Experimentally induced life-history evolution in a natural population. Nature (London) 346, 357-359. Reznick, D. N., and Endler, J. A. (1982). The impact of predation on life history evolution in Trinidadian guppies (Poecilia reticulata). Evolution 36, 160-177. Reznick, D. N., and Miles, D. B. (1989). A review of life history patterns in poeciliid fishes. I n “Ecology and Evolution of Livebearing Fishes (Poeciliidae)” (G. K. Meffe and F. F. Snelson, eds.), pp. 125-148. Prentice Hall, Englewood Cliffs, New Jersey. Rice, W. R., and Hostert, E. E. (1993). Laboratory experiments on speciation: What have we learned in 40 years? Evolufion 27, 1637-1653. Ridgway, M. S., and McPhail, J. D. (1984). Ecology and evolution of sympatric sticklebacks (Gasterosteus):Mate choice and reproductive isolation in the Enos Lake species pair. Can. J . Zool. 62, 1813-1818. Rodd, F. H. (1994). Phenotypic plasticity in the life history traits and sexual behaviour of Trinidadian guppies (Poecilia reticulata) in response to their social environment. Unpublished doctoral dissertation, York University, Canada. Rosen, D. E., and Bailey, R. M. (1963). The poeciliid fishes (Cyprinodontiformes), their structure, zoogeography, and systematics. Bull. Am. Mus. Nat. Hist. I%, 1-176. Ruzzante, D. E., and Doyle, R. W. (1991). Rapid behavioral changes in medaka (Oryzias latipes) caused by selection for competitive and noncompetitive growth. Evolution 45, 1936- 1946.

Ruzzante, D. E., and Doyle R. W. (1993). Evolution of social behaviour in a resource-rich, structured environment: Selection experiments with medaka (Oryzias latipes).Evolution 41,456-470.

Schluter, D., and Ricklefs, R. E. (1993). Species diversity: An introduction to the problem. I n “Species Diversity in Ecological Communities” (R. E. Ricklefs and D. Schulter, eds.). pp. 1-10. Chicago University Press, Chicago. Seghers, B. H. (1973). An analysis of geographic variation in the antipredator adaptations of the guppy, Poecilia reticulata. Unpublished doctoral dissertation, University of British Columbia. Seghers, B. H. (1974). Schooling behavior in the guppy (Poecilia reticulata): An evolutionary response to predation. Evolution 28,486-489. Seghers, B. H. (1978). Feeding behavior and terrestrial locomotion in the cyprinodontid fish, Rivulus hartii (Boulenger). Verh. Internat. Verein. Limnol. 20,2055-2059. Seghers, B. H. (1992). The rivers of northern Trinidad: Conservation of fish communities for research. I n “River Conservation and Management” (P. J. Boon, P. Calow, and G. E. Petts, eds.), pp. 81-90. Wiley, London. Shaw, E. (1970). Schooling in fishes: Critique and review. I n “Development and Evolution of Behavior” (L. R. Aronson, B. Tobach, D. S. Lehrman, and J. S. Rosenblatt, eds.), pp. 452-480. W. H. Freeman, San Francisco. Shaw, P. W., Carvalho, G. R., Magurran, A. E., and Seghers, B. H. (1991). Population differentiation in Trinidadian guppies (Poecilia reticulata): Patterns and problems. J. Fish. Biol. 39 (supplement A), 203-209. Shaw, P. W., Carvalho, G. R., Magurran, A. E., and Seghers, B. H. (1994). Factors affecting the distribution of genetic variability in a freshwater fish, the guppy Poecilia reticulata. J. Fish Biol. 45, 875-888. Shaw, P. W., Carvalho, G. R., Seghers, B. H., and Magurran, A. E. (1992). Genetic consequences of an artificial transplant of guppies, Poecilia reticulata in N. Trinidad. P r 0 c . R . SOC.Lond. B. 248, 111-116. Slatkin, M. (1993). Isolation by distance in equilibrium and non-equilibrium populations. Evolution 41, 264-279.

202

A. E. MAGURRAN

ET AL.

Smuts, B. B., and Smuts, R. W. (1993). Male aggression and sexual coercion of females in nonhuman primates and other mammals: Evidence and theoretical implications. Adu. Study Beh. 22, 1-63. Snelson, F. F. (1989). Social and environmental control of life history traits in poeciliid fishes. I n “Ecology and Evolution in Livebearing Fishes” (G. K. Meffe and F. F. Snelson, eds.), pp. 149-162. Prentice Hall, Englewood Cliffs, New Jersey. Stoner, G., and Breden, F. (1988). Phenotypic differentiation in female preference related to geographic variation in predation risk in the Trinidad guppy (Poecilia reticulata). Beh. Ecol. Sociobiol. 22, 285-291. Straws, R. E. (19%). Predation and life-history variation in Poecilia reticulata (Cyprinodontiformes: Poeciliidae). Enu. Biol. Fishes 27, 121-130. Theodarkis, C. W. (1989). Size segregation and the effects of oddity on predation risk in minnow school. Anim. Behau. 38,496-502. Thibault, R. E., and Schultz. R. J. (1978). Reproductive adaptations among viviparous fishes (Cyprinodontiformes: Poeciliidae). Evolution 32, 320-333. Thompson, J. D. (1991). Phenotypic plasticity as a component of evolutionary change. Trends Ecol. Euol. 6, 246-249. Thorpe, J. P. (1983). Enzyme variation, genetic distance and evolutionary divergence in relation to levels of taxonomic separation. I n “Protein Polymorphism: Adaptive and Taxonomic Significance” (G.S. Oxford and D. Rollinson, eds.), pp. 131-152. Academic Press, New York. Tinbergen, N. (1963). On aims and methods of ethology. Z. Tierpsychol. 20,410-433. Trexler, J. C. (1989a). Phenotypic plasticity in poeciliid life histories. In “Ecology and Evolution of Livebearing Fishes (Poeciliidae)” (G. K. Meffe and F. F. Snelson, eds.), pp. 201-214. Prentice Hall, Englewood Cliffs, New Jersey. Trexler, J. C. (1989b). Phenotypic plasticity in the sailfin molly, Poecilia latipinna (Pisces: Poeciliidae). 1, Field experiments. Evolution 44, 143-156. Trexler, J. C. (1989~).Phenotypic plasticity in the sailfin molly, Poecilia latipinna (Pisces: Poeciliidae). 2. Laboratory experiments. Evolution 44, 157- 167. Via, S . (1993). Adaptive phenotypic plasticity: Target or by-product of selection in a variable environment? Am. Nat. 142, 352-365. Waples, R. S. (1987). A multispecies approach to the analysis of gene flow in marine shore fishes. Evolution 41, 385-400. Ward, R. D., Woodwark, M., and Skibinski, D. 0. F. (1994). A comparison of genetic diversity levels in marine, freshwater and anadromous fishes. J . Fish Biol. 44,213-232. West-Eberhard, M. J. (1989). Phenotypic plasticity and the origins of diversity. Ann. Rev. EcoI. SysI. 20, 249-218. Wilson, D. S. (1989). The diversification of single gene pools by density- and frequencydependent selection. I n “Speciation and Its Consequences” (D. Otte and J. A. Endler, eds.), pp. 366-385. Sinauer, Cambridge, Massachusetts. Winemiller, K. 0. (1993). Seasonality of reproduction of livebem‘ng fishes in tropical rainforest streams. Oecologia 95, 266-276. Winer, L., and Boos, H. E. (1991). Agouti tozandoli: Faunain thedictionary ofTrinbagonian. Living World: Journal of the Trinidad and Tobago Field Naturalists’ Club 1991-1992, 25-28.

Winge, 0. (1937). Succession of broods in Lebistes. Nature (London) 140, 467. Wourms, J. P. (1981). Viviparity: The maternal foetal relationship in fishes. Amer. Zool. 21,473-575.

Zhuikov, A. Y. (1993). Avoidance learning and aggression in guppies. Anim. Behau. 45, 825-826.

ADVANCES IN THE STUDY OF BEHAVIOR. VOL. 24

Sociality, Group Size, and Reproductive Suppression among Carnivores SCOTTCREEL*?$AND DAVIDMACDONALD* *WILDLIFE CONSERVATION RESEARCHUNIT DEPARTMENT OF ZOOLOGY UNIVERSITY OF OXFORD OXFORD,ENGLAND ?FRANKFURT ZOOLOGICAL SOCIETY

SELOUS WILD DOGPROJECT SELOUS CONSERVATION PROGRAMME DARES SALAAM, TANZANIA

$FIELDRESEARCHCENTERFOR ECOLOGY AND ETHOLOGY ROCKEFELLER UNIVERSITY MILLBROOK, NEWYORK I. INTRODUCTION The order Carnivora is diverse in many respects. Carnivore body mass ranges from 80-g weasels (Mustela spp.) to 800-kg polar bears (Thelurctos muritimus), spanning five orders of magnitude (Ewer, 1973; Macdonald, 1992). Imagine a pack of 10,OOO weasels and you see the range of body mass found among carnivores. Their shapes and habits vary from the weasel, adapted for pursuit of rodents through tunnels, to the binturong (Nundinia binotara), a slow-moving arboreal frugivore, to the African wild dog (Lycaon pictus), adapted to agile pursuit of antelope through bushveld, to the explosive power of the lion (Punthera leo), to the massive bulk of the polar bear. Parallel to their physical diversity is broad variation in ecology. Fissiped carnivores (which exclude the seals and sea lions) can be found on and under land, among the trees, and in fresh- or seawater. They occur in habitats as open as Arctic tundra and shortgrass savannah (e.g., wolves Canis lupus; cheetahs Acinonyx jubatus) and in those as closed as rainforest (e.g., palm civets Paradoxurus hermaphroditus; coatis Nasua naricu). Although most carnivores share a pair of carnassial teeth, designed as a meat-shearing edge and inherited from Eocene ancestors (Martin, 1989), 203

Copyright 0 1995 by Academic Press, Inc. All rights of reproduction in any form reserved.

204

SCOTT CREEL AND DAVID MACDONALD

contemporary carnivores’ diets include nearly pure carnivory (many felids and mustelids), omnivory (ursids, many canids and solitary herpestids), insectivory (social herpestids), frugivory (some viverrids and the kinkajou, Potos j7auus), and herbivory (ailurids and ailuropodids, Ewer, 1973; van Valkenburgh, 1989). Recorded home range sizes vary from less than 0.04 km2for female stoats (Mustela erminea) in dense populations (Erlinge and Sandell, 1986) to more than 1,000 km2 in low-density African wild dog populations (Frame, Malcolm, Frame, and van Lawick, 1979). Intraspecific variation in their behavior is also great: the home ranges of red foxes, Vulpes vulpes, span between 10 and 1,0oO ha (Macdonald, 1983), the group sizes of Eurasian badgers vary between 2 and 29 individuals (Da Silva, Macdonald, and Evans, 1993), and goldenjackals (Canisaureus, Macdonald, 1979) or spotted hyenas (Crocuta crocufa,Kruuk, 1972) living in big groups on small territories adopt patterns of scent marking not seen among their conspecifics at lower population densities. Our aim in this chapter is to seek generalizations that organize the diversity within the Carnivora. We focus on carnivore sociality, and ask what selective pressures affect interspecific variation in group size and reproductive suppression, using phylogenetic regression (Grafen, 1989, 1992), one of several recently developed comparative methods. We first review the literature on carnivore societies, deriving several working definitions that categorize their social behavior. These categories then form the basis of three comparative analyses, the results of which shed light not only on the evolution of carnivore social organization, but also on some strengths and weaknesses of the method. The primary goal of our comparative analyses is to test the relative importance of various hypotheses for the evolution of sociality and reproductive suppression across the order. Before presenting the results of our analyses we review the following: aspects of social organization, the evolution of social breeding, and the evolution of reproductive suppression.

11. ASPECTSOF SOCIAL ORGANIZATION

A.

SOCIAL AND NONSOCIAL BREEDING

Given their physical and ecological diversity, it is not surprising that carnivores exhibit considerable variation in social organization. The majority of species (8540%) are solitary outside of mating periods or temporary associations between mothers and offspring (reviewed by Gittleman, 1989; Sandell, 1989; Macdonald, 1992). Considered family by family, most felids, mustelids, vivemds, ailurids, ailuropodids, and large herpestids

REPRODUCTIVE SUPPRESSION AMONG CARNIVORES

205

are solitary, in the sense that they spend most of their time alone. Most species of ursids and procyonids are also solitary under most conditions, although they may form temporary feeding groups at concentrations of food (e.g., brown bears, Ursus arctos, at spawning runs of salmon). A few species in these families show more permanent sociality (e.g., Eurasian badgers, Kruuk, 1978; female coatis, Russell, 1983). Monogamous pairs are the basic social units among canids (Kleiman and Malcolm, 1981), with a tendency toward larger groups at the two extremes of body size (Moehlman, 1986). The larger species tend to polyandry (Moehlman, 1986), as in grey wolves (Packard, Mech, and Seal, 1983), Ethiopian wolves, Canis simensis (Sillero-Zubiri, 1994), and wild dogs (Kuhme, 1965; Frame et al., 1979).The smaller species tend toward polygyny, as in red foxes (Vulpes vulpes, Macdonald, 1979, 1980) or stricter monogamy, as in Blanford’s foxes (Vulpes cana, Geffen and Macdonald, 1992). Breeding units larger than monogamous pairs are the general rule among smaller herpestids and larger hyaenids (Rood, 1986; Mills, 1989a, 1989b), and are seen in a few representatives of the predominantly solitary families (e.g., lions and coatis; see Section 11,B). As studies have accumulated for carnivores traditionally classified as solitary, incipient sociality has been detected among some species (e.g., white-tailed mongooses, Zchneumia albicauda, Waser and Waser, 1985; large felids, Sunquist and Sunquist, 1989; black-tipped mongooses, Herpestes sanguineus, Waser, Creel, and Lucas, 1994). Because many carnivores remain unstudied in the field, we may have seen only the tip of an iceberg of marginally social species: to be seen alone is an inadequate criterion for asociality. Descriptions of mating systems are further complicated as observations (e.g., aardwolf, Proteles cristatus, Richardson, 1987; Ethiopian wolves, Sillero-Zubiri, Gottelli, and Macdonald, submitted) and genetic studies (e.g., grey wolves, Lehman et al., 1992; badgers, Evans, Macdonald, and Cheeseman, 1989; Ethiopian wolves, Gottelli et al., 1994) disclose that extra-pair (and off-territory) copulations may be commonplace. Social monogamy may not imply genetic monogamy. Group size itself can be an ambiguous term (Gittleman, 1989). Foraging group size may differ from feeding group size (as in spotted hyenas, Crocuta crocuta, Kruuk, 1972;Mills, 1989a, 1989b;or lions, Packer, 1986), and often both are different from the size of the social group stable over the long term, dubbed the population group size by Gittleman (1989). For instance, brown hyena clans in the Kalahari average four adults, but 99% of animals seen foraging are solitary (Mills, 1989a). A similar situation occurs in red foxes, which generally travel alone, but are organized spatially into groups of one male and up to five adult females (Macdonald,

206

SCOTT CREEL AND DAVID MACDONALD

1980, 1987). In lions and male cheetahs, members of a long-term coalition may separate temporarily (Schaller, 1972; Car0 and Collins, 1987). In the comparative analyses of this chapter, we focus on breeding group size, defined as the number of adults that ( 1 ) share a home range, and (2) are in social contact throughout the period of mating and raising young. Breeding group size is equivalent to population group size for most species. We elaborate on our definition below. First, we define shared home ranges to include only those cases in which individuals have similar, essentially identical ranges (spatial groups as defined by Macdonald, 1983). We distinguish shared home ranges from encompassinghome ranges, where the home range of one individual (usually a male) completely encompasses that of another (usually a female), but also includes part or all of other (usually females’) home ranges: in the case of encompassing home ranges, we do not consider all individuals to belong to a single breeding group. Second, by specifying that individuals be in social contact, we distinguish between cases in which a stable monogamous pair mates and provides biparental care (as in many canids), and cases in which a pair may share a home range but the female cares for the young solitarily (as occurs when a male is monogamously mated in predominantly polygynous systems, e.g., many mustelids and felids; Sandell, 1989). Monogamy with biparental care clearly defines a breeding group size of two, but we have assigned a breeding group size of one to the case of monogamy without male care. This might appear paradoxical for sexually reproducing mammals, but it is necessary to distinguish between different systems of parental care. We consider social breeding groups to be distinct from both uniparental and biparental breeding groups. We define a social breeding group as one in which at least two adults of one sex breed in the same group with at least one adult of the other sex. Thus, the minimum group size for social breeding is three, reflecting a real biological dichotomy: social breeders exceed the minimum group size for sexual reproduction. Another purpose in defining social breeding in this way is to distinguish between situations in which all offspring on a territory have the same parentage (barring off-territory mating), and more complicated situations in which reproductive competition is likely to occur within the breeding group. Despite its lack of elegance, we call species that are not social breeders “nonsocial breeders” rather than ‘‘solitary breeders,” because the nonsocial category includes monogamous species with biparental care. In defining social breeding we do not distinguish between species in which all group members reproduce and those in which some are reproductively suppressed. Instead, we treat intragroup suppression as a separate dichotomy (see below).

REPRODUCTIVE SUPPRESSION AMONG CARNIVORES

207

Some species. particularly canids, breed sometimes in pairs and other times in larger groups (Malcolm, 1985). Such variation in breeding group size may occur within single populations (e.g., golden and silverbacked jackals (Cunis uureus and C . mesomelus, Moehlman, 1983). For comparative analyses, we have classified species as social if at least one study has reported that social breeding occurs commonly (it was not possible to define common quantitatively). Thus we have pooled obligately social breeders and species with the capacity to breed socially. As an illustration of our distinction between social and nonsocial breeding, and to provide continuity in terminology, we have applied our dichotomy to each of the mammalian breeding systems discussed by Davies (1991: Fig. 9.2). Davies’s categories that fall under our definition of social breeding are polyandry and polygyny in which all members reside on a single territory, including unimale and multimale polygyny, and female defense harem polygyny (if harems are stable over the breeding period). Nonsocial under our dichotomy are obligate monogamy, polygyny, and polyandry in which the multiple sex lives on separate territories, lekking, scramble polygyny and polyandry, and resource defense harem polygyny. Of Davies’s categories, the most common nonsocial type among the carnivores is polygyny with the multiple sex on separate territories (termed exclusive territorial polygyny by Macdonald, 1992, and well illustrated by tigers, Smith and McDougal, 1991). The most common of the social types among carnivores is multimale polygyny, which in the carnivore literature is more often called communal breeding (and illustrated by African wild dogs, Malcolm and Marten, 1982).

B.

BREEDINGGROUPSIZE

Although our definition of social breeding focuses on a real biological dichotomy, it is a simplification of a continuum in breeding group sizes. Breeding group size varies substantially within the order, even within single families (Gittleman, 1989; Macdonald, 1992). For instance, small canids typically breed in groups with few members beyond a monogamous pair (e.g., red foxes, Macdonald, 1979; blackbacked jackals and golden jackals, Moehlman, 1983; coyotes, Cunis latruns, Bekoff and Wells, 1982; cf. Camenzind, 1978; bat-eared foxes, Octocyon megalotis, Nel, Mills, and van Aarde, 1984; Maas, 1993). In contrast, most of the larger canids live in larger breeding groups (e.g., Ethiopian wolves, Cunis simensis, Gottelli and Sillero-Zubiri, 1992; dholes, Cuon alpinus, Johnsingh, 1982; wolves, Mech, 1970; African wild dogs, Kuhme, 1965; Fuller et ul., 1992). Inevitably, there are exceptions: the large maned wolf (23 kg) generally forages alone and is organized spatially in pairs (Dietz, 1984), whereas

208

SCOTT CREEL AND DAVID MACDONALD

the small bush dog, at 7 kg, lives in packs, and the 1.5-kg fennec fox is said to form groups. Large breeding groups typify the smaller members of the Herpestidae (Rood, 1986), for instance dwarf mongooses (Helogale paruula, Rood, 1983), banded mongooses (Mungos mungo, Rood, 1975), cusimanses (Crossarchus spp, Kingdon, 1977), and meerkats (Suricata suricatta, Ewer, 1973; Macdonald, 1986, 1992). The larger hyaenids also breed in relatively large groups, particularly the spotted hyena (Kruuk, 1972; Mills, 1989a, 1989b), with the largest recorded breeding groups among carnivores. Social species within predominantly solitary families sometimes breed in large groups, for example, lions and farm cats (Felis cattus) among the felids (Bygott, Bertram, and Hanby, 1979; Packer, 1986; Macdonald, Apps, Carr, and Kerby, 1987), coatis among the procyonids (Russell, 1983), and some populations of Eurasian badgers among the mustelids (Kruuk, 1978, 1989; Kruuk and Parish, 1982). These species contrast sharply with incipiently social species (e.g., white-tailed mongooses) thought to illustrate an intermediate step of an evolutionary route to welldeveloped sociality (Waser and Waser, 1985; Rood, 1986). C. REPRODUCTIVE SUPPRESSION Although it simplifies variation in group size, the dichotomy between social and nonsocial breeders does not address whether all adults within the group breed, or whether some adults are reproductively suppressed. We define reproductive suppression simply to mean that some individuals beyond the age of sexual maturity do not raise young. We have avoided a more restrictive definition because for most species it is unknown whether suppression is socially imposed or due to individual restraint; we also do not know how suppression is effected. Where it has been studied, suppression appears to be imposed on subordinates (young adults) by dominants (old adults). Although our definition of suppression does not specify socially imposed suppression, that is probably the general mechanism among carnivores (Rood, 1978, 1980; Macdonald, 1979; Bekoff and Wells, 1982; Macdonald and Moehlman, 1982; Malcolm, 1979; Reich, 1981; Moehlman, 1983; Vehrencamp, 1983; Packard et al., 1983; Packard, Seal, Mech, and Plotka, 1985). Failure to raise young might occur in many ways. Among females, physiological suppression of reproductive function can arise through failure of the estrogen buildup that stimulates attractiveness and receptivity to males, failure of the peak in luteinizing hormone that stimulates ovulation, failure of progestogen production by the corpus luteum necessary

REPRODUCTIVE SUPPRESSION AMONG CARNIVORES

209

to establish pregnancy, or failure of ovarian, placental, and uterine production of progestogens necessary to maintain pregnancy. Among males, physiological suppression can occur through insufficient follicle stimulating hormone and luteinizing hormone for spermatogenesis, or insufficient testosterone for normal reproductive behavior. This list is simplified, because only the final level in the endocrine control of each process is described. For instance, low testosterone levels may or may not be the product of low follicle stimulating hormone levels, which in turn depend on levels of gonadotropin releasing hormone. Regulatory feedback is also a consideration. For example, the ovulatory surge of luteinizing hormone may depend on a positive feedback loop between estrogens and mating rate. The few studies of endocrine suppression in female carnivores suggest that if endocrine suppression occurs, preovulatory stages are usually affected (wolves, Packard et al., 1983,1985; African wild dogs, van Heerden and Kuhn, 1985; dwarf mongooses, Creel, Monfort, Wildt, and Waser, 1991; Creel, Creel, Wildt, and Monfort, 1992). Among male carnivores, baseline androgens are not affected by social rank in spotted hyenas (Frank, Davidson, and Smith, 1985), dwarf mongooses (Creel et al., 1992; Creel, Wildt, and Monfort, 1993), or wolves (Packard et al., 1985). However male wolves with higher mating rates had greater circulating testosterone response when injected with LHRH (Packard etal., 1985). In Eurasian badgers, social groups contain several natal males and one immigrant male; both categories of male have high plasma testosterone titers early in the breeding season, but later in the season the immigrant males have significantly higher titers than do the natal males (and immigrants retain scrota1 testes, whereas those of natal males ascend into the abdomen) (Woodroffe and Macdonald, in press). Where the endocrine aspects of subordinates’ reproduction are normal, reproductive suppression may still be caused by behavioral mechanisms. Dominants may directly prevent subordinates from mating (wolves, Packard et al., 1983; African wild dogs, Reich, 1981; Malcolm, 1979; dwarf mongooses, Rasa, 1973; Rood, 1980; red foxes, Macdonald, 1979, 1980). If subordinates mate and produce young, dominants may kill subordinates’ young (African wild dogs, Frame et al., 1979; Malcolm and Marten, 1982; dwarf mongooses, Rasa, 1973; Rood, 1980; dingoes, Corbett, 1988). Behavioral and endocrine mechanisms of reproductive suppression can be closely intertwined. For instance, low social rank leads to both low baseline estrogen levels and low mating rates in dwarf mongoose females (Creel et al., 1992). It is sometimes suggested that harassment by dominants leads to social stress in subordinates, which in turn affects reproductive endocrinology or

210

SCO'lT CREEL AND DAVID MACDONALD

behavior (Packard et al., 1985). Although this is mechanistically plausible, the only study among carnivores that has demonstrated chronic elevation of corticosteroid (stress) hormones among subordinates was conducted with captive wolves, and involved repeated immobilization that led to social instability and fatal fighting (Packard et al., 1985). Under these conditions, stresses external to the normal mechanisms of suppression may arise. Among dwarf mongooses, physiological reproductive suppression is not associated with elevated corticosteroids (Creel et al., 1992). In addition to suppressing reproduction, chronic elevation of corticosteroids has generally debilitating effects (e.g., suppression of immune and histamine responses and neural death; reviewed by Sapolsky, 1992). It therefore seems likely that natural selection would favor direct mechanisms of reproductive suppression not mediated by corticosteroids. Subordinates faced with harrassment sufficient to curtail their reproduction via the stress response would benefit from a direct mechanism of suppression within the hypothalamic-pituitary-gonadal axis, because this direct mechanism would remove the inducement for dominants to place them under stress. If true, stress-mediated reproductive suppression would be evolutionarily unstable. Carnivores show wide variation in the extent of reproductive suppression within groups. At one extreme lie species in which all adults breed in each season (or at comparable rates for asynchronous and aseasonal breeders), for example, lions (Schaller, 1972; Packer and Pusey, 1983), coatis (Russell, 1983), and feral cats (Macdonald et al., 1987). Next along the continuum are species in wbich most adults in a group breed, but a few (usually adults not far beyond the age of maturity) do not. This uncommon system may be seen in the banded mongoose (Rood, 1978; Waser, Elliott, Creel, and Creel, in press). In at least one population of badgers most older females conceived, but a variable proportion went on to rear young depending on an interplay between their condition and social status as determined by interannual variation in food availability (Woodroffe and Macdonald, in press). In this population, individuals that bred in one year did not necessarily breed in subsequent years (Da Silva et al., 1993). At the other extreme are species in which a single pair monopolizes most or all reproduction. Species typified by reproductive suppression of subordinates include wolves (Packard et al., 1983, 1985), African wild dogs (Malcolm and Marten, 1982; van Heerden and Kuhn, 1989, Ethiopian wolves (Sillero-Zubiri and Gottelli, 1992; Sillero-Zubiri et al., submitted), dwarfmongooses (Rood, 1980; Creel etal., 1992; Keane et al., 1994), and most small canids (reviewed by Macdonald and Moehlman, 1982; Moehlman, 1986, 1989). It is unlikely that any species shows complete reproductive suppression

REPRODUCTIVE SUPPRESSION AMONG CARNIVORES

21 1

of all subordinates under all conditions. In most species characterized by reproductive suppression, subordinates reproduce at low rates (e.g., dwarf mongooses, Rood, 1980; Creel and Waser, 1991; wolves, Packard et al., 1983; African wild dogs, Malcolm, 1979; Reich, 1981; Fuller and Kat, 1990; red foxes, Macdonald, 1987). Subordinates outnumber dominants in many species with reproductive suppression (whenever breeding group size exceeds four), so that reproduction by subordinates can constitute a surprisingly large percentage of all offspring. Among dwarf mongooses, subordinate pregnancies comprised 27% of the total (Creel and Waser, 1991, 1994), although genetic studies indicate that subordinates produce a smaller share of the young raised (15-24%, Keane et al., 1994). In African wild dogs, 38% of packs in Masai Mara Reserve held more than one breeding female (Fuller et al., 1992). Many genetic studies of parentage are now under way, which will illuminate the true degree of reproductive suppression for many carnivores. Where genetic data now exist, they reinforce the caveat that breeding systems categorized as with reproductive suppression or with equal reproductive success may be more complicated. In lions, behavioral observations indicated that the first male to encounter an estrous female could normally monopolize matings with her, even when other males encountered the pair (Packer and Pusey, 1983). This justified the apparently reasonable assumption that male reproductive success has little bias within a male coalition (Packer et al., 1988). Subsequently, DNA fingerprinting revealed that bias in male reproductive success within coalitions can be considerable, despite the ownership convention for access to females, and despite near equality in female reproductive success (Packer, Gilbert, Pusey, and O’Bnen, 1991). In dwarf mongooses, DNA fingerprints confirmed that reproductive suppression is not absolute (Keane et al., 1994). Among females, the probability of producing young was directly related to social rank (and thus age), matching predictions from inclusive fitness equalities (Creel and Waser, 1991). Patterns of maternity did not match perfectly with patterns of pregnancy; subordinate pregnancies formed 27% of the total, but only 15% of offspring raised had subordinate mothers. Thus, even documentation of pregnancies may not be entirely accurate as a measure of the degree of reproductive suppression. Multiple paternity within litters has been found in all of the carnivores for which we are aware of genetic tests of paternity (lions, Packer et al., 1991; dwarf mongooses, Keane et al., 1994; Eurasian badgers, Da Silva et al., 1993; domestic dogs and cats, Georges, Lequarre, Castelli, Haset, and Vassart, 1988, Ethiopian wolf, Gottelli et al., 1994). This complicates assessment of the degree of reproductive suppression considerably. In

212

SCOTT CREEL A N D DAVID MACDONALD

lions, unsuspected within-group bias in male reproductive success was detected. On the other hand, bias in male dwarf mongoose reproductive success was less than previously assumed, and 25% of offspring had subordinate fathers. The effects of suppression on reproductive success are complicated by the growing realization that off-territory matings may be common in some species. For example, dominant female Ethiopian wolves appear to mate only with the dominant male within their own group, but mate with males of any status from neighboring groups (Sillero-Zubiri et al., in press). Similar complications may be widespread: territorial male red foxes make frequent excursions beyond their territories during the mating season, during which itinerant males also make incursions into territories (Macdonald, 1987), and the genetic consequences of such wandering in superficially unimale systems remain to be determined. However, Richardson’s (1987) discovery that among aardwolves a local despot may father cubs in several adjoining territories occupied by seemingly monogamous pairs illustrates the potential disparity between spatial or social patterns and genetic ones. Interspecific comparisons of the degree of reproductive suppression could be based on the percentage of the adult population that overtly fails to breed. However, in our interspecific comparative analyses, we have had to treat reproductive suppression dichotomously. We recorded socially breeding species as with reproductive suppression if studies noted that a single female within a group normally monopolized reproduction, or if nonbreeding adult helpers were reported (e.g., all canids except the bateared fox, Macdonald and Moehlman, 1982; Nel et al., 1984). Other social breeders and all nonsocial breeders were recorded as without reproductive suppression. (Because nonsocial breeders were recorded as without suppression, in our quantitative analyses of reproductive suppression we have controlled for sociality.) Our classification has two major limitations. First, it dichotomizes a continuous variable. At present, that is unavoidable, because quantitative data on individual reproductive success are limited, and are reported in ways that are not directly comparable. In short, no quantitative measure of bias in reproductive success is available for a large sample of species. Second, we base our classification on females only. Patterns of reproduction in females are commonly reported because pregnancy can be seen in the field, while patterns of male reproduction are less well known. Differences in the degree of reproductive suppression between the sexes within a single species do exist (e.g., dwarfmongooses, Creel et al., 1992), but the frequency and extent of sex differences remain to be seen.

REPRODUCTIVE SUPPRESSION AMONG CARNIVORES

213

111. THEEVOLUTION OF SOCIAL BREEDINGA N D GROUPSIZE

Several bodies of theory pertain to the evolution of sociality in carnivores. Some of these were initially developed with specific reference to carnivores (Kruuk, 1975; Lamprecht, 1978a, 1981; Macdonald, 1983) or to single families within the Carnivora (Gorman, 1979; Waser and Waser, 1985; Moehlman, 1986; Packer, 1986; Rood, 1986; Macdonald and Carr, 1989). Among these carnivore-specifichypotheses, those that pertain especially to small carnivores deal primarily with the distribution and renewal rates of food, and how these properties might affect the costs of tolerating conspecifics on a territory (Kruuk 1978; Macdonald, 1980, 1983; von Schantz, 1981; Sandell, 1989; Waser, 1981; Waser and Waser, 1985). Hypotheses directed at large carnivores have focused on potential benefits associated with cooperative hunting of large prey (Schaller, 1972; Kruuk, 1975);Mills, 1985; Packer et al., 1991; Caro, 1994), and on interspecific and intraspecific competition at kills (Kruuk, 1975; Lamprecht, 1981 ; Packer, 1986). Inter- and intraspecific predation have been invoked as factors selecting for sociality in carnivores as small as mongooses (Gorman, 1979; Rood, 1986) and as large as lions (Packer, 1986). Not surprisingly, many general hypotheses for the evolution of sociality (e.g., Alexander, 1974) have the same relevance for carnivores as for other taxa (Gittleman, 1989). For instance, the energetic costs of reproduction may influence the benefits of alloparental care and thus influence the evolution of social breeding (Kleiman, 1977; Gittleman, 1985a; Moehlman, 1986; Clutton-Brock, 1991). Exploitation competition may lead to increases in group size (Wrangham,Gittleman, and Chapman, 1993). Habitat saturation (Stacey, 1979), risky dispersal (Waser and Jones, 1983; Vehrencamp, 1983), or a biased sex ratio (Malcolm, 1979; Brown, 1987; Emlen, 1991) can all shift the evolutionary balance in favor of nondispersal of young, leading to a social breeding system (Brown, 1974). Below, we review five general mechanisms or conditionsthat may select for sociality in carnivores. Two general conditions either reduce the costs of tolerating conspecifics, or constrain the alternatives to tolerance: (1) Abundant prey (or other resources), rich or variable prey patches, or rapid prey renewal lead to low costs of tolerating conspecifics, in terms of foraging success; (2) Disperal opportunities may be constrained by lack of suitable habitat, low availability of mates, or inherent risks (e.g., intraspecific fights), favoring the retention of young past the age of maturity. Three general mechanisms identify benefits of tolerating conspec.$cs: (3) Several benefits may arise through ecological interactions. Hunting in

214

SCOlT CREEL AND DAVID M A C W N A L D

a group may increase foraging success where prey are large or difficult to kill. Groups may fare better in territorial defense and intra- and interspecific competition for food, especially at large kills; (4) Groups may be less vulnerable to intra- and interspecific predation; ( 5 ) Just as male parental care allows females to reproduce successfully where reproduction is risky or energetically costly, larger groups may help to meet the costs of reproduction through alloparental care. Below, we describe these five factors affecting sociality in more detail. A N D RENEWAL A. RESOURCE AVAILABILITY, DISPERSION,

Kruuk’s (1972) study of spotted hyena populations on the Serengeti plains and in the Ngorongoro crater showed that within a single species, resource availability can dramatically affect social organization. Ngorongor0 spotted hyenas are supported by unusually dense ungulate prey, yearround, resulting in stable hyena clans that stay within small, rigorously defended territories (see also Sillero-Zubiriand Gottelli, 1992).In contrast, Serengeti hyenas prey on mostly migratory ungulates. Territories are larger, and territorial defense is more lax. Hyenas from different clans follow prey concentrations for up to 70 km, crossing intervening territories along “hyena pathways” (Kruuk, 1972, 1975; Hofer and East, 1993). Hyenas from different clans often feed together, and they scavenge considerably more than in Ngorongoro. In the southern Kalahari, an arid area with much lower prey density than that of either Serengeti or Ngorongoro, spotted hyenas live in much smaller clans with much larger territories (Mills, 1989a, 1989b). Within a social species, it is not surprising that prey density should affect social organization, or that group size should be correlated with prey density. A logical extension is that sociality itself should be more likely to evolve where an increase in group size above one or two does not harm individuals’ foraging success. An alternative response to increased food availability is to remain solitary and reduce home range size, as in bobcats (Litvaitis, Sherburne, and Bissonette, 1986)and lynx (Ward and Krebs, 1985).Which alternative is favored by natural selection probably depends in part on prey types and hunting methods, for example, stealth in cats versus pursuit in dogs (Macdonald, 1992). Below, we discuss several hypotheses about carnivore sociality centering on resource availability. These hypotheses use logic similar to that underlying ideal free distributions of individuals within a habitat ( Fretwell and Lucas, 1970; Fretwell, 1972).The ideal free distribution can be loosely summarized as follows: if patches (territories) of varying quality exist, and resources within the patch (territory) are depleted more rapidly as

REPRODUCTIVE SUPPRESSION AMONG CARNIVORES

215

group size increases, then equal competitors who are free to choose among patches (territories) will settle so that everyone obtains equal foraging success, and all territories yield equal payoffs. Simple models of such behavior yield the habitat matching rule, in which the number of individuals in a patch is proportional to the quality of the patch (Parker, 1978; Pulliam and Caraco, 1984; Milinski and Parker, 1991). The resource dispersion hypothesis (RDH) argues that, for many carnivores, food resources are spatiotemporally patchy in distribution (Macdonald, 1983; Kruuk and Macdonald, 1985). A minimal home range will be of a size that allows an animal assurance of finding sufficient productive patches within its range (Carr and Macdonald, 1986; Macdonald and Carr, 1989), so that home range size is predicted to depend on the dispersion of patches. The model works rather differently depending on whether several patches, or a single rich patch, are required by an individual during a given feeding period (Carr and Macdonald, 1986). It is simplest to visualize where a single patch provides more food than a single individual can exploit: in this case group mates could be tolerated at little or no cost. Group size is therefore predicted to depend on patch richness (Macdonald, 1983; Carr and Macdonald, 1986; Macdonald nad Carr, 1989). This disarticulation of territory size and group size was first shown for badgers foraging upon patches of earthworms (Kruuk, 1978). The resource dispersion hypothesis pertains particularly to solitarily foraging species (Doncaster and Macdonald, 1992; Woodroffe and Macdonald, 1993). To explain why individuals should act as a cohesive unit or cohabit in a shared den rather than remaining solitary with a shared home range, other benefits must be invoked, for example, common use of rare denning sites (Doncaster and Woodroffe, 1993). The mathematics of the resource dispersion hypothesis have been explored by Bacon, Ball, and Blackwell (1991), and comparisons with similar hypotheses have been made by Bacon and Blackwell (1993; see Lindstrom, 1993) and Woodroffe and Macdonald ( 1993). The resource dispersion hypothesis has been used to explain social breeding in red foxes (Macdonald, 1981), arctic foxes (Hersteinsson and Macdonald, 1982), Blanford’s foxes (Geffen, Heffner, Macdonald, and Ucko, 1992), brown hyenas (Mills, 1982), and otters (Kruuk and Moorhouse, 1991). In Eurasian badgers, the dispersion of food (Kruuk, 1978, 1989; Woodroffe and Macdonald, 1993) and of dens (Doncaster and Woodroffe, 1992) has been invoked to explain social breeding. Substituting prey individual for patch, the hypothesis can be generalized to other carnivores (Macdonald, 1992). Notably, many solitary mustelids and felids feed primarily on vertebrate prey too small to be shared, although there are exceptions (e.g., tigers, Puntheru tigris, Sunquist, 1981).

216

SCOTT CREEL AND DAVID MACDONALD

The RDH can be applied to cases where the emphasis is on spatial heterogeneity of resources, or temporal heterogeneity, or both. An instance of the former case (depicted in Carr and Macdonald, 1986: Figure 8) was invoked by von Schantz (1981, 1984) to explain grouping among Scandinavian red foxes under circumstances where prey (voles) were uniformly distributed in space, but varied in abundance with stages of their population cycle. He suggests that with temporal variation in resources, annual crunch periods dictate the minimal group size for a given home range. During periods of abundance, additional individuals are allowed to join. This constant territory size hypothesis has been used to explain variation in group size in golden jackals (Moehlman, 1989). The spatial and temporal emphases are close conceptual relatives, focusing on two aspects of variability in a single parameter, food availability (Carr and Macdonald, 1986; Woodroffe and Macdonald, 1993).

A third aspect of food availability that may influence the costs of tolerating conspecifics is the renewal rate of a depleted patch (Waser, 1981; Carr and Macdonald, 1986). Where resources renew quickly, there is little cost to sharing, as demonstrated for white-tailed mongooses (Waser and Waser, 1985) and other small nocturnal carnivores in Serengeti (Waser, 1980). Experiments have shown that following complete removal of invertebrates from small (1 m2)grassland plots, density renewed far more quickly than typical patch-revisitationintervals by carnivores that depend on invertebrate prey (Waser, 1980, 1981). A review of sociality among mongooses (Rood, 1986) showed that, as predicted by renewal rates, social species (e.g., dwarf and banded mongooses, Mungos mungo; cusimanses, Crossarchus obscurus) depend primarily on invertebrate prey, while nonsocial species (e.g., great grey mongooses, Herpestes ichneumon; blacktipped mongooses, Herpestes sanguineus) depend primarily on vertebrates. Although quantitative comparative analyses have not previously demonstrated a significant association between prey type and sociality across the entire order (Gittleman, 1989; cf. results below), it is perhaps noteworthy that two of the most solitary families, the mustelids and felids, are also the most dependent on slowly renewing vertebrate Prey Patches that are variably rich from one feeding period to the next (especially if their mean richness is high), periods of plenty, and rapid renewal rates are all conditions that reduce the cost of an individual sharing its home range. These properties of resource availability do not suggest what benefits may accrue to individuals in groups. Possible benefits are discussed in Section III,C, III,D, and II1,E.

REPRODUCTIVE SUPPRESSION AMONG CARNIVORES

217

B. CONSTRAINTS ON DISPERSAL AND RETENTION OF ADULTOFFSPRING One evolutionary route to sociality is through nondispersal of young or natal philopatry (Brown, 1974; Waser and Jones, 1983). In many mammals (reviews in Waser and Jones, 1983, and Emlen, 1991), including most carnivores, social breeding groups do indeed form by offspring remaining beyond the age of sexual maturity (although groups in most socially breeding carnivores include adult immigrants of at least one sex). Nondispersers avoid the mortality risks of dispersal, which can be substantial (Waser et al., 1994). They may inherit a proven territory (Woolfenden and Fitzpatrick, 1984; Brown, 1987) and they may gain indirect fitness by helping relatives on their natal territory (Hamilton, 1963, 1964; Brown, 1987). The particular ecological or demographic constraints that favor nondispersal vary among species. Early work focused on habitat saturation, the situation in which suitable habitat is completely occupied (Selander, 1964; Brown, 1974; Stacey, 1979). Habitat saturation has been invoked to explain social breeding in carnivores (review in Macdonald and Moehlman, 1982), but, as discussed below, other constraints are likely to interact with habitat availability. A serious difficulty exists in testing for habitat saturation in the field: how can suitable habitat be identified, independently of whether it is occupied or not? Furthermore, avian studies have revealed instances in which apparently suitable habitat remained unoccupied despite a nearby pool of potential dispersers (Rabenold, 1984; Stacey and Ligon, 1987). A parallel among carnivores is provided by badger groups from which individuals did not disperse into adjoining vacant territories (Cheeseman, Mallinson, Ryan, and Wilesmith, 1993). Such problems spurred avian population biologists to address continuous variation in habitat quality, rather than classifying habitats as suitable or not (Koenig and Pitelka, 1981 ; Stacey and Ligon, 1987). Emlen (1991)has used the term generalized constraints to broaden the basic concept of habitat saturation: ecological or demographic constraints may be such that dispersal leads to lower reproductive success than does nondispersal. In the broader context of generalized constraints on independent breeding, several factors can affect the expected payoff to dispersers. If vacant habitat is of poor quality, the reproductive success expected from an independent breeding attempt may be sufficiently low to tip the balance in favor of nondispersal (Koenig and Pitelka, 1981). In some social carnivores, helpers are necessary for successful reproduction (e.g., African wild dogs, Reich, 1981; Malcolm and Marten, 1982; dwarf mongooses, Creel, 1990). Lack of a labor force precludes independent breeding in

218

SCOTT CREEL AND DAVID MACDONALD

these species, and would selectively favor nondispersal in any species where helpers increase reproductive success (Brown, 1987: some authors consider the labor force an extended aspect of territory quality [e.g., Woolfenden and Fitzpatrick, 19841). The necessity for a ready-made labor force, together with the benefits of strength in numbers in taking over a new group may explain why in some social carnivores members of at least one sex may disperse in groups (male and female wild dogs, Fuller et al., 1992; male and female meerkats, Macdonald, 1992; female badgers, Woodroffe and Macdonald, 1993; male and female dwarf mongooses, Rood, 1987; male cheetahs, Car0 and Collins, 1987). Nondispersal and dependence on helpers for reproduction may form a positive feedback loop, so that increasing energetic dependence on helpers may create selection for increasing group size (Creel and Creel, 1991) and expansionism (sensu Kruuk and Macdonald, 1985). Finally, a biased sex ratio can limit breeding opportunities for dispersers of the more abundant sex (e.g., wild dogs, Frame et al., 1979; Reich, 1981; Malcolm and Marten, 1982; see also Reyer, 1980). A consequence of all these constraints on dispersal is that individuals may try to hedge their bets. Among canids, for example, male crab-eating foxes may disperse, establish a breeding territory close to their natal range, and, following the death of their mate, return home to tend the next generation of their parents’ cubs (Macdonaldand Courtenay, submitted). Female African wild dogs sometimes return to their natal group following a protracted dispersal, presumably because they failed to find a breeding opportunity (S. Creel, personal observation). Among carnivores, dispersal itself may be inherently risky, regardless of demographic conditions or habitat saturation. Small carnivores are highly susceptible to predation, both by raptors (Rood, 1990) and larger carnivores (Sunquist and Sunquist, 1989;Rood, 1990),and dispersal probably amplifies predation risk (Waser et al., 1994). Intraspecifically, carnivores’ adaptations for killing prey can be put to fatal use in fights between residents and immigrants. Serious injuries or fatalities from intraspecific fights have been documented in species as small as 350-g dwarf mongooses (loss of digits and perhaps limbs, S. Creel, personal observation), 1-kg banded mongooses (fatalities, M. Murray and J. Rood, personal communication), 22-kg African wild dogs (disembowelment. Frame and Frame, 1981),35-kg wolves (fatalities. E. Klinghammer, personal communication) 50-kg cheetahs (fatalities. Caro, 1994)and 165-kg lions (fatalities. Packer, 1986). Most of the above constraints on dispersal are difficult to quantify in practice. Estimates of dispersal risk are almost absent from the literature (Chepko-Sade and Halpin, 1987;Johnson and Gaines, 1990; Waser et al., 1994). Sometimes a habitat constraint is clearcut, such as the restricted

REPRODUCTIVE SUPPRESSION AMONG CARNIVORES

219

Afro-alpine meadows occupied by Ethiopian wolves (Sillero-Zubiri, 1994), but habitat suitability is often difficult to assess objectively, and estimates of habitat quality depend on identifying and measuring all relevant variables. Proxy variables such as rainfall or habitat type are common (e.g., Vehrencamp, Koford, and Bowen, 1988; Waser et al., in press). Beyond these practical problems, interactions between factors are likely to have a strong effect on the overall strength of constraint on dispersal. For instance, where habitat is fairly saturated, the likelihood of aggressive encounters with conspecifics will be high, but the likelihood of remaining without a labor force will be low. Following Emlen (1991), all constraints can be quantitatively incorporated by taking the product of three quantities: (1) the probability of surviving dispersal; (2) the probability of securing a mate; and (3) expected reproductive success following successful dispersal. Where this product (plus any indirect fitness accrued after dispersal; Creel and Waser, 1994) is less than the inclusive fitness accrueing to nondispersers, selection favors remaining in the natal group (e.g., dwarf mongooses of some age-sex classes; Creel and Waser, 1994).

C. GROUPHUNTING A N D INTRASPECIFIC AND INTERSPECIFIC COMPETITION AT KILLS 1 . Group Hunting

Among carnivores that hunt large prey, cooperative hunting has often been proposed to increase hunting success and therefore favor sociality (Ewer, 1973; Kruuk, 1975; Lamprecht, 1978a, 1981; Gittleman, 1989). Three components of hunting success may be improved with an increase in group size: (1) a number of intraspecific studies have shown an increase in captures per hunt with increasing group size, including spotted hyenas hunting wildebeest (Kruuk, 1972), blackbacked and golden jackals hunting Thomson’s gazelle (Wyman, 1967, in Moehlman, 1989), and wild dogs hunting wildebeest or impala (Fanshawe and Fitzgibbon, 1993; Fuller and Kat, 1990); (2) the range of prey that can be taken may also increase. Coyotes prey on elk and mule deer only in large groups (Camenzind, 1978; Bekoff and Wells, 1980, 1982), Kalahari spotted hyenas take adult gemsbok only when hunting in groups (Mills, 1989a), and Etosha spotted hyenas kill zebra only in groups (Gasaway, Mossestad, and Stander, 1991). Solitary lions rarely hunt cape buffalo, an important prey species for larger prides (Scheel and Packer, 1991). In miombo woodland, small packs of African wild dogs rarely kill wildebeest, but larger packs can kill up to five yearling wildebeest in a single hunt (Creel and Creel, in press).

220

SCOlT CREEL AND DAVID MACDONALD

Generalizing, comparative analysis has shown that among meat-eating carnivores, group size is associated with prey size (Gittleman, 1989); (3) finally, the likelihood of making multiple kills may be increased, although data are sparse (lions, Caraco and Wolf, 1975; Packer, 1986; African wild dogs, Fanshawe and Fitzgibbon, 1993; Creel and Creel, in press). These benefits have not been found in all studies. For instance, Kalahari spotted hyenas gain no increase in capture rate with increasing group size when hunting gemsbok calves or wildebeest (Mills, 1985, 1989a, 1989b). Such results have led several authors to emphasize that we should not assume that sociality in large carnivores is favored through cooperative hunting (Kruuk, 1975; Lamprecht, 1978a; Packer, Scheel, and Pusey, 1990; Caro, 1994). Even where benefits to cooperative hunting exist, per capita feeding rates (kg food/individual/day) may be unchanged or decrease with increasing group size (Messier and Barrette, 1982; Packer er al., 1990) so that social hunting actually imposes a cost. Furthermore, hunting effort itself may be a function of group size, confounding correlations between group size, and hunting success. These complications can seriously alter the conclusions of analyses of the costs/benefits of social hunting. Serengeti lions were long held as an example of the benefits of cooperative hunting (Schaller, 1972; Kruuk, 1975; Caraco and Wolf, 1975; van Orsdol, Hanby, and Bygott, 1989, but reanalysis called these benefits into question for Serengeti lions (Packer, 1986). Packer et al. (1990) showed that hunting group size did not affect daily foraging success during seasons of abundance, and that solitaries do better than typical-sized hunting groups during seasons of prey scarcity. Still later, the benefits of social hunting have been confirmed for lions in a more open environment (Etosha National Park; Stander and Albon, 1993). In addition, there may be differences among individuals in the degree to which communal hunting is beneficial or harmful. For example, a poor hunter may benefit from joining a good hunter, although the good hunter suffers a loss (Caro, 1994). So, although increases in capture rate and maximum prey size have been reported for many species (Gittleman, 1989), the energetic profitability of social hunting remains equivocal for all but a few populations (Creel and Creel, in press). 2 . Competition at Kills Large carnivores often kill relatively large prey, which can take a considerable time to eat; tigers may revisit kills over several days (Sunquist, 1981). In open habitats such as savannah or tree-savannah, it is common for a kill to be detected by other carnivores before it is consumed (Packer, 1986; Mills, 1989a; Gasaway el al., 1991). Spotted hyenas in the southern Kalahari detect prey by smell over a mean distance of 1.1 km, and by

REPRODUCTIVE SUPPRESSION AMONG CARNIVORES

22 1

sound over a mean distance of 2.4 km (Mills, 1989a). Many carnivores watch vultures and approach quickly when a vulture makes its characteristic plummet to a kill (Packer, 1986). Because of intra- and interspecific competitors’ abilities to detect carcasses, competition for kills is common, particularly in open habitats or high-density populations (Sunquist and Sunquist, 1989). Data on the percentage of kills lost through direct interactions at kills are available for a guild of predators on the shortgrass plains of southeastern Serengeti. Blackbacked jackals lost up to 30% of Thomson’s gazelle kills, primarily to spotted hyenas (Lamprecht, 1978b). Wild dogs lost portions of 86% of their kills to spotted hyenas (Fanshawe and Fitzgibbon, 1993). Cheetahs lost portions of 12% of their kills, primarily to lions (Schaller, 1972). Lions frequently appropriated spotted hyena kills, but even lions lost portions of 44% of large kills to spotted hyenas (Schaller, 1972).That interspecific competition can be serious even for the largest carnivores was confirmed in Chobe NP, where female and young lions lost 20% of their food to spotted hyenas (Cooper, 1991). We present these figures simply to illustrate that interference competition at kills can be considerable. The numbers themselves should be treated with caution, because carcasses often change ownership after most of the profitable flesh has been consumed (Packer, 1986; Mills, 1989a; Fanshawe and Fitzgibbon, 1993). Also, species differ in their abilities to exploit carcasses; a bone that goes unused by a jackal will be split for the marrow by a wild dog, and completely digested by a spotted hyena. The intensity of interspecific scavengingalso varies with habitat. For example, spotted hyenas make 91% of their own kills in Ngorongoro, 75% in Etosha, 59% in southern Kalahari, and 55% in Serengeti (Kruuk, 1972; Mills, 1989a; Gasaway et al., 1991). Finally, in some interactions, carcasses normally change hands after they have been abandoned, so that direct fighting is not common (Mills, 1989a, 1989b). Where fights over ownership of carcasses are common, competition will favor sociality and evolutionary increases in group size, if larger groups are more likely to win (Kruuk, 1975; Lamprecht, 1981). Intraspecific scavenging may lead to arms-race style evolution in group size, until other factors limiting group size provide balancing selection (Packer, 1986). Considerable evidence suggests that the outcome of competition at kills is generally affected by group size. Despite being lighter by a factor of two, African wild dogs can defend their kills from spotted hyenas, provided the dogs outnumber the hyenas (Estes and Goddard, 1967; Malcolm, 1979; Reich, 1981; Fanshawe and Fitzgibbon, 1993). The outcome of competition between lions and hyenas also depends on relative numbers (Kruuk, 1972; Schaller, 1972; Mills, 1989a). Packs of dholes are capable

222

SCOTT CREEL AND DAVID MACDONALD

of driving single tigers, ten times their size, away from kills (Schaller, 1967). Intraspecific clashes between packs of dwarf mongooses (Rood, 1983), meerkats (Macdonald, 1992), banded mongooses (Rood, 1978), and Ethiopian wolves (Sillero-Zubiri, 1994) almost always result in retreat by the smaller pack. In addition to selecting for larger groups, interspecific competition may simply limit the density or distribution of the weaker competitor (Schoener, 1974). For instance, the regional density of spotted hyenas has been suggested to affect the distribution and abundance of brown hyenas (Mills and Mills, 1982) and African wild dogs (Frame et al., 1979). African wild dogs avoid areas heavily used by lions in the Selous Game Reserve and in Kruger National Park, probably because the dogs’ pups are killed by lions (S. Creel, unpublished data; G. Mills, personal communication). Among solitary carnivores, also, it seems likely that interspecific clashes have widespread influence on species density (e.g., red vs. Arctic foxes, Hersteinsson and Macdonald, 1992). Solitary carnivores may reduce losses to competitors by means other than active defense (Ewer, 1973; Sunquist and Sunquist, 1989). These alternative methods include bolting food (e.g., cheetahs, Caro, 1994), caching food (e.g., red foxes, Macdonald, 1976; Macdonald, Brown, Canbolat, and Yerli, in press), and hiding food in brush (e.g., tigers, Sunquist, 1981) or up trees (e.g., jaguars, Panthera onca, Schaller and Vasconelos, 1978; leopards, Panthera pardus, Hamilton, 1976). These behavioral responses reduce intra- and interspecific competition at kills, while increasing group size probably exacerbates competition. (There is a parallel to behind-the-scenes diplomacy versus arms races in humans.) Whether behavioral adaptations or increases in group size are favored probably depends on a broad range of physical, ecological, and demographicpreconditions.

D. GROUPVIGILANCEAND PREDATOR DEFENSE Particularly among the small carnivores, sociality may be favored by improving predator detection (e.g., dwarf mongooses, Rood, 1978; Rasa, 1986) or predator defense (e.g., banded mongooses, Rood, 1975; meerkats, Macdonald, 1992). A comparative study of herpestids showed that sociality was associated with diurnality, small body size, and open environments (Rood, 1986). These traits increase either the likelihood of being detected by predators or vulnerability once detected (Rood. 1986). Among social herpestids, the sole exception to these patterns WiIS the cusin~i~nse. highly social despite being a forest dweller. Current theory for the evolution ot’ sociiility of iiioiigoiws ccml~ii1c.s the antipredator benefits of group vipiliitlcc aiitl tldciisc \\ ilti hw L-‘cwls

REPRODUCTIVE SUPPRESSION AMONG CARNIVORES

223

of tolerating others due to abundant and rapidly renewing prey (Gorman, 1979; Waser and Waser, 1985; Rood, 1986). Both elements of the theory seem necessary. For example, the mustelids are generally even smaller than the mongooses, yet they are solitary and they prey on relatively less abundant and more slowly renewing vertebrates (Macdonald, 1992). These differences between the mongooses and mustelids provide an illustration of the potential effects of phylogeny on social organization; two families faced with the same problem may evolve different solutions that depend on preexisting conditions (Macdonald, 1992).

E. BENEFITSOF ALLOPARENTAL CARE Research on mating systems and parental care (Lack, 1968; Emlen and Oring, 1977; Clutton-Brock, 1991; Davies, 1991) suggests that monogamy with biparental care evolves where juveniles are energetically costly to produce and rear, in harsh or dangerous environments, or where intraspecific competition for resources is strong (Clutton-Brockand Harvey, 1977; Kleiman and Malcolm, 1981; Malcolm, 1985; Clutton-Brock, 1991). Under these conditions, the care of two parents increases juvenile survival beyond the level that a single parent could accomplish. Biparental care is the rule among birds (Lack, 1968), probably due to high energetic costs of feeding nestlings (Clutton-Brock, 1991), but is relatively uncommon among mammals (4% of all species, Kleiman, 1977; Kleiman and Malcolm, 1981). Carnivores are one of three mammalian orders (together with primates and perissodactyls) in which biparental or communal care is relatively common (10- 15% of species, Clutton-Brock, 1991; Kleiman, 1977).

Extending the logic of arguments about the evolution of biparental care, Gittleman, (1985a; Gittleman and Oftedal, 1987) showed that among carnivores, communal care was associated with energetically costly reproduction. Total litter mass and litter growth rate were significantly higher in species with biparental or communal care than in species with female care only (Gittleman, 1985a). More detailed patterns can be found within the general association of social breeding with costly reproduction. Allometric regressions show that communally breeding canids not only produce unusually heavy litters but these litters are composed of a large number of small neonates (Moehlman, 1986). This trend is exemplified by African wild dogs, which can produce litters of up to 16 pups (G. Mills, personal communication), and which have the heaviest litters among social carnivores, after standardization for allometry (Creel and Creel, 1991). A large litter of small young is likely to require high postnatal energeticinvestment in total litter growth.

224

SCOTI' CREEL AND DAVID MACWNALD

Altricial young are also quite vulnerable to environmental stresses and predation (Moehlman, 1986). In socially breeding carnivores, most or all adults in a group typically help to offset reproductive costs and risks. Direct alloparental care takes many forms, including grooming the young (e.g., banded mongooses, Rood, 1978), guarding them (e.g., brown hyenas, Mills, 1983; Owens and Owens, 1984), feeding them (red foxes, Macdonald, 1979; jackals, Moehlman, 1979; Ethiopian wolves, Sillero-Zubiri, 1994), cross-suckling (review in Packer, Lewis, and Pusey, 1992), and suckling by nonbreeders (dwarf mongooses, Rood, 1980; Creel et al., 1991). Group members may invest in the young through the mother, by feeding her during gestation or lactation (e.g., silverbacked jackals, Moehlman, 1989; African wild dogs, Malcolm and Marten, 1982). This allows the mother to guard the young more continuously (Malcolm and Marten, 1982), to direct more energy into gestation and lactation (Gittleman, 1985a; Oftedal and Gittleman, 1989),or to increase her foraging time (Rood, 1978; Creel and Creel, 1991). Among small carnivores that can breed several times annually, these benefits may allow a female to produce additional litters (Rood, 1990; Emlen, 1991). Considerable evidence suggests that alloparental help usually results in improved reproductive success for the recipient (Macdonald, 1979; Moehlman, 1979; Malcolm and Marten, 1982; Bekoff and Wells, 1982; Bygott et al., 1979; Rood, 1990), although no benefits have been shown in a few studies (e.g., for Ethiopian wolves, Sillero-Zubiri, 1994;badgers, Da Silva et al., 1993). Within-group relatedness is high in all social carnivores for which it has been quantified (0.26 in brown hyenas, Mills, 1989a, 1989b; 0.31 in spotted hyenas, Mills, 1985, 1989b; 0.33 in dwarf mongooses, Creel and Waser, 1994; 0.25 to 0.5 in pridemate lionesses, Packer et al., 1991; 0.38 in red foxes, Macdonald, 1980). Similar social structures suggest that within-group relatedness is probably close to 0.25-0.35 in many carnivores for which quantitative data are unavailable. High r values, together with effects of help on reproductive success, suggest that indirect fitness benefits may be generally important in the evolutionary maintenance of social breeding in carnivores. Delayed benefits, both direct and indirect, may also be important. Cost/benefit analyses with dwarf mongooses show that both direct and indirect fitness benefits help to maintain social breeding, with their relative importance varying among age-sex classes (Creel and Waser, 1994).

F. SUMMARY A broad range of ecological, life history, energetic, and behavioral variables have been hypothesized or demonstrated to favor sociality or

REPRODUCTIVE SUPPRESSION AMONG CARNIVORES

225

increased group size among carnivores. Some mechanisms are specific to carnivore biology (e.g., effects of group size on tackling large or dangerous prey). Other mechanisms are more general (e.g., effects of group size on meeting high costs of reproduction). Some have been tested across a broad range of species or in comparative analyses, while tests of others have been taxonomically limited. A primary goal or our comparative analyses (below) is to test the relative importance of several hypotheses for the evolution of social breeding across the order Carnivora as a whole. OF REPRODUCTIVE SUPPRESSION Iv. THEEVOLUTION

In some social carnivores, lack of bias within groups in reproductive success is reflected by lack of an obvious dominance hierarchy (e.g., sea otters, Enhydra lutra, Estes, 1989; otters, Lutru lutru, Kruuk and Moorhouse, 1991; lions, Packer et al., 1988). A complication is illustrated by farm cats, which are organized in matrilines among which all females breed, but members of those matrilines occupying the periphery of the farmyard, farthest from food sources, consistently have lower reproductive success (Kerby and Macdonald, 1988). In a few species, all (or most) adult females normally breed despite clear dominance relationships (e.g., spotted hyenas, Frank, 1986). Also, in a few social species some females are reproductively suppressed and others are not (e.g., banded mongooses, Rood, 1975; Waser et al., in press; badgers, Woodroffe and Macdonald, in press). And in the majority of social carnivores (16 [57%] of 28 species in our quantitativeanalyses), only the dominant female normally breeds, while subordinates are typically reproductively suppressed (we qualify this below). Discussions of cooperative breeding in birds (the largest such literature) have carefully distinguished the question of why subordinates do not disperse from the question of why subordinates tolerate reproductive suppression (Brown, 1987; Emlen, 1991). Following this distinction, we restrict our analyses of reproductive suppression to species that are already social breeders. The antecedent question of why groups form is treated separately (above). A.

VEHRENCAMP’SFITNESS BIAS MODEL

Vehrencamp (1983; following Hamilton, 1963, 1964) modeled the evolution of reproductive suppression so that egalitarian and despotic species formed the end points of a continuum of bias in reproductive success among group members. This conceptual framework applies well to carnivores, because they show considerablevariation in the degree of reproduc-

226

SCO’M’ CREEL AND DAVID MACDONALD

tive suppression, even within single species (see Section 11). For instance, subordinate females tend not to breed in high-density red fox populations, where interactions with the dominant vixen are high, but may breed successfully in low-density populations, possibly because they are less harrassed (Macdonald, 1980, 1987). Vehrencamp’s model assumes that a dominance hierarchy exists, and that the dominant will impose the degree of reproductive suppression on subordinates that maximizes her own inclusive fitness. The limit on how far she may bias reproduction in her own favor (maximum allowable bias) is determined by two measures: ( I ) the indirect fitness that a subordinate female obtains by tolerating reproductive suppression and helping her group mates; and (2) the reproductive success a subordinate could expect if, rather than tolerating reproductive suppression, she dispersed to compete for a breeding position elsewhere. When the first quantity exceeds the second, females will do better by tolerating reproductive suppression than by dispersing. In species without dominance (of which there may be none among social carnivores), the model predicts no bias in reproductive success, because there is no individual with leverage over the others. There is a circularity to the questions of why dominance has evolved and why reproductive suppression has evolved that is difficult or impossible to resolve, post hoc. Because dominance sometimes exists without reproductive suppression, but not vice versa, we follow Vehrencamp (1983) in taking the existence of dominance as a starting point from which to model the evolution of reproductive suppression of subordinates. Vehrencamp’s model is described in two cases. In one case, either all subordinates remain and tolerate suppression, or all disperse en masse. In the other, subordinates make their decisions individually. Intuitively, these cases should make the same predictions, but they were originally reported to differ (Vehrencamp, 1983). The differences arise because of an error in the equations for the first case, which attribute to each subordinate the entire increase in the dominant’s reproductive success due to all subordinates’ help. This results in an overestimate of the indirect fitness benefits of helping, and the model overestimates the maximum allowable bias. When this problem is corrected, the two cases produce identical results (S. Creel and S. Vehrencamp, unpublished simulations). The model predicts that reproductive suppression will be most common and most complete under two general conditions.

I . Reproductive Suppression Is Favored When Indirect Fitness Accrued by Subordinates through Helping Is Substantial Indirect fitness can be large where relatedness between subordinates and dominants is high (Hamilton, 1963, 1964), often a property of small

REPRODUCTIVE SUPPRESSION AMONG CARNIVORES

227

group size (Packer et al., 1992). As discussed above, relatedness within carnivore groups is generally high. The average reported value is 0.32 2 0.06 S E ( n = 4), and many other species probably have similar values. Where reproduction is energetically costly or risky, the relationship between breeders’ reproductive success and the number of helpers is likely to be steep, providing a large inclusive fitness benefit to helpers (Gittleman, 1985b, Brown, 1987; Creel and Creel, 1991). Whether helpers actually cause increases in breeders’ reproductive success has been controversial (Brown, 1987; Emlen, 1991; Mumme, 1992). Brown (1987) and Emlen (1991) have reviewed correlations between group size and offspring raised per litter, concluding that in most species there is a significant association. This generalization holds true among carnivores (see Section 111,E; Jennions and Macdonald, 1994), despite several potential exceptions (e.g., Da Silva et al., 1993; Cresswell, Harris, Cheeseman, and Mallinson, 1992; Sillero-Zubiri, 1994). Other possible benefits include more litters per season, improved breeder or helper survivorship, or improved future reproductive success for the helper or its relatives (Brown, 1987; Mumme, Koenig, and Ratnieks, 1989; Emlen, 1991). Correlations between reproductive success (or survival) and group size do not prove that the increase in reproductive success is caused by help. Experimental evidence for the benefits of help rely on helper removals. In birds, two of three experiments have shown a significant decrease in reproductive success following reduction in group size (Brown, Brown, Brown, and Dow, 1982; Leonard, Horn, and Eden, 1989; Mumme, 1992). However, such experiments can subtract more than the alloparental contribution of the individuals that are removed; their contribution to cooperative hunting or vigilance would also be modified, as would their role in the social dynamics of the manipulated groups (Jennions and Macdonald, 1994). Creel and Waser (1994) suggest an alternative approach, by regressing annual changes in reproductive success on annual changes in group size. This approach uses removals of helpers (through death and emigration) and additions (through recruitment and immigration) as a natural experiment. This is a test of Granger causation, as used in econometrics (Johnston, 1984). In dwarf mongooses, this approach suggested that the association between reproductive success and group size was indeed causal. Another approach is to include group size as an independent variable in a multiple regression, together with independent variables measuring aspects of ecology, and test the partial regression of reproductive success on group size (Vehrencamp et al., 1988; Emlen and Wrege, 1989), a method not yet applied to carnivores.

228

SCOTT CREEL AND DAVID MACDONALD

2. Reproductive Suppression Is Favored When Alternatives to Tolerating Suppression Are Constrained Vehrencamp’s model shows that constraints on dispersal will also favor bias in reproductive success within groups; where subordinates’ alternatives are poor, dominants will have greater leverage. Because constraints on dispersal were discussed at length above (see Section III,B), we mention them only briefly here. A low probability of dispersing successfully is a strong inducement to tolerating reproductive suppression, even when the payoff to successful dispersers is high (Waser et al., 1994). Habitat saturation (Selander, 1964) or biased sex ratios (Brown, 1987) may decrease the probability of successfully joining a breeding group, even where dispersal itself is not inherently risky. Where dominance is dependent on age or body mass, young or small subordinates may not greatly improve their expectation of obtaining a dominant breeding position by dispersing. Reproductive success following dispersal into a breeding position may be low, particularly if dispersers typically join small groups (as may apply to social mongooses) or form new groups in order to obtain dominance (as may apply to many canids). In such cases, lack of a labor force may constrain the reproductive success of dispersers. Dispersers’ reproductive success will also be low if only low-quality territories are available. B. SUMMARY Vehrencamp’s (1983) model uses inclusive fitness equalities to illustrate the conditions that favor reproductive suppression. These include factors that increase the indirect fitness of subordinates, either by increasing relatedness within the group or by increasing the effectiveness of help. Factors that constrain independent breeding opportunities for dispersers also favor reproductive suppression, by giving the dominant leverage over subordinates.

V. COMPARATIVE ANALYSES OF SOCIALITY AND REPRODUCTIVE SUPPRESSION

Recent years have seen an explosion in applications of quantitative comparative methods to questions about carnivore behavior, social organization, and ecology. Comparative analyses encompassing the entire order have examined the correlates of home range size (Gittleman and Harvey,

REPRODUCTIVE SUPPRESSION AMONG CARNIVORES

229

1982; Sandell, 1989; Gompper and Gittleman, 1991), body size (Gittleman, 1985b; McNab, 1989), sociality andgroup size (Bekoff, Daniels, and Gittleman, 1984; Gittleman, 1989), life history patterns (Bekoff et al., 1984; Gittleman, 1986; 19931, metabolic rate (Gittleman and Harvey, 1982; McNab, 1989), reproductive energetics (Gittleman and Oftedal, 1987; Oftedal and Gittleman, 1989), patterns of reproductive suppression (Creel and Creel, 19911, and patterns of parental care (Gittleman, 1985a).

Other comparative analyses have examined social organization within single carnivore families, including the Herpestidae (Rood, 1986), Felidae (Packer, 1986; Sunquist and Sunquist, 1989), Canidae (Bekoff, Diamond, and Mitton, 1981; Moehlman, 1986, 1989), and Hyaenidae (Mills, 1989a 1989b). Most of these analyses have been qualitative or nonnumeric (Geffen, Gompper, Gittleman, Macdonald, and Wayne, submitted). Simultaneous with comparative studies of carnivores, the past decade has seen revolutionary change in the comparative method itself. New methods have been stimulated by the realization that many pairs of species will not vary independently in the value of a trait; common ancestry normally produces covariance among species (Clutton-Brockand Harvey, 1977; Harvey and Mace, 1982; Felsenstein, 1985,1988). Asaresult, quantitative analyses that treat related species as independent will inevitably overestimate residual degrees of freedom. Simulation studies confirm that analyses that treat species as independent normally yield erroneous statements of significance (Felsenstein 1985; Grafen, 1989; Martins and Garland, 1991; Harvey and Pagel, 1991). The problem is at its worst when phylogenetic associations are strong (i.e., among closely related taxa), when the null hypothesis of no association between a pair of traits is false, and at high nominal significance levels (Grafen, 1989). In the face of problems arising from covariation among species due to phylogeny, several comparative methods have been developed that take phylogenetic covariation into account when testing whether a pair of traits are associated with one another (Harvey and Mace, 1982; Ridley, 1983; Cheverud, Dow, and Leutenegger, 1985; Felsenstein, 1985, 1988; Grafen, 1989, 1992; Pagel and Harvey, 1988, 1989; Gittleman and Kot, 1990; Harvey and Pagel, 1991; Garland, Harvey, and Ives, 1992; Pagel, 1992). Several of these methods have been reviewed and compared through simulations by Pagel and Harvey (1988, 1992), Grafen (1989, 1992), Harvey and Pagel (1991), Harvey and Purvis (1991), and Gittleman and Luh (1992, in press). Of these methods, we have applied phylogenetic regression (Grafen, 1989, 1992) in our analyses. Phylogenetic regression applies the general principle of identifyingindependent evolutionarytransitions by identifying independent linear contrasts of trait values (following Felsenstein, 1985).

230

SCOTT CREEL AND DAVID MACDONALD

In this regard, phylogenetic regression resembles another well-known method, “comparative analysis by independent contrasts” (Purvis, 1991 ; Pagel 1992; Harvey and Pagel, 1991). As discussed by Pagel and Harvey (1992) and Grafen (1992), both methods deal with incompletely resolved radiations (multiple nodes) by modeling the expected pattern of covariation among species within the multiple node. The methods diverge in the way in which this covariation is modeled. Phylogenetic regression employs generalized least squares regression (Johnston, 1984) to model the covariance uii between a pair of species, i and j, as u.. u = (1 - hop).

(1) In this equation, h, is the height in the phylogeny at which species i and jdiverge, with species assigned a height of 0 and the highest node assigned a height of 1 . If a pair of species has diverged recently, h, will be near 0, and the covariance will be large. If the pair diverged long ago, near the highest node, h, will be near 1, and the covariance small. Grafen (1989) describes a number of methods for assigning node heights, including a simple rule that assigns node heights proportional to the number of species below a given node. In Equation (I), p is a parameter that encapsulates the degree of similarity due to common ancestry by selectively stretching or compressing the phylogenetic tree at different heights (Grafen, 1989: Fig.3). Large values of p represent strong phylogenetic constraint, and covariances are increased. In the phylogenetic regression, p is fitted simultaneously with the regression coefficients by maximum likelihood, allowing the data themselves to modify the heights assigned to nodes. Garland et al. (19921, Grafen (1989, 1992), and Harvey and Pagel (1991; Pagel and Harvey, 1992) discuss the limitations of this and other methods of estimating patterns of covariance among species. We have used phylogenetic regression because it has a detailed analytic proof (Grafen, 1989), as well as simulations verifying error rates close to nominal values (Grafen, 1989). Phylogenetic regression, “comparison using independent contrasts,” and autogression methods (Cheverud et al., 1985; Gittleman and Kot, 1990) each have their proponents. Although it is primarily their differences that have been emphasized (Gittleman and Luh, 1992, in press), they are similar in that they adjust for covariation due to phylogeny so that tests for associations among variables independent of phylogeny can be made. Autoregression methods adjust for phylogeny using a weighting matrix, the “phylogenetic connectivity matrix” (Cheverud et al., 1985; Gittleman and Kot, 1990). Phylogenetic regression accomplishes this by modeling the variance-covariance matrix in a generalized least squares regression (as described above), a method well characterized and widely used in econometrics (Johnston, 1984). Simulations suggest that indepen-

REPRODUCTIVE SUPPRESSION AMONG CARNIVORES

23 1

dent contrast methods and autoregression methods perform equally well when sample size is large and taxonomic data are used to control for common ancestry, as in our analyses (Gittleman and Luh, 1992, in press). Most of the early comparative analyses of carnivore behavior, ecology, and life history relied on species or genus means regressions (but see Gittleman, 1993), and simulations have shown these methods to have potentially inflated significancelevels (Grafen, 1989;Martins and Garland, 1991; Harvey and Pagel, 1991). We reanalyzed five published carnivore data sets using phylogenetic regression and found that conclusions were altered dramatically in some cases, and unaffected in others. We conclude that, although significance levels are probably lower than stated for many comparative studies of carnivores (including our own, Creel and Creel, 19911, there is no standard rule for reinterpretation of a given analysis.

VI. PHYLOGENETIC REGRESSION ANALYSESOF SOCIALITY, GROUP SIZE,A N D REPRODUCTIVE SUPPRESSION A.

METHODSFOR PHYLOGENETIC REGRESSIONS

The logic underlying phylogenetic regression is discussed briefly above. The logic and mathematical justification are given at length by Grafen (1989, 1992). One of the strengths of phylogenetic regression is that it can be applied to incompletely resolved phylogenetic trees. Where the exact pattern of bifurcation is not known, species within an unresolved node can still be included in the regression, provided they are attached to the correct node. Molecular and biochemical data on phylogenetic relationships are becoming available for carnivores (Collier and O’Brien, 1985; O’Brien, Nash, Wildt, Bush, and Benveniste, 1985; O’Brien et al., 1987; Wayne and O’Brien, 1987; Wayne, Benveniste, Janczewski, and O’Brien, 1989; Geffen et a / . , 1992; Geffen et a / . , submitted). Although these data provide precise phylogenies for species that have been subjected to molecular or biochemical analyses (e.g., Wayne et al., 1989),many more species remain unplaced. Because the level of phylogenetic resolution varies so greatly among carnivore species, and because the method allows for an incompletely resolved phylogeny, we have used Wozencraft’s (1989a, 1989b, following Ewer, 1973) taxonomy as a working phylogeny. This approach is conservative; by admitting considerable ignorance about the true phylogeny we reduce the statistical power of the analysis, but increase validity (Grafen, 1989). We implemented phylogenetic regressions in GLIM 3.77 (generalized linear interactive modeling) software, using the program PHY LO.REG

232

SCOTT CREEL AND DAVID MACDONALD

provided by Alan Grafen (version 1.03, see Grafen, 1989). We entered the phylogenetic information as a series of taxonomic vectors coding superfamilies,families, subfamilies, genera, and species. Using this taxonomic vectors method, Grafen’s program includes an algorithm for assigning identities to higher nodes of the phylogeny, which we checked by drawing the phylogeny. We assigned initial node heights (later modified by maximum likelihood estimation of p ) using Grafen’s (1989; Fig. 2) counting rule (see Section V). Considerable comparative data have been compiled for carnivores. We initially compiled data from a number of recent reviews, including Ewer (1973), Kingdon (1977), Bekoff et al. (1981), Gittleman and Harvey (1982), Bekoff et al. (1984), Macdonald (1984), Gittleman (1985a, 1985b, 1986, 1989), Moehlman (1986), Oftedal and Gittleman (1989), Creel and Creel (1991), and Packer et al. (1992). Where values differed among reviews, or seemed questionable, we referred to primary literature or personal communication. The data are presented in Table I. We analyzed three dependent variables (see Section I1 for explanations of dichotomies): ( 1 ) socially breeding species versus nonsocially breeding species (dichotomous); (2) mean breeding group size (ordinal); (3) groups with reproductive suppression versus groups without suppression (dichotomous). Phylogenetic regression was originally developed for continuous dependent variables, whereas ours are dichotomous or ordinal. Because the range in group size is considerably wider than its increments, treating group size as a continuous variable is likely to be valid (Steel and Tome, 1980). Dichotomous dependent variables are more problematic for phylogenetic regression, because they increase the probability of losing phylogenetic degrees of freedom. This problem, discussed at length by Grafen (1989), was minor in our analyses because we have used a continuous variable (body mass) as a control variable in all regressions (A. Grafen, personal communication). The remaining effects of using qualitative dependent variables were the same as in ordinary least squares regression. Namely, we have used a linear probability model where it would have been preferable to fit a logistic function (logit analysis) or cumulative normal function (probit analysis). The linear model was used because no parallel to logit or probit analysis is available for phylogenetic regression (A. Grafen. personal communication). Substitution of linear for logistic or cumulative normal functions is a common practice in standard regression, and has two consequences (Johnston, 1984). The major consequence is that error terms become heteroskedastic, so that ordinary least squares regression cannot be used. This problem was circumvented using Box-Cox transformation

REPRODUCTIVE SUPPRESSION AMONG CARNIVORES

233

prior to analysis (results are presented after retransformation). The second consequence is that although generalized least squares regression remains valid, R’ becomes a poor indicator of goodness of fit. We therefore do not report R2 values in our results. For each of the three dependent variables, we tested the effects of 11 independent variables. We broke these 1 1 variables into three categories. Ecological variables included: (1) Diet, broken into five classes: carnivore, insectivore, omnivore, frugivore, herbivore. Following Gittleman (1989), omnivores were those species for which none of the other categories comprised 60% or more of the diet; (2) Home range size, measured in km’. We used values for females where sex differences were reported. As noted by Gittleman and Harvey (1982), the time over which home range size is estimated varies widely between studies, and where ranges are unstable this may affect the realism of estimates (e.g., Doncaster and Macdonald, 1991). Methodological differences are also ignored. Life history variables included: ( 3 ) Life-span, measured in months. For this variable only, we gave precedence to estimates from captivity. Values from the wild were relatively uncommon and probably biased by study durations; (4) Age at onset of breeding, measured in days, using values reported as typical, for females where sex differences were reported; (5) Sexual dimorphism, expressed as the ratio of male body mass to female body mass. Variables measuring costs of reproduction included: (6) Neonate mass, the mean mass in kilograms of a single neonate; (7) Litter size, the mean number of offspring recorded at first count. The interval between birth and the first count of young is variable, and early juvenile mortality is variable, so precount reductions in litter size are also likely to vary; (8) Duration of gestation, mean value, in days; (9) Duration of lactation, mean value, in days, from birth to weaning, as opposed to the date at which offspring begin eating solid food. Weaning can take place long after solids begin to be eaten. Two variables derived from these were also tested: (10) Litter mass, the product of neonate mass and litter size; (11) Prenatal litter growth rate, equal to litter mass divided by the duration of gestation. A stepwise regression procedure would normally be used to determine which independent variables were significant predictors of the dependent variable, but no analogue to stepwise regression exists for phylogenetic regression. Running many independent tests leads to Type 1 error (Steel and Torrie, 1980), so we minimized the number of tests by using a scheme similar to stepwise regression. First, we broke the dependent viiriables into the above categories. For each dependent vilriilhlc, wc tcstcd each independent variable ( I V ) singly. I t there were no signilicxnt prcdictiws

TABLE I COMPARATIVE DATAUSEDI N PHYLOCENETIC REGRESSIONS

Species

w

bJ

*

Canidae Canis aureus Canis lupus Canis latrans Canis adustus Canis mesomelas Canis simensis Canis fam. dingo Lycaon pictus Cuon alpinus Alopex lagopus Vulpes vulpes Vulpes bengalensis Vulpes chama Vulpes velox Fennecus zerda Otocyon megalotis Urocyon cineroargentus Dusicyon culpaeus Dusicyon gymnocercus Cerdocyon thous Chrysocyon brachyurus Speothos venaticus Nyctereutes procyonoides Felidae Felis sylvestris Felis libyca

la

2

3

1

1

2

2

9.0 31.1 9.7 8.3 7.4 13.4 12.5 25.0 13.8 2.9 3.9 1.8 3.1 1.9 1.5 3.9 3.3 6.7 4.2 6.0 23.0 6.0 7.5

2 2

2 2

4.3 3.9

1

1 1 2 1

1 1 1 2 1

1 I 1 1

1

I I I 1 1

1 I -1

-I -1 1

I 2 2

-I

-1 -1

-1 1

2 2 2 -1

2 2

2 2

4

-1 0.14 0.40 0.16 0.11

5.7 6.0 5.4 5.4 5.7 4.4 5.4 8.8 8.5 7.1 4.8 3.5 4.0 4.5 2.8 3.5 3.8 5.0 4.0 3.1 2.0 3.8 5.0

0.14 -1

3.3 2.8

-Ib 0.43 0.22 -1

0.16 0.40 0.30 0.37 0.28 0.07 0.11 -1 -1

0.04 0.04 -1

0.12 0.17

6

5

1.15 1.30 1.20 1.13 1.12 1.80 1.16 1.00 1.21 1.70 1.24 1.67 1.00 1.16 1.50

-I 1.30 1.15

-I

7 2.7 9.0 4.4 2.0 2.9 7.0 6.0 11.1

8.3 3.0 3.0 2.0 -1

2.0 6.0 3.5 2.0 -1 -1

1.60 1.50

3.5 2.0 4.0 4.5

1.16 1.21

1.o 1.O

-1

1.00

8 2.5 390 42 74 19 6.4 27 500 20 12.5 4.1 -1 -1 3.2 -1 1.o 15.5 -1 -1 7.0 25

10

9 330 660 365 330 392 730 730 540 365 3 LO 285

63 63 62 63

60 60 63 71 62 53

0.2

365 730 365 305

55 53 -1 53 51 60 63 58 58 56 63 61 62

1.o 0.7

313 280

67 57

-1

-1 -1

300 255 365 305 365

-I

11 63 63 42 63 61 63 63 77 60 63 49 -1 -1

45 66 84 56 31 -1 90 99 56 30 84 84

12

13

2 1 1 2 2

156 177 -I 132 168

1

-1 -1

1 1

1 1 2 2 2 2 2 3 2 2 1 2 2 1

2 1

2

132 186 108 144 -1 -I 156 141 -I 180 -1

163 -I 162 120 132 -1 180

Felis chaus Felis serval Felis bengalensis Felis rubiginosa Felis viverrinus Felis caracal Puma concolor Leopardus pardalis Leopardus geoffreyi Lynx lynx Lynx rufus Panthera leo Panthera tigris Panthera pardus Panthera onca Panthera uncia Acinonyx jubatus

Ursidae Ursus arctos Ursus americanus Thelarctos maritimus Selenarctos thibetanus Helarctos malayanus Melursus ursinus Tremarctos ornatus Ailuropoda melanoleuca

2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2

3.3 1.3 6.3 14.8 39.6 10.8 2.2 17.8 5.2 135.5 131.0 50.0 77.6 32.5 45.0

2 2 2 2 2 2 2 2

2 2

299 97

2

240

2 78 2 4 6 2 102 2 -1 2 120

-I

2

2

3.7

-1

2 2

2 2

0.9 2.0

2 2 2 2 2 2 2 2 2 2 2 1

2 2 2 2 2

6.7

0.14 0.14 0.08

11.0 -1

0.09

-1 0.40 0.25 0.07 0.07 0.31 1.65 1.26 0.55 0.82 0.44 0.23 1.00 0.29 0.64

0.32

-1 0.31 0.11

2.9 2.4 2.5 2.5 2.5 3.0 3.5 2.5 2.0 2.3 3.2 2.6 2.5 2.6 2.5 2.8 4.2

1 .o

0.98 1.18

1.o 2.0 1.o 1.o 1.o 1.o

-I -1

1.oo 1.10 1.62 1.19 1.OO 1.17 1.37 1.22 1.45

1 .o -1 1 .o 1 .o

8.7

1 .o 1.o

1.20

I .23 1.OO 1.20

I .o 1.o

2.0 2.5 I .9 2.0 2.0 2.0 2.0 1.5

1S O 1.29 2.00 1.65 1.20 -1 I .80 1.10

1.o 1.o 1.o

2.0

1 .00

3.0 1.5

1.11 1.05

1.o

.o

1

1 .o 1 .o 1.o 1.o

L

-1

330

1.5 -1

-1

-1

65 71 67 -I 93 74 90 73 70 68 63 106 104 98 105 97 91

102 -1 25 -1 53 123 -I 49 63 113 60 240 165 139 115 -1 98

1 I I 1 I 1 -1 1 1 1 1 1 1 -1 1

162 156 216 207 264 -1 -1 204

63 74 -1

-1

730 168 -1 119 -1 -I -1 180

2 2 1 2 2 2 5 5

304 270 408 3% -1 360 252 216

570

90

150

5

158

300 -1

52 107

120 -I

1 4

96 228

750 -I

-1

-1

-1

450 913 653 480 690 593 1620 1643 1187 1110 730 645

48.8 -1 -1 22.4 30.7 240 71.4 23.6 258 100 67.5 53 56 -1 -1 -1 -1

-1 2.5

1338 1834 1734 1186 -1 -1 -1 1620

%

-1 -1 -1

-1 1

144

180 150 -1 -1 204 -1 -1 -1

Ailuridae Ailurus fulgens Procyonidae Bassariscus astutus Potos flavus

0.03 0.17

-I

-1

1.o 1.o

1.4 -1

(continues)

IOZI 9SI 081 882

I-

9p

E I

9s

z E

E I

0 9 z

I I E I I I I I I

OSI

I

I-

z I z

981 I-

I

IPPI

IIII1-

ZP 56

ILL

9s IIOL

6P 9p 9p

IIIE9 OL

ZP f9 ZP ZP 891

ZE 9E It 9 SE 82

I80E S6E SZS I1-

IIIOf9

OfL

LL

611

z

I

LE

9s 9s E9 PZ

1 1

IIIIII-

ZI

009

S9E

LZ Of z9 IP

62 Of 62 op PZ

1-

E

I-

PPI

EI

IIIOff Off

opS opS

59 PL

IIZSZ II9EE

SE

E9

9E

ZE

11

01

6

O'L L' I 0' I 0' I I01

IIII-

O'P

00'1

Z'I

E'P 0'9

1W'I

9'P

1S'P P'S

6'0 I-

Sop I6'52 9'I 5' I IO'E I 6'I 1-

IIIIZI'O 01'0

9'I 9'0 8

01

0' I 0' I 0' I 0' I 5' I I0' I 0' I 0' I 0' I 0' I 0' I 0' I 0' I 0'1 0' I 0'8

L

6510'0

EP'O OO'Z Of '9 6'01 6S'L

Z

O'P

00'1

8'E

16'0

EI'O

LIOO'O

S'P

LL'I

SO'O 90'0

OE00'0

8'5

IOOO'O

8'P

IL9'I

op'0 E 1'0 EZ'O

O'P

Z6.l

0'5

58'1

W'O

IIIIEOO'O

0'9

00'1

S'P

89'1

19'0

1-

08'0

Z

Z

Z

Z

OZ'I ES'O

II-

Z

LL'O

9'Z E'E 0'9 0'9 0'5

86'1

Z

SZ'Z

Z Z Z Z

Z Z Z Z Z Z

E0'1

I Z

I Z

Z Z Z

Z Z

EI'I EI'I 8P'I

PP'I 88'1 01'1

ox

S'Z O'Z E'Z O'Z S'E

L'Z

EE'Z

OX

IE'I

8'2

SZ'1 9Z'I 00'1

9Z'I 85'1

9

5

OEEO'O ISEOI ' 0

Iopoo-0 OSIO'O I-

I2660.0 OEEO'O 08Z0.0 08zoo 1-

PE'0 OE'I 08' I op'P 9-01

Z Z

Z Z

Z Z

Z Z

Z

Z

Z

Z

Z Z

Z

2

el

Z

Z 1

Z Z

L'9 0'5

Z Z

Z Z

Z

Z

E

P

11'0 PI'O

sapadg

Lutra lutra Lutra canadensis Lutra maculicollis Lutrogale perspicillata Aonyx capensis Enhydra lutris Herpestidae Herpestes ichneumon Herpestes sanguineus Herpestes auropunctatus Herpestes edwardsii Herpestes smithii Herpestes fuscus Herpestes vitticollis Herpestes urva Mungos mungo Crossarchus obscurus Helogale parvula Ichneurnia albicauda Atilax paludinosus Cynictis penicillata Paracynictis selousi Suricata suricatta Galidia elegans Vivemdae Nandinia binotata Paradoxurus hermaphroditis Arctictis binturong Fossa fossa Eupleres goudoti

-1 -1 -1 0.2100

2.5 3.0 3.0 1.5 3.0 1.0

1.73 1.10 1.25 1.41 1.11 1.39

3.5 -1 4.5 2.0 2.0 6.0

2.90 0.57 0.50 1.00 1.30 0.80 1.70 2.00 1.60 1.30 0.35 3.50 3.30 0.60 1.60 0.70 0.80

-1 -1 -1 -1 -1 -1 -1 -1 0.02 -1 0.02 -I -1 -1 -1 0.03 0.05

2.5 2.5 3.2 3.0 2.5 3.5 3.0 3.0 3.8 4.0 3.9 2.5 2.5 2.5 1.5 4.0 1.0

1.14 1.24 -1 1.56 1.62 -1 -1 1.39 1.03 0.90 1.00 1.23 1.25 1.00 1.13 1.09 1.03

1.o 1.o 1.o 1.o 2.0 3.0 1.o 1.o 15.0 15.0 8.9 1.0 1.o 5.0 2.0 12.0 3.0

2.00 2.70 13.0 1.60 2.10

0.06 0.10 0.32 0.08 0.15

1.8 3.3 3.0 1.0 1.5

1.00 1.22 -1 1.25 1.11

I .o

1 -1 1 2 2 1

2 -1 2 2 2 2

5.20 7.80 3.50 7.30 18.0 19.9

2 2 2 2 2 1 2 2 1 1 1 2 2 1 2 1 1

2 2 2 2 2 2 2 2 2 1 1 2 2 2 -1 1 2

2 2 2 2 -1

2 2 2 2

-I

0.2850 -1

1.o -1 2.0 -1

1 1 1 1 2 1

49 -1 -1 180 -1 -1

49 -1 -1 42 -1 56 56

1 2 3 2 2 -1 -1 2 3 2 3 2 2 3 3 2 2

156 -1 120 -1 -1 -1 -1 120 108 % -1 -1 -1 -1 -1 -1 730

-1 -1 56 52 63

4 2 4 2 2

-1 168 216 -1

913 730 730 -1 -1 1095

62 56 56 62 63 65

112 93 -1 126 -1

0.75 0.3 1 -1 -1 -1 -1 -1 2.40 -1 0.27 8.00 -1 -1 -1 -1 0.23

-1 365 365 270 -1 -1 -1 -1 730 -1 450 -1 -1 -1 -1 -1 730

63 -1

-1 -1 32 -1 -1 -1 -1 -1

0.73 -1 -1 1.00 -1

365 -1 -1 365 -1

29.2 -1 -1

7.0 -1 2.5 -1

46 61 -1 -1

-1 61 59 70 51 -1 -1

46 -1 77 83

64 -1 90 85 -1

64

-I -I

-I

(continues)

TABLE I (Continued) Species

la

2

3

4

5

6

7

Crocuta crocuta Hyaena brunnea Proteles cristatus Hyaena hyaena

1

2 2 2 2

51.9

1.50 0.69 0.30

2.0

0.9

3.0

1.o 1.o

8.0 3.7

8

12

13

360

1

276

360

2

150

-1

3

144

60

2

282

9

10

I1

913 -1 -1

110 -1 loo

821

87

Hyaenidae 1

2 2

43.4

10.0 35.1

-I

2.8 2.5

I .o

560

1.3

308 I .o

2.0

289

a Columns are as follows: (1) SO: I , social breeding; 2, nonsocial breeding; (2) SU: 1, reproductive suppression typical; 2, suppression not typical: (3) BM: adult female body mass, in kg; (4) NM: individual neonate mass, in kg; ( 5 ) LS: litter size: (6) DI: ratio of adult male to female body mass: (7) GS: group size, excluding dependent young; (8) HS: home range size, in km2;(9) BR: age at first breeding for females, in days; (10) GES: duration of gestation, in days; (11) LAC: duration of lactation, in days; (12) DIET: 1, carnivorous; 2, omnivorous; 3, insectivorous; 4, frugivorous; 5, herbivorous; (13) LIFE: maximum captive (or wild) life-span, in months.

Denotes missing data.

REPRODUCTIVE SUPPRESSION AMONG CARNIVORES

239

within a category, we ran no further regressions for independent variables in that category. If one or more IVs were significant, we paired the best single IV with each other IV within the category. If none of the resulting pairs of IVs was significant we proceeded no further, but if one or more pairs of IVs was significant, we formed triplets of IVs by joining the best pair of IVs to each other IV within the category. We proceeded in the same fashion for quadruplets. This scheme imperfectly mimics stepwise regression, but substantially reduces the number of tests performed, relative to running all possible regressions. Two IVs, litter mass and prenatal litter growth rate, were tested as single IVs but were not considered for pairs, triplets, and so on, because they were formed by combining other variables; pairing these would have created strong collinearity. All analyses controlled for variation in female body mass, as many of the independent variables have been shown to be allometrically related to mass (see Harvey and Pagel, 1991). Analyses of reproductive suppression also control for sociality.

B. RESULTS OF PHYLOGENETIC REGRESSIONS The results of the phylogenetic regressions are presented in Tables 11, 111, and IV. The results are expressed as F tests, after controlling for body mass. We also give the phylogenetic degress of freedom for residuals. Total phylogenetic degrees of freedom are equal to the number of higher nodes (36 for our tree). In a given regression, some degrees of freedom may be lost due to missing data, and some due to lack of variation within a node in the dependent variable (especially where the dependent variable is dichotomous). In our analyses, these two causes reduced phylogenetic degrees of freedom by seven, on average, mostly because of missing data. We give the value of p (although it is biased: Grafen, 1989) to allow rough comparisons of the strength of phylogenetic constraint among analyses; higher values of p indicate stronger phylogenetic constraint. Because phylogenetic regressions pass through the origin, we report only the sign of the slope of the regression (Harvey and Pagel, 1991), although the estimate of the slope is unbiased (Grafen, 1992; cf. Pagel and Harvey, 1992). 1 . Social and Nonsocial Breeding

Table I1 shows the results of analyses of the dichotomy between social and nonsocial breeders, controlling for body mass. Strikingly, none of the independent variables was a significant predictor of sociality. Although this result was unexpected, it is not unprecedented. Using different methods, Gittleman (1989)found no correlation between group size and activity

240

SCOlT CREEL AND DAVID MACDONALD

TABLE I1 PHYLOGENETIC REGRESSION RESULTS WITH SOCIAL VERSUS NONSOCIAL BREEDING AS THE DEPENDENT VARIABLE Independent variablesa Energetic variables NM LS LM LD GES LAC BM Life history variables BR LIFE DI Ecological variables HS DIET

Number of species

Sign of slope

68 105 68 62 89 73

P

df

F

P

.21

I ,26 1.31 1.25 I ,23 1.29 I ,26

105

dab nla nla nla nla nla nla

1,17

3.61 1.21 I .54 1.13 0.14 2.16 0.06

nsc ns ns ns ns ns ns

68 56 94

nla nla nla

.30 .25 .I5

1.25 1.23 1.29

1.82 0.51 0.17

ns ns ns

60 I05

nla nla

.20

1,24 4,28

3.10 1.56

ns ns

.I5 .02 .49 .35

.I7 .I5

.I5

Note: Results are given for all single independent variables regardless of significance, but pairs, triplets, and so on are presented only if significant in this and subsequent tables. a For this and subsequent tables: NM = neonate mass; LS = litter size; LM = litter mass; LD = prenatal litter growth rate; GES = duration of gestation; LAC = duration of lactation; BM = body mass; BR = age at first breeding; LIFE = life-span; DI = ratio of male to female body mass; HS = home range area; DIET = diet class. Not applicable. Not significant.

pattern, zonation, diet, age at independence, or litter size. In Gittleman’s review, group size did vary with vegetation, with larger groups in more open habitats (Gittleman, 1989). On first appraisal, the lack of correlation between sociality and all of the independent variables might cast doubt on the general importance of several hypotheses for carnivore sociality reviewed above. For instance, Waser’s (1981) hypothesis that rapidly renewing prey favors sociality would predict an association between sociality and diet. The resource dispersion hypothesis (Macdonald, 1983) and its variants (e.g., von Schantz, 1981, 1984) would also predict an association between sociality and diet. Hypotheses about the benefits of alloparental care (e.g., Gittleman, 1985a; Creel and Creel, 1991) suggest that costly reproduction and social breeding would be associated.

REPRODUCTIVE SUPPRESSION AMONG CARNIVORES

24 1

We do not believe that lack of significance in the above regressions constitutes failure of these hypotheses to pass a strong test, because group size does correlate with many of these independent variables (Table I11 and below).This suggests that the lack of significant correlations with social breeding may be due to the fact that our dichotomous analysis ignores variation in group size among social species. Other factors reduce the power of these regressions, particularly the use of indirect measures of ecology as independent variables, a point we will return to in the discussion. The few single-species studies that have carefully examined the effects of diet on social organization suggest that it plays a critical role (Macdonald, 1983). Unfortunately, these studies also suggest that the relationship between social organization and resource dispersion is quite complex, dimming the prospect of meaningful quantitative comparisons, despite the emergence of new methods. 2. Breeding Group Size Table 111 shows the results of regressions using group size as the dependent variable, controlling for body mass. In the life history category, neither age at first breeding, life-span, nor the degree of sexual dimorphism correlated with group size. In the ecological category, group size was positively correlated with home range size. A priori, the direction of causation in this relationship is difficult to know. Perhaps individualgroups expand their territories as their group size increases (expansionism,Kruuk and Macdonald, 1985). Equally plausibly, groups may be large in species in which individuals must defend large areas to maintain year-round assurance of food (contractionism, Kruuk and Macdonald, 1985). Upon pairing IVs, diet also entered the regression. We used one-way ANOVA to examine the effect of diet in greater detail. Breeding group size is largest in the insectivores, significantly larger than omnivores or carnivores (Fishers protected LSD, p < .05). Frugivores and herbivores tend to be solitary, in significantly smaller groups than omnivores or carnivores (Fishers protected LSD,p < .05). These results lend support to the hypothesis that resource characteristics influence sociality: (1) Species feeding on scarce (fruit) or low quality (vegetation) foods are normally nonsocial breeders; (2) The largest mean group size is among species that feed on invertebrates, a relatively abundant, rapidly renewing prey that often occurs in large patches; (3) Carnivores and omnivores have intermediate group sizes, and eat foods that are less patchy (for omnivores), less abundant (for carnivores), or less rapidly renewing (for carnivores) than is true for insectivorous species. Four measures of the costs of reproduction were positively related to

242

SCOTT CREEL AND DAVID MACDONALD

TABLE 111 PHYLOGENETIC REGRESSION RESULTSWITH GROUP SIZE AS THE DEPENDENT VARIABLE Independent variables Energetic variables NM LS LM LD GES LAC BM LS,NM LS,GES LS,NM,GES LS ,NM,LAC LS,NM ,GES,LAC Life history variables BR LIFE DI Ecological variables HS DIET HS,DIET

Number of species 65 100 65 60 86 69 100 65 86 60 54 52

66 55

90 61 100 61

Sign of slope

P

df

F

P

I ,25 1.29 I .25 I ,22 1,27 1,23 1,24 2,24 2.26 3.20 3,19 4.16

5.91 9.68 18.10 15.68 0.46 0.11 0.73 9.10 4.77 5.66 6.13 5.00

.05 .01 .01 .01

+,+ +,+,+,+,+,+,+,+

.29 .20 .29 .26 .26 .I8 .I8 .29 .26 .26 .30 .33

nla' nla nla

.I8 .33 .22

1,24 1.22 I ,27

2.02 1.77 1.98

ns ns ns

+

.08

1.25 4.26 5,21

9.20 1.88 9.08

.005

+ + + + nla nla nla +,-b

nla

+,d

.20 .08

n sa ns ns .005 .05 .01 .01 .01

ns .01

Not significant. For pairs, triplets, and so on, signs of slopes are in same order as independent variables at left. Not applicable. See text for relationship of group size and diet.

group size (Table 111). Larger litters, larger neonates, heavier litters, and more rapid prenatal litter growth were all associated with larger groups. Large litters was the strongest single predictor of group size. All pairs, triplets, and the quadruplet of variables measuring reproductive costs were significant, except one (the pair of litter size and duration of lactation). The durations of gestation and lactation were not correlated with group size when tested as single IVs. In combinations of IVs, the sign of the relationship between the durations of gestation and lactation with group size sometimes changed, and was always near zero. Thus, increasing the total energetic cost of producing young is associated with breeding in larger

243

REPRODUCTIVE SUPPRESSION AMONG CARNIVORES

groups, but changes in the rate of investment seem less important. These results give support to the hypothesis that costly reproduction favors sociality because alloparental care becomes more beneficial (Gittleman, 1985a; Brown, 1987), perhaps even necessary (Creel, 1990; Creel and Creel, 1991). 3. Reproductive Suppression Table IV shows the results of analyses of the dichotomy between species with and without reproductive suppression of subordinates, controlling for body mass and sociality. None of the variables in the ecological or life history categories correlated with reproductive suppression, although group size (negatively associated) and the age at onset of breeding (positively associated) approached significance.

TABLE IV PHYLOGENETIC REGRESSION RESULTSWITH PRESENCE VERSUS ABSENCEOF REPRODUCTIVE SUPPRESSION AS THE DEPENDENT VARIABLE Independent variables Energetic variables NM LS LM LD GES LAC BM LS,GES LS,LAC LS,GES,NM LS,GES,LAC LS,GES,NM,LAC Life history variables BR LIFE DI Ecological variables HS DIET GS

Number of species 68 104 68

Sign of slope

+ +

df

F

P

.I0

1,25 1,30

1,25 1.22 1,28 1.25 1,9 2,27 2,24 3.20 3,21 4.16

4.26 4.79 2.08 0.21 0.13 2.54 0.52 5.06 5.31 4.71 7.02 8.41

.05 .05 ns' ns ns ns ns .02 .03 .03

P

89 73 62 70 54

+,-' +,-,+,+,+,+

.I2 .I0 .I8 .22 .I2 .I4 .22 .II .I8 .I8 .29

68 56 93

nla nla nla

.I9 .I5 .14

1.24 1.22 1.28

3.56 1.24 0.10

ns ns ns

60 104 97

nla nia nla

.25 .I2 .29

1,23 4,27 1,27

0.03 0.32 3.56

ns ns ns

62 89 73 104

nla' nla n/a nla nla

+,-

+,-,+

'See footnotes in Table 111 for explanation.

.01

.01

244

SCOTT CREEL AND DAVID MACDONALD

Among the variables measuring the cost of reproduction, large litters and heavy neonates were significantly associated with reproductive suppression. The durations of prenatal or postnatal investment and the rate of investment were not significant predictors. As with group size, large litters was the single best predictor of reproductive suppression. All pairs, triplets, and quadruplets among the variables measuring reproductive cost were significant, except one (the pair of litter size and neonate mass). Collectively,these results lend support to the hypothesis that reproductive suppression is found where reproduction is energetically costly (Creel and Creel, 1991). Hypotheses for the evolution of reproductive suppression have focused on ecological and demographic constraints limiting the benefits of attempting to breed independently, which are difficult to test comparatively, but have considerable support from single-species studies (Emlen, 1991; Brown, 1987).Our results suggest that the energetic costs of reproduction may also be a determinant of reproductive suppression. When costs are high, they are more likely to exceed the benefits of attempting to breed as a subordinate, at which point total suppression should be tolerated. The intuitive force of this argument is particularly clear when watching small groups of social mongooses; in one group of four meerkats, the lactating female devoted all her time to foraging, leaving another individual to baby-sit all day, another to make up the food debt incurred by baby-sitting all the previous day, and the last member to take almost the entire responsibility for vigilance (D. Macdonald, personal observation). Under these circumstances the energy demands on such a small group were patently debilitating. C. SUMMARY OF PHYLOGENETIC REGRESSIONS Our comparative analyses do not address all of the hypotheses we have reviewed (see Section VI,D), but do allow us to confirm the general importance of several hypotheses for the entire order Carnivora. The association between large groups and insectivory (and to a lesser degree omnivory and carnivory) suggests that group living is favored by dependence on foods that are shared at little cost to a territory owner (Waser, 1981; Macdonald, 1983; Carr and Macdonald, 1986; Macdonald and Carr, 1989; Rood, 1986). The result does not let us distinguish the relative importance of food abundance, dispersion, and renewal rate, because these variables are confounded with one another across diet types. We can conclude that one or more is likely to favor sociality. Energetically costly reproduction is also associated with life in larger groups, lending indirect support to the hypothesis that sociality is favored in part through the benefits of alloparental care (Gittleman, 1985a). Where

REPRODUCTIVE SUPPRESSION AMONG CARNIVORES

245

singletons or pairs have difficulty meeting the costs of reproduction, breeders will benefit directly from the presence of alloparents. If relatedness among adult group members is high, as appears common among carnivores (see Section III,E), alloparents will gain indirect fitness through their efforts. Costly reproduction is also associated with reproductive suppression of subordinates. Opportunities for subordinates to breed are often limited to marginal habitat, without a labor force of helpers, following risky dispersal (Brown, 1987; Emlen, 1991; Waser and Jones, 1983; Waser et al., 1994). Together with such constraints, energetically costly reproduction raises the probability that the cost will exceed the expected benefit of a subordinate’s breeding attempt, and suppression should be tolerated (Creel and Creel, 1991). D.

LIMITATIONS

Comparing our regression results with the preceding review of carnivore sociality and reproductive suppression, it is apparent that some hypotheses can be given strong comparative tests (e.g., that alloparents are associated with costly reproduction, Gittleman, 1989). Other hypotheses can be tested indirectly (e.g., that resource renewal rates affect the costs of sociality, Waser, 1981). For yet other hypotheses, it is difficult to devise a quantitative test using the comparative data available (e.g., the hypothesis that sociality is favored by cooperative hunting or cooperative prey defense). These last hypotheses are best evaluated by qualitatively reviewing the results of single-species studies, which also allows a better look at functional and mechanistic relationships among variables. Hypotheses that center on ecological variables in particular tend to be weakly tested by quantitative comparative analyses (Western, 1979; Moehlman, 1989). For example, the subtleties of food dispersion and renewal have not yet been adequately measured for most species (Woodroffe and Macdonald, 1993). Ecological variables usually must be treated categorically (e.g., diet in our analyses), and proxy variables are common (in our analyses diet proxies for food distribution, abundance, and renewal rate), both of which weaken the test. Physical variables and life history variables tend to be more easily quantifiable, and less variable, so that they are more suited to strong quantitative comparisons. Perhaps this partially underlies the result that relationships among life history traits are more often significant than relationships between life history and ecological variables (Read and Harvey, 1989; Lessels, 1991). The last two decades of field study have unveiled fascinating intricacy in carnivore societies. These observations, together with burgeoning theory,

246

SCOTT CREEL AND DAVID MACDONALD

have allowed many new hypotheses about carnivore social organization to be phrased. The parallel development of the comparative method has provided steadily more valid and powerful means of testing for predicted patterns within the variation found among carnivores. An important limitation on comparative analyses is the validity of the phylogenies on which they are based. Molecular techniques are causing rapid advances in this regard. Using Wozencraft’s phylogeny and Grafen’s regression technique, our analyses have confirmed some such patterns. However, we have also found that some hypotheses are poorly tested with this approach; it is not a panacea. Nonsignificant results can arise from the phylogenetic regression for at least three practically important reasons. First, variables are often inadequately measured for comparative tests to retain in practice the power they have in theory. Second, existing interspecific patterns may be obscured by the substantial intraspecific variation that typifies carnivores. And finally, the process of controlling phylogenetic effects can (and should) leave the method with little power to detect effects that are not dramatic. While comparative methods have an important role in testing broad hypotheses about social organization, these caveats serve as a reminder that we also cannot do without detailed field studies of single species.

VII. SUMMARY The variations within the Carnivora in social and nonsocial breeding, breeding group size, and reproductive suppression are reviewed. Evidence exists to suggest that these facets of carnivore behavior have evolved in response to one or more of at least five categories of selective pressures. These include ecological factors (such as the patterns of resource availability, dispersion and renewal, and constraints upon dispersal and philopatry) and sociological factors (such as cooperation in hunting, in defense against predators and competing conspecifics, and in alloparental care). There is evidence that reproductive suppression is favored by factors that increase the indirect fitness of subordinates (such as greater relatedness within the group or greater effectiveness of helping behavior), and by factors that constrain dispersal. Against the background of this review we undertook comparative analyses using phylogenetic regressions of sociality, group size, and reproductive suppression. This revealed an association between large groups and insectivory, suggesting that group living is favored by dependence on foods that are shared at little cost to a territory owner. However, although one or more of food abundance, dispersion, or renewal rate is likely to favor sociality, we cannot determine which because these

REPRODUCTIVE SUPPRESSION AMONG CARNIVORES

247

variables are confounded across diet types. The comparative analysis also revealed an association between larger groups and energetically costly reproduction, which is in accord with the hypothesis that sociality is favored by the benefits of alloparental care. Costly reproduction is also associated with reproductive suppression of subordinates. We conclude that, while quantitative comparativeanalyses are powerful tools for testing some hypotheses, others (particularly those centering on ecological variables) tend to be weakly tested by this approach and are best evaluated through detailed field study.

Acknowledgments We thank N. Creel, E. Geffen, J. Gittleman, H. Kruuk, C. Sillero-Zubiri, P. Waser, and R. Woodroffe for helpful comments, and A. Grafen for supplying software to run the phylogenetic regressions (PHYLO.REG 1.03) and for advice on its implementation. S . R. Creel was supported by an NSF-NATO postdoctoral fellowship and the Frankfurt Zoological Society, Project I 1 12/90.

References Alexander, R. D. (1974).The evolution of social behavior. Ann. Rev. Ecol. Syst. 5,325-383. Bacon, P. J., Ball, F. G., and Blackwell, P. G. (1991). A model for territory and group formation in a heterogeneous habitat. J . Theoret. Biol. 148, 433-444. Bacon, P. J., and Blackwell, P. G. (1993). A comparison of models for the evolution of group living. (Research rep.). United Kingdom: University of Sheffield, Department of Probability and Statistics. Bekoff, M., Diamond, J . , and Mitton, J. B. (I981).Life-history patterns and sociality in canids: Body size, reproduction and behavior. Oecologia 50, 386-390. Bekoff, M., Daniels, T.J., and Gittleman, J. L. (1984). Life history patterns and thecomparative social ecology of carnivores. Ann. Reu. Ecol. Syst. 15, 191-232. Bekoff. M., and Wells, M. C. (1980). The social ecology ofcoyotes. Sci. Amer. 242,130-148. Bekoff. M.. and Wells, M. C. (1982). The behavioral ecology of coyotes: Social organization, rearing patterns, space use, and resource defense. Z. Tierpyschol. 60, 281-305. Brown, J. L. (1974). Alternate routes to sociality in jays-with a theory for the evolution of altruism and communal breeding. Amer. Zoo. 14, 63-80. Brown, J. L. (1987). “Helping and Communal Breeding in Birds.” (Princeton University Press, Princeton. Brown, J. L., Brown, E. R., Brown, S. D., and Dow, D. D. (1982). Helpers: Effects of experimental removal on reproductive success. Science 215,42122. Bygott, J. D., Bertram, B. C. R., and Hanby, J. P. (1979). Male lions in large coalitions gain reproductive advantages. Nature (London) 282, 838-840. Camenzind, F. J. (1978).Behavioral ecology ofcoyotes on the National Elk Refuge, Jackson, Wyoming. In “Coyotes: Biology, Behavior and Management” (M. Bekoff ed.), pp. 267-294. Academic Press, New York. Caraco, T., and Wolf, L. L. (1975). Ecological determinants of group sizes of foraging lions. Am. Nut. 109, 343-352.

248

SCO’IT CREEL AND DAVID MACDONALD

Caro, T. M. (1994). “Cheetahs of the Serengeti Plains: Group Living in an Asocial Felid.” University of Chicago Press, Chicago. Caro, T. M., and Collins, D. A. (1987). Male cheetah social organization and territoriality. Ethology 74, 52-64. Carr, G. M., and Macdonald, D. W. (1986). The sociality of solitary foragers: A model based on resource dispersion. Anim. Behav. 34, 1540-1549. Cheeseman, C. L., Mallinson, P. J., Ryan, J., and Wilesmith, J. W. (1993). Recolonisation by badgers in Gloucestershire. In “The Badger” (T. J. Hayden, ed.), pp. 78-93. Royal Irish Academy, Dublin, Ireland. Chepko-Sade, S. D., and Halpin, Z. T. (1987). “Mammalian Dispersal Patterns: the Effects of Social Structure on Population Genetics.” University of Chicago Press, Chicago. Cheverud, J. M., Dow, M. M., and Leutenegger, W. (1985). The quantitative assessment of phylogenetic constraints in comparative analyses: Sexual dimorphism on body weight among primates. Evolution 39, 1335-1351. Clutton-Brock, T. H. (1991). “The Evolution of Parental Care.” Princeton University Press, Princeton, New Jersey. Clutton-Brock, T. H., and Harvey, P. H. (1977). Primate ecology and social organization. J . 2001.Lond. 183, 1-39. Collier, G. E., and O’Brien, S. J. (1985). A molecular phylogeny of the Felidae: Immunological distance. Evolution 39, 473-487. Cooper, S. M. (1991). Optimal hunting group size: The need for lions to defend their kills against loss to spotted hyaenas. 4 f r . J . Ecol. 29, 130-136. Corbett, L. K. (1988). Social dynamics of a captive dingo pack: Population by dominant female infanticide. Ethology 78, 177-178. Creel, S. R. (1990). How to measure inclusive fitness. Proc. Roy. SOC. Lond., Ser. 241, 229-23 1.

Creel, S . R., and Creel, N. M. (1991). Energetics, reproductive suppression and obligate communal breeding in carnivores. Beh. Ecol. Sociobiol. 28, 263-270. Creel, S. and Creel, N. M. (in press). Communal hunting and pack size in African wild dogs (Lycaon pictus). Anim. Behav. Creel, S. R.. Creel, N. M., Wildt, D. E.. and Monfort, S. L. (1992). Behavioral and endocrine mechanisms of reproductive suppression in Seregeti dwarf mongooses. Anim. Behau. 43,231-245. Creel, S. R., Monfort, S. L., Wildt, D. E., and Waser, P. M. (1991). Spontaneous lactation is an adaptive result of pseudopregnancy. Nature (London) 351,660-662. Creel, S . R., and Waser, P. M.(1991). Failures of reproductive suppression in dwarfmongooses Helogale paruula: Accident or adaptation? Behav. Ecol. 2, 7-15. Creel, S. R., and Waser, P. M. (1994). Inclusive fitness and reproductive strategies in dwarf mongooses. Behav. Ecol. 5,339-348. Creel, S. R., Wildt, D. E., and Monfort, S. L. (1993). Androgens, aggression and reproduction in wild dwarf mongooses: A test of the challenge hypothesis. A m . Nar. 141, 816-825.

Cresswell, W. J., Harris, S., Cheeseman, C. L., and Mallinson, P. J. (1992). To breed or not to breed: An analysis of the social and density-dependent constraints on the fecundity of female badgers (Meles meles). Proc. Roy. SOC.Lond., Ser. B 338, 393-407. Da Silva, J., Macdonald, D. W., and Evans, P. G. H. (1993). Net costs of group living in a solitary forager, the Eurasian badger. Behav. Ecol. 5(1). Davies, N. B. (1991). Mating systems. I n “Behavioural Ecology,” ( J . R. Krebs and N. B. Davies, eds.), 3rd ed., pp. 263-294. Blackwell, London. Dietz, J. M. (1984). Ecology and social organization ofthe maned wolf. Smirhsonian Contrib. ZOO^. 392, 1-51.

REPRODUCTIVE SUPPRESSION AMONG CARNIVORES

249

Doncaster, C. P., and Macdonald, D. W. (1991). Drifting temtoriality in the red fox Vulpes vulpes. J . Anim. Ecol. 60,423-439. Doncaster, C. P., and Macdonald, D. W. (1992). Optimal group size for defending heterogenous distributions of resources: A model applied to red foxes, Vulpes uulpes, in Oxford city. J . Theorer. Biol. 159, 189-198. Doncaster, C. P., and Woodroffe, R. B. (1993). Den site can determine shape and size of badger territories: Implications for group living. Oikos 66, 88-93. Emlen, S. T. (1991). Evolution of cooperative breeding in birds and mammals. In “Behavioural Ecology” ( J . R. Krebs and N. B. Davies, eds.), 3rd ed., pp. 301-337. Blackwell, London. Emlen, S. T., and Oring, L. W. (1977). Ecology, sexual selection and the evolution of mating systems. Science 197, 215-223. Emlen, S. T., and Wrege, P. (1989). A test of alternate hypotheses for helping behavior in white-fronted bee-eaters of Kenya. Beh. Ecol. Sociobiol. 23, 305-315. Erlinge, S., and Sandell, M. (1986). Seasonal changes in the social organization of male stoats, Mustela erminea: An effect of shifts between two decisive resources. Oikos 47, 57-62.

J$tes, J. A . (1989). Adaptations for aquatic living by carnivores. In “Carnivore Behavior, Ecology and Evolution” ( J . Gittleman, ed.). pp. 242-282. Cornell University Press, Ithaca, New York. Estes, R. D., and Goddard, J. (1967). Prey selection and hunting behavior of the African hunting dog. J . Wildl. M g m t . 31, 52-70. Evans, P. G. H., Macdonald, D. W., and Cheeseman, C. L. (1989). Social structure of the Eurasian badger (Meles meles): Genetic evidence. J . Zool. (Lond.)218, 587-595. Ewer, R. F. (1973). “The Carnivores.” Cornell University Press, Ithaca, New York. Fanshawe, J. H., and Fitzgibbon. C. D. (1993). Factors influencing the hunting success of an African wild dog pack. Anim. Behav. 45,479-490. Felsenstein, J. (1985). Phylogenies and the comparative method. Am. Nar. l25, 1-15. Felsenstein, J. (1988). Phylogenies and quantitative characters. Ann. Rev. Ecol. Sysr. 19, 445-47 I . Frame, G. W., and Frame, L. H. (1981). “Swift and Enduring: Cheetahs and Wild Dogs of the Serengeti.” E. P. Dutton, New York. Frame, L. H., Malcolm, J. R., Frame, G. W., and van Lawick, H. (1979). Socialorganisation of African wild dogs Lycaon picrus on the Serengeti plains, 1%7-1978. Z. Tierpschol. 50,225-249.

Frank, L. G. (1986). Social organization of the spotted hyaena Crocura crocura. I. Demography. Anim. Behav. 34, 1500-1509. Frank, L. G., Davidson, J. M., and Smith, E. R. (1985). Androgen levels in the spotted hyaena Crocura crocuta: The influence of social factors. J . Zool., Lond. 206,525-531. Fretwell, S . D. (1972). “Populations in a Seasonal Environment.” Princeton University Press, Princeton, New Jersey. Fretwell. S. D., and Lucas, H. J. (1970). On territorial behavior andotherfactors influencing habitat distribution in birds. Acta Biotheor. 19, 16-36. Fuller, T. K., and Kat, P. W. (1990). Movements, activity and prey relationships of African wild dogs Lycuon pictus near Aitong, southwest Kenya. Afr. J. Ecol. 28,330-350. Fuller, T. K., Kat, P. W., Bulger, J. B., Maddock, A. H., Ginsberg, J. R., Burrows, R., McNutt, J. W., and Mills, M. G. L. (1992). Population dynamics of African wild dogs. I n “Wildlife 2001: Populations” (D. R. McCullough and R. H. Barret, eds.), pp. 1125-1 139. Elsevier Applied Science, New York. Garland, T., Harvey, P. H., and Ives, A. R. (1992). Procedures for the analysis of comparative data using phylogenetically independent contrasts. Syst. Biol. 41, 18-32.

250

SCOTT CREEL AND DAVID MACDONALD

Gasaway, W. C., Mossestad, K. T., and Stander, P. E. (1991). Food acquisition by spotted hyaenas in Etosha N. P., Namibia: Predation versus scavenging. Afr. J. Ecol. 29,64-75. Geffen, E., Gompper, M., Gittleman, J. L., Macdonald, D. W., and Wayne, R. K. (submitted). A comparative analysis of canid reproduction. Am. Nut. Geffen, E., Heffner, R., Macdonald, D. W., and Ucko, M. (1992). Habitat selection and home range use in the Blanford’s fox Vulpes cana: Compatability with the resource dispersion hypothesis. Oecologia 91, 75-81. Geffen, E., and Macdonald, D. W. (1992). Small size and monogamy: Spatial organization of Blanford’s foxes, Vulpes cana. Anim. Behav. 44, 1123-1 130. Georges, M., Lequarre, A. S., Castelli, M., Haset, R., and Vassart, G. (1988). DNA fingerprinting in domestic animals using four different minisatellite probes. Cytogenet. Cell. Genet. 47, 127-131. Gittleman, J. L. (1985a). Functions of communal care in mammals. I n “Evolution: Essays in Honour of John Maynard Smith” (P. H. Harvey and M. Slatkin, eds.), pp. 187-205. Cambridge Univ Press, Cambridge. Gittleman, J. L. (1985b). Carnivore body size: Ecological and taxonomic correlates. Oecologia 67, 540-554. Gittleman, J. L. (1986). Carnivore life history patterns: Allometric, phylogenetic and ecological associations. Am. Nut. 217, 744-771. Gittleman, J. L. (1989). Carnivore group living: Comparative trends. In “Carnivore Behavior, Ecology and Evolution” ( J . L. Gittleman, ed.), pp. 183-207. Cornell Univ Press, Ithaca, New York. Gittleman, J. L. (1993). Carnivore life histories: A re-analysis in the light of new models. In “Mammals As Predators” (N. Dunstone and M. Gorman, eds.). Symp. Zool. SOC. Lond. ~ , 6 5 - 8 6 . Gittleman, J. L., and Harvey, P. H. (1982). Carnivore home-range size, metabolic needs and ecology. Beh. Ecol. Sociobiol. 10, 57-63. Gittleman, J. L., and Kot, M. (1990). Adaptation: Statistics and a null model for estimating phylogenetic effects. Syst. Zool. 39, 227-241. Gittleman, J. L., and Luh, H-K. (1992). On comparing comparative methods. Annu. Rev. Eco~.Syst. 2, 383-404. Gittleman, J. L., and Luh, H-K. (in press). Phylogeny, evolutionary models, and comparative methods: A simulation study. I n “Phylogenetics and Ecology” (P. Eggleton and D. Vane-Wright, eds.). Academic Press, London. Gittleman, J. L., and Oftedal, 0. T. (1987). Comparative growth and lactation energetics in carnivores. Symp. Zool. SOC.Lond. 57, 41-77. Gompper, M. E., and Gittleman, J. L. (1991). Home range scaling: Intraspecific and comparative trends. Oecologia 87, 343-348. Gorman, M. L. (1979). Dispersion and foraging of the small Indian mongoose, Herpestes auropunctatus Carnivora: Viverridae, relative to the evolution of social viverrids. J. Zool. Lond. 187, 65-73. Gottelli, D., and Sillero-Zubiri, C. (1992). The Ethiopian wolf An endangered endemic canid. Oryx 26,205-214. Gottelli, D., Sillero-Zubiri, C., Applebaum, D., Girman, D., Roy, M., Garcia-Moreno, J., Ostrander, E., and Wayne, R. K. (1995). Molecular genetics of the most endangered canid: The Ethiopian wolf, Canis simensis. Molecular Ecology 3. Grafen, A. (1989). The phylogenetic regression. Phil. Trans. R. SOC.Lond., B 326, 119-157. Grafen, A. (1992). The uniqueness of the phylogenetic regression. J. Theoret. Biol. 156, 405-423.

Hamilton, P. H. (1976). The movements of leopards in Tsavo national park, Kenya, as

REPRODUCTIVE SUPPRESSION AMONG CARNIVORES

25 1

determined by radio-tracking. Unpublished master’s thesis, University of Nairobi, Nairobi, Kenya. Hamilton, W. D. (1963). The evolution of altruistic behavior. Am. Nat. 97, 354-356. Hamilton, W. D. (1964). The genetical evolution of social behavior. J . Theore/. Biol. 7, 1-52. Harvey, P. H., and Mace, G. M. (1982). Comparisons between taxaand comparative trends: Problems of methodology. I n “Current Problems in Sociobiology” (King’s College Sociobiology Group, eds.), pp. 343-361. Cambridge University Press, Cambridge. Harvey, P. H., and Pagel, M. D. (1991). “The Comparative Method in Evolutionary Biology.” Oxford University Press, Oxford. Harvey, P. H., and Purvis, A. (1991). Comparative methods for explaining adaptations. Nature (London) 351, 619-624. Hersteinsson. P., and Macdonald, D. W. (1982). Some comparisons between red and arctic foxes, Vulpes vulpes and Alopex lagopus, as revealed by radio tracking. Symp. Zool. SOC.Lond. 49, 259-289. Hersteinsson, P., and Macdonald, D. W. (1992). Interspecific competition and the geographical distribution of red and arctic foxes. Vulpes vulpes and Alopex lagopus. Oikos 634, 505-5 15. Hofer, H., and East, M. (1993). The communting system of Serengeti spotted hyaenas: How a predator copes with migratory prey. I. Social organization. Anim. Behau. 46, 547-557. Jennions, M. D.. and Macdonald, D.W. (1994). Cooperative breeding in mammals. TREE 9(3), 89-93. Johnsingh. A . J. T. (1982). Reproductive and social behaviour of the Dhole, Cuon alpinus Canidae. J . Zool. (Lond.) 198, 443-463. Johnson, M. L., and Gaines, M. S. (1990). Evolution of dispersal: Theoretical models and empirical tests using birds and mammals. Ann. Rev. Ecol. Syst. 21,449-480. Johnston, J. (1984). “Econometric Methods,” 3rd ed. McGraw-Hill International Editions, Tokyo. Keane, B., Waser, P. M., Creel, S. R., Creel, N. M., Elliott, L. F., and Minchella, D. J. (1994). Subordinate reproduction in dwarf mongooses. Anim. Behau. 47, 65-75. Kerby, G., and Macdonald, D. W. (1988). Cat society and the consequences of colony size. 1n “The Domestic Cat-The Biology of its Behaviour” (D. C. Turner and P. Bateson, eds.), pp. 67-81. Cambridge University Press, Cambridge. Kingdon, J. (1977). “East African Mammals, Volume 3: Carnivores.” Academic Press, London. Kleiman, D. G. (1977). Monogamy in mammals. Q. Rev. Biol. 52, 39-69. Kleiman, D. G., and Malcolm, J. R. (1981). The evolution of male parental investment in mammals. In ”Parental Care in Mammals” (D. J. Gubernick and P. H. Klopfer, eds.), pp. 347-387. Plenum Press, New York. Koenig, W. D., and Pitelka, F. A. (1981). Reproductive success, group size and the evolution of cooperative breeding in the acorn woodpecker. Am. Nat. 117, 421-443. Kruuk, H. (1972). “The Spotted Hyaena: a Study of Predation and Social Behavior.” University of Chicago Press, Chicago. Kruuk, H. (1975). Functional aspects of social hunting by carnivores. I n “Function and Evolution in Behavior” (G. Baerends, C. Beer, and A. Manning, eds.), pp. 119-141. Clarendon Press, Oxford, England. Kruuk, H. (1978). Spatial organisation and territorial behaviour of the European badger, Meles meles. J . Zool. Lond. 184, 1-19. Kruuk, H. (1989). “The Social Badger.” Oxford Univ Press, Oxford.

252

SCOTT CREEL AND DAVID MACDONALD

Kruuk, H.,and Macdonald, D. W. (1985). Group territories of carnivores: Empires and enclaves. I n “Behavioral Ecology: Ecological Consequences of Adaptive Behavior” (R. M. Sibley and R. H. Smith, eds.), pp. 521-536. Blackwell, Oxford. Kruuk, H., and Moorhouse, A. (1991). The spatial organization of otters (Lutra lurra) in Shetland. J . Zool. Lond. 224, 41-57. Kruuk, H., and Parish, T. (1982). Factors affecting population density, group size and territory size of the European badger. J . Zool. Lond. 196, 31-39. Kuhme, W. D. (1%5). Freilandenstudien zur soziologie des hyanahundes Lycaon pictus lupinus Thomas 1902. Z. Tierpsychol. 22,495. Lack, D. (1968). “Ecologial Adaptations for Breeding in Birds.” Methuen, London. Lamprecht, J. (l978a). The relationship between food competition and foraging group size in some larger carnivores: A hypothesis. Z . Tierpsychol. 46, 337-343. Lamprecht, J. (1978b). On diet, foraging behavior and interspecific food competition of jackals in Serengeti National Park, East Africa. Z . Saugetierk. 43, 210-223. Lamprecht, J. (1981). The function of social hunting in larger terrestrial carnivores. Mammal. Rev. 11, 169-179. Lehman, N., Clarkson, P., Mech, L. D., Meier, T. J., and Wayne, R. K. (1992). The use of DNA fingerprinting and mitochondria1 DNA to study genetic relationships within and among wolf packs. Behav. Ecol. Sociobiol. 30,83-94. Leonard, M. L., Horn, A. G., and Eden, A. F. (1989). Doesjuvenile helping enhance breeder reproductive success: A removal experiment on moorhens. Beh. Ecol. Sociobiol. 25, 357-361. Lessels, C. M. (1991). The evolution of life histories. I n “Behavioural Ecology: An Evolutionary Approach” (J. R. Krebs and N. B. Davies, eds.), 3rd ed., pp. 32-68. Blackwell, Oxford. Lindstrom, E. R. (1993). Group formation and group persistence: On realism, elegance and simplification. Anim. Behau. 46, 824-826. Litvaitis, J. A., Sherburne, J. A., and Bissonette, J. A. (1986). Bobcat habitat use and home range size in relation to prey density. J . Wildl. Mgmt. 50, 110-117. Maas, B. (1993). Behavioural ecology and social organization of the bat-eared fox in the Serengeti National Park, Tanzania. Unpublished doctoral dissertation, Cambridge University, Cambridge, United Kingdom. Macdonald, D. W. (1976). Food caching by red foxes and some other carnivores. Z. Tierpsychol. 42, 170-185. Macdonald, D. W. (1979). “Helpers” in fox society. Nature (London) 282, 69-71. Macdonald, D. W. (1980). Social factors affecting reproduction amongst red foxes. In “The Red Fox: Symposium on Behavior and Ecology” (E. Zimen, ed.), pp. 131-183. Junk, The Hague. Macdonald, D. W. (1981). Resource dispersion and the social organization of the red fox, Vulpes uulpes. In “Proceedings of the World-Wide Furbearer Conference” (J. A. Chapman and D. Ursley, eds.). pp. 918-949. Frostburg, Maryland. Macdonald, D. W. (1983). The ecology of carnivore social behaviour. Nature (London)301, 379-384. Macdonald, D. W. (1984). “The Encyclopedia of Mammals.” Facts on File, New York. Macdonald, D. W. (1986). A meerkat volunteers for guard duty so its comrades can live in peace. Smithsonian Magazine, 54-65. Macdonald. D. W. (1987). “Running with the Fox.” Unwin Hyman Press, London. Macdonald, D. W. (1992). “The Velvet Claw.” BBC Books, London. Macdonald, D. W., Apps, P. J., Carr, G . , and Kerby, G. (1987). Social behaviour, nursing coalitions and infanticide in a colony of farm cats. Ado. Ethol. 28, 1-66.

REPRODUCTIVE SUPPRESSION AMONG CARNIVORES

253

Macdonald, D. W., Brown, L., Canbolat, H. F., and Yerli, S. (in press). The behaviour of red foxes, Vulpes vulpes, caching the eggs of loggerhead turtles. Carreta carreta. J . Mamm. Macdonald, D. W., and Carr, G. M. (1989). Food security and the rewards of tolerance. Spec. Publ. British. Ecol. SOC. 8, 75-99. Macdonald, D. W.. and Courtenay, 0. (submitted). Enduring social relationships in a widely dispersing population of crab-eating foxes (Cerdocyon thous) in Amazonian Brazil. J. Zool. (Lond.). Malcolm, J. R. (1979). Social organization and the communal rearing of pups in African wild dogs Lycaon pictus. Unpublished doctoral dissertation, Harvard University, Cambridge, Massachusetts. Malcolm, J. R. (1985). Socioecology of bat-eared foxes Otocyon megalotis. J . Zool. (Lond.) 208,457-467. Malcolm, J. R., and Marten, K. (1982). Natural selection and the communal rearing of pups in African wild dogs Lycaon pictus. Beh. Ecol. Sociobiol. 10, 1-13. Martin, L. D. (1989). Fossil history of the terrestrial Carnivora. In “Carnivore Behavior, Ecology and Evolution” ( J . Gittleman, ed.), pp. 536-568. Cornell University Press, Ithaca, New York. Martins, E. P., and Garland, T. (1991). Phylogenetic analyses of the correlated evolution of continuous characters: A simulation study. Evolution 45, 534-557. McNab, B. K. (1989). Basal rate of metabolism, body size, and food habits in the order Carnivora. In “Carnivore Behavior, Ecology and Evolution” (J. Gittleman, ed.), pp. 335-354. Cornell University Press, Ithaca, New York. Mech, L. D. (1970). “The Wolf.” Doubleday Natural History Press, New York. Messier, F., and Barrette, C. (1982). The social system of the coyote Canis latrans in a forested habitat. Can. J . Zool. 60, 1743-1753. Milinski, M., and Parker, G. A. (1991). Competition for resources. In “Behavioural Ecology: An Evolutionary Approach” ( J . R. Krebs and N. B. Davies. eds.), 3rd ed., pp. 137-168. Blackwell, Oxford. Mills, M. G. L. (1982). Factors affecting group size and territory size of the brown hyaena, Hyaena brunnea. in the southern Kalahari. Z. Tierpsychol. 48, 113-141. Mills, M. G. L. (1983). Mating and denning behavior of the brown hyaena, Hyaena brunnea, and comparisons with other hyaenidae. Z. Tierpsychol. 63,331-342. Mills, M. G. L. (1985). Related spotted hyaenas forage together but do not cooperate in rearing young. Nature (London) 316,61-62. Mills, M. G. L. (1989a). “Kalahari Hyaenas: The Behavioural Ecology of Two Species.” Unwin Hyman Press, London. Mills, M. G. L. (1989b). The comparative behavioral ecology of hyaenas: The importance of diet and food dispersion. In “Carnivore Behavior, Ecology and Evolution” (J. L. Gittleman, ed.), pp. 125-142. Cornell University Press, Ithaca, New York. Mills, M. G. L., and Mills, M. E. J. (1982). Factors affecting the movement patterns of brown hyaenas, Hyaena brunnea, in the southern Kalahari. S . Afr. J . Wildl. Res. 12, I 1 1-1 17. Moehlman, P. D. (1979). Jackal helpers and pup survival. Nature (London) 277, 382383. Moehlman, P. D. (1983). Socioecology of silverbacked and goldenjackals (Canis mesomelas and Canis aureus). In “Advances in the Study of Mammalian Behavior” (J. F. Eisenberg and D. G. Kleiman, eds.), pp. 423-453. American Society of Mammalogists, Lawrence, Kansas. Moehlman, P. D. (1986). Ecology of cooperation in canids. In “Ecological Aspects of Social

254

SCOTT CREEL AND DAVID MACDONALD

Evolution: Birds and Mammals” (D.I. Rubenstein and R. W. Wrangham, eds.), pp. 64-86. Princeton University Press, Princeton, New Jersey. Moehlman, P. D. (1989). Intraspecific variation in canid social systems. In “Carnivore Behavior, Ecology and Evolution” ( J . L. Gittleman, ed.). pp. 143-163. Cornell Univ Press, Ithaca, New York. Mumme, R. N. (1992). Do helpers increase reproductive success: An experimental analysis in the Florida scrub jay. Beh. E c d . Sociobiol. 31, 319-328. Mumme, R. N., Koenig, W. D., and Ratnieks, F. L. W. (1989). Helping behaviour, reproductive value, and the future component of fitness. Anim. Behau. 38, 331-343. Nel, J. A. J., Mills, M. G. L., and van Aarde, R. J. (1984). Fluctuating group size in bateared foxes Otocyon megalotis in the south-westem Kalahari. J. Zool. (Lond.) 203, 294-298. O’Brien, S. J., Collier, G. E., Benveniste, R. E., Nash, W. G., Newman, A. K.,Simonson, J. M., Eichelberger, M. A., Seal, U. S., Bush, M. E., and Wildt, D. E. (1987). Setting the molecular clock in Felidae: The great cats. In “Tigers of the World” (R.L. Tilson, ed.). pp. 10-26. Noyes Publications, Park Ridge, New Jersey. O’Brien, S. J., Nash, W. G., Wildt, D. E., Bush, M. E.. and Benveniste. R. E. (1985). A molecular solution to the riddle of the giant panda’s phylogeny. Nature (London)317, 140- 144. Oftedal, 0. T., and Gittleman, J. L. (1989). Patterns of energy output during reproduction in carnivores. I n “Carnivore Behavior, Ecology, and Evolution” ( J . L. Gittleman, ed.), pp. 353-378. Cornell Univ. Press, Ithaca, New York. Owens, D. D., and Owens, M. J. (1984). Helping behaviour in brown hyaenas. Nature (London)308,843-845. Packard, J. M., Mech, L. D., and Seal, U. S. (1983). Social influences on reproduction in wolves. I n “Wolves in Canada and Alaska: Their Status, Biology and Management” (L. N. Carbyn, ed.), pp. 78-85. Canadian Wildlife Service, Edmonton. Packard, J. M., Seal, U. S., Mech, L. D., and Plotka, E. D. (1985). Causes of reproductive failure in two family groups of wolves. Canis lupus. Z. Tierpsychol. 68, 24-40. Packer, C. (1986). The ecology of sociality in felids. In “Ecological Aspects of Social Evolution: Birds and Mammals” (D. I. Rubenstein and R. W. Wrangham, eds.), pp. 429-45 1. Princeton University Press, Princeton, New Jersey. Packer, C., Gilbert, D. A., Pusey, A. E., and O’Brien, S. J. (1991). A molecular genetic analysis of kinship and cooperation in African lions. Nature (London)351, 562-565. Packer, C., Herbst, L.. Pusey, A. E., Bygott, J. D., Hanby, J. P., Cairns. S. J., and Borgerhoff Mulder, M. (1988). Reproductive success in lions. In “Reproductive Success” (T. H. Clutton-Brock, ed.). pp. 363-383. University of Chicago Press, Chicago. Packer, C., Lewis, S., and Pusey, A. (1992). A comparative study of nonoffspring nursing. Anim. Behav. 43,265-281. Packer, C., and Pusey, A. E. (1983). Adaptations of female lions to infanticide by incoming males. Am. Nat. El, 716-728. Packer, C., Scheel, D., and Pusey, A. E. (1990). Why lions form groups: Food is not enough. Am. Nat. l36, 1-19. Pagel, M. D. (1992). A method for the analysis of comparative data. J. Theoret. Biol. 156, 431-442. Pagel, M. D., and Harvey, P. H. (1988). Recent developments in the analysis of comparative data. Q. Rev. Biol. 63, 413-440. Pagel, M. D., and Harvey, P. H. (1989). Comparative methods for examining adaptation depend on evolutionary models. Folia. Primat. 53, 203-220.

REPRODUCTIVE SUPPRESSION AMONG CARNIVORES

255

Pagel, M. D., and Harvey, P. H. (1992). On solving the correct problem: Wishing does not make it so. J. Theoret. B i d . 156,425-430. Parker, G. A. (1978). Searching for mates. I n “Behavioural Ecology: An Evolutionary Approach” ( J . R. Krebs and N. B. Davies, eds.), pp. 214-244. Blackwell, Oxford. Pulliam, H. R., and Caraco, T. (1984). Living in groups: Is there an optimal group size? I n “Behavioural Ecology” ( J . R. Krebs and N. B. Davies, eds.), 2nd ed. pp. 122-147. Sinauer, Sunderland, Massachusetts. Purvis, A. (1991).Comparative analysis by independent contrasts: A statistical package f o r the Apple Macintosh. Private distribution, Department of Zoology,Oxford University, United Kingdom. Rabenold, K. N. (1984). Cooperative enhancement of reproductive success in tropical wren societies. Ecology 65, 871-885. Rasa, 0. A. E. (1973). Intrafamilial sexual repression in the dwarf mongoose Helogale undulata rufula. Naturwissenschaften 60,303-304. Rasa. 0. A. E. (1986). Coordinated vigilance in dwarf mongoose family groups: The “watchman’s song” hypothesis and the costs of guarding. Z. Tierpsychol. 71, 340-344. Read, A., and Harvey, P. H. (1989). Life history differences among the eutherian radiations. J . ZOO^. (Lond.)219, 329-353. Reich, A. (1981). The behaviour and ecology of the African wild dog. Lycaon pictus, in the Kruger National Park. Unpublished doctoral dissertation, Yale University, New Haven, Connecticut. Reyer, H. U. (1980). Flexible helper structure as an ecological adaptation in the pied kingfisher (Cetyle rudis). Beh. Ecol. Sociobiol. 6, 219-227. Richardson, P. R. K. (1987). Aardwolf mating system: Overt cuckoldry in an apparently monogamous mammal. S . Afr. J . Sci. 83,405-410. Ridley, M. (1983). “The Explanation of Organic Diversity: The comparative Method and Adaptations for Mating.” Oxford University Press, Oxford. Rood, J. P. (1975). Population dynamics and food habits of the banded mongoose. E. Afr. Wildl. J . 13, 89-1 11. Rood, J. P. (1978). Dwarf mongoose helpers at the den. Z. Tierpsychol. 48, 277-287. Rood, J. P. (1980). Mating relationships and breeding suppression in the dwarf mongoose. Anim. Behav. 28, 143-150. Rood, J. P. (1983). The social system of the dwarf mongoose. I n “Advances in the Study of Mammalian Behavior” ( J . F. Eisenberg and D. G . Kleiman, eds.), pp. 454-488. American Society of Mammalogists, Lawrence, Kansas. Rood, J. P. (1986). Ecology and social evolution in the mongooses. I n “Ecological Aspects of Social Evolution: Birds and Mammals” (D. I. Rubenstein and R. W. Wrangham, eds.), pp. 131-152. Princeton University Press, Princeton, New Jersey. Rood, J. P. (1987). Dispersal and intergroup transfer in the dwarfmongoose. In “Mammalian Dispersal Patterns: The effects of Social Structure on Population Genetics” (R. D. Chepko-Sade and Z. P. Halpin, eds.), pp. 85-103. University ofChicago Press, Chicago. Rood, J. P. (1990). Group size, survival, reproduction, and routes to breeding in dwarf mongooses. Anim. Behav. 39, 566-572. Russell, J. K. (1983). Altruism in coati bands: Nepotism or reciprocity? In ”Social Behavior of Female Vertebrates” (S. K. Wasser, ed.), pp. 263-290. Academic Press, New York. Sandell, M. (1989). The mating tactics and spacing patterns of solitary carnivores. I n “Carnivore Behavior, Ecology and Evolution” ( J . L. Gittleman, ed.), pp. 164-182. Cornell University Press, Ithaca, New York. Sapolsky, R. M. (1992). Neoroendocrinology of the stress-response. I n “Behavioral Endocri-

256

SCOTT CREEL AND DAVID MACWNALD

nology” (J. 9. Becker, S. M. Breedlove, and D. Crews, eds.), pp. 287-324. MIT Press, Cambridge, Massachusetts . Schaller, G. B. (1%7). “The Deer and the Tiger.” University of Chicago Press, Chicago. Schaller, G. 9. (1972). “The Serengeti Lion: A Study of Predator-Prey Relations.” University of Chicago Press, Chicago. Schaller, G. 9.. and Vasconelos, J. M. C. (1978). Jaguar predation on capybara. Z. Saugetierk. 43, 2%-301. Scheel, D., and Packer, C. (1991). Group hunting behaviour of lions: A search for cooperation. Anim. Behav. 41,697-710. Schoener, T. W. (1974). Resource partitioning in ecological communities. Science 185, 27-39.

Selander, R. K. (1964). Speciation in wrens of the genus Campylorrhinchus, Univ. Cal. Pub/. Z O O / . 74, 1-224. Sillero-Zubiri, C. (1994). The behauioural ecology of the Efhiopian WOKCanis simensis. Unpublished doctoral dissertation, Oxford University, Oxford, United Kingdom. Sillero-Zubiri, C., and Gottelli, D. (1992). Population ecology of spotted hyaena in an equatorial mountain forest. J. Afr. Ecol. 30, 292-300. Sillero-Zubiri, C., Gottelli, D., and Macdonald, D. W. (in press). Dispersal and mating system of the Ethiopian wolf, Canis simensis. Beh. Ecol. Sociobiol. Smith, J. L. D., and McDougal, C. (1991). The contribution of variance in lifetime reproduction to effective population size in tigers. Cons. Bio. S(4), 484-490. Stacey, P. 9. (1979). Habitat saturation and communal breeding in the acorn woodpecker. Anim. Behau. 27, 1153-1 166. Stacey, P. B., and Ligon, J. D. (1987). Territory quality and dispersal options in the acorn woodpecker, and a challenge to the habitat saturation model of cooperative breeding. Am. Nut. 130,654-676. Stander, P. E., and Albon. S. D. (1993). Hunting success of lions in a semi-arid environment. Symp. Zool. SOC.Lond. 65, 127-143. Steel, R. G. D., and Torrie, J. H. (1980). “Principle and Procedures of Statistics.” McGrawHill, New York. Sunquist, M. E. (1981). The social organization of tigers Panthera tigris in Royal Chitawan National Park, Nepal. Smithsonian Cnfrib. Zool. 336, 1-98. Sunquist, M. E., and Sunquist, F. C. (1989). Ecological constraints on predation by large felids. In “Carnivore Behavior, Ecology and Evolution” ( J . L. Gittleman, ed.), pp. 283-301. Cornell University Press, Ithaca, New York. van Heerden, J., and Kuhn, F. (1985). Reproduction in captive hunting dogs, Lycaon pictus. S . Afr. J . Wildl. Res. 15, 80-84. van Orsdol, K.G.. Hanby, J. P., and Bygott, J. P. (1985). Ecological correlates of lion Panthera leo social organization. J . Zool. (Lond.) 206, 97-112. an Valkenburgh, 9. (1989). Carnivore dental adaptations and diet: A study oftrophic diversity within guilds. In “Carnivore Behavior, Ecology and Evolution” (J. L. Gittleman. ed.), pp. 410-436. Cornell University Press, Ithaca, New York. Vehrencamp, S. L. (1983). A model for the evolution of despotic versus egalitarian societies. Anim. Behav. 31,667-682. Vehrencamp, S . L., Koford, R. R., and Bowen, 9. F. (1988). The effect of breeding-unit size on fitness components in groove-billed anis. In “Reproductive Success” (T. CluttonBrock, ed.), pp. 291-304. University of Chicago Press, Chicago. von Schantz, T. (1981). Female cooperation, male competition and dispersal in red fox. Vulpes vulpes. Oikos 31, 63-68.

REPRODUCTIVE SUPPRESSION AMONG CARNIVORES

257

von Schantz, T. (1984). Carnivore social behaviour-does it need patches? Nature (London) 307, 388-390. Ward, R. M. P., and Krebs, C. J. (1985). Behavioural responses of lynx to declining snowshoe hare abundance. Can. J. Zool. 63, 2817-2824. Waser, P. M. (1980). Small nocturnal carnivores: Ecological studies in the Serengeti. Afr. J . EcoI. 18, 167-185. Waser, P. M. (1981). Sociality or territorial defense: The influence of resource renewal. Beh. Ecol. Sociobiol. 8, 231-237. Waser, P. M., Creel, S. R., and Lucas, J. R. (1994). Death and disappearance: Estimating mortality risks associated with philopatry and dispersal. Beh. Ecol. 5, 135-141. Waser, P. M., Elliott, L. F., Creel, N. M., and Creel, S. R. (in press). Habitat variation and mongoose demography. I n “Serengeti 11: Research, Management and Conservation of an Ecosystem” (A. R. E. Sinclair and P. Arcese, eds). University of Chicago Press, Chicago. Waser, P. M.. and Jones, W. T. (1983). Natal philopatry among solitary mammals. Q. Rev. Biol. 58, 355-390. Waser, P. M., Keane, B., Creel, S. R., Elliott, L. F., and Minchella, D. J. (1994). Possible male coalitions in a solitary mongoose. Anim. Behau. 47, 289-294. Waser, P. M., and Waser, M. S. (1985). Ichneumia albicauda and the evolution of mongoose gregariousness. Z. Tierpsychol. 68, 137-151. Wayne, R. K., Benveniste, R. E., Janczewski, D. N., and O’Brien, S. J. (1989). Molecular and biochemical evolution of the carnivora. I n “Carnivore Behavior, Ecology and Evolution” ( J . L. Gittleman, ed.), pp. 465-494. Cornell University Press, Ithaca, New York. Wayne, R. K., and O’Brien, S. J. (1987). Allozyme divergence within the Canidae. Syst. Zool. 36,339-355. Western, D. (1979). Size, life history, and ecology in mammals. Afr. J. Ecol. 17, 185-204. Woodroffe, R., and Macdonald, D. W. (1993). Badger sociality-models of spatial grouping. Symp. Zool. SOC. Lond. 65, 145-169. Woodroffe, R., and Macdonald, D. W. (in press). Female/female competition in European badgers (Meles meles): Effects on breeding success. J Anim. Ecol. 64(1). Woolfenden, G. E., and Fitzpatrick, J. W. (1984). “The Florida Scrub Jay: Demography of a Cooperative-Breeding Bird.” Princeton University Press, Princeton, New Jersey. Wozencraft, W. C. (1989a).The phylogeny of the recent carnivora. I n “Carnivore Behavior, Ecology and Evolution” (J. L. Gittleman, ed.), pp. 495-535. Cornell University Press, Ithaca, New York. Wozencraft, W. C. (1989b). Classificationof the recent carnivora. I n “Carnivore Behavior, Ecology and Evolution” ( J . L. Gittleman, ed.), pp. 569-593. Cornell University Press, Ithaca, New York. Wrangham, R. W., Gittleman, J. L., and Chapman, C. A. (1993). Constraints on group size in primates and carnivores: Population density and day-range as assays of exploitation competition. Beh. Ecol. Sociobiol. 32, 199-209. Wyman, J. (1967). The jackals of the Serengeti. Animals 10, 79-83.

This Page Intentionally Left Blank

ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 24

Development and Relationships: A Dynamic Model of Communication ALAN FOGEL DEPARTMENT OF PSYCHOLOGY UNIVERSITY OF UTAH SALT LAKE CITY, UTAH

I.

INTRODUCTION

A. PURPOSE

In this chapter I ask two questions: (1) what are relationships, and (2) how do relationships develop? The goal is to describe a theoretical conception of the process by which relationships develop. I offer definitions of relationships and development that form the basis of an interpretive scheme by which research in this area can be evaluated and from which conclusions are drawn. I refer primarily to social relationships between conspecifics. The approach is especially suited to species that have a highly developed repertoire of social behavior, particularly species in which communication is not constrained by obligatory signals but retains a degree of flexibility. B. RELATIONSHIPS A relationship is a set of ongoing connections and attachments. Relationships reflect the history of transactions between the coparticipants and imply a continued series of transactions into the future. I assume that relationships are the primary units of analysis, regardless of whether the focus is on social relationships or relationships with the physical environment. It is always the case that we are in relation to someone or something. Every action derives its form, meaning, and function from its relation to the environment, whether we are talking about the behaviorally proximal or the evolutionarily distal environment. This relarional perspective has its roots in a variety of late twentieth century intellectual movements including sociocultural psychology, postmodern feminism, systems thinking, and ecology. I have reviewed some 259

Copyright 8 1995 by Academic Press. Inc. All rights of reproduction in any form reserved.

260

ALAN FOGEL

of these lines of thought elsewhere (Fogel, 1993), and will not discuss them here. Instead, I focus of the implications of a relational perspective for the study of behavioral development. If the units of analysis in theory and research are defined as relational rather than individual units, then we can find consistency and order to behavioral and developmental processes that do not appear if actions are treated as belonging to individuals (Oyama, 1989; Weiss, 1969). In the traditional perspective, individuals are said to “have” a relationship, or to be “involved” in one, or to be “affected” by a relationship. Because these terms suggest that relationships are possessions or things, they imply that, like Cartesian objects, they exist as conceptually separate from the individual. The objectification of participants as autonomous individuals suggests that relationships are nothing more than the sum of each individual’s actions in response to the other. As an example of the consequences of taking a relational as opposed to an individual perspective, consider the research on mother-infant attachment. The development of attachment relationships has a similar course across most mammalian species. Attachment begins with mutual responses to simple signals that are part of the infant’s and mother’s behavioral repertoire. It grows toward a system of mutual proximity maintenance that transcends the presence or absence of particular signals. Bowlby’s (1969) original theoretical view was that mature attachment is the emergence of a higher order relational control system that is mutually regulated and coconstructed. This view gives a palpable reality to the relationship as an observable unit of analysis. Relatively few human or primate scholars of attachment have incorporated the conceptual features of Bowlby’s relational perspective into their empirical work. Although attachment security is defined as a quality of the parent-infant relationship, the measures of security often are attributed to the infant as an individual trait. Furthermore, the antecedent “causes” and consequent “outcomes” of attachment security at 12 months, for example, are typically attributed to “infant” or “mother” or “ecological” factors. Examples of an individual-centered view can be seen in the trait like interpretation of recent advances in the understanding of the etiology of variability in the quality of attachment. Maternal depression has been shown to have an immediate and lasting negative impact on attachment and on other indices of infant social and emotional behavior (Bettes, 1988; Fleming, Ruble, Flett, and Shaul, 1988; Gelfand and Teti, 1990). Infant temperamental withdrawal also appears to have direct effects on attachment and social interaction (Thompson, Connell, and Bridges, 1988). Al-

DEVELOPMENT AND RELATIONSHIPS

26 1

though there is evidence that individual characteristics such as depression and temperament may be partially alleviated through the relationship, depression and temperament have been interpreted as persistent and pervasive in their effects regardless of the nature of the relationship. Maternal depression, however, explains poor attachment outcomes for only about 10% of the population, and stable and severe temperamental withdrawal occurs in only about 5% of infants, both in human and nonhuman primates (Kagan, Reznick, Snidman, Gibbons, and Johnson, 1988; Suomi, 1987). Thus, most of the individual variation in attachment security cannot be explained by traitlike measures. In addition, even if depression and some temperamental dimensions can be shown to have a genetic component, a relational model is still essential in understanding the environmental pathways by which such genetic susceptibilitiesbecome manifest in the phenotype (Fogel, 1990a; Gottlieb, 1992; Plomin, 1990; Oyama, 1989; Schneirla, 1956). From a relational perspective, breakdowns and pathologies of attachment can be attributed to problems in the jointly regulated relationship system. Breakdowns may be factors associated with the balance of control between infant and parent for proximity maintenance, or the misinterpretation of communicative information within the system. One should be searching for the source of relationship pathologies in the developmental history of the relationship itself, recognizing its dynamics and organization, and not in the fixed traits of individual contributors (Sameroff and Emde, 1989; Sroufe, 1989). I recognize the need in the measurement process to objectify and reify complex processes into simple units. I also recognize that once defined and measured, constructs take on a reality of their own as shown by the perverse historical fates of scholarly and social definitionsof mental illness (Foucault, 1965) and intelligence (Gould, 1981), which have caused untold harm to countless individuals. Science is a social enterprise and all scientific decisions have ultimate moral implications because they take part in and influence the sociocultural context (Fogel, 1993; Hermans and Kempen, 1993). A fully relational perspective is a radical shift from an individualcentered one because it suggests that the objectivity of western science, the conceptual and mathematical tools we have all learned so well to apply, is likely to be the main culprit in preventing a more parsimonious view of behavior and development than we have today (Markova, 1990). Relationships have their own integrity that is not reducible to the sum of their contributing parts. I will suggest that simple tabulations of social action sequences and response probabilities, no matter how detailed, will fall short of a description of a relationship. Relationships as systems are

262

ALAN FOGEL

properly construed only at a level of analysis that goes beyond individual actions. C. DEVELOPMENT The individual’s ontogenetic history has been implicated in the explanation of action in observational and experimental studies across a wide variety of species. In spite of the impressive list of correlates and consequences, we are, with rare but notable exceptions, far from an understanding of the process of development change. The developmental unit of analysis is the individual-environment relationship in which both individuals and environments are seen as open systems and both are changing with respect to the other. Development is the result of a dynamic coregulation among the various components of the individual (from neural to behavioral) and the environment (from physical to social) such that neither individuals nor environments are defined fully in advance. Rather, the emergence of structure in both individuals and environments is the result of the process of their mutual relationship (Fogel, 1993; Gottlieb, 1992; Oyama, 1989; Thelen, 1989b). Development is not something that “happens to” an individual, nor is it “caused” by any completely specifiable set of mechanisms, nor is it a “readout” of an epigenetic program. The conceptual bottleneck in understanding development, analogous to the problem of understanding relationships, is with the individual-centered language of description. Typically we assume that individuals are objective entities subject to the interplay of equally objective internal and environmental forces. Because I view relationships as the fundamental units of analysis, what develops is not the individual’s actions, but rather the relationship between the individual and the environment. Recent advances in the field of motor development, for example, have established that skills and muscles change over time, but that these changes are more readily comprehensible when we consider them as a developmental process of the refinement of the tuning of perception and action systems that mediate the person-environrnent relationship (Kugler, Kelso, and Turvey, 1982; Thelen, 1989b). What might be seen as discontinuous or puzzling developmental changes within the individual motor system can be described at the level of the person-environment relationship as the change of a small number of tuning parameters. Consider the following example, After human infants acquire reaching, at around the fourth month of life, they prefer to reach with both arms. Just before they acquire the skill to sit independently (between 5 and 7

DEVELOPMENT AND RELATIONSHIPS

263

months), however, they begin to use one arm for reaching, and once they can sit stably (after 7 months) they are just as likely to use one or two arms for reaching (Rochat and Senders, 1990). This ontogeny is puzzling from the individualistic perspective because one is forced to invent a hypothetical neural change, perhaps the emergence of hemispheric dominance. The development of distinct hemispheric differences, however, seems to occur later in the first year, around the time of emergence of wordlike sounds (Ramsay, 1984). From the person-environment relationship perspective one can examine the infant’s action in relation to both internal changes and external constraints (Thelen, 1989). There is the relationship of the arm to the trunk, and there is the relationship of the trunk to the sitting surface and gravitational field. One-arm reaches appear at between 5 and 7 months because, quite simply, the infant needs the other arm to provide postural support while sitting. This observation further suggests that the form of even the first reaches, while lying supine or held by a parent, are incomprehensible without a consideration of gravity, posture, and additionally the nature of the object: its size, color, speed, distance, and the like (von Hofsten and Ronnqvist, 1993;Thelen, Corbetta, Kamm, Spencer, Schneider, and Zernicke, 1993). Actions are fundamentally embedded within the individual-environment system. In addition, the environments themselves change systematically with changes in the infant. As infants acquire postural skills, for example, parents can spend less time providing postural supports and more time enriching the playing field. As infants develop in their motor and perceptual skills, new objects and increasingly complex tasks are introduced by parents (Fogel, 1990b, 1993). Environments in socially living species are inherently dynamic and created as a result of the interaction with the individual. I can summarize these introductory comments in the following way. Relationships are the fundamental units of analysis, not just social relationships but any form of individual-environment connection. The study of development is made comprehensible by the conceptualizationof relationship, rather than individual, change. These points lead to the conclusion that relationships are developing systems, developing systems are inherently relational, and developmental process must be a fundamental constitutive property of any relationship in a living system. In what follows I give a more complete definition and description of the terms relationshipsand development; they are complementarydescriptions of the same biological system. I elaborate these discussions with examples. Finally, I suggest a way in which a relational conceptualization

264

ALAN FOGEL

can lead us to search for a wide range of new phenomena that appear only when one approaches the traditional problems from a relational level of analysis.

11. RELATIONSHIPS

A. COMMUNICATION

Relationships are systems in which the history of communication is reflected in ongoing patterns of co-action (Fogel, 1993). This definition of relationship implies that communication is at the core of any relationship, but not any kind of communication. In order for a communication process to constitute a relationship, it must evolve from the prior set of encounters of these participants. Relationships are communicative systems that develop and have a unique history. The most widely accepted model of communication was invented to explain electrical signal transmission between transmitters and receivers and computer information flow (Shannon, 1963; von Neumann, 1958). In this view, there is a fixed set of codes, symbols, or messages that receivers and transmitters can send and interpret. Communication failures are due, not to ambiguity in the messages, but to noise in the transmission process. The behavior of the receiver must change in some way in order to say that information has been transmitted, but the change in behavior in this model is only an alteration of the pattern of preset messages in the response. I have argued at length elsewhere (Fogel, 1993) that this discrete message model of communication could not be the basis for a relationship as defined here. Relationships imply more than just an alternation in the flow pattern of messages. Rather, relationships require the modification of the message units themselves in both their form and their meaning. In addition, the modification of the message units becomes the hallmark of the uniqueness of the particular relationship history out of which the modification evolved.

I. Example: Human Mother-Infant-Object Interaction In humans, infants acquire reaching at between 3 and 5 months of age. The presumption is that reaching and other sensorimotor skills are acquired through the infant’s own active exploration with the nonsocial environment (Burnham and Dickinson, 1981; Piaget, 1952; Thelen et al., 1993; von Hofsten and Ronnqvist, 1993). My own work has revealed, however, that early human sensorimotor development is embedded within

DEVELOPMENT AND RELATIONSHIPS

265

the mother-infant relationship system (Fogel, 1990b; Reimers and Fogel, 1992; West and Fogel, 1990). At the age of three months, human infants have none of the postural control that would support independent reaching. In the traditional (i.e., nonsocial) studies of reaching, infants are supported upright in a hightech infant seat designed especially to facilitate the free movement of their arms, while giving them the feeling of supportive comfort. Researchers also must adjust the distance, size, color, shape, and weight of the objects to what is “reachable” and “graspable” for the infant. These researchers have, without explicit recognition, created a sociocultural context of postural support and object choice in which infant reaching emerges. In the natural situation, a considerably more dynamic sociocultural context is cocreated with the parent. Our research shows that many features of sensorimotor skill are inherently communicative, not individualistic, because in some way the infant has to let the parent know what to provide. The infant’s initial reaches and the way in which the infant manipulates and attends to objects are socially constructed and depend on the unique qualities of the particular relationship the infant has with the parent. We videotaped 13 mother-infant dyads weekly between the ages of 1 and 12 months. For this analysis, we chose six weekly sessions before and six sessions following the onset of infant visually directed reaching. Infants were supine on the floor of a laboratory playroom. One camera was focused on the infant’s body and another on the mother and infant. Both cameras were mixed with a special effects generator into a composite image. The mother was instructed to play with the infant as she might at home, and she was free to use any of a set of age-appropriate infant toys available in the room. Videotapes were coded for mother and infant actions on objects as well as infant gaze direction. The following narrative descriptions compare two mother-infant dyads that differ in their communication patterns about objects. Both infants are 4 months of age. Jerry: The mother presents objects to the baby while the baby lies on his back, watches

and tries to reach for the objects. The mother alternates between demonstrating the object’s properties and then moving the object so the baby can easily contact the object with his hands. She helps the infant grasp the object and helps to pick it up when dropped. She seems focused on highlighting the object’s physical features and on maintaining the infant’s attention to the object. Her voice is distinctly different for each of these object-related actions, helping to mark the different pieces of action for the infant. The infant remains alert, attentive to the object, and smiles and brightens in a modulated way that is co-regulated with the mother’s actions and voice. She seems content to wait until the infant loses interest in an object before introducing

266

ALAN FOGEL

another one. When she introduces objects to the baby, she rarely persists for more than ten seconds unless the baby shows some interest. (Fogel, 1993, pp. 111-112) Andrew: Like Jerry’s mother, this mother also demonstrates the physical affordances of the objects, but in a rather different manner. She moves the object continuously in the same position as the infant gets excited, but rarely follows up by putting the object in the infant’s hand. She touches the infant’s face and body with the object and when Andrew loses visual contact, he becomes visibly frustrated and over-agitated, calming only when the mother puts the object down. The mother’s voice is similar in its tonal characteristics regardless of the activity. Even when Andrew is holding an object, his mother touches or distracts him with a different object, a theme change that gives the child little opportunity for focused and socially supported visual and manual exploration. (Fogel, 1993, p. 112)

As a result of the differences in the communication system, Andrew’s knowledge about and skill with objects appears to be more fragmentary than Jerry’s. For Andrew, visual attention to objects is not systematically linked to the manual exploratory system. It is as if the nature of the communication system is fundamental to the infant’s experience of the world and his action upon it. In this view, Andrew and Jerry are developing, through their relationship with their mothers, a very different experience of the world, and perhaps for each child a different set of meanings or functions becomes related to the same set of objects. Using a more quantitative approach (Fogel and West, in preparation), we divided infants into two groups of six: those whose attention to objects was coordinated with their actions on objects (high focal attention) and those who were less likely to gaze at objects on which they were acting (low focal attention). The twelve weekly sessions for each dyad were divided into three age periods of four sessions each: prereach, transition, and postreach. The focal attention groups were created by examining the mean duration during the postreach period of all instances of focal attention, that is, when infants were looking at an object they were holding. The median infant from the original 13 in the study was dropped. The mean duration of focal attention for the six infants above the median during the four postreach sessions was 7.55 s, compared to 1.60 s for the low focal attention group. Thus, the infants in the high group were focally attending almost five times as long, on the average, as the infants in the low group. During the postreach period, the high focal attention infants were likely to distribute their time across all the toys and they used a greater variety of adaptive actions (shaking a rattle, squeezing a soft toy). The low focal attention infants concentrated primarily on two objects, soft toys that they grasped with two hands and spent a relatively large percentage of time mouthing without looking at them, rather than engaging in visually directed manual exploration.

DEVELOPMENT AND RELATIONSHIPS

267

Table I shows the duration and percentage of co-occurrences between mother and infant actions for each of the three age periods and for each group of infants. Co-occurrences are defined as the joint occurrence of both the mother and infant action. The table shows the duration of cooccurrence for instances when the infant was gazing at the toy object, that is, focally attentive. Behavior definitions are as follows. Mother support refers to any attempt to assist the infant in reaching for or manupulating an object. Demonstrate is to show the infant the appropriate use of the toy, such as shaking a rattle. Social play refers to any instance of play in which the focus is not on objects, but on the mother-infant face-to-face interaction. For the infants, no toy is any period in which the infant is not holding a toy, although the infant may be watching the mother support or demonstrate a toy she is holding. Explore refers to any manual action on a toy, while mouth refers to oral contact with the toy. Reach is any instance in which the infant attempts to retrieve a toy with arm extended. The durations in the cells of Table I fit a log-linear model in which all the main and interaction effects were significant. All the factors (age period, focal attention group, infant action, and mother action) and their interactions must be taken into account to explain the data in the table. During the prereach period, the mothers in the low attention group were more active in all categories when the infant was not holding a toy. Mothers in the high attention group were four times more likely, during the prereach period, to support infant attempts to reach and six times more likely to demonstrate objects during infant object exploration periods. In the transition period, these patterns were somewhat reversed. High attention group mothers were more active when the infant was not holding an object, especially for the category of support. Mothers of low attention infants were more likely to support while the infant explored. During this period, low attention infants spent almost three times as long mouthing objects than the high attention infants, and their mothers attempted more social play when the infant was mouthing. During the postreach period, there were once again large differences in the duration of infant mouthing between the two groups, and mothers of low attention infants were primarily social during this action. The distribution of maternal support was especially revealing. Mothers of low focal attention infants were more likely to support the infant’s object exploration, while mothers of high attention infants supported reaching attempts and allowed the infants the opportunity to explore more independently at this age. There were few group differences in the patterns of maternal demonstration. The overall picture is one in which the developmental change in the

s t4

TABLE I TOTALDURATIONS, SUMMED ACROSS SUBJECTS, OF CO-OCCURRENCE (IN S) AND WITHIN-GROUP ROW PERCENTAGES ACTIONSWHILETHE INFANTIS GAZINGAT THE TOYOBJECT

FOR

MOTHERAND INFANT

Infant action Age period hereach

Mother action SUPP0l-t Duration Row percent Demonstrate Duration Row percent Social play Duration Row percent

No toy 31 1

High focal attention Explore Mouth

Reach

No toy

Low focal attention Explore Mouth

61 14

2 0

82 20

465 85

61

302 12

49 2

68

1906

83

3

96

8% 78

161 17

23 4

II I

574 83

66

1928

Reach

0 0

5 25

68 2

14

53

0

2

57 13

0 0

4

10

1

Transition

Postreach

support Duration Row percent Demonstrate Duration Row percent Social play Duration Row percent SUPPOrt Duration Row percent Demonstrate Duration Row percent Social play Duration Row percent

606 48

359 30

8 I

279 22

265 30

522

52

19 2

132 16

1064

398 22

51

98

70

3

5

983 61

508 26

83 6

153 8

954 27

1603 64

136 4

120 5

899 26

1421

325 14

88 4

287 24

42 1 35

21 2

423 38

115 13

51

62

39 5

151 20

550 32

942 48

76 4

289 16

363 30

649 49

97 7

176 14

1280 20

1391 63

544

577 10

632 14

2513 59

1187 23

175 4

7

.

56

270

I

ALAN FOGEL

high attention dyads is from early maternally supported object play to later more independent object play to which the infants are more visually attentive and explore a greater variety of objects. In the low attention group, mothers were more active in the early period in relation to their infants’ relatively lower object-directed actions, and they persisted in their supportive involvement with the objects in the later periods as the infants became less visually attentive in relation to relatively few objects in which they specialized. It is not clear from these data if either mother or infant is the cause. What is clear is that mother-infant-object relationships are complex relational systems that over time tend to create a small number of stable patterns of joint action. There is converging evidence from other research that early attention to and manipulation of objects is socially constructed. Gray (1978) reported differences in object play between mother-infant dyads similar to the ones we found. Infants are more skilled at object play if they have been exposed to more frequent adult touching, carrying, and holding (White and Castle, 1964). Preterm infants show more accelerated sensorimotor development when their mothers play more object games with them (Landry, Chapieski, and Schmidt, 1986). Finally, infant attention to objects is more focused if mothers time their interventions with objects to periods in which the infant shows some prior interest in the object (Parinello and Ruff, 1988). These implicitly causal findings, however, were based on research in which measures of mother and infant were taken out of the context of the relationship process. In this example, the important point is that there is not a predefined set of meaningful signals by which mothers and infants communicate about objects. On the contrary, dyads can take almost anything as communicative and incorporate that into the relationship. Small variations in vocal intonation become meaningful for one dyad and not for another. The objects themselves become meaningful to infants and mothers not merely for their physical-perceptualfeatures, but because of the manner in which objects are used: as ways of satisfying curiosity, as a means to get attention, as a source of pleasure or distress. 2. Other Examples This type ofmutual coregulation of the social system toward the creation of a unique pattern of joint action is remarkably robust across species. Flexibility and creativity in the communication of higher social animals are what distinguish it from the current state of the art in electronic and machine communication. Individual differences in behavior have been observed to emerge during

DEVELOPMENT AND RELATIONSHIPS

27 1

the early history of mother-infant relationships in rhesus monkeys. Rhesus infants who are most frequently rejected by their mothers are the most distressed by experimental separations. Separation distress also depends on the social circumstances of the separation, such as if the infant is left with other adults, with age-mates, or alone (Hinde, 1977; Suomi, Mineka, and Delizio, 1983). Food and nipple preferences in kittens and rats depend on the patterns developed in the course of the mother-infant relationship, as well as on contextually mediating factors (Rosenblatt, 1963,1983). The access to power and resources in naturally living chimpanzee groups is regulated by a complex and subtle network of social processes embodying clear evidence of relationship history in patterns of brooding, retaliation, revenge, and allegiance (de Waal, 1989; Goodall, 1971). Even in species developing without parental care there are a growing number of examples in the literature suggesting that mutually regulated social tuning processes are required to explain their communicative behavior over and above the assumption of any set of innate discrete signals. In the cowbird, for example, social experience with females motivates males to change the songs they had developed in isolation (West, King, and Duff, 1990). The traditional theory of animal communication suggests that particular stimuli act as releasers of an obligatory communicative act (Smith, 1977). Recent work has shown, to the contrary, that even communicative acts thought to be directly tied to emotional arousal, and therefore presumed to be involuntary, are nonobligatory and sensitive in their display to the social context (Cheney, Seyfarth, and Smuts, 1986; Hinde, 1985; Marler and Mitani, 1988; Provine and Fischer, 1989). Roosters will emit alarm calls more frequently to experimentally introduced eagle displays if there is another conspecific visible in the same room (Marler and Mitani, 1988). These so-called audience effects have also been found for a variety of human emotional displays including smiling, motor mimicry, and vocalization (Bavelas, Black, Lemery, and Mullett, 1986; Chovil, 1991 ; Fridlund, 1991; Kraut and Johnston, 1979; Rivera and Grinkis, 1986) and have even been observed for smiling in infants during the first year (Fogel et al., 1992; Jones and Raag, 1989; Trevarthen, 1993). It is not the case, therefore, that an individual “has” a state of arousal and simply transmits it. This research would suggest that not only the communicative action is altered by the social context but possibly also the internal state itself. My colleagues and I have argued that emotions, from this perspective, are not universal motivational states that unilaterally cause behavior. Instead, emotions are emergent from the dynamic

272

ALAN FOGEL

relational system; action and emotion codetermine each other (Fogel et al., 1992). Similar views of emotion as a process can be found in the literatures of sociology and anthropology (cf. Lutz, 1988).

B. COREGULATION 1 . Discrete versus Continuous Actions

These examples illustrate that communication, even in relatively simple situations, is not adequately described by a discrete signal-response model. The research on audience effects shows clearly that the individual is capable of modifying the “signal” with respect to the receiver, and possibly the internal state, even before the signal is emitted. In this section I propose a mechanism, based on the concept of continuous process in communication, by which such ongoing mutual modifications can be explained. In an analysis of animal aggression, Hinde (1985) raises the question of why individuals show threat displays before attacking. An animal that is concerned about the risk of injury might simply flee the encounter. If, on the other hand, the animal intends to defend offspring or territory, why not attack suddenly and without warning before the opponent can mount a defense? Hinde concludes that the threat display is one way interacting individuals can assess the partner’s intentions. Threat interactions most typically end in standoff or retreat rather than fighting. According to Hinde, “such signals are thus to be seen as involving negotiation with the rival as well as expression of internal state. The term negotiation does not necessarily imply manipulation but emphasizes the continuous interaction between the two individuals involved” (p. 1 1 1). Similar processes have been observed in human teasing where insults varying from serious to playful are continuously modulated by the partner’s co-actions and returned insults (Labov, 1972; Pawluk, 1989). It is quite likely in these examples that the ongoing mutual modification of action also involves concomitant changes in motivation and emotion (Fogel et al., 1992). Human kissing is another example of continuous communication processes. Kendon (1975) showed that each partner’s face is continuously in comotion with respect to the other during the approach phase. Discrete signals, such as smiling, appear during the process but these occur as embedded within the continuous comovement, and their effect is on the alteration of the movement pattern from initial to final approach. During comotion in the final approach, lip parting occurs in varying degrees and one partner may remain still while the other completes the approach. Research on the perception of social action suggests that individuals parse partner’s behavior into units based on the functional coherence of

DEVELOPMENT AND RELATIONSHIPS

273

ongoing action, rather than the presence or absence of a universal signal. In many cases what is perceived is related to the social coordination process itself, that is, the degree to which the partners are mutually attuned or whether they have achieved shared goals or shared understanding (Branco and Valsiner, 1992; Newtson and Engquist, 1976; McArthur and Baron, 1983; Valenti and Good, 1991). The theoretical importance of these examples is that communication is a continuously changing and highly variable process in which individuals are fundamentally embedded, not as discretely different units but as open systems being continuously modified. Signals are not, in themselves, compelling or obligatory. Rather, they depend on the context of continuously evolving co-actions for their meaning. Indeed, the signals themselves may be dynamic displays that can vary with subtle shades of movement of the effector muscles (Fogel, 1993), for example, a kiss with the lips pursed versus one with the lips slightly or widely parted. There is considerable evidence that human speech has both discrete and gradient qualities. Speech varies in timing, pitch, intensity, drawl, and other paralinguistic features that allow communicators to construct continuously matched patterns of co-action during conversations (Heath, 1984). The continuous process of communication can be described with the concept of coregulation. Coregulation is a social process by which individuals dynamically alter their actions with respect to the ongoing and anticipated actions of their partners. Because of the pervasiveness of continuously variable and coconstructed communication processes, to continue to describe the reality of communication with bounded terms like signal and response does not dojustice to the apparent complexity of the process. Furthermore, discrete state models of communication, as I show in the next section, are unable to encompass the developmental aspects of relationships. 2. Framing It is a common observation that all biological processes, and social processes in particular, are hierarchically organized. Thus, within relationships there are divisions of rest and activity. Within the active phases there are different types of identifiable social routines such as feeding, fighting, mating, and playing. Finally, each of these routines is composed of different phases of co-activity such as stage setting, negotiation, consummation, and conclusion. Following the work of other scholars in the area of communication (Bateson, 1955; Garfinkel, 1967; Goffman, 1974; Kendon, 1985), I refer to each of the identifiable social routines asfrarnes. Traditional communi-

274

ALAN FOGEL

cation models suggest that frames are organized from the top down, by genetically driven species mandates or by sociocultural rules. My own view is that the patterns of co-action can also be sustained by means of the negotiated consensus between partners that is the result of a coregulated process. The emphasis, therefore, is on the sense of the word “frame” as a verb rather than as a noun. To frame means to fashion, compose, construct, or invent and the word is chosen for these connotations. In contrast, words such as social routine or social ritual connote a structure that is guided by a set of fixed rules that generate behavior. In my view, framing represents the outcome of negotiation, a settlement rather than a prescription. Two children can decide to “play house,” for example. There are cultural models and no doubt physiologically based cognitive processes that make symbolic pretend play possible. Nevertheless, what is striking from even a casual observation of multiple instances of playing house is that they are rarely alike; they are creative and negotiated affairs. Children have to decide whether and how to accept another’s invitation to play, they have to decide when and where play will occur. They need to concur on which objects and actions “count” as part of the game, on the roles each will play, and on whether other children are “allowed” to join in. Research suggests that children do not slavishly follow cultural rules, but rather they make it up as they go along and explicitly discuss the nature of rules that themselves change dynamically during the course of play (Branco & Valsiner, 1992; Fogel, Nwokah, and Karns, 1993; Haight and Miller, 1993). Regularities can occur in complex systems without the guidance of a set of rules or plans but as the result of the constraints that are imposed on partners in the process of coregulated negotiation. As mutual effects occur through co-action, it limits the degrees of freedom allowed to each partner. If a child walks past another who wants to play, both children’s possibilities become immediately constrained to what can occur together rather than in solitary play. Opening moves limit possible rejoinders and after a period of initial tentativeness, the partners settle into a recognizable pattern of co-action and turn taking that is sustained by what each does in relation to the other (Fogel, 1993). Observers of such coregulated framing patterns can infer that they have rules and that the rules are responsible for the generation of action, but we should be clear that although the concept of rule may be heuristic for descriptive purposes, it may be only a metaphor rather than an explanation of cause (Fogel et al., 1993). From the perspective of the participants, frames (games, arguments, mating, and the like) rarely have the feeling of following rules. They more typically feel compelling and creative.

DEVELOPMENT AND RELATIONSHIPS

275

Frames that are experienced as rulelike by the participants, I have argued, may be perfunctory rituals done out of habit or compulsion. Such frames are often experienced as aversive (Fogel, 1993). On occasion, however, in both human and nonhuman communication, the process of communication and its presumed rules become the topic of communication. Metacommunication is defined as communication about the communication process itself. One example of metacommunication is play fighting between juveniles in a number of different species. A variety of nonverbal metacommunicative devices, such as exaggerated gestures and interrupted movements, are used by participants to distinguish play from serious fighting (Bateson, 1955; Mitchell, 1991). It is as if by these gestures, the participants are negotiating a consensual frame that the communication is to be a simulation. The frame is dynamic and abstract, not prescriptive of the details of action sequences, and therefore not rulelike. In human friendships and romantic couples, for example, there are two types of metacommunication: episodic and relational. In episodic metacommunication, partners highlight particular actions, such as commenting on the partner’s emotional behavior or redirecting the topic of conversation. In relational metacommunication, partners verbally discuss the relationship: what is means and where it is moving in the future. Examples are discussing levels of intimacy, and proclaiming love or affection (Duck, 1991; Shafer, 1990; Wilmot, 1980). There are few reported instances of metacommunication in the nonhuman literature but there is likely to be a great deal of it. Episodic metacommunication may occur during play when one partner winces, growls, or bears the teeth toward a too aggressive playmate. Relational metacommunication is likely to occur in mating and dominance structures in socially living groups. In chimpanzee groups, for example, there is a good deal of third party mediation, usually by dominant females, between male competitors. This mediation, and the males’ response to it, suggest that the relationships in the group are being negotiated and regulated in an explicit way (de Waal, 1989). Even in these instances where the topic of communication is the relationship itself, there is little evidence that the rules are ever explicitly articulated or that they ever take on the character of top-down organizers in the absence of the bottom-up dynamics of the coregulation process. Indeed, the theory of continuously coregulated communication suggests that language and symbols are not needed to maintain the rather complex communication processes observed in many species, and the theory can also explain the role of communication in the development of action during

276

ALAN FOGEL

the prelinguistic phase of human infancy. These conclusions are possible when one conceptualizes communication as a complex and multilayered system of simultaneous as well as sequential events that are cocreated, rather than a single stream of signals and responses. In the discrete model of communication, since there is no assumption of co-occurringprocesses, one requires a symbolic medium as the only way to “step outside” of the communication in order to mediate it. A N D COREGULATED COMMUNICATION PROCESSES C. RELATIONSHIPS

Relationships are systems of communication in which the history of the social process is embodied in action in the present. If we use coregulation as the theoretical description of the communication process, then we can see that all forms of coregulated communication are inherently historical. Because coregulated action is adapted to the ongoing mutual constraints of the communicative partners, every action of each individual is imbued with the results of the past negotiations between all participating individuals. When one begins an initial approach to kiss an intimate partner, there is already in the form of that action a reflection of the negotiation process by which kissing has been agreed upon to occur in the past between these partners. Every coregulated action is inherently relational in the sense that it is created then and there in relation to the partner, it is created in the present context, and with respect to prior encounters in similar contexts. This historical aspect of relational action is more than simple conditioning of responses. Without a doubt, reinforcement and punishment affect the likelihood of a social action with a partner. Nevertheless, conditioning processes alone do not account for the inherently relational, and hence openly creative, features of every action in the present (Fogel, 1993; Hinde, 1978). Furthermore, one need not assume that there is necessarily a cognitive or symbolic representation of the rules of action in a context. Action can be conceptualized as created out of the constraints of coregulation with a minimum of cognitive baggage necessary only to get the action motivated and select some relevant strategies. The final form of the action is created and emergent rather than slavishly guided by represented rules; it is procedural rather than generated (Fogel, 1993; Rogoff, 1982; Valsiner, 1987; Wertsch, 1985). Coregulated action not only embodies its past, it also embodies its future. In the next section I argue that coregulated relationships are inherently developmental in the sense that they function as open systems carrying the seeds of their own change.

DEVELOPMENT AND RELATIONSHIPS

277

111. DEVELOPMENT

A.

RELATIONSHIPS AREDEVELOPING SYSTEMS

In the previous section I tried to establish that coregulation is a mechanism by which relationships carry their own history into the present encounter. This is because coregulated action embodies the results of a series of mutual negotiations toward the establishment of consensual frames. In this section, I suggest that coregulation also points toward the future so that at any given point in time the social interaction is a living process that is irreversible, leading from past to future. According to Hinde (1978),

. . . relationships involve properties not present in the interactions which constitute them, but emergent from the patterning of those interactions over time . . . to describe a relationship fully, we must describe also the cognitive and affective components which accompany and transcend the behavioral ones. These have reference to a broad span of time. (p. 374) Similar points have also been made by Carvalho (1988). Sander (1977) writes, We believe it useful to consider the infant and caretaking environment together as a biological system and to focus on the aspects of the regulation of exchange in the system . . . We take for granted the notion of the life process as an ongoing synthesis in an open-ended dynamic, one which is resolving basic polarities but leaving us with an "in-between" in the open endedness of the present moment-an open endedness that our living activity resolves in new organization. (pp. 152-153)

There is, in these theoretical statements, the ideal that social action suggests its own future in a concrete manner that does not require symbolic representation of a future. Consider again the example of animal threat displays. These displays seem not so much to advertise a current emotional state as to announce the intention to fight if necessary. Thus there is, in the current display, de facto evidence of an intention for future action. This affords no absolute guarantee, however, that the animal will attack, stand ground, or flee. Indeed, studies across a variety of species suggest that there is little in the way of cues from current behavior to predict who is likely to attack and who is likely to win a confrontation(Hinde, 1985). When one individual moves to avoid another that is displaying threat,

. . . it is not necessary to assume that the individual animal has a representation or imagination of the threat. Instead, there is information available in the present situation

278

ALAN FOGEL

that calls forth in the animal some anticipatory directionality, either to approach or to avoid . . . In greetings, individuals anticipate a communicative engagement with the person they are saluting, yet the greeter remains vigilant with respect to the possibility that the interaction will not develop, that the person being greeted, on closer approach, will turn out to be mistaken for another more familiar person. (Fogel, 1993. pp. 124-125)

Guilford and Dawkins (1991), in a discussion of the importance of the receiver’s psychology in the signaling of the emitter, say the following: Receivers can predict the future behavior of emitters because the emitter unwittingly gives away its intentions. The receiver therefore can exploit the emmitter, because it cannot help giving away what it is about to do next. (p. 10)

The Guilford and Dawkins quote overstates the point I am trying to make. They seem to equate intention, future orientation, and the ability to predict the future. These concepts, in my view, are not the same and the differences between them are theoretically important. A system can be open to change and intentional, while at the same time it can be unpredictable with respect to its future. This distinction between intention, as an open-ended property of ongoing action, and prediction is a crucial theoretical point. It provides a model for action as directed within the constraints of the social process, and at the same time provides the opportunity for creativity, innovation, and change (Merleau-Ponty, 1962; Vedeler, 1987). In other words, the directed, intentional quality of action is inherently developmental. Action carries within it the seeds of its own change. This open-ended, yet directed property of action in general and social action in particular has been noted by a number of scholars. There is a reasonably well-developed body of theoretical and philosophical work suggesting that action is the source of its own development (Bergson, 191 1 ; Brentano, 1973; Markova, 1990; Mead, 1934; Valsiner, 1993). This view suggeststhat we do not need to look outside of action for a mechanism or source of developmental change; rather, we need to examine carefully the nature of action and, particularly, the nature of co-action and coregulation. B. RELATIONSHIP CHANGE PROCESSES Change in social-relational systems has been hypothesized to occur through a number of different processes. Generally speaking, change in any system arises because of the introduction of novelty or innovation into the system. The innovation, however, is only the starting point of

DEVELOPMENT AND RELATIONSHIPS

279

the change process. Work in both sociology and anthropology on complex social systems reveals that innovations, no matter how clever they may appear, must undergo a process of diffusion through the system in order to be adopted into general practice (Bandura, 1986). Diffusion refers to the adoption of the innovation in different units and subsystems of the larger social system. There are many examples of inventions and innovations that never become generally adopted because they were not readily diffused throughout the system. Some innovations may become diffused many years after their introduction (in the United States, civil rights legislation in the 1860s and again in the 1960s was not followed immediately by compliance and there are still many struggles and confusions). Other innovations are rapidly diffused, such as changes in the economic system (tax laws, inflation). The work on diffusion suggests that change does not always follow upon innovation, and that the innovation must be coregulated into the system by a sometimes lengthy process. On a microgenetic level of change, that is, change within a single relationship, changes over time occur in the frames of communication. One form of change is the sudden emergence of new frames and the disappearance of existing frames. Another form of change is change within frames (Fogel, 1993). I discuss the latter below. Change within frames, or consensual reframing, could take the form of changes in the meaning of co-actions, changes in partners’ roles, elaboration, or simplification of previous forms of co-action within the frame. Just as in the diffusion examples, change grows out of innovation through mutually coregulated negotiation and not by virtue of one partner imposing the conditions for change on the other. Just because one partner “wants” to change does not mean that these wishes will be immediately ratified and incorporated into the relationship frames. Relationship frames have an inertial quality. Any parent or spouse will tell you this if they have tried unsuccessfully to alter interaction patterns in their relationships with their children or partners. The suggestion for change may come too early, or it may be presented in the wrong way, or it may be presented as an ultimatum rather than as grounds for negotiation. 1 . Innovation and Consensual Reframing in

Mother-Infant Relationships With respect to the idea that the source of development is within the field of coregulated action in a relationship, my students and I have been studying changes over time within mother-infant dyads. Relationships are constituted when two or more individuals communicate over repeated occasions. Thus, we collected a set of videotaped mother-infant interaction episodes from 13 dyads observed weekly in free play during the

280

ALAN FOGEL

infants’ first year of life and every two weeks in the second year of life. Although the videotaping procedures, the physical setting, and the toys were the same for each dyad, the couple was left to develop their own unique relationship over time. It is important to recognize that these dyads were extremely comfortable in what for them was a familiar environment. Both mothers and infants looked forward to their play time in the laboratory each week. In one study based on these data we focused on change processes within mother-infant frames occurring between 9 and 15 months of age. The development of reframing discussed here is the movement from nonconsensual action within the frame to consensual action. In the particular cases to be discussed, infants shift from treating things as physical objects to using them in more symbolic ways. It is the symbolic use of the object that the mothers innovate in the frame, but it takes some time for the symbolic use to become consensual. Two examples will be used: mother-infant play with a red toy telephone, and play with a toy baby doll and bottle. We found that both TELEPHONE play (observed in one dyad) and BABYIBOTTLE play (observed in four dyads) passed through similar developmental phases in the approach toward a consensual reframe (see Table 11). In both examples the mothers were attempting to change the infant’s more direct physical play with the objects into a cultural pattern more suited to the adult’s symbolic uses of the objects. In the case of both TELEPHONE and BABY/BOTTLE play, mother never, in our observations, used the object in a nonsymbolic manner; for the mother, the telephone was for having a pretend conversation and the bottle was for pretending to feed the baby doll. Thus, from the perspective of the relationship, these symbolic uses of the objects for pretend were innovations, and the process of change from physical to symbolic co-action can be thought of as a process of consensual reframing of the innovation. In our data, this reframing process lasted anywhere from 10 to 30 weeks. Our observations of the TELEPHONE game (Reinecke and Fogel, 1994) began when the infant (Hannah) chose to explore the physical properties of the toy telephone, at about 9 months of age (nonconsensual co-action). Several weeks later, Hannah’s mother began to take the telephone gently out of Hannah’s hands in order to have a pretend conversation. Hannah was insistent in demanding the return of the phone and was apparently uninterested in the mother’s use of the object. After four weeks of this reluctant release of the phone, Hannah allowed her mother to take it (preconsensual responsiveness). Hannah would watch the phone in the mother’s hand, watch her mother put the phone up to her ear to pretend to talk, and finally crawl over and demand the return of the phone. After

DEVELOPMENT AND RELATIONSHIPS

PHASES IN

THE

28 1

TABLE I1 DEVELOPMENT OF MOTHER-INFANT PLAY

Nonconsensual co-action: Infants explore the toy with respect to its physical properties only: banging, squeezing, sucking. Preconsensualresponse to requests: Infant responds to maternal request for symbolic action, but only briefly and skeptically. The object is still physical for the infant, but part of a Communicative frame. Preconsensualinitiatives: Infant initiates symbolic action while observing mother. The infant is more willing to submit to the mother’s uses of the object and will observe her as she uses the object. The object is still physical for the infant, but the communicative frame becomes more elaborated. Consensual co-action: The object transcends its physical characteristics to become a symbol. The communicative frame changes to pretend play accompanied by smiling and laughter.

several more weeks, Hannah would voluntarily offer the phone to her mother and wait to watch what mother did with it (preconsensual initiative). A good many microinnovations occurred during this process with respect to initiation (giving and taking) and forms of activity (active manipulation of the phone vs. waiting and watching mother’s actions). These innovations took as little as several minutes or as long as several weeks to become consensual. Finally, we recorded the following as indicative of the onset of the consensual coaction phase at 44 weeks. Hannah picks up the phone and looks at it, then at her mother. She offers the phone to her mother in the same manner that she had done in previous weeks. Mother takes the phone, puts it to her ear as in previous instances and says, “Hi, grandma!” At that point, Hannah looks and smiles at mother and reaches out to request the phone. Her mother offers the phone to Hannah, who puts it to her ear (the first time we have observed Hannah to do this) and says, “Ha-0.” Hannah again looks at mother and offers the phone to her for more pretend talk (Reinecke and Fogel, p. 184).

We observed similar developmental sequences in BABY /BOTTLE play (Dedo, 1993). At first the bottle was a squeezable and suckable toy for the infants. Mothers, in every case, did not want the infant to suck on the bottle. Mothers made disgust faces, used prohibitive language, or took the bottle away from the infant. In all cases, however, infants persisted in sucking the bottle for many weeks and seemed uninterested in the mother’s intended uses for the bottle (nonconsensual action). Later, infants put the bottle on the doll’s face, responding to mother’s requests, but only briefly and reluctantly, while mothers continued to voice their strong approval or disapproval (preconsensual responsiveness). A typical

282

ALAN FOGEL

observation from our protocols of one subject pair in this preconsensual period is the following: The mother offered the bottle to the infant who put the bottle in his mouth. The mother strongly disapproved of the infant’s mouthing by physically removing the bottle from his mouth, saying “No, no,” and “You silly, you are stealing the bottle from the baby.” The mother then demonstrated the pretend use of the bottle to feed the doll and gave the bottle to the infant. This time the infant opened his mouth as the mother offered the bottle to which the mother responded with prohibitive statements. Once again the mother pretended to feed the doll, and once again the infant took the bottle and put it in his mouth (Dyad I , 49 weeks).

In the next phase, infants alternated putting the bottle in their own mouths with putting the bottle in the baby doll’s mouth. Mothers typically ignored the infant or changed the direction of play when the infants put the bottle into their own mouths, and mothers created narrative stories about the conventionally appropriate uses of the bottle (preconsensual initiation). The mother begins by demonstrating the pretend feeding of the doll. The infant takes the bottle and puts it in his mouth which meets mother’s disapproval. although other instances of mouthing in this session were ignored by the mother. The infant initiated a series of sequences in which he put the bottle in his mouth, offered it to the mother, and then put the bottle in the doll’s mouth. The mother pretended to drink from the bottle, becoming a play partner rather than a teacher. After that the infant fed the doll accompanied by sucking sounds from the mother and infant laughter (Dyad 1, 54 weeks).

Finally, infants completely stopped sucking on the bottle as play became pretend/symbolic. When putting the bottle into the doll’s mouth, the infants would smile at their mothers and their mothers would voice approval (consensual co-action). For the BABY/BOTTLE play, in which we have more instances than for the TELEPHONE game, we can see that these phases were highly variable with respect to age of onset (Table 111).

TABLE 111 INFANT AGES(WEEKS) AT EACHPHASEOF BABYIBOTTLE GAME‘ Dyad 1

Dyad 2

Dyad 3 ____

Nonconsensual co-action Preconsensual responsiveness Preconsensual initiation Consensual co-action Adapted from Dedo (1993).

42 51 53 62

41 43 44 52

42 58 70 71

Dyad 4 ~ _ _ _ _

41 46 54 72

DEVELOPMENT AND RELATIONSHIPS

283

2. Innovation and Consensual Reframing as Developmental Process In conclusion, there are several points regarding these observed developmental changes in the process of reframing. a . The consensual reframing of innovations occurs in the context of existing relationship frames. In each case, innovations were introduced and reframed with respect to an ongoing relationship frame, such as mutual play with a particular object. Consensual reframing makes sense only as a continuous elaboration of the dyad’s history of coregulated action within each particular frame. The developmental gradient, from nonconsensual to consensual co-activity, can be observed at any age and at any time during the development of relationships around particular themes. b. Consensual reframing takes time to elaborate. Innovations cannot be said to be reframed until they are adopted as consensually coregulated action by all the participants. Until such a time, each partner may appear to be using the same object but in different ways and with different purposes. The existence of joint goals is created and elaborated through coregulated negotiations; it is the result of development rather than its source (Branco and Valsiner, 1992; Fogel, 1993; Lyra and RossettiFerreira, 1994). I believe that some scholars of socially mediated learning fail to recognize the subtlety and difficulty of such negotiations. Particularly as children get older and the cultural activities become more complex and demanding, shared purposes of co-action may take years to achieve and may not ever be achieved. An example is Goodnow’s (1988) studies of the negotiation of household chores in which parents and children persist in having different reasons and levels of compliance for such activities over many years. c . Consensual reframing may be accompanied by emotional markers. In both of the types of play reported in this paper, open expressions of smiling and laughter did not begin until infants and parents reached the level of consensual co-action with shared understanding of the meaning and purpose of the activity. Before then, there were often struggles for control, anger over disagreements, and distress at the disappointing lack of co-activity on the part of both mother and infant. Once again, imagine the experience of getting children to comply with household chores. In our research on emotion, we have found that expressions are tied to specific phases of co-activity between partners, both in real-time processes and with respect to developmental changes in frames. During a peek-a-boo game, for example, smiles occur only at particular junctures of the game. When peek-a-boo is first observed, between 9 and 12 months of age, infant smiles occur following the adult’s uncovering actions. After some experience with the game, infant smiles are more likely to occur when the infant uncovers, or during the prior covering phase in anticipation

284

ALAN FOGEL

of the uncovering (Fogel et al., 1992). This suggests that facial expressions and body movements are fully embedded within and derive their meanings from the social communicative process. IV. RESEARCH IMPLICATIONS If one considers the relationship as the unit of conceptual analysis, and if furthermore one views relationships as inherently historical and developmental systems, then research designs for the study of behavior and its development should reflect these views (Fogel, 1993). This can be accomplished by following a few simple guidelines. a . Change over time should be an explicit feature of the study design. All action, as explained in this paper, is more than its behavioral manifestation in the present. Action is fundamentally dynamic, leading from the past and pointing toward the future. The simplest unit of action is the sequencing of behavior in real time, such as the sequence of acts during a mating encounter. Behavioral codes representing events should not be simply summed across an observation. Rather, they should be counted in relation to prior, co-occuring, and subsequent acts. The simplest action unit, therefore, has three elements representing the immediate past, the present, and the immediate future (Markova, 1990). The next simplest unit of action analysis is the examination of action across repeated observations, such as observing every mating encounter between two individuals over the course of a single mating season. The most comprehensive unit of analysis is the observation of the same mating pair developmentally, across many years of mating seasons. In these two cases, the focus of research is on changes over observations in the cooccurrences and sequences of acts, including both communicative and metacommunicative actions. It is crucial for the unit of developmental analysis to be the relationship, and change within relationships to be the comparison metric. The goal is to understand the historical transformation of the function or meaning of the actions for the participants (Fogel, 1993; Oyama, 1989; Thelen, 1989a,b; Thorngate, 1987). 6 . Experimental designs should preserve the relationship unit of analysis. Animal models are used frequently when experimental paradigms introduce a perturbation into an ongoing relationship that would not be ethically permitted in human populations. Separation and fostering studies of attachment in nonhuman species are examples of this research approach. Putting aside the ethical issues in the experimental use of these animal models, one can ask about their scientific purpose as a tool for understandingrelationshipformation and development. Separation studies

DEVELOPMENT AND RELATIONSHIPS

285

have clearly demonstrated that relationships continue even in the absence of the partner, which supports the view that action in the present is fundamentally configured by the past. Yet, separation research offers little insight into the process of relationship development since, by its nature, the relationship under study has ended the social communicative phase of its development. Experimental designs can be used for the study of relationship development only if the social communication is allowed to continue in some form. Thus, separations can be useful in the context of a subsequent reunion with the same partner. Experiments that change the size, shape, color, or timing of features of interactive situations/partners may also serve to illuminate the essential features of the relationship process. Such studies will be different from the ethological work on signals and releasers if they are done with respect to the historicaUtempora1features of ongoing relationship process, since the same action is likely to have different functions or meanings depending on prior history and context. V. SUMMARY

New approaches for the study of development and individualdifferences can arise by taking a relational, rather than an individual, perspective on developmental change. Relationships should be the unit of analysis, and they should be studied from a developmental perspective in order to understand how they create and preserve their own history within the cultural and ecological context of the relationship system. Relationships, defined as ongoing historical social processes, are inherently developmental. The open-endedness of coregulation and innovation within a relationship creates the seeds for its own development. One need not postulate the existence of a mechanism for development outside of the relationship process itself. Even when change seems to originate within one or the other individual, or within the cultural or ecological context, that change must nevertheless be negotiated into the relationship system. This negotiation of innovation is fundamental to the change process and determines whether the relationship will survive over time as an intact system in the face of externally or internally induced changes. The lesson of this theoretical perspective is that relationships are more than the sum of their behavioral parts. There is both a history and a future that are only implicit in the present features of social action. Research on social development and relationships should incorporate change over time as an essential component of the research design. This could be done by observing the unfolding of action over a single observation period, instead

286

ALAN FOGEL

of simply coding and summing activity as if each instance had the same meaning to the participants. Repeated observations of relationships over more extended periods of time are also required. In either research design, analytical procedures should be adopted that show the historical transformation of meaning of the same actions over time. A quantitative analysis, in which the frequency counts are taken out of the context of the social frame from which they derive their meaning, is less likely to yield important clues to developmental process. Acknowledgments This work is supported by grants from the National Institute of Health (ROlHD21036) and the National Science Foundation (BNS9006756).I am grateful to Angela Branco, Wendy Haight, Maria Lyra, Jay Rosenblatt, and two anonymous reviewers for their critical comments on earlier versions of this paper. References Bandura, A. (1986).“Social foundations ofThought and Action: A Social Cognitive Theory.” Prentice-Hall, Englewood Cliffs, New Jersey. Bateson, G. (1955). The message: “This is play.” I n “Group Processes” (B. Schaffner, ed.), Vol. 2. Madison Printing Co, Madison, New Jersey. Bavelas, J. B., Black, A., Lemery, C. R., and Mullett, J. (1986). “I show how you feel.” Motor mimicry as a communicative act. Journal of Personality and Social Psychology 50,322-329. Bergson, H. (1911). “Creative Evolution.” Henry Holt & Co., New York. Bettes, B. A. (1988). Maternal depression and motherese: Temporal and intonational features. Child Development 59, 1089-1096. Bowlby, J. (1969). “Attachment and Loss: Vol. 1: Attachment.” Basic Books, New York. Branco, A. U., and Valsiner, J. (1992).Development of convergence and divergence in joint actions of preschool children within structured social contexts. Paper presented at the 25th International Congress of Psychology, Brussels, Belgium. Brentano, F. (1973). Psychology from an empirical standpoint (trans. L. L. McAlister). Routledge and Kegan Paul, London. (Original work published 1874). Burnham, D. K., and Dickinson, R. G.(1981). The determinants of visual capture and visual pursuit in infancy. Infant Behavior and Development 4, 359-372. Carvalho, A. M. A. (1988). Algumas reflexoes sobre o us0 da categoria “interacao social.” Annals of XVIII Meeting of Brazilian Psychological Society, 51 1-515. Cheney, D. L., Seyfarth, R. M., and Smuts, B. (1986). Social relationships and social cognition in nonhuman primates. Science 234, 1361-1366. Chovil, N. (1991). Social determinants of facial displays. Journal of Nonverbal Behavior 15(3), 141-154. de Waal, F. (1989). “Chimpanzee Politics: Power and Sex among apes.” Harper & Row, New York. Dedo, J. (1993). Changes in mother and infant behaviors in the transition to symbolic play: A longitudinal case study. Presented at Society for Research in Child Development Conference, March, New Orleans, Louisiana.

DEVELOPMENT AND RELATIONSHIPS

287

Duck, S. (1991). “Understanding Relationships.” The Guildord Press, New York. Fleming, A. S., Ruble, D. N., Flett, G. L., and Shaul, D. L. (1988). Postpartum adjustment in first-time mothers: Relations between mood, maternal attitudes, and mother-infant interactions. Developmental Psychology 24, 71-81. Fogel, A. (1990a). The process of developmental change in infant communicative action: Using dynamic systems theory to study individual ontogenies. In “Individual Differences in Infancy: Reliability, Stability and Prediction” (J. Colombo and J. Fagen, eds.). Erlbaum, Hillsdale, New Jersey. Fogel, A. (1990b).Sensorimotor factors in communicative development. In “Sensory-Motor Organizations and Development in Infancy and Early Childhood” (H. Bloch and B. I. Bertenthal, eds.), pp. 75-88. Kluwer Academic, Boston, Massachusetts. Fogel, A. (1993). “Developing through Relationships.” Harvester Wheatsheaf, London, and University of Chicago Press, Chicago. Fogel, A., Nwokah, E., Dedo, J., Messinger, D.. Dickson, K. L., Matusov, E., and Holt, S. A. (1992). Social process theory of emotion: A dynamic systems perspective. Social Development 1, 122-142. Fogel, A., Nwokah, E., and Karns, J. (1993). Parent-infantgamesasdynamic socialsystems. In “Parents and Children Playing” (K. B. MacDonald, ed.). SUNY Press, Albany, New York. Foucault, M. (1965). “Madness and Civilization.” Pantheon, New York. Fridlund, A. J. (1991). Sociality of solitary smiling: Potentiation by an implicit audience. Journal of Personality and Social Psychology 60, 229-240. Garfinkel, H. (1967). Studies in ethnomethodology. Prentice-Hall, Englewood Cliffs, New Jersey. Gelfand, D. M., and Teti, D. M. (1990). The effects of maternal depression on children. Clinical Psychology Review 10,329-353. Goffman, E. (1974). “Frame Analysis: An Essay on the Organization of Experience.” Harvard University Press, Cambridge, Massachusetts. Goodall, J. (1971). “In the Shadow of Man.” Dell Publishing, New York. Goodnow, J. J. (1988). Parents’ ideas, actions, and feelings: Models and methods from developmental and social psychology. Child Development 59, 286-320. Gottlieb, G. (1992). “Individual Development and Evolution: The Genesis of Novel Behavior.” Oxford University Press, New York. Gould, S. (1981). “The Mismeasure of Man.” Norton, New York. Gray, H. (1978). Learning to take an object from the mother. In “Action, Gesture and Symbol: The Emergence of Language” (A. Lock, ed.), pp. 159-182. Academic Press, New York. Guilford, T., and Dawkins, M. S. (1991). Receiver psychology and the evolution of animal signals. Animal Behaviour 42, 1-14. Haight, W., and Miller, P. (1993). “The Development of Pretend Play in Sociocultural Context: A Longitudinal Study of Everyday Pretend from 1 to4 Years.” SUNY, Albany, New York. Heath, C. (1984). Talk and recipiency: Sequential organization in speech and body movement. In “Structures of Social Action: Studies in Conversation Analysis” (J. M. Atkinson and J. Heritage, eds.), pp. 247-265. Cambridge University Press, New York. Hermans, H., and Kempen, H. (1993). “The Dialogical Self.” Academic Press, New York. Hinde, R. A. (1977). Mother-infant separation and the nature of inter-individual relationships: Experiments with rhesus monkeys. Proc. R. SOC.Lond. B. 1%,29-50. Hinde, R. A. (1978). Interpersonal relationships: In quest of a science. Psychological Medicine 8, 373-386.

288

ALAN FOGEL

Hinde, R. A. (1985). Expression and negotiation. I n “The Development of Expressive Behavior: Biology Environment Interactions” (G. Zivin, ed.), pp. 103-1 16. Academic Press, New York. Jones, S. S., and Raag, T. (1989). Smile production in older infants: The importance of a social recipient for the facial signal. Child Development 60, 81 1-818. Kagan, J., Reznick, J. S., Snidman, N., Gibbons, J., and Johnson, M. 0. (1988). Childhood derivatives of inhibition and lack of inhibition to the unfamiliar. Child Development 59, 1580- 1589.

Kendon, A. (1975). Some functions of the face in a kissing round. Journal of the International Association for Semiotic Studies 14, 99-334. Kendon, A. (1985). Behavioral foundations for the process of frame attunement in face-toface interaction. I n “Discovery Strategies in the Psychology of Action” (G. P. Ginsburg, M. Brenner, and M. von Cranach, eds.). Academic Press, London. Kraut, R. E., and Johnston, R. E. (1979). Social and emotional messages of smiling: An ethological approach. Journal of Personality and Social Psychology 37, 1539-1553. Kugler, P. N., Kelso, J. A. S., and Turvey, M. T. (1982). On coordination and control in naturally developing systems. I n “The Development of Movement Coordination and Control” ( J . A. S. Kelso and J. E. Clark, eds.), pp. 5-78. Wiley, New York. Labov, W. (1972). “Language in the Inner City.” University of Pennsylvania Press, Philadelphia. Landry, S. H., Chapieski, M. L., and Schmidt, M. (1986). Effects of maternal attentiondirecting strategies on preterms’ response to toys. Infant Behavior and Development 9,257-268.

Lutz, C. (1988). “Unnatural Emotions: Everyday Sentiments on a Micronesian Atoll and Their Challenge to a Western Theory.” University of Chicago Press, Chicago. Lyra, M., and Rossetti-Ferreira, M. (194). Transformation and construction in social interaction: A new perspective on analysis of the mother-infant dyad. In “Child Development within Culturally Structured Environment” ( J. Valsiner, ed.), Vol. 3. Ablex Publishing, Norwood, New Jersey. Markova, I. (1990). Why the dynamics of dialogue? I n “The Dynamics of Dialogue” (I. Markova and K. Foppa, eds.), pp. 1-22. Springer-Verlag, New York. Marler, P., and Mitani, J. (1988). Vocal communication in primates and birds: Parallels and contrasts. I n “Primate Vocal Communication” (D. Todt, P. Goedeking, and D. Symmes, eds.). Springer-Verlag. Berlin and Heidelberg. McArthur, L. Z., and Baron, R. M. (1983). Toward an ecological theory of social perception. Psychological Review 90, 215-238. Mead, G. H. (1934). “Mind, Self, and Society.” University of Chicago Press, Chicago. Merleau-Ponty, M. (1%2). “Phenomenology of Perception” (trans. C. Smith). Routledge and Kegan Paul, London. Mitchell, R. W. (1991). Bateson’s concept of “metacommunication” in play. New Ideas in Psychology 9,73-87. Newtson, D., and Engquist, G. (1976). The perceptual organization of ongoing behavior. Journal of Experimenral Social Psychology 12,436-450. Oyama, S. (1989). Ontogeny and the central dogma: Do we need the concept of genetic programming in order to have an evolutionary perspective? I n “Systems and Development” (M. Gunnar and E. Thelen, eds.), Vol. 22, pp. 1-34. The Minnesota Symposia on Child Psychology. Erlbaum, Hillsdale, New Jersey. Paninello, R. M., and Ruff, H. A. (1988). The influence of adult intervention on infants’ level of attention. Child Development 59, 1125-1 135.

DEVELOPMENT AND RELATIONSHIPS

289

Pawluk, C. J. (1989).Social construction ofteasing. Journal for the Theory of SocialBehavior 19, 146-167. Piaget. J. (1952). “The Origins of Intelligence in Children.” International Universities Press, New York. Plomin, R. (1990). The role of inheritance in behavior. Science 248, 183-188. Provine, R. R., and Fischer, K. R. (1989). Laughing, smiling, and talking: Relation to sleeping and social context in humans. Ethology 83,295-305. Ramsay, D. S. (1980). Onset of unimanual handedness in infants. Infant Behavior and Development 3,377-386. Reimers, M., and Fogel, A. (1992). The evolutions of joint attention to objects between infants and their mothers: Diversity and convergence. Analise Psicologia. Reinecke, M.. and Fogel, A. (1994). The development of referential offering in the first year. Early Development and Parenting 3, 181-186. Rivera, J., and Grinkis, C. (1986). Emotions as social relationships. Motivation and Emotion 10, 351-369. Rochat, P., and Senders, S. J. (1990). Sitting and reaching in infancy. Paper presented at the Seventh International Conference on Infant Studies, April, Montreal. Rogoff, B. (1982). Integrating context and cognitive development. In “Advances in Development Psychology” (M. E. Lamb and A. L. Brown, eds.), Vol. 2, pp. 125-170. Erlbaum, Hillsdale, New Jersey. Rosenblatt, J. S. (1963). The basis of synchrony in the behavioral interaction between the mother and her offspring in the laboratory rat. I n “Determinants of Infant Behavior 111” (B. M. Foss, ed.), pp. 3-45. Methuen & Co., London. Rosenblatt, J. S. (1983). Olfaction mediates developmental transition in the altricial newborn of selected species of mammals. Developmental Psychobiology 16(5), 347-375. Sameroff, A. J., and Emde, R. N. (1989). “Relationship Disturbances in Early Childhood: A Developmental Approach.” Basic Books, New York. Sander, L. W. (1977). The regulation of exchange in the infant-caretaker system and some aspects of the context-content relationship. In “Interaction, Conversation, and the Development of Language” (M. Lewis and L. A. Rosenblum, eds.), pp. 133-147. Wiley & Sons, New York. Schneirla. T. C. (1956). Interrelationships of the innate and the acquired in instinctive behavior. In “L’Instinct dans le Comportement des Animaux et de I’Homme” (P. P. Grasse, ed.). pp. 387-452. Mason et Cie, Paris. Shafer, K. (1990).Metacommunication behaviors used by adult best friends. Paper presented at the annual meeting of the Western States Communication Association, Sacramento, California. Shannon, C. E. (1963). The mathematical theory of communication. In “The Mathematical Theory of Communication (C. E. Shannon and W. Weaver, eds.), pp. 29-125. University of Illinois Press, Urbana. Smith, W. J. (1977). “The Behavior of Communicating: An Ethological Approach” Harvard University Press, Cambridge, Massachusetts. Sroufe, L. A. (1989). Relationships, self, and individual adaptation. In “Relationship Disturbances in Early Childhood: A Developmental Approach” (A. J. Sameroff and R. N . Emde, eds.), pp. 70-94. Basic Books, New York. Suomi, S. J. (1987). Genetic and maternal contributions to individual differences in rhesus monkey biobehavioral development. In “Perinatal Development: A Psychobiological Perspective” (N. A. Krasnegor, E. M. Blass, M. A. Hofer, and W. P. Smotherman, eds.), pp. 397-420. Academic Press, New York.

290

ALAN FOGEL

Suomi, S. J., Mineka, S., and Delizio, R. D. (1983). Short- and long-term effects ofrepetitive mother-infant separations on social development in rhesus monkeys. Developmental Psychology 19,770-786. Thelen, E. (1989a). The (Re)discovery of motor development: Learning new things from an old field. Developmental Psychology 25(6), 946-949. Thelen, E. (l989b). Self-organization in developmental processes. Can systems approaches work? I n “Systems in Development” (M. Gunnar, ed.), pp. 77-1 18. Erlbaum, Hillsdale, New Jersey. Thelen, E., Corbetta, D.. Kamm, K., Spencer, J. P., Schneider, K., and Zernicke, R. F. (1993). The transition to reaching: Mapping intention and intrinsic dynamics. Child Development 64, 1058-1098. Thelen, E., and Ulrich, B. D. (1991).Hidden skills. Monographs of the Society for Research in Child Development 56, 6-97. Thompson, R. A., Connell, J. P., and Bridges, L. J. (1988). Temperament, emotion, and social interactive behavior in the strange situation: A component process analysis of attachment system functioning. Child Development 59, 1102-1 110. Thorngate, W. (1987). The production, detection, and explanation of behavior patterns. In “The Individual Subject and Scientific Psychology” ( J . Valsiner, ed.), pp. 71-93. Plenum Press, New York. Trevarthen, C. (1979). Communication and cooperation in early infancy: A description of primary intersubjectivity. In “Before Speech: The Beginning of Interpersonal Communication” (M. Bullowa. ed.), pp. 321-347. Cambridge University Press, New York. Trevarthen, C. (1993). The function of emotions in early infant communication and development. I n “New Perspectives in Early Communicative Development” (J. Nadel and L. Camaioni, ed.), pp. 48-81. Routledge, New York. Valenti, S. S., and Good, J. M. (1991). Social affordances and interaction I: Introduction. Ecological Psychology 3, 77-98. Valsiner, J. (1987). “Culture and the Development of Children’s Action” Wiley, New York. Valsiner, J. (1993). Irreversibility of rime and the construction of historical developmental psychology. Paper presented at the XI1 Biennial Meetings of the International Society for the Study of Behavioral Development, Recife, Brazil. Vedeler, D. (1987). Infant intentionality and the attribution of intentions to infants. Human Development 30, 1-17. von Hofsten, C., and Ronnqvist, L. (1993). The structuring of neonatal arm movements. Child Development 64, 1046-1057. von Neumann, J. (1958). “The Computer and the Brain.” Yale University Press, New Haven, Connecticut. Weiss, P. A. (1969). The living system: Determinism stratified. I n “Beyond Reductionism: New Perspectives in the Life Sciences” (A. Koestler and J. R. Smythies, eds.), pp. 3-55. Beacon Press, Boston, Massachusetts. Wertsch, J. V. (1985). “Vygotsky and the Social Formation of Mind.” Harvard University Press, Cambridge, Massachusetts. West, J. M., King, A. P., and Duff, M. A. (1990). Communicating about communicating: When innate is not enough. Developmenial Psychobiology 23, 585-598. West, L.. and Fogel, A. (1990). Maternal guidance of object interaction. Paper presented at International Conference on Infant Studies, April, Montreal. White, B. L., and Castle, P. W. (1964). Visual exploratory behavior following postnatal handling of human infants. Perceptual and Motor Skills 18, 497-502. Wilmot. W. W. (1980). Metacommunication:A re-examination and extension. In “Communication Yearbook 4“ (D. Nimmo, ed.), pp. 61-69. Transaction Books, New Jersey.

ADVANCES IN THE STUDY OF BEHAVIOR. VOL. 24

Why Do Females Mate with Multiple Males? The Sexually Selected Sperm Hypothesis LAURENT KELLER ZOOLOGISCHES INSTITUT BERN UNIVERSITY HINTERKAPPELEN, SWITZERLAND AND

INSTITUT DE ZOOLOGIE ET D’ECOLOGIE ANIMALE UNIVERSITY OF LAUSANNE LAUSANNE, SWITZERLAND

HUDSONK . REEVE SECTION OF NEUROBIOLOGY A N D BEHAVIOR CORNELL UNIVERSITY ITHACA, NEW YORK

I. INTRODUCTION One of the more debated issues in evolutionary biology is why in many species females mate with multiple males. For females, the primary function of copulation is to fertilize their ova. If one mating provides females with sufficient sperm, and the males provide little but gametes in mating, there are no obvious reasons why females should benefit from further matings (Daly, 1978; Parker, 1979). Mating entails costs such as: (1) time and energy costs devoted to courtship and copulation; (2) increased risk of predation while mating; and (3) risk of disease from parasite transmission, (4) negative effects of some seminal fluid products transferred by males during mating, all of which may shorten the female’s life-span (Daly , 1978; Thornhill and Alcock, 1983; Sherman, Seeley, and Reeve, 1988; Arnqvist, 1989; Fowler and Patridge, 1989; Chapman, Hutchings, and Partridge, 1993, 1995; Sheldon, 1993; Chapman, Liddle, Kalb, Wolfner, and Partridge, 1995; Keller, 1995). However, despite the potential costs, females of many species mate with several males (e.g., Thornill and Alcock, 1983; Austad, 1984; Smith, 1984a; Eberhard, 1985; Ridley, 1990; 29 1

Copyright 0 1995 by Academic Press. Inc. All rights of reproduction in any form reserved.

292

LAURENT KELLER AND HUDSON K. REEVE

Westneat, Sherman, and Morton, 1990; Birkhead and Mgller, 1992). Several hypotheses have been put forward to explain why females mate multiply despite the apparent costs. These hypotheses include the suggestions that: (1) females replenish depleted sperm supplies (e.g., Thornill, 1976; Lewis and Austad, 1994); (2) females receive nutrients or other paternal investment from males (Thornill and Alcock, 1983; Pitnick, Markow, and Riedy, 1991; but see Chapman, Trevitt and Partridge, 1994); (3) females increase probability of mating with high-quality males through extra-pair copulation (Birkhead and Mgller, 1992, 1993; Kempenaers et al., 1992; Mgller, 1992); (4) females bet hedge against sterility of the first male (Walker, 1980); (5) sperm competition allows a female to gain better genes for her offspring if there is a correlation between sperm and offspring quality (Madsen, Shine, Loman, and Hakansson, 1992; Parker, 1992; 01sson, Gullberg, Tegelstrom, and Shine, 1994a; but see Capula and Luiselli, 1994, Keller, 1994; Olsson, Madsen, Shine, Gullberg, and Tegelstrom, 1994b); (6) females increase genetic variability of the offspring (e.g., Walker, 1980; Hamilton, 1987; Sherman et al., 1988; Ridley, 1993; Keller and Reeve, 1994); and (7) genetic correlation between the sexes results in females’ tendency to mate multiply because there is strong selection for this behavior in males (Halliday and Arnold, 1987; but see Sherman and Westneat, 1988; Arnold and Halliday, 1988; Cheng and Siegel, 1990; Schwartz and Boake, 1992). However, few studies have attempted to discriminate among the hypotheses for the advantages of multiple mating (here defined as mating with several males), and thus the benefits of multiple mating remain unclear in many species (e.g., Parker, 1992). An intriguing hypothesis for the widespread occurrence of multiple mating was proposed by Harvey and Bennett (1985; see also Cox and Le Boeuf, 1977). They suggested that sperm competition might contribute to the evolution of multiple mating because, by mating multiply and mixing the sperm of several males, females increase the probability that their eggs are fertilized by the more competitive sperm, which thereby will increase the chances that their sons will produce competitive sperm. Using a simple, two-locus genetic model, Curtsinger ( 1991) investigated the validity of this hypothesis. In his model, one locus influenced sperm competition and the other one affected female mating behavior. He concluded that the conditions for sperm competition to promote the evolution of multiple mating in females are very restrictive; there must be tight linkage between the two loci, positive linkage disequilibrium, and no fitness cost associated with multiple mating. However, as Curtsinger (1991) himself pointed out, different genetical assumptions of the model

WHY DO FEMALES MATE MULTIPLY?

293

might yield different conclusions. Specifically, one critical assumption of his model is that the allele conferring high sperm competitiveness rapidly attains a selective equilibrium, at which point evolution of multiple mating ceases (or even reverses if multiple mating is costly), because the genetic correlation between mating frequency and sperm competitiveness decays. When there is no such genetic correlation, the multiple mating allele can no longer increase in frequency by hitchhiking with an allele conferring higher sperm competitiveness. However, it is not necessarily the case that alleles affecting sperm competitiveness will reach a selective equilibrium (e.g., go to fixation). For example, random mutations will tend to lower and maintain the average degree of sperm competitiveness beneath the optimum or selectively equilibria1 value. The effect of random mutation on sexually selected traits has been discussed in detail by Pomiankowski, Iwasa, and Nee (1991). They showed that random mutations are likely to have a directional effect, tending to create an excess of poorly ornamented males. Indeed, significant genetic variability in male sexual traits has been reported in several species (Hedrick, 1988; Mgller, 1991 ;Houde, 1992; Bakker, 1993; Wilkinson and Reillo, 1994). A recent review including data on 36 species showed that there is significant additive genetic variance in the sexual characters of most species (Pomiankowski and Mgller, in press). Similarly, it is likely that biased mutation will maintain additive genetic variance in sperm competitiveness, creating the possibility that the genetic correlation driving the evolution of multiple mating never decays to zero. Females that mate multiply will always tend to produce sons that possess both a (nonmutant) allele for high sperm competitiveness and an allele for multiple mating, continually establishing the required genetic correlation. Thus, multiple mating could spread to fixation or even become further elaborated. A nondecaying coupling of mating frequency and sperm competitiveness also can be created by the oscillating evolution of multiple sperm types that form a nontransitive “dominance” hierarchy. For example, hypothetical type A sperm may be more efficient than hypothetical type B sperm, which in turn may be more efficient than type C sperm, but hypothetical type C sperm may be more efficient than type A sperm. Thus, oscillatory evolutionary dynamics such as B + A + C + B, and so on, might result (see Appendix). In such a situation, multiple mating would always increase the probability that a female receives the currently more competitive sperm and, because she therefore produces sons that tend to possess both the multiple mating allele and a “highly” competitive sperm allele, the genetic correlation never decays and multiple mating can continue to evolve.

294

LAURENT KELLER AND HUDSON K. REEVE

11. THESPERMSEXUAL SELECTION HYPOTHESIS We reexamine and expand the Harvey and Bennett (1985) hypothesis, here referred to as the sperm sexual selection (SSS) hypothesis, to account for the widespread occurrence of multiple mating in females. We argue that multiple mating by females may rapidly spread when initially a small fraction of the females mate multiply, and if there is a heritable difference among males in one or several of the four following characteristics: (1) the quantity of sperm they produce; (2) the success of their sperm in reaching and fertilizing an egg; (3) their ability to displace the sperm that females stored during previous matings; and (4) their ability to prevent any other male from subsequently introducing sperm (e.g., differential efficiency of mating plugs). Females that mate multiply increase the probability that their eggs are fertilized by a male with high fertilization efficiency, and thus genes predisposing multiple mating can rapidly spread because the sons of such females will tend to carry both these genes and also genes specifying sperm with relatively high fertilization efficiency. Moreover, females should provide opportunities for increased competition among sperm of different males, such as by increasing the length or complexity of their reproductive tracts. Multiple mating (together with female reproductive morphology) and male fertilization efficiency thus may coevolve via a Fisherian selection process even though mating is random with respect to observable male phenotypes. This argument is exactly parallel to those purporting to explain the evolution of exaggerated male secondary sexual characteristics (such as peacocks’ tails) by means of a Fisherian process (Maynard Smith, 1991). An initial low frequency of multiple mating may occur because of variable direct advantages of multiple mating to females or because of forced copulations by males (Parker, 1979; Walker, 1980; Austad, 1984; Smith, 1984a; Crozier and Page, 1985; Birkhead and Mgller, 1992). Once multiple mating occurs at a small frequency, regardless of its cause, an opportunity is created for between-male sperm competition, that is, selection should favor males with the higher fertilization efficiency. Indeed, since Parker’s seminal paper on sperm competition (1970), numerous studies have been devoted to showing that many male adaptations (behavioral, morphological, and physiological) relate to enhancing the success of self’s sperm against a rival male’s sperm (Parker and Stuart, 1976; Waage, 1979; Smith, 1979, 1984b; Eberhard, 1985; Baker and Bellis, 1989; Schwagmeyer and Parker, 1990; Radwan and Witalinski, 1991). Much of sperm structure and organization can also be interpreted in terms of intermale (as opposed to within-male) gamete competition (Sivinski, 1984; Silberglied, Shepherd, and Dickinson, 1984; see below).

WHY DO FEMALES MATE MULTIPLY?

295

Females that increase their number of matings will increase the probability that their sons will themselves have high fertilization efficiency. For example, in the case of variability in sperm competitiveness, increased number of matings by females will increase the probability that their sons will produce the exceptionally competitive sperm if, as seems likely, between-male variation in sperm quality exceeds within-male variation in sperm quality. For example, suppose that an allele B, which occurs in a total proportion m of males, enables B-bearing sperm to always outfertilize b-bearing sperm, where b is the alternative allele. A bb female that mates randomly one time produces Bb sons with frequency m, whereas a bb female that mates two times produces Bb sons with higher frequency 1 - ( 1 - m)2 = m(2 - m).This situation is logically the same as that under which classical Fisherian sexual selection occurs. Polyandry (here defined as mating with several males) is analogous to active female choice (since it provides a mechanism by which sperm are chosen), and genes for polyandry are expected to become associated with genes specifying the favored male trait (in this case, sperm that are more competitive), because sons of polyandrous females will tend to produce the more competitive sperm as well as possess genes for polyandry. As models of classical Fisherian selection suggest (O’Donald, 1962; Lande, 1980; Kirkpatrick, 1982; Pomiankowski et al., 1991), the association between polyandry and sperm-competitiveness genes can thus lead to Fisherian sexual selection enhancing both female mating frequency and sperm competitiveness, with positive selection for competitive-sperm alleles causing genetically correlated alleles for polyandry to increase in frequency (Fig. 1). Thus, sperm sexual selection coupled with Fisherian selection may increase the frequency of multiple mating once it is established at even a small frequency and maintain multiple mating even when its initial advantages (if any) are no longer present. Since longer or more complex female reproductive tracts might also in effect increase female “choosiness” for higher quality sperm, the same process may lead to a coevolution of sperm competitiveness and length or complexity of female reproductive tracts. Polyandry may be costly to females (e.g., by increasing the exposure to predation and the risk of disease from parasite transmission; Sherman et al., 1988; Sheldon, 1993); however, this need not lead to a decay in polyandry if, as discussed above, “biased mutation” tends to lower the population-average degree of sperm competitiveness beneath the optimum value (Pomiankowski et al., 1991). The assumptions of the SSS hypothesis are that males differ in their fertilization success, and there is a heritable component for variation in male fertilization success. Only a few studies have focused on intraspeciJc variability in male fertilization success in cases of multiple mating, and

296

LAURENT KELLER AND HUDSON K. REEVE

Benefits of

Forced copulations

(Initial frequency of rnuitiple-mating)

\

>--I> Increased polyandry, exaggerated female reproductive tract

Fisherian coevolution

+

Selection for males with high fertilization efficiency

FIG. 1. Diagrammatic representation of sperm sexual selection. A small initial frequency of multiple mating by females (due to direct benefits of polyandry to females or occasional forced copulations) leads to intermale sperm competition. Genes for multiple mating become associated with genes for high male fertilization efficiency because the sons of multiply mating females will tend to produce males with higher fertilization efficiency. Thus, selection for high fertilization efficiency will tend to promote the spread of multiple mating, which in turn leads to intensified sperm sexual selection. This positive feedback process can trigger a Fishenan coevolution of female mating frequency and male fertilization efficiency.

evidence suggests high variability. Lewis and Austad (1990) systematically examined sources of variation in sperm precedence in Tribolium castaneum. They showed that consistent differences among males represent a significant percentage (17.8%) of the total variation in sperm precedence. Reviewing the literature on sperm precedence, these authors found that in virtually all studied species there is an extremely wide range in the proportion of matings secured by the second male (see also Cordero and Miller, 1992). Differences in fertilization success have also been found for different strains of insects and mammals (Wilkes, 1966; Levine, 1976; Holmes, 1974; Childress and Hartl, 1972; Rout and Bundgaard, 1977; Gromko, Gilbert, and Richmond, 1984). These differences may arise from differences among males in the quantity of sperm they produce, the probability that their sperm will reach and fertilize an egg, their ability to displace the sperm females stored during previous matings, and their ability to prevent any other male from subsequently introducing sperm. In many species the number of sperm delivered in a single ejaculate has been shown to vary between males (Baker and Bellis, 1989, 1993; Cordero and Miller, 1992; Keller and Passera, 1992) and to influence the probability of paternity (Huck, cited in Ginsberg and Huck, 1989). Sperm of different males also differ in their probability of fertilizing an egg. For example, when species as

WHY DO FEMALES MATE MULTIPLY?

297

diverse as chickens, mice, rabbits, swine, insects, and cattle are artificially inseminated with similar numbers of sperm from two or three males, sperm from particular animals are consistently much more effective in obtaining fertilization (e.g., R. G. Edwards, 1955; Beatty, 1960, 1975; Beatty, Bennett, Hall, Hancock, and Stewart, 1969; Martin and Dziuk, 1977). These differences might be associated with the differences in size, shape, and metabolism of spermatozoa of different males (e.g., Napier, 1961; Woolley and Beatty, 1967; Kirby and Froman, 1991; Krzanowska, Wabik-sliz, and Rafinski, 1991; Ward and Hauschteck-Jungen, 1993), which in turn may influence sperm motility, sperm longevity, or the ability of sperm to penetrate the ovum successfully (Napier, 1961; Bellis, Baker, and Gage, 1990a; Bellis, Baker, Matson, and Chew, 1990b; Gomendio and Roldan, 1991, 1993). Finally, males differ in their ability to displace the sperm stored by females from a previous mating (Waage, 1979; Birkhead and Hunter, 1990). It has, for example, been shown that males may vary in their ability to remove sperm from a female storage organ (Corder0 and Miller, 1992; Simmons and Parker, 1992). Genetic variability in male fertilization efficiency is supported by differences in fertilization success among males of different strains of various animal species (Napier, 1961; Woolley and Beatty, 1967; Kirby and Froman, 1991; Krzanowska et al., 1991). Breeding experiments on domestic ungulates also showed that variation in testicular size and number of sperm produced is highly heritable and can be rapidly manipulated through artificial selection (Neeley, Johnson, Dillard, and Robison, 1982; Toelle, Johnson, and Robison, 1984).Furthermore, a genetic and heritable component in the probability that a sperm will fertilize an egg has been demonstrated, for example, in mice (R. G. Edwards, 1955;Beatty, 1975).Heritability of sperm dimensions, shape, and structure has also been found in mice and rabbits (Napier, 1961; Woolley and Beatty, 1967; Krzanowska et al., 1991). Son-sire regressions showed high heritability (h2)values for several morphological characters of sperm, for example, 0.72 for the length of the spermatozoan head in rabbits (Napier, 1961) and 0.97 for the midpiece length in mice (Woolley and Beatty, 1967).Artificial selection experiments on length of the mouse midpiece also demonstrated that changes in the length were caused by changes in the number of mitochondria1gyres surrounding the sperm axis (Wooley, .1970), suggesting that differences in size are possibly associated with differences in metabolism. More detailed information on genetic variability in male fertilization efficiency has been recently provided by Clark, AguadC, Prout, Harshman, and Langley (1995). Sperm displacement abilities of males were compared among 152 lines of Drosophila melanogaster that had been made homozygous for the second and/or third chromosomes. Highly significant differences among lines were found in both males’ abilities to displace

298

LAURENT KELLER AND HUDSON K. REEVE

sperm that females had stored during previous matings and males’ abilities to prevent any other male from removing the sperm they transferred to females. Clark et al. also determined the genotypes at seven accessory gland protein (Acp) genes of males in lines that exhibited naturally occumng variation in sperm displacement. Acp genes encode proteins that are transmitted to the females in the seminal fluid, and thus probably influence sperm displacement. Particular alleles at four of the seven Acp loci were significantly associated with males’ ability to resist sperm displacement by sperm from males that subsequently mated with the same females. (No clear association was found between the genotype at Acp and a male’s ability to displace sperm that females stored during previous matings.) Alleles of the four genes are not in linkage disequilibrium, indicating that each of the four Acp genes (or linked genes) influences the ability of males to resist sperm displacement. It is unlikely that each of the four Acp genes is linked to one or several genes influencing males’ ability to resist sperm displacement. This, plus the fact that Acp genes are known to encode proteins that are transmitted to the females during mating, suggest that Acp genes directly influence resistance to sperm displacement. The most likely effect of Acp gene products is probably linked to the role of Acp protein on sperm transport (Clark et al., 1995). Theoretically, heritable variation in sperm phenotype and fertilization efficiency may be maintained by mutation-selection balance at polygenic loci (Lande, 1980), or perhaps by oscillating evolution of multiple sperm types that form a nontransitive “dominance” hierarchy, as discussed above. Intraspecific variation in sperm phenotypes has indeed been found in several species (Fig. 2; see Sivinski, 1984). 111. PREDICTIONS A N D EVIDENCE If SSS is a selective force promoting multiple mating, then one would predict that females should provide opportunities for increased sperm competition to occur (Bellis and Baker, 1990). This is because their sons’ fertilization efficiency will be on average higher the more intense the competition among sperm from different males. There are several means by which females may increase the probability that their eggs will be fertilized by a male with high fertilization efficiency, including the following. 1 . Providing a Challenging Genital Environment

If sperm of different males differ in their success in fertilizing an egg, females may increase this variability by producing a challenging genital

WHY DO FEMALES MATE MULTIPLY?

299

FIG. 2. Three examples of strong within-species sperm polymorphism, with castes of specialized gametes. (A) The apyrene sperm ( 1 ) of the mollusk, Clathrus cluthrus, and its fertile eupyrene passengers (2). (B) Another atypical molluscan (Cinctesculu euscultpa) sperm ( I ) with typical sperm attached (2). (C) Spermatostyle (1) and associated spermatozoa (2) of a gyrinid beetle, Dineutus sp. Modified from Sivinski (1984).

environment in which only the most “vigorous” sperm could reach and fertilize an egg. Birkhead, Mgller, and Sutherland (1993a)indeed suggested that females of mammals and birds have evolved particularly hostile reproductive tracts as a means of selecting “good” sperm. For example, if differences among sperm result in differences in their motility, then females may increase the length of the tract that sperm have to migrate to reach the egg. The incredibly complex female genitalia exhibited by some species are consistent with this argument. Eberhard (1985) states that “in

300

LAURENT KELLER AND HUDSON

K. REEVE

some groups, . . . it is clear that the lengths of “overlong” female ducts do not correlate with the lengths of male intromittent organs.’’ For example, the male hamster deposits sperm in the uterus, but the oviduct is extremely long and convoluted (Yanagimachi and Chang, 1963). Tortuous reproductive tracts have also been reported in several other taxa including insects and spiders (Fig. 3; see Davey, 1965; Eberhard, 1985). Furthermore, coevolution between sperm size and shape and the morphology of the female genital structure has been reported in beetles of the genus Bambara, and evidence suggests this may be true in mammals and birds (Dybas and Dybas, 1981; Eberhard, 1985; Briskie and Montgomerie, 1992; Gomendio and Roldan, 1993; Dixson, 1993). That females provide a challenging environment to favor competition between sperm from different males is supported by the sexual behavior and genital morphology of chimpanzees (Pan troglodytes). Female chimpanzees typically mate repeatedly with several males. For example, Goodall(l986) observed a female mating 84 times with 7 males in 8 days. As expected from the intense sperm competition, males have large testes, relative to body weight (Short, 1979). Adult male chimpanzees also possess a remarkably enlongated and filiform penis that can measure up to 18 cm when erect (14.4 f 2.0 cm; E ? SD;N = 11) (Dixson and Mundy, 1994). Careful measurements of 11 males and 19 females (Dixson and Mundy,

FIG. 3. Examples of tortuous reproductive tracts in two species of beetles of the genus Meropathus. The male’s aedeagus has a long filiform extension that may penetrate at least partway up the ducts. Modified from Ordnish (1971). Reprinted with permission of Bishop

Museum Press, Bishop Museum, Honolulu, Hawaii.

30 1

WHY DO FEMALES MATE MULTIPLY?

1994) showed that in 10 out of the 1 1 males penile length exceeded vaginal depth in 14 out of 19 females (Fig. 4; data for females when sexual skin is not swollen). Maximum intromission during mating with females might provide a reproductive advantage for males, since this would allow males to deposit spermatozoa close to, or actually inside, the female’s 0s cervix (Smith, 1984b). There is, however, a further twist to the story. Dixson and Mundy (1994) showed that the length of the female vagina varies according to the menstrual cycle, reaching the maximum size at the time of ovulation (due to swelling of the pink sexual skin). The increase in

20

-

18 -

1

.

-

i

16 -

Y

2 -

r:

s

14

-

I

.

.

.

t ;

. .

n

12

-

10 -

#2

FIG. 4. Measurement of erect penis and vaginal length in adult chimpanzees. Data are individual values with the mean and standard deviation indicated by the histogram and bar. Modified from Dixson and Mundy (1994).

302

LAURENT KELLER AND HUDSON K. REEVE

vaginal length can be up to 50% in some females. As a result of this increase of vaginal length, only few males are capable of intromission to the 0s cervix when most copulations occur: 7 out of the 11 males would be unable to reach the 0s cervix of any of the females and the other four males would be able to reach the 0s cervix of only some of them (Fig. 4). Thus, increased vaginal length at the time of ovulation is consistent with females increasing competition among sperm from different males by making it more difficult for sperm to reach the 0s cervix. Increasing length of the reproductive tract is only one of the means by which more “vigorous” sperm might be selected by females. Other mechanisms that function as barriers to sperm in mammals and birds are reviewed in detail by Birkhead et al. (1993a). (Note that the primary function of some of these barriers is certainly to prevent bacteria from infecting the uterus, but they can secondarily evolve so as to select for more vigorous sperm.) In mammals, for example, barriers to sperm are diverse at the location of the cervix. First, there are both physical and chemical barriers. Second, phagocytosis is a threat to sperm, with the cervix being populated by more than 50 million leucocytes in some species. Finally, there can be important immune responses against sperm, with high concentrations of antisperm antibodies occurring in the cervical mucus of a number of species, including humans. That all these barriers are indeed effective against sperm is demonstrated by the fact that only a tiny proportion of the large number of sperm inseminated by males ever reach the vicinity of the ova (see Birkhead et al., 1993a, 1993b). Finally, females can also increase competition among sperm by producing ova that are difficult to penetrate (selection then occurs among the sperm that have been successul in the first round of selection to reach the ovum). There are many morphological and physiological modifications that may increase the difficulty for sperm to penetrate the ovum. Therefore, if the ovum’s morphology and physiology have evolved so as to make penetration by sperm difficult, one would predict great interspecific variability in the mechanisms impeding sperm penetration (as in the case for secondary sexual characters in general). Evidence indeed suggests this might be true. In many species, sperm have a special structure, the acrosome, that is responsible for penetration into the ovum, and there is high interspecific variability in the role of the acrosome in permitting sperm penetration, sperm binding, and gamete fusion (Tilney, 1985). Furthermore, it has been repeatedly noted that sperm exhibit tremendous interspecific variation in morphology, particularly in the acrosomal part (Fig. 5 ; Fawcett and Phillips, 1970; Baccetti and Afzelius, 1976). This variability is so great that most species can be identified by the morphology of their sperm alone. This has led many authors to question the causes of this high interspecificvariability (e.g., Fawcett and Phillips, 1970;Afzel-

A

303

WHY DO FEMALES MATE MULTIPLY?

I1

‘

I1

Macaaue

Chinchilla

Guinea pig

Ground squirrel

Mouse

FIG. 5. Drawing showing the great variation in size and shape of the mammalian acrosoma1 sperm cap. All are midsagittal sections except the mouse sperm, which is sectioned in the transverse axis of the head. From Fawcett and Phillips (1970).

ius, 1970). For example, Fawcett and Phillips (1970) ask whether “the specific shapes of the acrosome have any significance or are they meaningless accidents of evolution?” Similarly, speaking of the tremendous interspecific diversity of sperm acrosomes, Tilney (1985) wonders “why evolution has varied this cell type so extensively.” This variability could be explained if sperm of different males differ in their success in penetrating an egg, and if females increase this variability by producing ova so difficult to penetrate that only the most “vigorous” sperm can fertilize the eggs. Since there are probably many ways in which females can increase the difficulty of sperm penetration, it is not surprising that species may rapidly diverge in sperm characteristics. 2 . Controlling Sperm Displacement Two selective forces may work in species in which males are able to displace sperm. First, the probability that a male with relatively “vigorous” sperm will fertilize eggs is higher when females acquire similar amounts of sperm than when they acquire uneven amounts of sperm from different males. Second, females should also want their eggs to be fertilized by males that are efficient in displacing the sperm of other males. The outcome is that females should let the second male try to displace sperm, but make it so difficult that only “good” males can achieve this feat.

304

LAURENT KELLER AND HUDSON K. REEVE

Since there will be strong selection on males to develop more efficient appendages to remove sperm, this arms race is likely to result in rapid evolution in the genitalia of both sexes with females storing sperm in a way that makes it more diMicult to be displaced by males. Consistent with this prediction is the incredible diversity of male genitalia (Fig. 6; R. Edwards, 1993) and the observation that sperm storage organs have gradually moved higher up in the female reproductive tracts during evolution in several taxa (Eberhard, 1985, p. 106). However, as mentioned by Eberhard, this shift of storage organs may also be an adaptation to prevent

FIG. 6. A sample of mammalian penes (all flaccid, drawn to different scales), showing how forms are varied and complex. Those in the top three rows are all from primates. Modified from Eberhard (1985) and Dixson (1987). Reprinted with permission of Harvard University Press and Oxford University Press.

WHY DO FEMALES MATE MULTIPLY?

305

immediate fertilization of the eggs after each mating, which of course is also consistent with females increasing competition among ejaculates of males. 3. Resisting Males’ Attempts to Prevent Them from Remating Arguments similar to those developed above can apply to species in which males deposit a “plug” of one form or another to prevent females from remating. Females should prefer to mate with males depositing “efficient copulatory plugs.” To this end females should create conditions in which only efficient plugs prevent remating. One would thus predict that in many species mating plugs do not totally prevent subsequent matings. There is little information on the efficiency of plugs, but the evidence suggests that, as predicted, in most species copulatory plugs do not completely preclude subsequent copulations (e.g., Fenton, 1984;see Eberhard, 1985, pp. 122-125; Barker, 1994). The evolution of any of the above-mentioned mechanisms increasing sperm competition will depend on the nature of the causes underlying variation in male fertilization efficiency. For example, the evolution of complex reproductive tracts may evolve only if the sperm of different males differ in their ability to travel in the reproductive tract. Then SSS may favor females having reproductive tracts with shape, length, or pH that maximizes sperm competition among males. There will, however, be an upper limit on how hostile the reproductive tract is, this limit depending on the benefits of increasing the probability that the eggs are fertilized by a male with high fertilization efficiency balanced by the cost of incomplete fertilization (see Birkhead et al., 1993a).

IV. CONCLUSIONS By mating multiply, females may increase the probability that their sons will have high fertilization efficiency (sperm sexual selection). Thus, multiple mating by females (in the context of SSS) is analogous to female choice in classical sexual selection, since it provides a mechanism for choosing males with greater fertilization efficiency. Note that if genetically superior males have a greater fertilization efficiency (e.g., by producing more vigorous or greater numbers of sperm) multiple mating could also serve as a mechanism of female choice for good-quality males. The relationship found by Madsen et al. (1992) between the number of matings of female adders and the mean viability of their offspring is consistent with a correlation between male quality and fertilization efficiency. However, as mentioned by Parker (1992), several alternative explanations may

306

LAURENT KELLER AND HUDSON K. REEVE

also account for the relationship found by Madsen et al. (1992; see also Capula and Luiselli, 1994). Further experiments are needed to test the interesting hypothesis that genetically superior males may have greater fertilizationefficiency, with multiple mating by females serving as a mechanism for obtaining “good genes.” For SSS to operate it is necessary to have additive genetic variance in males’ fertilization efficiency. If this variation disappears, there is no longer any benefit for females to mate multiply, as has been shown by Curtsinger (1991). The problem is identical to classical sexual selection with regard to genetic variability in male sexual traits. If there is no additive variance for male sexual traits or fitness, there is no advantage for females to choose among males. The question of why genetic variability should be maintained for traits under strong directional selection has been an often repeated concern about sexual selection models. However, recent empirical studies have shown that this variability exists (Hedrick, 1988; Mgller, 1991; Houde, 1992; Bakker, 1993; Wilkinson and Reillo, 1994) and it has been suggested that it might be maintained by biased mutation (Pomiankowski et al., 1991; see also Iwasa and Pomiankowski, 1994). Similar mechanisms to those maintaining genetic diversity in male sexual traits are likely also to maintain genetic variation in a male’s fertilization efficiency. In addition, nontransitivity of sperm-type dominance might be an additional mechanism maintaining variability in prevailing sperm fertilization efficiency. Overall, the evidence indeed suggests that genetic variability exists for male fertilization efficiency. The available data are also consistent with females providing opportunities for increased competition among sperm of different males, as would be expected if SSS is a selective force promoting and maintaining multiple mating by females. Critical support for the SSS hypothesis would be evidence that tortuous reproductive tracts, eggs difficult to penetrate, and other means by which females increase sperm competition are more common in multiply mating than in singly mating species. Unfortunately, the lack of data on the reproductive systems of species exhibiting variable numbers of matings does not currently permit such a comparison. Another critical prediction is that females mating multiply should on average produce sons with higher fertilization efficiency than those mating singly. This could be tested by controlling the number of matings by females in a polyandrous species and comparing the fertilization abilities of their sons. Finally, a general assumption of the SSS hypothesis is that there is (or has been) a positive genetic correlation in nature between the tendency of females to mate multiply and the fertilization efficiency of their sons or brothers (the equivalent of the genetic correlation between male ornaments and female prefer-

WHY DO FEMALES MATE MULTIPLY?

307

ences for these ornaments in classical “Fisherian” models of sexual selection). Remating rate has been shown to be heritable in some species such as Drosophila melanogaster (Gromko and Newport, 1988). Whether there is a genetic correlation between remating rates of females and fertilization efficiency of their sons or brothers could be tested by using standard breeding experiments, as has recently been done for female choice in sticklebacks (Bakker, 1993). Such experiments might not only shed light on the factors promoting multiple mating in animals, but they may also provide an alternative way to test models of sexual selection in general.

V. SUMMARY

One of the more debated issues in evolutionary biology is why in many species females mate with multiple males; several hypotheses have been put forward, yet the benefits of multiple mating (here defined as mating with several males) remain unclear in many cases. We develop the sperm sexual selection (SSS) hypothesis, to account for the widespread occurrence of multiple mating in females. We argue that multiple mating by females may rapidly spread when initially a small fraction of the females mate multiply, and if there is a heritable difference among males in one or several of the four following characteristics: ( 1 ) the quantity of sperm they produce; (2) the success of their sperm in reaching and fertilizing an egg; (3) their ability to displace the sperm that females stored during previous matings; and (4)their ability to prevent any other male from subsequently introducing sperm (e.g., differential efficiency of mating plugs). Genetic variability in the above characteristics might be maintained by ( 1 ) mutation-selection balance or (2) nontransitivity of sperm-type dominance. Indeed, there is evidence that such genetic variability exists. Females that mate multiply increase the probability that their eggs are fertilized by a male with high fertilization efficiency, and thus genes predisposing multiple mating can rapidly spread because the sons of such females will tend to carry both these genes and also genes specifying traits for relatively high fertilization efficiency. Moreover, females should provide opportunities for increased competition among sperm of different males, such as by increasing the length or complexity of their reproductive tracts. Multiple mating (together with female reproductive morphology) and male fertilization efficiency thus may coevolve via Fisherian sexual selection, even though mating may be random with respect to observable male phenotypes.

308

LAURENT KELLER AND HUDSON K. REEVE

APPENDIX:OSCILLATING EVOLUTION OF MULTIPLE SPERMTYPES

The operation of SSS requires that sperm competitiveness fails to reach a selective equilibrium. One way this can happen is if there is oscillating evolution of alternative sperm types, that is, competing sperm types exist, and no single sperm type or mixture of sperm types is evolutionarily stable against invasion by all other types. This scenario requires that sperm polymorphisms exist, which is certainly the case for at least some species (R. G. Edwards, 1955; Napier, 1961; Lee and Wilkes, 1965; Woolley and Beatty, 1967; Krzanowska et al., 1991; see also Dajoz, Till-Bottraud, and Gouyon, 1991, and TillBottraud, Venable, Dajoz, and Gouyon, 1994, for a discussion of how intraspecific variation in pollen morphology can be maintained), but it is important to note that the existence of a sperm polymorphism by itself is not sufficient for demonstrating the required oscillating evolution. For example, Curtsinger (1991) showed that sperm polymorphisms maintained by heterozygote advantage fail to provide the impetus for the evolution of multiple mating. In addition, interindividual sperm polymorphisms representing mixed evolutionarily stable strategies (ESSs) due to frequency-dependent selection (e.g., Maynard Smith, 1982) will not promote multiple mating, because, at the mixed ESS, the expected fertilization success of alternative sperm types will be equal. For example, suppose that, in a population exhibiting a mixed ESS for sperm types, a mutant female arises that mates more times than does the average female in the population. She will either have sons that produce alternative sperm types in the same ratio as do other males in the population, in which case the mutant sons do not reproduce more than wild-type males (and thus increased mating frequency is not favored), or she will have sons that produce sperm types in ratios different from that of the population. In the latter case, the expected reproductive success of the sons must be equal or lower than that of the wild-type males, since wild-type males exhibit (on average) sperm-type ratios that are uninvadable, by assumption; again, the mutant gene for enhanced mating frequency does not profit. How, then, might the necessary oscillating evolution of sperm types occur? Maynard Smith (1982) has pointed out that evolutionary games with three or more pure strategies may have no ESS and display only cyclical dynamics, like the “rock-scissors-paper’’ game. We thus consider a simple evolutionary game involving three sperm types: “Fast,” “Penetrator,” and “Average” sperm. We assume that there is a tradeoff between swimming speed and ability to penetrate an egg once an egg

309

WHY DO FEMALES MATE MULTIPLY?

is encountered. Fast (F) sperm can outswim both Penetrator (P) sperm (which have the lowest swimming ability) and Average (A) sperm (which have intermediate swimming ability), but have the least ability to penetrate eggs. Penetrator sperm have a higher probability of penetrating eggs than do either Average sperm (which are intermediate in this ability) or Fast sperm. We also make the additional assumption that, when only Fast sperm are in a female reproductive tract, they have a slightly higher probability of fertilizing all of the eggs than when sperm compete in other combinations (e.g., larger sperm of the P and A types interfere with each other to some small degree, resulting in a small risk that not all eggs will be fertilized; let the mean fraction of unfertilized eggs equal Z -e S O when relatively large sperm are present). Let the probability that a sperm will penetrate the egg in one time unit (while at the egg) be constant: pf for fast sperm, pa for average sperm, and pp for penetrator sperm ( p , > pa > p f ) . F sperm arrive at the egg t , time units before A sperm, and t , units before P sperm (t, > t l ) . A sperm arrive at the egg t2 units before P sperm. Because of mutual travel interference between A and P sperm, we assume that t2 < t3 - t l . Payoffs are mean probabilities that an egg will be fertilized. In this simplistic evolutionary game, A sperm invade a population of F sperm if Pa

P-Pffl-

+ Pf

Pa

X > 0.50 + 2’

P sperm invade A sperm if e-Pat2

pD Pp + Pa

> 0.50,

and F sperm invade P sperm if

It is possible to choose values for the parameters that lead to cycling evolution. For example, choosing tl = 1, t2 = 0.05, t, = 2 , pp= 2 , pa = 1, and pf = 0.30. The resulting, approximate payoff matrix follows. Fast Fast Average Penetrator

0.50 + 5 2 0.57 0.48

Average

Penetrator

0.43

0.52

0.50

0.37 0.50

0.63

310

LAURENT KELLER AND HUDSON K. REEVE

The above game has no pure ESS. The only condition for a mixed ESS is to produce fast sperm with probability 13/22, average sperm with probability 2/22, and penetrator sperm with probability 7/22. However, this mixed strategy will have a payoff less than 0.50 + ?/2 against itself, and thus will be vulnerable to invasion by the pure fast-sperm strategy. Even if there is no incomplete fertilization when large sperm are involved (i.e., X = O), the mixed ESS can be readily invaded by any of the other pure strategies through genetic drift, setting the stage for a new round of selective invasion. Under the latter conditions, oscillating evolution will result, creating the possibility that multiple mating will continually evolve via the association of alleles for multiple mating with the sperm-type alleles currently undergoing positive selection. Acknowledgments We thank Steven Austad. The0 Bakker, Giorgina Bernasconi, Tim Birkhead, Belinda Chang, William Eberhard. David Haig. Manfred Milinski. Geoff Parker, Nicolas Perrin, Naomi Pierce, Andrew Pomiankowski, Peter Slater, Peter Vogel, Paul Ward, Claus Wedekind, and E. 0. Wilson for comments on the manuscript and support. This work was funded by the Swiss National Science Foundation (Grants No. 823A-0283650.3 1-35584.92. and 3 136907.93). the Janggen-Pohn Stiftung (L. Keller). and a Junior Fellowship from the Harvard University Society of Fellows (H.K. Reeve).

References Afzelius, B. A. (1970).Thoughts on comparative spermatology. In “Comparative Spermatology” (B. Baccetti, ed.), pp. 565-573. Academic Press, New York. Arnold, S. J., and Halliday, T. R. (1988). Multiple mating: Natural selection is not evolution. Anim. Behau. 36, 1547-1548. Arnqvist, G. (1989). Multiple mating in a water strider: Mutual benefits or intersexual conflict? Anim. Behau. 38, 749-754. Austad, S. N. (1984). Evolution of sperm priority patterns in spiders. In “Sperm Competition and the Evolution of Animal Mating Systems” (R.L. Smith, eds.), pp. 223-249. Academic Press, London. Baccetti, B. and Afzelius, B. A. (1976). “The Biology of the Sperm Cell.” Karger, Basel. Baker, R. R., and Bellis, M. A. (1989). Number of sperm in human ejaculates varies in accordance with sperm competition theory. Anim. Behau. 37, 867-869. Baker, R. R., and Bellis, M. A. (1993). Human sperm competition: Ejaculate manipulation by females and a function for the female orgasm. Anim. Behav. 46, 887-909. Bakker, T. C . M. (1993). Positive genetic correlation between female preference and preferred male ornament in sticklebacks. Nature (London) 363, 255-257. Barker, D. M. (1994).Copulatory plugs and parternity assurance in the nematode Caenorhabditis elegans. Anim. Behav. 48, 147-156. Beatty, R. A. (1960). Fertility and mixed semen from different rabbits. J . Reprod. Fertil. 1, 52-60.

WHY DO FEMALES MATE MULTIPLY?

311

Beatty, R. A. (1975). Genetics of animal spermatozoa. I n “Gamete Competition in Plants and Animals (D. L. Mulcahy, ed.). pp. 61-68. North-Holland Publishing Company, Amsterdam, The Netherlands. Beatty, R. A., Bennett, G. H., Hall, J. G., Hancock, J. L., and Stewart, D. L. (1969). An experiment with heterospermic insemination in cattle. J. Reprod. Fertil. 19, 491-501. Bellis, M. A., and Baker, R. R. (1990). Do females promote sperm competition? Data for humans. Anim. Behav. 40,997-998. Bellis, M. A., Baker, R. R., andcage, M. J. G. (1990a). Variationinratejaculatesconsistent with the kamikaze-sperm hypothesis. J. Mammal. 71,479-480. Bellis, M. A., Baker, R. R., Matson, P., and Chew, J. (1990b). A guide to upwardly mobile spermatozoa. Andrologia 22, 397-399. Birkhead, T. R.. and Hunter, F. M. (1990). Mechanism of sperm competition. Trends Ecol. E d . 5, 48-52. Birkhead, T. R., and Mgiller, A. P. (1992). “Sperm Competition in Birds. Evolutionary Causes and Consequences.” Academic Press, London. Birkhead, T. R., and Moiler, A. P. (1993). Female control of paternity. Trends Ecol. Evol. 8, 100-104. Birkhead, T. R., Mgiller, A. P., and Sutherland, W. J. (1993a). Why do females make it so difficult for males to fertilize their eggs? J . Theoret. Biol. 161, 51-60. Birkhead, T. R., Pellatt, E. J., and Fletcher, F. (1993b). Selection and utilization of spermatozoa in the reproductive tract of the female zebra finch Taeniopygia guttata. J. Repr. Fertil. 99, 593-600. Briskie, J. V., and Montgomerie, R. (1992). Sperm size and sperm competition in birds. Proc. R . SOC.Lond. B 247, 89-95. Capula, M., and Luiselli, L. (1994). Can female adders multiply? Nature (London)369,528. Chapman, T., Hutchings, J., and Partridge, L. (1993). No reduction in the cost of mating for Drosophila melanogasrer females mating with spermless males. Proc. R . SOC.Lond. B 253, 211-217. Chapman, T., Liddle, L., Kalb, J., Wolfner, M., and Partridge, L. (1995). Male accessory gland products cause the cost of mating in Drosophila melanogaster females. Nature (London) 373,41-244. Chapman, T., Trevitt, S. , and Partridge, L. (1994). Remating and male-derived nutrients in Drosophila melanogaster. J. Evol. Biol. 7, 51-69. Cheng, K. M., and Siegel, P. B. (1990). Quantitative genetics and multiple mating. Anim. Behau. 40,406-407. Childress, D., and Hartl, D. L. (1972). Sperm preference in Drosophila melanogaster. Genetics 71,417-427. Clark, A. G., Aguadt, M., Prout, T., Harshman, L. G., and Langley, C. H. (1995). Variation in sperm displacement and its association with accessory gland protein in Drosophila melanogaster. Genetics 139, 189-207. Cordero, A., and Miller, P. L. (1992). Sperm transfer, displacement and precedence in Ischura graellsii (Odonata: Coenogrionidae). Behav. Ecol. Sociobiol. 30, 261-267. Cox, C. R., and Le Boeuf, B. J. (1977). Female incitationofmalecompetition: A mechanism in sexual selection. Am. Nat. 111, 317-335. Crozier, R. H., and Page, R. E. (1985). On being the right size: Male contributions and multiple mating in social Hymenoptera. Behau. Ecol. Sociobiol. 18, 105-1 15. Curtsinger, J. W. (1991). Sperm competition and the evolution of multiple mating. Am. Nut. 138, 93-102. Dajoz, I., Till-Bottraud, I., and Gouyon, P. -H. (1991). Evolution of pollen morphology. Science 253, 66-68.

312

LAURENT KELLER AND HUDSON K. REEVE

Daly, M. (1978). The cost of mating. A m . Nat. 112, 771-774. Davey, K. G. (1%5). “Reproduction in Insects.” Oliver and Boyd, London. Dewsbury, D. A. (1975). A diallel cross analysis of genetic determinants of copulatory behavior in rats. J . Comp. Psychol. 88, 713-722. Dixson, A. F. (1987). Observations on the evolution of the genitalia and copulatory behaviour in male primates. J . Zool. (Lond) 213, 423-443. Dixson, A. F. (1993). Sexual selection, sperm competition and the evolution of sperm length. Folia Primatol. 61, 221-227. Dixson, A. F., and Mundy, N. I. (1994). Sexual behaviour, sexual swelling, and penile evolution in chimpanzees (Pan rroglodytes). Arch. Sexual Behav. 23, 267-280. Dybas, L. K., and Dybas, H. S. (1981). Coadaptation and taxonomic differentiation of sperm and spermatheca in feathewing beetles. Evolution 35, 168-174. Eberhard, W. G. (1985). “Sexual Selection and Animal Genitalia.” Harvard University Press, Cambridge. Eberhard, W. G. (1990). Inadvertent machismo. Trends Ecol. Euol. 5 , 263. Eberhard, W. G. (1991). Copulatory courtship and cryptic female choice in insects. Eiol. Rev. 66, 1-31. Eberhard, W. G. (1993). Evaluating models of sexual selection: Genitalia as a test case. Am. N a t . 142, 564-571. Edwards, R. (1993). Entomologicaland mammalogical perspectives on genital differentiation. Trends Ecol. Evol. 8,406-409. Edwards, R. G. (1955). Selective fertilization following the use of sperm mixtures in the mouse. Nature (London) 175, 215-223. Fawcett, D. W., and Phillips, D. M. (1970). Recent observations on the ultrastructure and development of the mammalian spermatozoon. I n “Comparative Spermatology” (B. Baccetti. ed.), pp. 13-28. Academic Press, New York. Fenton, M. B. (1984). Sperm competition? The case of Vespertilinid and Rhinolophid bats. I n “Sperm Competition and the Evolution of Animal Mating Systems” (R. L. Smith, ed.), pp. 573-587. Academic Press, London. Fowler, K., and Partridge, L. (1989). A cost of mating in female fruitflies. Nature (London) 338,760-761.

Ginsberg, J. R., and Huck, U. W. (1989). Sperm competition in mammals. Trends Ecol. Euol. 4, 74-79. Gomendio, M., and Roldan, E. R. S. (1991). Sperm competition influences sperm size in mammals. Proc. R. SOC. Lond. B 243, 181-185. Gomendio, M., and Roldan, E. R. S. (1993). Coevolution between male ejaculates and female reproductive biology in eutherian mammals. Proc. R. SOC.Lond. B 252, 7-12.

Goodall, J. (1986). “The chimpanzees of Gombe: Patterns of Behavior.” Belknap Press, Cambridge, Massachusetts. Gromko, M.H., Gilbert, D. G., and Richmond, R. C. (1984). Sperm transfer and use in the multiple mating system of Drosophila. In “Sperm Competition and the Evolution of Animal Mating Systems” (R. L. Smith, ed.), pp. 371-426. Academic Press, London. Gromko, M.H., and Newport, M. E. (1988). Genetic basis for remating in Drosophila melanogaster. 11. Response to selection based on the behaviour of one sex. Behau. Genet. 18,621-632. Halliday. T . R., and Arnold, S. J. (1987). Multiple mating by females: A perspective from quantitative genetics. Anim. Behau. 35, 939-941. Hamilton, W. D. (1987). Kinship, recognition, disease, and intelligence: Constraints of social evolution. In “Animal Societies: Theories and Facts” (Y.It8, J. L. Brown, and J. Kikkawa, eds.), pp. 81-102. Japan Sci. SOC.Press, Tokyo.

WHY DO FEMALES MATE MULTIPLY?

313

Harvey, P. H., and Bennett, P. M. (1985). Sexual dimorphism and reproductive strategies. I n “Human Sexual Dimorphism” ( J . Ghesquire, R. D. Martin, and F. Newcombe, eds.), pp. 43-59. Taylor and Francis, London. Hedrick, A. V. (1988). Female choice and the heritability of attractive male traits: An empirical study. Am. Nut. 132, 267-276. Helversen, D., and von Helversen, 0. (1991). Pre-mating sperm removal in the bushcricket Metaplasfes ornatus Ramme (I93 1) (Orthoptera, Tettigonoidea, Phaneropteridae). Behau. Ecol. Sociobiol. 28, 391-396. Holmes, H. B. (1974). Patterns of sperm competition in Nusoniu uifripennis.Can. J . Genet. Cytol. 16, 789-795. Houde, A. E. (1992). Sex-linked heritability of a sexually selected character in a natural population of Poecilia reticulafa (Pisces: Poeciliidae) (guppies). Heredity 69,229-235. Iwasa, Y.,and Pomiankowski, A. 1994. The evolution of costly mate preferences for multiple sexual ornaments. Evolution 48, 853-867. Keller, L. (1994). Rewards of promiscuity. Nature 372, 229-230. Keller, L. (1995). All’s fair when love is war. Nature (London) 373, 190-191. Keller, L., and Passera, L. (1992). Mating sytem, optimal number of matings, and sperm transfer in the Argentine ant Iridomyrmex humilis. Behau. Ecol. Sociobiol. 31,359-366. Keller, L., and Reeve, H. K. (1994). Genetic variability, queen number and polyandry in social insects. Evolution 38, 694-704. Kempenaers, B., Verheyen, G. J., Van den Broeck, M., Burke, T., Van Broekhoven, C., and Dhondt, A. A. (1992). Extra-pair paternity results from female preference for highquality males in the blue tit. Nature (London) 357,494-4%. Kirby, J. D., and Froman, D. P. (1991). Comparative metabolism of spermatozoa from subfertile Delaware and Wyandotte roosters. J. Repr. Fert. 91, 125-130. Kirkpatrick, M. (1982). Sexual selection and the evolution of female choice. Evolution 36, 1-12. Krzanowska, H., Wabikisliz, B., and Rafinski, J. (1991). Phenotypic and fertilizing capacity of spermatozoa of chimaeric mice produced from two strains that differ in sperm quality. J. Repr. Fert. 91, 667-676. Lande, R. (1980). Sexual dimorphism, sexual selection and adaptation in polygenic characters. Euolution 34, 292-305. Lee, P. L., and Wilkes, A. (1965). Polymorphic spermatozoa in the Hymenopterous wasp Dahlbonius. Science 147, 1445-1446. Levine, L. (1976). Sexual selection in mice. IV. Experimental demonstration of selective fertilization. Am. Nut. 101, 289-294. Lewis, S. M., and Austad, S. N. (1990). Sourceofintraspecificvariationinspermprecedence in red flour beetles. Am. Nut. l35, 351-359. Lewis, S. M., and Austad, S. N. (1994). Sexual selection in flour beetles: The relationship between sperm precedence and male olfactory attractiveness. Behau. Ecol. 5,219-224. Madsen, T., Shine, R., Loman, J., and Hakansson, T. (1992). Why do female adders copulate so frequently? Nature (London)355, 440-441. Martin, P. A., and Dziuk, P. J. (1977). Assessment of relative fertility of males (cockerels and boors) by competitive mating. J. Reprod. Ferril. 49, 323-329. Maynard Smith, J. (1982). “Evolution and the Theory of Games.” Cambridge University Press, Cambridge. Maynard Smith, J. (1991). Theories of sexual selection. Trends Ecol. Euol. 6, 146-151. M@ller,A. P. (1991). Sexual selection in the monogamous barn swallow (Hirundo rustica). I. Determinants of tail ornament size. Evolution 45, 1823-1836. M@ller,A. P. (1992). Frequency of female copulations with multiple males and sexual selection. Am. Nut. 139, 1089-1101.

314

LAURENT KELLER AND HUDSON K. REEVE

Napier, R. A. N. (1961). 111. Estimation of spermatozoan quality by mixed insemination, and the inheritance of spermatozoan characters. J. Reprod. Ferr. 2, 273-289. Neeley, J. D., Johnson, B. H., Dillard, E. U., and Robison, 0. W. (1982). Genetic parameters for testes size and sperm number in hereford bulls. J. Anim. Sci. 55, 1033-1040. O'Donald, P. (1962). The theory of sexual and natural selection. Heredity 22,499-518. Olsson, M., Gullberg, A., Tegelstrom, H., and Shine, R. (1994). Can female adders multiply? Nature 369, 528. Olsson, M., Madsen, T., Shine, R., Gullberg, A., Tegelstrom, H. (1944b). Rewards of promiscuity. Nature 372, 230. Ordnish, R. G. (1971). Entomology of the Aucklands and other islands south of New Zealand: Coleoptera: Hydraenidae. Pac$c Insects Monographs 27, 185-192. Otronen, M., and Siva-Jothy, M. T. (1991). The effect of postcopulatory male behaviour on ejaculate distribution within the female sperm storage organs of the fly, Dryomyza anilis (Diptera: Dryomyzidae). Behav. Ecol. Sociobiol. 29, 33-37. Parker, G. A. (1970). Sperm competition and its evolutionary consequences in the insects. Biol. Rev. 45, 525-567. Parker, G. A. (1979). Sexual selection and sexual conflict. In "Sexual Selection and Reproductive Competition in Insects" (M. S. Blum and N. A. Blum, eds), pp. 123-166. Academic Press, London. Parker, G. A. (1992). Snakes and female sexuality. Nature (London) 355,395-396. Parker, G. A., and Stuart, R. A. (1976). Animal behaviour as a strategy optimizer: Evolution of resource assessment strategies and optimal emigration thresholds. Am. Nat. 110, 1055-1076.

Pitnick, S . , Markow, T. A., and Riedy, M. F. (1991). Transfer of ejaculate and incorporation of male-derived substances by females in the nannoptera species group (Diptera: Drosophilae). Evolution 45, 774-780. Pomiankowski, A,, Iwasa, Y.,and Nee, S. (1991). The evolution ofcostly mate preferences I. Fisher and biased mutation. Evolution 45, 1422-1430. Pomiankowski, A., and MPler, A. P. (in press). A resolution of the lek paradox. Proc. R . SOC. Lond. B. Prasad, M. R. N. (1974). Mannliche Geschlechsorgane. Handbuch der Zoologie, Band 8, Lieferung 51(2) 1-150. Prout, T.,and Bundgaard, J. (1977). The population genetics of sperm displacement. Genetics 85,95-124.

Radwan. J., and Witalinski, W. (1991). Sperm competition. Nature (London)352,671-672. Ridley, M. (1988). Mating frequency and fecundity in insects. Biol. Rev. 63, 509-549. Ridley, M. (1990). The control and frequency of mating in insects. Funct. Ecol. 4, 5-84. Ridley, M. (1993). Clutch size and mating frequency in parasitic Hymenoptera. Am. Nut. 142, 803-910.

Schwagmeyer, P. L., and Parker, G. A. (1990). Male mate choice as predicted by sperm competition in thirteen lined ground squirrels. Nature (London) 348, 62-64. Schwartz, J. M., and Boake, C. R. B. (1992). Sexual dimorphism in remating in Hawaiian Drosophila species. Anim. Behav. 44,231-238. Sheldon, B. C. (1993). Sexually transmitted disease in birds: Occurrence and evolutionary significance. Phil. Trans. R . Soc. Lond. 339, 491-497. Sherman, P. W., Seeley, T. D., and Reeve, H. K. (1988). Parasites, pathogens, and polyandry in social Hymenoptera. Am. Nut. 131, 602-610. Sherman, P. W., and Westneat, D. F. (1988). Multiple mating and quantitative genetics. Anim. Behav. 36, 1545-1547. Short, R. V. (1979). Sexual selection and its component parts, somatic and genital selection. as illustrated by man and the great ape. Adv. Stud. Behav. 9, 331-158.

WHY DO FEMALES MATE MULTIPLY?

315

Shrenker. P., and Maxson. S. C. (1983). The genetics of hormonal influence on male sexual behavior of mice and rats. Neurosci. Behau. Rev. 7 , 349-359. Silberglied, R. E., Shepherd, J. G., and Dickinson, J. L. (1984). Enuchs: The role ofapyrene sperm in Lepidoptera? Am. Nut. l23,255-265. Simmons. L. W., and Parker, G. W. (1992). Individual variation in sperm competition success of yellow dung flies, Scatophaga stercoraria. Evolurion 46, 366-375. Sivinski, J. (1984). Sperm in competition. In “Sperm Competition and the Evolution of Animal Mating Systems” (R. L. Smith, ed.), pp. 86-1 15. Academic Press, London. Smith, R. L. (1979). Repeated copulation and sperm precedence: Paternity assurance for a male brooding water bug. Science 205, 1029-1031. Smith, R. L. (Ed.). (l984a). “Sperm Competition and the Evolution of Animal Mating Systems.” Academic Press, London. Smith, R. L. (1984b). Human sperm competition. I n “Sperm Competition and the Evolution of Animal Mating Systems” (R. L. Smith, ed.), pp. 601-659). Academic Press, London. Thornill, R. (1976). Sexual selection and parental investment in insects. Am. Nar. 110, 152- 163. Thornill, R., and Alcock, J. (1983). “The Evolution of Insect Mating Systems.” Harvard University Press, Cambridge. Till-Bottraud, I., Venable, L. D., Dajoz, I., and Gouyon, P. -H. (1994). Selection on pollen morphology: A game theory model. Am. Nar. 144, 395-411. Tilney, L. W. (1985). The acrosomal reaction. In “Biology of Fertilization” (C. B. Metz and A. Monroy, eds). pp. 157-213). Academic Press, London. Toelle, V. D., Johnson, B. H.. and Robison. 0. W. (1984). Genetic parameters for testes traits in swine. J . Anim. Sci. 59, 967-973. Waage, J. K. (1979). Dual function of the damselfly penis: Sperm removal and transfer. Science 203, 916-918. Walker, W. F. (1980). Sperm utilization strategies in nonsocial insects. Am. Nar. 115, 780-799. Ward, P. I . , and Hauschteck-Jungen, E. (1993). Variation in sperm length in the Yellow Dung Fly Scatophaga stercoraria (L). J . Insect Physiol. 39, 545-547. Westneat, D. F., Sherman, P. W., and Morton, M. L. (1990). The ecology and evolution of extra-pair copulations in birds. Current Ornithol. 7 , 331-369. Westneat, D. F. (1987). Extra-pair fertilization in a predominantly monogamous bird: Observations of behaviour. Anim. Behau. 35, 865-876. Wilkes, A. (1966).Sperm utilization following multiple insemination in the wasp Dahlbominus fuscipennis. Can. J . Genet. Cyrol. 8, 451-461. Wilkinson, G. S., and Reillo, P. R. (1994). Female choice response to artificial selection on an exaggerated male trait in a stalked-eyed fly. Proc. R . SOC.Lond. B. 255, 1-6. Woolley, D. M. (1970). The midpiece of the mouse spermatozoon: its form and development as seen by surface replication. J . Cell Sci. 6, 865-879. Woolley, D. M., and Beatty, R. A. (1967). Inheritanceofmidpiece lengthin mouse spermatozoa. Nature (London) 215,94-95. Yanagimachi, R., and Chang, M. C. (1963). Sperm ascent through the oviduct of the hamster and rabbit in relation to the time of ovulation. J . Repr. Fert. 6,413-420.

This Page Intentionally Left Blank

ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 24

Cognition in Cephalopods JENNIFER A. MATHER DEPARTMENT OF PSYCHOLOGY THE UNIVERSITY OF LETHBRIDGE LETHBRIDGE, ALBERTA, CANADA

I. INTRODUCTION The cephaloped molluscs offer a challenge to the study of animal cognition, if cognition is viewed as “all the processes by which the sensory input is transformed, reduced, elaborated, stored, recovered and used” (Neisser, 1976). The cephalopods are known for their dependence on learning, which is clearly required for cognition, as it is unusual to find the equivalent of a fixed action pattern in any species (see Mather, 1986). Many learning studies done in the 1960s and 1970s focused on learning as a phenomenon and so were not always linked to the natural history of the animals themselves. For instance, a major effort was launched to define the features of a visual stimulus that Octopus vulgaris could discriminate in object recognition (Wells, 1978). The same species forages mainly by chemotactile search in the ocean (Mather, 1991a), leaving researchers to evaluate for what reason visual processing of figures might be used. It may be used in spatial memory, as Octopus uulgaris maintains a central home and forages in different directions out from it (Mather, 1991a). In the laboratory, octopuses can easily learn to go to visual landmarks for food (Mather, 1991a), so perhaps the visual learning ability evolved in this context. Two current influences make the study of cephalopod cognition now a possibility. The first is the expanded scope of animal behavior in general. Recent work, with focuses from animal cognition and behavioral ecology to constraints on learning and animal communication, has given us a much wider view of what animals can do. For example, studies of the limitations and abilities of bees and rodents in spatial memory could be compared with those in cephalopods. A second influence, increased knowledge of the ecology and physiology behind the behavior of cephalopods, has also helped this advance. Researchers have begun to study how octopuses and squid behave in their marine environment and to understand how they 317

Copyright 0 1995 by Academic Press. Inc. AU rights of reproduction in any form reserved.

318

JENNIFER A. MATHER

are adapted to it (Boyle, 1983, 1987). These two areas of information can supply a background for broadening our knowledge base, from knowing about learning to understanding its function for the animal. Robinson (1990) suggests that the coral reef environment, because of its complexity, is one of the two areas in which evolution exerted major pressure for intelligence. How might the soft-bodied marine cephalopod have been adapted to use intelligence because of this environment, and how might the end result differ from that of vertebrates whose intelligence evolved in the context of social groups (Humphrey, 1976)? The study of cephalopod behavior offers constraints as well as rewards. Dawkins (1993) discussed the search for animal consciousness, and concluded that similarity of animals’ and humans’ cognitive ability should lead us to hypothesize that “lower” vertebrates possess such internal processing. Griffin (1993) used similarities in physiological processing in the brain as an argument for consciousness in animals. We cannot use phylogenetic similarity in the search for specializations of information processing in cephalopods, as they are evolutionarily far from us and we cannot expect similarly organized behavior. Because of the difficulty of using phylogenetic similarity as evidence, Yoerg and Kamil (1991) suggested that we should cease studying that which we cannot know, consciousness, and study that which we can, information processing. Using this approach, we can pose answerable questions about octopuses and squid. What do cephalopods use visual object recognition for? What limits their spatial memory or programs their migrations, and do they form cognitive maps? Is their manipulative capacity and tool use limited by their apparent lack of knowledge of arm position? Can they make detours? Do they have a visual language? In the process of answering such questions about the cephalopods, so like us in their dependence on learning, yet so unlike us in their short solitary life without bones and with visual signals flashing on their skins, we may gain the reward of learning more about the evolution of that learning and about the adaptation of behavioral capacity to the lifestyle of the animals that evolved it.’

‘

A reviewer of this chapter, and several participants at a recent workshop on learning and cognition in cephalopods ( J . Mather and D. Mather, unpublished manuscript), objected to the term “cognition,” particularly as applied to cephalopods. Perhaps the title should instead refer to “information processing,” but this term is often used in a more limited way. It suggests a step-by-step analysis of what a passive animal does when presented with information. Piaget (1%2) pointed out in detail that humans are exploratory, and that children in particular seek out information and attempt mastery of their environment. Using the term cognition may therefore help us both to see an animal as similarly actively seeking information about its environment and to identify specializations and limitations in receiving, processing,

COGNITION IN CEPHALOPODS

3 19

11. BACKGROUND ON CEPHALOPODS At first glance, the cephalopods do not look like a promising group for complex information processing. They are molluscs, and the basic molluscan nervous system is a group of four small, paired ganglia. The small size of these ganglia and the fact that they are separated means that much molluscan behavior is stereotyped and some of its control is strictly local. But the coleoid cephalopods (all the modern ones except Nautilus) are a different branch off the molluscan stem. Nautilus, the only remnant genus of a group that flourished in the Paleozoic, retains the ancestral shell and a slow-moving,scavenging way of life (Saunders and Landsman, 1988) and does not appear to have produced much complex behavior. In most coleoid cephalopods, however, a complex nervous system has developed. A brachial nerve cord leads to each of the eight arms, and within each arm is a centralized system of axial nerve cords, ganglia, intramuscular nerve cords, and a ganglion above each sucker (Graziadei, 1971). There is also a central brain, described in detail by Young (1971), containing some areas that appear mainly involved in learning. The brain :body ratio is fairly large, somewhere between that of fish and that of birds and mammals (Packard, 1972). This cephalopod brain retains the regionalization that stems from its ganglionic derivation, with the more dorsal areas concerned with control of input and processing, and the ventral with output; but it is a brain nevertheless. Since the cephalopods are different in many ways from vertebrates, something of their physiology must be described as a basis for discussing their behavior. One striking aspect of coleoid sensory physiology is the visual system. The structure of vertebrate and cephaloped eyes is used as the classic example of convergent evolution, although there are functional differences. The only major difference in structure between the two is that the vertebrate eye has the receptor cells buried behind the layers of bipolar, amacrine, and ganglion cells, whereas the coleoid arrangement has rhabdomes out front in the eyes and neurons for processing the visual information in the optic lobe. Thus, the cephalopods have not needed to evolve a mechanism to filter out stabilized images and generalize over the blind spot as vertebrates must (Schiffman, 1990). Instead, the large optic -

and using such information. Nevertheless, the term cognition is not a simple one, even when used for human information processing, and Beer (1992) points out some of the problems in the history of the use of this term. But such an approach may be necessary in the study of animals quite different from ourselves, and, if it forces us to look more broadly at behavior (as with the similar approach by Dyer, 1994, to spatial cognition in insects), then cognition is the appropriate term.

320

JENNIFER A. MATHER

lobes (75% of the brain) contain all the neurons that process visual information. The structure of this area is known, but no structure-function linkage like that for cats (Hubel and Weisel, 1962) has been proposed for the optic lobe. Cephalopod eyes are placed laterally, with a small binocular overlap and an area of higher cell density similar to but not as concentrated as the mammalian retina. Octopus visual acuity is at least as good as 10 min of arc (Muntz and Gwyther, 1988), and octopuses have dark adaptation, accomplished by the migration of pigment along retinal cells. Cephalopods apparently lack color vision (Messenger, Wilson, and Hedge, 1973) despite their ability to match the background with the color of their skin. Members of a recent workshop (J. Mather and L. Mather, unpublished manuscript) expressed concern over this anomaly and agreed that the presence of color vision must be reevaluated with modern, sophisticated techniques. Cephalopods can also discriminate the plane of polarization of light (Moody and Parriss, 1960), an ability that aids bee navigation but is of unknown use in cephalopods. This formidable list of competencies of the eye has stimulated studies of one area of brain function: investigation of how the brain behind the eyes “transforms, reduces and elaborates’’ (Neisser, 1976) visual information (Sutherland, 1963). The process is still not completely understood. As well as having good vision, octopuses have excellent tactile and chemical sensitivity (Graziadei, 1964). Three morphologically different types of receptors are found within the skin of the flexible arms, particularly in the adhesive suckers on the inner surface. Their physiology is unstudied. Cephalopods also have a balance system built in a threedimensional structure similar to semicircular canals in vertebrates and similarly responsive to angular and linear acceleration (Budelmann, Sachse, and Staudigl, 1987). In addition, squid have a lateral line analogue and can “hear” (Budelmann and Bleckmann, 1988). Study of the linkage between structure and function for all these systems has scarcely begun. In contrast to most animals whose motor output has been described, cephalopods’ skeletal support ranges from an internal shell for flotation (Sepia) to no hard skeleton at all (Octopus). In the course of evolution, they have lost the protective external molluscan shell. The abandonment of built-in protection has meant both the physical vulnerability and the great flexibility of movement that correlate with the evolution of intelligence in cephalopods. The fluid-filled mantle cavity acts as a hydrostatic skeleton, and contraction of the mantle musculature results in jet propulsion by expulsion of water through the flexible funnel (O’Dorand Webber, 1986); dark ink may be mixed with the water jet for concealment. The movement of fluid, used for respiration in the basic molluscan plan, has been adapted and modified (Wells, 1990), and octopuses and squid use

COGNITION IN CEPHALOPODS

32 1

the water jet for a number of purposes. These include jet escape and prey capture, as well as alteration of the landscape for concealment (Von Boletzky and Von Boletzky, 1970; Mather, 1994), removing feces and unwanted food debris from their vicinity, and repulsing scavenging fishes (Mather, 1992). The flexibility of motor output reaches its peak in the movement of body parts by octopuses: their eight long arms, each with radial, circular, and oblique muscles, can theoretically move in almost any possible pattern. Such limitlessness is only theoretically true, and the author is collecting observations of arm postures and movements to look at actual limits on actions. Crawling along the substrate can be in any direction relative to the body, can be by sucker attachment and pulling or pushing by some of the eight arms, or by “tiptoe” movement of suckers only. It can also be assisted to a greater or lesser extent by jet propulsion. The arm movements cannot be described by conventional movement notations such as EscholWachman (S. Pellis, personal communication).How can you describe movement around fixed points when there is no fixed point on the arm? This analytical problem needs to be resolved, since cephalopods explore their environment, find and handle prey, and contact potential mates with these arms. Arms are lined on the inner surface by one or two rows of suction cups, smaller toward the arm tips in an orderly progression; some squid also have hooks. Suckers are normally set on flexible stalks and can either grasp and hold onto a smooth surface or can be folded sideways to form the equivalent of the pincer grasp of apes (Kier and Smith, 1990). This almost limitless ability at manipulation means octopuses can lift many times their own weight (Dilly, Nixon, and Packard, 1964) or can untie knots in the finest surgical silk ( J . Mather, personal observation). Such capacities have hardly been studied. Several octopus species, including 0. joubini, offer an opportunity to study local coordination of the arm. The arms are autotomized readily; arm detachment and the ensuing movement of that arm may produce a “throwaway” to predators like the tail of lizards (Arnold, 1988). A detached arm can be maintained in seawater for hours, performing such coordinated activities as stepping and moving food along the arm from sucker to sucker. Eventually, the animal forms a new arm by regeneration. The arm movements indicated above may not always be controlled by the brain. Cephalopods have a nerve ring circling above the arms, with an axial ganglion at the base of each; there are twice as many neurons in the arms of octopuses (3.5 x lo8) as in the brain. Early research (Ten Cate, 1928) suggested that some control of movement may be reflexive, as much organization is local and response to stimulation spreads to neighboring arms. It is possible that information about arm position is simply

322

JENNIFER A. MATHER

too extensive and complex to be completely processed by the brain, and so much of it is dealt with peripherally (Wells, 1978). Such local control might limit an animal’s ability to learn to manipulate objects. Fiorito, von Planta, and Scotto (1990) found that octopuses would learn to open a glass jar to get at a lobster inside, yet learning did not reduce the duration of exploratory behavior by the arms in later trials. Filming of animals in a similar exploratory situation might reveal the behavioral changes that take place. A step back from the behavior itself is the mode of life of the animals from which it derives, and the cephalopods’ physiology and natural history have only recently been explored. The subclass Coleoidea probably evolved to be quite different from the ancestral mollusc structure in outand-out competition with the bony fishes (Packard, 1972). O’Dor and Webber (1986) argued that in terms of energetics the fishes are winning the competition, as the water-ejection model of locomotion that squid use is not nearly as efficient as the lateral body flexion of fish. The natural history of coleoid cephalopods may be defined in terms of this interaction with fish. Cephalopod skin camouflage may be adapted to color-sensitive vertebrate eyes (Parkard, 1988a) and the irregular movement patterns of the octopuses may be devised to prevent vertebrate predators from learning their location (Mather and O’Dor, 1991). The cephalopod pattern of development may be an alternative to that of fishes as well. Unlike fish and Nautilus,coleoid cephalopods mostly live less than two years. They are semelparous, having a short life and producing many eggs at its end, whereas many fish live a long time and reproduce again and again. Their different lifestyle may maintain populations in the face of pressure from fishes, but the lack of integenerational contact (female octopuses brood their eggs but die when the tiny eggs hatch) means no cross-generational transmission of information. To support their explosive growth, cephalopods have a high conversion efficiency (50%) of food intake to body weight.2 They accumulate no lipids but only body proteins, and convert these to eggs and sperm at the end of their life-span (O’Dor and Wells, Their high metabolic rate means that cephalopods can be difficult to keep, but advances in mariculture have resulted in several species being reared to adulthood (see Hanlon, 1987). Methods for maintaining cephalopods have recently been addressed by Boyle (1991) and aquaria are an untapped source for the study of these animals (Anderson, 1987). Squid must have enough room to swim, which can be provided by a circular tank. One of the enduring problems of research on octopuses is their tendency to escape. Their lack of solid skeleton, abandonment of even a “good” home range after a short stay (Mather and O’Dor, 1991). and muscular strength make locked lids on aquaria with strips of Astroturfm around the edges essential. Some will still escape. Many species (such as 0.joubini and the sepioid squid) are nocturnal and quite small (maturing at 15 g or less); as these become better known they will provide more research possibilities.

COGNITION IN CEPHALOPODS

323

1987). Because the young are both quite small and numerous, nearly all do not live to maturity. Such a “live fast and die young” life-style should result in a high investment of time and effort in foraging: the fact that the high production of young and short life does not do so in the cephalopods suggests that other constraints, possibly predation pressure, may affect the life-style and thus behavior (Mather and O’Dor, 1991). None of the major features of their life-style, such as competition for niches, predation pressure by fish and invertebrates, and short life-span, would suggest a group of animals likely to evolve heavy dependence on learning and cognitive specializations. Yet it may still be the case that such a set of demands could only be answered by behavioral flexibility. Another constraint that might shape cephalopod learning specializations is a social one. Humphrey (1976) suggested that the group-living primates evolved their extensive cognitive capacity to solve social problems, and Galef (1976) describes their social transmission of acquired behavior. However, cephalopods do not live in this context. Squid are social but in the relatively primitive social groups of schools, in which members are equal units (Mather and O’Dor, 1984). The influence of such group living on their behavior has scarcely been studied, except for the study of visual skin signals, documented for Sepia by Hanlon and Messenger (1988) and suggested as a visual language in Seopioteuthis by Moynihan (1985). Octopuses are solitary in their natural environment (Mather, Resler, and Cosgrove, 1985), though they have the capacity to form dominance hierarchies in the laboratory (Mather, 1980). Nevertheless, no cephalopod species appears to have a complex social organization, and their learning capacity is probably used mainly to solve environmental problems such as navigation, home “building,” dealing with prey, and predator avoidance, rather than social ones. Thus, the cephalopods present a series of problems for understanding animal cognition, which are best approached through studying their information processing (Yoerg and Kamil, 1991). Octopuses and squid use visual information, though the natural context for its use is not always known. Octopuses especially have a wide range of movement, yet much may not be under central control and so not accessible for modification by learning. Coleoid cephalopods have a two-phase life-span, the first phase devoted to surviving and gaining weight and the second to maturing, finding a mate, and reproducing. They are not very social and are everywhere at risk from predators. Why should all this result in a heavy dependence on learning, and what constraints do these pressures bring to cephalopod use of information? The following sections of this chapter seek answers to these questions, focusing first on what we know oftheir learning and then on particular areas of behavior that may involve cognitive pro-

324

JENNIFER A. MATHER

cessing. With so little information, some of the evaluation is speculative, but it also poses questions that need to be answered by future research. 111. LEARNING STUDIES A.

VISUALCATEGORIZATION

During the 1950s and 1960s, a long series of experiments was carried out, mostly at the Stazione Zoologica in Naples, on the learning capacity of Octopus vulgaris (see Sanders, 1975). Much of this research was not focused on the learning itself. Instead, learning was used as a vehicle to understand the abilities and limitations of the octopus in dealing with sensory information and to identify the areas of its brain involved in the control of learning. The studies were camed out before the introduction of more sophisticated methods in the last two decades. Other studies of perception and learning in the first half of the century were poorly controlled, so most of these are not mentioned here (though see Sanders, 1975; Wells, 1978). Bitterman (1975) has criticized the lack of proper controls in some of the Naples experiments, and Boa1 (1994) recently evaluated several cephalopod learning experiments and concluded that none was free of inadvertent cuing. Few used statistical analyses to assess the significance of results. Despite these problems, some general conclusions appear sound. Learning capacity is partly situation dependent, so the paradigm in which learning is set is important, as has been shown in the studies of spatial learning by bees foraging for nectar begun by von Frisch (1967). Researchers at the Stazione studying octopus learning developed a stan-

MAJORSOURCES OF

TABLE I AEOUTCEPHALOPODS BEHAVIOR STUDIES

INFORMATION

FOR

Authors

Year

Focus

Young Sanders Wells Boyle (Ed.) Boyle (Ed.) Boyle Messenger and Hanlon

1971 1975 1978 1983 1987 1991

Nervous system anatomy Learning Behavior and Physiology Species-specific description Physiology and natural history Maintenance Behavior

199s

COGNITION IN CEPHALOPODS

325

dard paradigm using the octopod habit of sheltering in protective homes and coming out to investigate novel stimuli (Boycott, 1954). Each octopus occupied a shelter at one end of an aquarium and was presented with stimuli lowered into the other end. Food was used as a reward for contacting the positive stimulus and a small electric shock as punishment for touching the negative one. The shock probably paralleled in intensity the nematocyst sting of sea anemones, which are sometimes carried on the gastropod shell occupied by a hermit crab and which often prevent the octopus from obtaining the crab (Ross, 1970). This paradigm led early researchers to a description of octopus behavior in terms of an approach-avoidance dichotomy (e.g., Maldonado, 1963),and an assumption that they were passive, which was natural for research at the time but may have limited understanding of the scope of their behavior. If we ask a simple question we will get an answer to it, but a simple approach may not fully explore an animal’s behavioral capacity. Octopuses are not reactive sit-and-waitpredators, but forage using an apparent “win-switch” strategy around their home (Mather, 1991a), and they may remember previous foraging directions. They move to a new home partly in response to falling food supply (Mather, 1994) and may assess food patch quality. An advance-retreat approach to behavior will not explore this wider ability Some of the Naples studies assessed how sensory information was “transformed, reduced and elaborated” (Neisser, 1976).As mentioned in Section 11, the octopus has a large area of the brain, called the optic lobe, dedicated to processing visual information and arranged in cell layers like the mammalian visual cortex (Young, 1971). Does the rectilinear array of rhabdomelike receptor cells mean that octopuses assess a mosaic of features of items such as extent and edges, like insects (Barth, 1989, or is their visual processing more holistic, as in vertebrates? A long series of studies by Sutherland (e.g., 1963) produced no simple answer to this question. Octopuses easily learned to discriminate a vertically oriented shape from a horizontally oriented one, but had much more difficulty discriminating a pair of figures oriented obliquely. Sutherland (1957) suggested from this that octopuses might assess figures by calculating their vertical and horizontal extent. The preponderance of horizontal and vertical dendritic connections found in the second-order cells of the optic lobe (Young, 1965)also pointed to such feature detection. However, octopuses could discriminate mirror-image figures that had the same vertical and horizontal extent, although they learned up-down inversions faster than left-right ones (Messenger, 1973a; Sutherland, 1960). Analysis by extent alone could not thus account for their visual information processing. Other studies by Sutherland (1963) continued to look for simple rules

3 26

JENNIFER A. MATHER

by which octopuses evaluated visual input. Another dimension of a figure on which visual information processing might be based is the ratio of edge to area. As octopuses discriminated between figures shaped like V and W, which have different edge-to-area ratios but the same vertical and horizontal extent, they were not limited to this simple assessment rule either. They could discriminate between a figure and itself simply rotated 90°, although not as reliably as some of the “easier” horizontal-vertical comparisons. They could distinguish pairs of square figures consisting of internal black and white bars that were either aligned vertically or horizontally (Sutherland, Mackintosh, and Mackintosh, 1963). Such figures cannot be discriminated by the simple rule of extent or by edge assessment. Much more complex processing rules suggested by Deutsch (1960), who proposed that neurons assessed edges but were inhibited by vertical and excited by horizontal extent, would be needed to discriminate these figures. The limited scope for understanding perceptual processing by expecting simplicity can be seen from this work. What an animal can discriminate when you test its learning in a restricted situation is not necessarily what it uses when it has a complex stimulus situation and is free to choose; note the complexity and fine discrimination that Ristau (1991) found for plovers assessing potential human threats to their young. Later work with the octopus proved this clearly. Muntz (1970) devised a pair of complex shapes (see Fig. 1) that octopuses could discriminate and that did not

A2

A1

81

82

83

B4

c1

c2

c3

c4

FIG. 1. Visual figures (Al, A2) and transfer shapes that Octopus uulgaris could discriminate but whose features did not match any of those proposed to be critical by theorists. From Muntz (1970). 0 The Experimental Psychology Society.

COGNITION IN CEPHALOPODS

327

differ on any of the projections earlier theorists considered important, and therefore did not fit any of the six hypotheses that they had advanced to explain visual processing. He concluded that, while vertical and horizontal extent might be important for assessing visual information and were coded in brain cells (as in primates; see Mansfield, 1974), the octopuses did not have any single or simple rule for encoding visual information. The experiments led to others that assessed what might be called simple concept formation. Messenger and Sanders (1972) trained octopuses with two relevant cues, brightness and orientation of shapes. Octopuses given both cues at once learned the discrimination faster than those given only one, and transfer tests showed that 22 of the individuals depended more on brightness and 6 depended more on orientation. When another group of octopuses was trained for a prolonged period on an orientation discrimination, the animals subsequently took longer to learn a shape discrimination (Mackintosh and Mackintosh, 1963). Octopuses could learn a discrimination between parallelograms that was too fine to be made originally if they were trained on closer and closer approximationsto the small angular difference (Sutherland, Mackintosh, and Mackintosh, 1965), an ability recently discovered for bees with different figures (Zhang and Srinivasan, 1994). This research, while starting with the assumption that octopuses use simple cue analysis for visual assessment, has led to a more complex but more interesting model of their information processing. If a concept is defined as an internal category of information that is activated by an external instance, do octopuses have concept formation?Watanabe, Lea, and Dittrich (1993) discussed the extensive studies on pigeon visual category discrimination, and tentatively concluded that pigeons do have concepts. Since octopuses can learn both to attend to cues not previously discriminable and to choose among these cues when they are presented to them, it can be argued that cephalopods, too, can form concepts. If this definition is used, perhaps this ability should also be recognized in bees (Ronacher, 1992). Octopuses were not merely making fine immediate sensory discriminations but were both “bottom-up” and “top-down’’ processors at once, as has been amply proven for humans (Schiffman, 1990). Thus, if cognition is defined as the rules by which the sensory information is processed, stored, and then used, following Neisser (1976), the first step in cephalopod cognition may be their ability to learn rules about what sensory information is relevant.

B.

EVALUATION OF TACTILE INFORMATION

While octopuses can learn tactile discriminations, the flexibility shown in visual processing does not appear to be present in the touch system.

328

JENNIFER A. MATHER

Wells and Wells (1957) modified the basic visual learning paradigm of earlier experiments by presenting plastic cylinders with incised patterns to the arms of blinded octopuses. The animals could easily process texture, learning the discrimination between a smooth cylinder and a grooved one. But octopuses had much more difficulty discriminating textures from one another, never mastering the difference between a horizontally and a vertically grooved surface or between linear grooves and a grid. Perhaps the animals used touch involving stimulation of mechanoreceptors, and not haptics (active touch), which also requires monitoring of the position of body parts. The distinction between passive touch and haptics has only relatively recently been clarified in humans (see Lederman, Browse, and Klatzky, 1988), and it is important to realize that haptics uses a complex of several sources of sensory information. Haptics relies on information about the position of the exploratory surface, in the case of humans, usually the position of hands and fingers through proprioceptors in muscles, joints, and tendons, as well as that from surface mechanoreceptors. Recordings from interneurons in the octopus arm showed that some tactile information was passed to the brain of the octopus when the arm was touched (Rowell, 1966), but this may only be touch receptor input and not haptics. Wells (1978)suggested that octopuses might discriminateby touch solely on the basis of distortion of the surface of the flexible suckers. Octopuses were not able to discriminate cylinders on the basis of weight (Wells, 1961), which would require haptics, but could discriminate cylinders of different diameters, perhaps due to the amount of angular distortion (Wells, 1964a). They could also discriminate a cube from a sphere of the same volume, but this distinction could not be transferred to a cube with rounded-off edges, suggestingthat the abrupt change in angle was signaling that the figure was a cube (see Fig. 2). Such a distinction may indicate some haptic processing in addition to tactile discrimination. An interesting experiment in this series raises the possibility that octopuses might have some primitive ability to assess amount. They were capable of discriminating cylinders on the basis of number of grooves incised in them, reaching criterion on three discriminations, 0 versus 13, 0 versus 7, and 7 versus 13. Even though they did not reach criterion on 4 versus 7 grooves, the animals nevertheless showed some interesting generalizations. Those for which 4 was the positive stimulus also readily responded to 2, whereas those for whom 7 was positive accepted 10 easily. When 7 was positive and 14 negative, animals also accepted 2 and 4, but this waned over time. Sanders (1975) described this as a “stimulus generalization,” but what is being generalized and how? Perhaps they were evaluating amount of distortion but, if they discriminate only abrupt

329

COGNITION IN CEPHALOPODS P1

P4

00

PC

+

PS

PCG

PSG

4

Cylinder cross-sections

38

: I :

PSL

FIG.2. Some of the perspex cylinders used in discrimination of objects using tactile information by Octopus uulgaris. From Wells (1964a). 0 The Company of Biologists Ltd.

edges, the paradigm could be useful to test their ability to evaluate amount (Wells and Young, 1970) or a limited capacity to count, though one certainly short of that which has been shown in birds (Pepperberg, 1994) and in rats (Davis and Bradford, 1986). The cephalopod tactile system may have limited capacity for central processing of information, or the octopus may not pass much tactile information to the brain because the arms process it locally and send little centrally. Experiments on this capacity may allow us to find out the basis on which such categorization of objects works. A complete assessment of how cephalopods process and store nonvisual information (their “rules of use” in the sense of Neisser, 1976) clearly remains to be made. For instance, Bogdany (1978) found a tight linkage between time, color, and scent for honeybees learning about flowers. Similar, but chemotactile, linkage might guide octopus learning. No studies have examined intersensory transfer or the effect of combining cues in more than one modality on learning by octopuses, since those used for studies of tactile learning were blinded. Cues relevant to natural problems for the octopus, such as size and shape of shelter or position relative to a rewarded location (see Walker, Longo, and Bitterman, 1970), may also elicit learning better. Ethological analyses of sites of penetration into molluscan shells, discussed in Section IV,C, suggest that haptic information is used in guidance of shell drilling.

330

JENNIFER A. MATHER

C. BRAINORGANIZATION AND LEARNING A completely separate interest in studies of cephalopod learning has been to discover how learned information is processed within their brains compared to those of vertebrates (see Bullock, 1993, for a plea for understanding the evolution of brains). Cephalopods are similar to vertebrates in being bilaterally symmetrical, and much of the brain organization of both is rooted in that symmetry. Assessment of information storage in the brain has been limited to studies using ablation because nervous control of blood vessels of cephalopods allows them to shut off the area exposed by an operation, causing the cells to die. Recording from the brain was impossible until recently (see work on Sepia by Bullock and Budelmann, 1991, and on Lolliguncula by Dubas, Hanlon, Ferguson, and Pinsker, 1986). Ablation experiments demonstrated that one area, the vertical lobe, was involved in visual learning, and that another, the subfrontal, was involved in touch learning (see Sanders, 1975; Wells, 1978, for reviews). One principle discovered was that information storage was limited to a general region, yet distributed within it. The optic lobes were necessary for visual learning by octopuses, and when a lobe was ablated learning was lost, but when half was removed, a lower proportion of correct responses was made (Parriss, 1963). Information was also found to be first stored locally and then distributed more generally. Muntz (1963) made a cut vertically through the octopus optic lobe and then projected images onto the front or back half of the retina. This led information to be stored only in one half of the lobe. Proof of this was that an octopus trained on a discrimination with one half of the eye could not recognize it when the image fell on the other half and was processed by the other half of the optic lobe. As octopus vision is mostly monocular, animals can also be trained to make a discrimination with one eye. When tested with the other, their performance was less efficient unless they were trained for a long period (Muntz, 1961). Birds may fail completely at the same kind of transfer (Remy and Watanabe, 1993). If the left-right connections in the brain were split after learning, the octopus could still respond correctly using the “untrained eye,” but if the split took place before training the octopus could not (Muntz, 1963), which is equivalent to the situation for split-brain humans. This primacy of local storage might account for octopuses’ behavior on detour tests. Wells (1964b, 1967) noted that the animals alternated “leading” around the walls with one and theh the other eye focused on the barrier. Octopuses with split brains could still crawl around a barrier normally but appeared to be in conflict, alternately moving into the maze and returning to their shelter. Such behavior was much reduced when

COGNITION IN CEPHALOPODS

33 1

one eye was blinded, perhaps removing the conflict in incoming visual information. Sanders (1975) assessed these studies in detail and pointed out problems with such a conclusion. Commissurotomized octopuses completed fewer detours but had a similar proportion of successes to intact animals. Because they succeeded only when maintaining contact with the “correct” wall, perhaps body orientation information was used in parallel with visual input to guide the detour. Detour tests of octopuses in which variables such as sight of the goal, visual information about the route, and tactile contact with the wall are varied, and the octopuses’ actions videotaped, could reveal what is going on in this experimental situation. Such studies would allow assessment of the relative contribution of sensory information from different modalities to how the animal determines its position. Nevertheless, lateralized storage of information appears to be a general principle, as split-brain octopuses could not perform a tactile discrimination with arms on one side if they had learned it with those on the other (Wells and Young, 1966). While these studies have revealed much about how information is stored in octopus brains (Young, 1965, 1991), they have two inevitable limitations. The first one is procedural: ablation studies presume complete localization, so that once an area of the brain is removed the behavior programmed only within it is gone. The second one is situational: laboratory research has shown what the octopus can discriminate on the basis of touch and vision, but the emphasis on establishing capacity meant that little thought was given to use, how these abilities relate to octopuses’ behavior in the natural environment. The synthesis of laboratory studies of learning and ethological observation of natural behavior, so well addressed in a review by Gould and Marler (1984), was not possible because the ethological basis was lacking. Octopuses do not normally use visual shape perception to recognize prey, but hunt predominantly by chemotactile search (Yarnall, 1969; Mather, 1991a). The sophisticated visual processing capacity uncovered by laboratory tests with a feeding paradigm is probably used more in other situations, such as navigation (Mather, 1991b) or the avoidance of predators (Packard, 1988a). One small series of studies illuminates the relationship between brain and behavior in the development of learning. The cuttlefish Sepia has a relatively fixed three-phase attack on small prawns (but see also the discussion in Section IV,C): Attention (including eye movements, color changes, and arm postures); Pivoting until the prey is straight ahead (Sepia fixates prey binocularly); and Attack by shooting out the extensible tentacles (Messenger, 1968). The Attack phase is an open loop but the first two phases are dependent on visual feedback. Cuttlefish struck at prawns in a glass tube, and could learn not to attack prey after repeatedly striking

332

JENNIFER A. MATHER

their tentacles on the tube, though they learned much more slowly when the tentacles were ablated and no feedback about impact was received (Messenger, 1973b).This learning was abolished by ablation of the vertical and superior frontal lobes of the brain, which probably demonstrated a localized control of learning like that of the octopus. Analysis of the three stages of the attack confirmed different dependence on learning. When the prawn was in a tube, Attack declined dramatically, Positioning less so, and Attention little if at all in the course of 20 min (Messenger, 1973b). The decline was due at least in part to carrying out the actions, as a cuttlefish blocked by a screen from coming close enough for Attack persisted in Positioning for many minutes. An intriguing result was obtained in the behavior of young cuttlefish, which are physically similar to adults though much smaller. Newly hatched animals would attack only a limited range of visual stimuli, those that were elongated horizontally and moving along their long axis, as their crustacean prey Mysis would do (Wells, 1958). As they matured they started to attack a greater range of prey, so that the early stereotypy seems to be a mechanism to restrict their prey choice to suitable species. Young cuttlefish never learned not to attack Mysis in a test tube, but this inflexibility gradually waned (Messenger, 1973b). Their increase in learning was correlated with brain development, as the vertical and superior frontal areas of the Sepia brain were poorly developed at birth and enlarged considerably in the first few months of life. It is rare to know about both the changes from stereotyped to flexible behavior and the accompanying brain changes as animals mature. This may be a parallel to the sensitive periods found in vertebrate behavioral development; further investigation of this topic should help in understanding the relationship between brain and behavior during development.

D. LEARNING LEVELS A N D TYPES Learning is sometimes presumed to follow universal rules across animal groups (see Mason, 1984). If such rules did apply, types of learning could be ranked depending on the capacity needed to carry them out (Thomas, 1980). Most animals are capable of simple learning (Levels 1-4 in Thomas’s terminology),though this has not always been clearly shown. Cephalopods show learning at these levels. Boycott (1954) demonstrated habituation of octopuses to visual stimuli, and Wells and Wells (1957), to tactile ones. Messenger’s (1973b) studies of Sepia (see Section 1II.C) are often presumed to demonstrate habituation of tactile strike responses to prawns in a glass tube. However, Sanders (1975) pointed out that the waning of response (see also Sanders and Young, 1940)occurred when only tentacle

COGNITION IN CEPHALOPODS

333

strikes were permitted, so this was actually associative learning. Operant conditioning has been demonstrated for octopuses by Crancher, King, Bennett, and Montgomery (1972), who conditioned arm extension up a tube out of water; the attempt by Dews (1959) to condition lever pressing was less successful. Classical conditioning was shown in the experiments discussed earlier. Some studies of octopuses have also demonstrated learning “set” (Harlow, 1949). These indicated competence on Levels 5 and 6, for which an animal must not only discriminate stimuli but also learn what has been called an Absolute Class Concept, reacting to stimulus membership in not one but a class of situations. This is concept formation, as described above for visual categorization, but the concept is a more abstract one (see Herrnstein, 1990, for a discussion of levels of concept formation). One situation that tests for learning set is reversal learning. Once an animal has learned to approach a positive stimulus and reject a negative one, their valence is reversed. Many mammals learn this quickly and respond correctly to reversals within one trial, but the situation is not so clear for octopuses. Mackintosh and Mackintosh (1964a) trained animals, using simultaneous presentation of visual stimuli, for six reversals. When the criterion of successful discrimination was that 70% of responses were correct, the octopuses significantly decreased the number of trials to criterion over the experiment. When the criterion was a more stringent 80% the animals seldom reached criterion, and thus did not demonstrate reversal. The problem may have been a lack of irrelevant cues, because a subsequent group, trained with such cues and then given 100 overtraining trials, demonstrated reversal (Mackintosh and Mackintosh, 1964b). Sanders (unpublished, in Sanders, 1975) also produced learning of reversals for octopuses given tactile stimuli. The paradigm is a difficult one, and octopuses’ tendency to prefer specific orientations (vertical) and textures (smooth),as well as their low positive response rate (Papini and Bitterman, 1991) may make experimental control difficult. A similar problem arises in testing octopuses for the learning set called oddity. For this, an animal is presented with three or more stimuli and rewarded for touching the one that is different, requiring not a recognition of stimulus characteristics but the “concept” of oddity to be formed. Boa1 (1991) tested octopuses on oddity, lowering sets of three molluscan shells into their tanks. They learned several different sets of comparisons and demonstrated transfer of inference, Thomas’s (1980) Level 5 . But they did not reach the stringent criterion of choosing the odd stimulus on the first trial, as would be needed to demonstrate Level 6. The octopuses’ variable response pattern may have prevented them from demonstrating such an evaluative ability, and more research with both standardized

334

JENNIFER A. MATHER

presentation and stringent control of external variables will be necessary to test for this capacity. A separate and more recent approach to learning within different animal groups has made it clear that every group has limits on learning. Thus, honeybees have constraints both on the situations in which they will learn and on the type of learning they use (Bogdany, 1978). This restriction is also true for cephalopods, though much less explored. Early researchers found that octopuses could be trained only with difficulty to treat crabs as a negative stimulus (Boycott, 1954) and had stimulus-specific preferences. It is not clear whether crabs were a Prepared Search Image for octopuses as flowers apparently are for bees (Menzel, 1989, which can learn variations in flowers from cues about them much more easily than they learn variations from cues for other situations. In contrast to vertebrates, octopuses responded poorly to partial reinforcement (Papini and Bitterman, 1991). Reversal on overlearning and poor response to partial reinforcement may represent a constraint on learning that is adaptive for animals with a foraging strategy that must track a limited and changing food supply (see Stephens and Krebs, 1986). Octopuses of several species have been found not to set up and defend a territory, but to move to a new home range after a variable but fairly short period (Hartwick, Ambrose, and Robinson, 1984; Mather, 1994). Availability of preferred food, as measured by remains in middens, was a predictor of duration of stay for 0. uulgaris (Mather, 1994). Perhaps these octopuses were using a winstay strategy for a limited duration, staying in the same area if prey was abundant. Thus, if the amount or type of reward fell below a particular threshold, or if duration exceeded some limit, they might cease to respond to rewarded stimuli and instead move to a new home range. Apparent “inabilities” of animals such as octopuses may also be due to a lack of fit to the sort of simple experimental situation that is suited to the laboratory rat. Fentress (1992) noted the larger repertoire of adaptive behavior in his pet wolf compared to his pet dog, and Mason (1984) described a problem-solving capacity in wild-reared monkeys that was wider than that in lab-reared monkeys. All octopuses that have been used in learning experiments were captured as adolescents and had thus survived against high odds (each female 0. uulgaris lays tens of thousands of eggs). Their behavior in experimental trials of learning was often far more complex than was revealed by simple tests involving positive or negative choices, demonstrating “recovery and use” (Neisser, 1976) of information about capture techniques. Boycott and Young (1957) noted “cautious” behavior when octopuses had been shocked for attacking negative stimuli. When the experimental apparatus was brought to their tank, some animals investigated the remains of previously consumed crabs

COGNITION IN CEPHALOPODS

335

or attacked air bubbles on the surface or shadows on the tank; one octopus attempted to climb right out of the tank. Variable behavior may be adaptive for reaching a goal such as food. Octopuses stung by an anemone when trying to capture a hermit crab used what Boycott (1954) again called cautious behavior. Instead of descending from above in a spread-web attack, they might approach slowly along the tank bottom and extend a single arm to the crab, avoiding the anemone. Some octopuses pulsed jets of water, as if attempting to dislodge the anemone or move the crab. The octopuses might also “walk” along the side of the aquarium and reach down to catch the crab from above. Such “versatile adaptability to novel challenges” is one of Griffin’s (1993) reasons to infer animal consciousness, although consciousness does not necessarily imply any type of cognition. Flexibility of behavior also suggests sophisticated information processing by the standards of Yoerg and Kamil(l991). This ability to recover and use information contrasts with the apparent failure of octopuses at “higher” levels of learning, as defined by Thomas (1980). One type of learning recently demonstrated for octopuses, which requires cognitive processing, is observational learning. Fiorito and Scotto (1992) taught one octopus to approach one of two colored balls in its aquarium. After four trials, another octopus, which had simply observed the rewarded situation, was found to perform at a 70% correct level, whereas the “demonstrator” was only performing at 51% (a chance level). This sophisticated means of gathering information has generally been demonstrated only in “higher” vertebrates and is unexpected because most octopuses do not live in social groups, although they can form dominance hierarchies (Mather, 1980; Cigliano, 1993). Why should octopuses have the capacity to perform observational learning? Perhaps it is retained from an ability that was useful to the ancestral social squid, or perhaps it tests a more general ability to elaborate and store information. These experiments are preliminary, and extension and elaboration of testing for this ability may yield interesting information.

Iv. OTHER DEMONSTRATIONS OF CAPACITY A.

COMMUNICATION A N D “LANGUAGE”

Our knowledge of animal cognition has expanded dramatically with increased understanding of the complexity and subtlety of animal communication (see Cheney and Seyfarth, 1990, for studies on monkeys). The suggestion by Moynihan (1985) that the visual skin patterns of the sepioid squid Sepioteuthis formed a sophisticated visual communication pattern

336

JENNIFER A. MATHER

offers a promising area of investigation, on which work has started but much more is needed. A large part of the motor control system of coleoid cephalopods is concerned with control of skin color. There is direct nervous connection from the brain to muscles that expand elastic pigment-filled chromatophore sacs on the skin surface, producing changes in the red, yellow, and black components in milliseconds (Packard, 1988b).In addition, reflective layers of white leucophores and green irridophores deeper in the skin reflect light when the chromatophores contract, and muscles in the skin raise papillae when appropriate. Three extra components of pattern, skin texture and arm and body posture, help generate a remarkable flexibility of appearance (see Fig. 3). Even though cephalopods are probably colorblind, many can match their surroundings nearly perfectly. Some complete

FIG.3. Some of the combinations of skin shading, and arm and body postures that present different appearances in Sepioteurhis sepioidea. From Moynihan (1985).

COGNITION IN CEPHALOPODS

337

patterns not used as camouflage, such as the Dymantic (pale skin, spread body surface and web between arms, dark outer edges, and central “eye spots,” orienting toward the receiver), are found throughout the coleoids (Moynihan, 1975). Others, such as background matching (see plates in Hanlon, 1988), appear almost infinitely variable. Packard (1988a) pointed out that the extraordinary allocation of resources to skin appearance suggests heavy selection by predation. Detailed analysis of the components of skin displays of Sepioteuthis revealed the extent of the repertoire. Visual patterns could be produced over most of the body (major components)or only part of it (minor components), aided by texture and position changes (Moynihan and Rodaniche, 1982).These authors found 31 major componentsand concluded that there could be 279 permutations of the four aspects of display (major, minor, texture, position), although actual combinations were not catalogued. As Sepioteuthis is a schooling species, patterns could be used as signals to conspecifics (in the school or during pairing off for mating), as well as for avoidance of predators or to cause changes in behavior of prey. This repertoire is not unusual for a coleoid; Sepia has perhaps an equally large set of skin colors and patterns (Hanlon and Messenger, 1988), although their combinations and signal value have not been explored. Moynihan (1985) theorized that, as a communication system, this pattern production includes several of the design features specified by Hockett (1960) for a language. It has broadcast transmission and directional reception (as signals can be general or on only one side), could continue for a long time, or fade rapidly (some were sudden flashes of much less than a second). It might have duality, in that components could be meaningless, while a combination had meaning. In addition, he postulated that the characteristics necessary for a language of interchangeability, specialization, semanticity, arbitrariness, and discreteness might be produced by this system. Moynihan and Rodaniche (1982) suggested that such a language as that expressed on Sepioteuthis skin had three classes of components. The first of these were Signalers, which they likened to the nouns and verbs of human language. The second were Modifiers, which appeared only on small areas of the body surface and could be compared with adjectives and adverbs. The third were Positionals, the postures and movements that best served to present the communication. The idea of squid body-surface visual language is a fascinating one that, if proven, would enlarge our understanding of what constitutes a language and expand the known scope of the intellectual ability of cephalopods. Studies on octopuses make it clear that combinations of visual cues can be processed (see Section 111,A). However, that such displays constitute a sophisticated communication system is far from proven. While Moyni-

338

JENNIFER A. MATHER

han and Rodaniche (1982) demonstrated a repertoire of patterns, they did not explore the complete repertoire or the situations in which it is produced. The detailed descriptions of such components by Hanlon and Messenger (1988) for the cuttlefish, by Packard and Sanders (1971) for Octopus vulgaris, and by Hanlon (1982) for Loligo plei show that an extensive repertoire of skin patterns might be common in coleoids. However, most of the observation of skin pattern production has been anecdotal and researchers have seldom demonstrated in what situations particular patterns were produced (see Mather and Mather, 1994, for a preliminary description of this in Octopus vulgaris). Developmental changes in the repertoire are also nearly unexplored, although Packard and Sanders (1971) found that the arm-curled Flamboyant of juvenile octopuses is replaced in older animals by the spread-web Dymantic. All these omissions must give considerable pause to any description of skin colors and patterns as a communication system. Since cephalopods have extensive local control of motor output (e.g., in the arms), communication might be the result of local responses and not central control by the brain. Yoerg and Kamil(l991) remind us that complexity and sensitivity of behavioral responses do not necessarily imply intentionality or mental calculation by the animal involved. We must first know much more about how such information is “received, transformed, recovered and used” (Neisser, 1976) before we can make assumptions about cognition. Cephalopod communication is a promising field for such an investigation, with experiments either in the field or in the laboratory. Cheney and Seyfarth (19%) elegantly demonstrated on vervet monkeys how field experiments can explore complex behavior, and Clark and Uetz (1990) showed how video could be used in the laboratory to standardize the presentation of visual stimuli to spiders. As no such studies have been carried out so far on coleoids, their communication in a languagelike fashion through skin patterns remains only an intriguing but unexplored possibility.

B. SHELTER CHOICEAND PRODUCTION Predation pressure may be so extreme on the soft-bodied cephalopods (Packard, 1972), in which many eggs result in a few adults, that skin color patterning is only one of several behavioral mechanisms that have evolved to deal with it. Many of the benthic octopus species dig into the substrate or seek out shelters in it. Aronson (1986) found an abundance of 0. briareus in Sweetings Pond, which is sheltered from the open sea, and almost none on the reefs outside it. There appeared to be a shortage of appropriate hiding places in some situations, such as on a sandy substrate, as 0. joubini were found only in eel grass beds where mollusc shells

COGNITION IN CEPHALOPODS

339

were available for occupation (Mather, 1982a). The shortage of shelter sometimes resulted from competition with other species when 0.joubini, hermit crabs, and Bleniid fish all competed for gastropod shells (Mather, 1982a). Lack of appropriate shelter may stimulate dominance interactions (Cigliano, 1993), and shell size limits fecundity (Iribarne, 1990). Some early authors romanticized shelter use of octopuses as ‘building of homes,” though Bierens de Haan (1926) claimed it was at a level far below the complexity this suggests: simply a thigmotaxis in which octopuses reflexly surround themselves with whatever is available, including transparent pieces of glass. Nevertheless, several authors have shown that different octopus species in the field select size-appropriate homes (Hartwick et al., 1984; Ambrose, 1982; Mather, 1982b). They do this on the basis of at least two characteristics: absolute size and size of the entrance (Mather, 1982b; Aronson, 1986). For the soft-bodied octopus, which can squeeze through small openings, small entrance size may eliminate competition from fish and crabs, and discarded beer cans are excellent homes. The decision of which shelter to use involves more than simple stimulus assessment (note Neisser’s, 1976, stages of information “elaboration, storage, and recovery”). Mather (1994) observed octopuses as they foraged across the rocky ocean floor and stopped to choose a home site; they explored little before choosing, and the homes were subsequently extensively modified. Some common characteristics were found in homes examined in two different years and in two locations 1 km apart. The opening angle was close to vertical and tended to face away from the nearest shoreline. The homes were also found over a small range of depth, averaging 20 cm below low-tide level (the intertidal area is rich in prey species, but has the disadvantage of drying out twice a day). One of the ways in which octopuses altered these sites was by bringing stones and shells to partially block the entrance, and there was a significant correlation of number of stones with entrance size. Octopuses were apparently not assessing what was merely in front of them, but also, perhaps on the basis of stored information, what could be. While 0.joubini showed preferences in terms of size of both home and aperture when presented with choices (Mather, 1982a),these 0. vulgaris constructed homes to meet such expectations. The strongest predictor of long duration of occupancy was home volume (Mather, 1994), suggesting that this characteristic was also assessed in “decisions” about moving to a new home or staying in the present one. Some actions in making the modifications involved meet the definition of tool use given by Beck (1980): “a manipulation of an object free of any fixed connection, outside the user’s body and held or carried prior to manipulation; the user must establish proper orientation between it and

340

JENNIFER A. MATHER

the incentive, thus altering the form, position or condition of user, object or organism.” He pointed out that use of tools did not, of course, prove intelligence. Nevertheless, Griffin (1993) argued that versatile use of tools, especially when tool preparation is separate from the use of the tool itself, is suggestive of intelligence. Mather (1994) found octopuses to block home entrances with several different items, including shells of prey brought home for consumption, stones removed from the home during modification, and rocks selected from the substrate near the home and brought back during a trip for that purpose. All of these could be regarded as tools. Octopuses also use the jet of water from their flexible funnel as a tool, blowing sand and shell fragments out of the cavity they have selected. This is part of a complex pattern of behavior that also includes carrying larger fragments out under the outstretched arm web and moving rocks around by pulling and pushing with the arms. Hausseknecht and Kuenzler (1990) argued that what appears to be intelligent behavior may just be response chaining, and reported such chaining in home construction by cichlid fishes. Shelter construction that seems complex may indeed only involve repeated responses with “simple” rules, as Franks, Wilby, Silverman, and Tofts (1992) modeled for construction by ants. Manipulation of the environment is a promising field in which to investigate the behavioral capacities of octopuses; the possibility that response chaining occurs can be investigated by seeing how they respond when the end results of each step are changed. The flexibility behind the apparently stereotyped action of Sepia cuttlefish digging into the sand for concealment (Mather, 1986) is relevant in this context. Several sepioid species use species-typical patterns of water jets and arm movements to bury themeselvesjust below the sand surface (Von Boletzky and Von Boletzky, 1970). Cuttlefish burying was normally achieved by a short sequence of four water jets underneath the body, alternately aimed forward and backward, followed by a lateral mantle Wiggle, which probably distributed the sand more evenly. When cuttlefish were on sand too shallow for complete burial they produced significantly morejets, and when they were simply uncovered by a stream of water they produced fewer. With a substrate of opaque plastic beads they changed the sequence of acts, producing more water jets but omitting the final mantle Wiggle; either tactile or visual feedback may have informed them of the abnormal situation. Before any digging action began, cuttlefish hovered over the substrate and performed tactile exploration with the outstretched arms, rejecting a gravel substrate in favor of sand when given a choice, and not digging but settling onto the gravel surface and assuming a camouflage coloration if gravel was the only substrate available. Much of the information assessment stressed by Yoerg and Kamil(1991)was apparently taking place both before and during the digging sequence.

COGNITION IN CEPHALOPODS

34 1

C. RESPONSE FLEXIBILITY DURING PREDATION Cephalopods, with the obvious exception of newly hatched Sepia (see Section W C ) , are generally unselective predators; Ambrose (1984) recorded 55 species of molluscan and crustacean prey for Octopus bimaculatus. Cephalopods must have evolved flexible and varied strategies for handling this variety. Many of these strategies have been observed but not tested in experimental situations, so evidence about information processing during predation is sometimes indirect. Examination of the visual attack of cuttlefish in an invariant situation by Messenger ( 1968) suggested a stereotyped response. However, Sepia could capture prey either by fast ejection of their small tentacles or by grabbing with their sucker-equipped arms, and they chose the behavior that suited the prey type. Small and slow-moving crabs were caught by tentacles and large ones and fast prawns by the arms. The cuttlefish also learned to orient approaches to crabs from behind to avoid the pinching claws (Duval, Chichery, and Chichery, 1984: Boulet, 1958, 1989). When a crab on a string was jerked away by the experimenter, the cuttlefish changed its technique from grabbing to tentacle strike, and when the tentacle tips were removed, some, though not all, of the capture attempts changed to grabs. On a sandy bottom crabs bury themselves in the substrate, and cuttlefish were found to switch their means of location to blowing sand off crabs with jets of water from the funnel, then capturing them with the aid of vision. Such flexibility is probably also common for squid, as Foyle and O’Dor (1988) reported that Zllex captured smaller, slow rnumrnichog fish by rotation and fast approach, but added a slow tracking phase when capturing faster trout. Sepia is a promising genus for analysis of behavior. Individuals hatch from large eggs and young are adultlike; animals are also relatively tolerant of variations in water quality and are slow moving, so that they can live in the confined conditions of captivity. Manipulation of variables such as velocity, trajectory, and size in prey models could aid in systematically exploring reactions to prey in Sepia. Octopuses demonstrate their flexibility in a different aspect of predation: penetration of the hard shells of other mollusc species. Prey animals are generally disabled by injection of a paralytic neurotoxin from the posterior salivary gland (Ghiretti, 1960). To gain access to hard-shelled molluscs, octopuses can pull valves apart or pull a snail out of its shell, break the shell by force, chip at the edge, or drill a small hole in the shell and inject the toxin through it. Because the octopus surrounds the shell with its arms while carrying out these procedures, it is not possible to observe the sequence of actions. However, casual observation and examination of the remains of prey in middens (Hartwick, Thorarinsson, and Tulloch,

342

JENNIFER A. MATHER

1978) suggested that smaller and presumably weaker bivalves were pulled apart and larger and stronger ones were drilled, though statistical analysis was lacking, Wodinsky (1%9, 1973) conducted systematic studies of the choice of technique used by Octopus vulgaris fed on snails in the laboratory. Dead snails, or those with their retractor muscle cut, were simply pulled out of the shell; this action appeared to be the first choice. Blockage of the aperture (which would be true of many snail species with a protective operculum) led to an attempt to pull out the blockage, and then to drilling if pulling was unsuccessful. Wodinsky attempted to prevent drilling by coating shells with various substances. When the coating was rubber the octopuses pulled the rubber off and then drilled, but when the coating was penetrable aluminum they drilled through both coating and shell. When only the apex of the shell (the target for most drilling) was covered by impenetrable dental plastic, two thirds of the octopuses drilled elsewhere and one third pulled the snail out; when the complete shell was covered with plastic, pulling was the only strategy used. When female octopuses mature and lay eggs their salivary gland degenerates, and they have little or no toxic saliva. Wodinsky (1978) found that they fed little but sometimes accepted snails as prey; in this situation, they would pull the snail out of the shell rather than drill and inject. This suggests that the octopuses were monitoring the results of their actions and perhaps also their internal physiology, demonstrating sophisticated “recovery and use” (Neisser, 1976) of stored information. A similar variety of techniques (pull, break, chip, and drill) was also true for penetration into bivalves by Octopus dojleini (Mather and Anderson, 1994). These octopuses usually broke the weak shells of Mytilus mussels, pulled apart the stronger Tapes, and often drilled into even stronger Protothaca clams. Possibly as a result of penetration difficulty, they consumed more Tapes and Mytilus than Protothaca when the clams were presented intact. However, when offered clams opened “on the half shell,” they preferred the Protothaca. The location of a hole drilled in the shell may be important because the paralytic neurotoxin will be diluted by the water in the mantle cavity and hence less effective than if it is delivered into the mollusc’s body. Both Arnold and Arnold (1969) and Wodinsky (1969) found that drilling was carried out in a relatively small apical area of the snail shell near the insertion of the retractor muscle, and nearly all of the penetrations of Strombus conch shells were opposite the snail’s enlarged lip (Arnold and Arnold, 1969). When the lip was sawed off, this orientation was lost. Similar selectivity of location was true for drill holes in bivalve shells examined by Ambrose and Nelson (1983). The preferred drilling locations may be species specific: central in 0. dofleini, at the anterior and posterior near the adductor muscles in 0. rubescens, and along the dorsal margins

COGNITION IN CEPHALOPODS

343

in 0. vulgaris (Mather and Anderson, in preparation). This may be an example of a combination of predisposition and learning, as in the guided learning of bees, which easily learn color and odor cues about flowers, though the cues used are somewhat different for different bee species (Menzel, 1985). Cross-species comparison of the use of tactile cues that have been studied in 0. vulgaris (Section II1,B) might also show differences. The fact that octopuses orient to and attack specific locations on molluscan prey again brings up the question of what restrictions there are on the learning of position information by cephalopods. Wells (1964b) argued for limitations on the basis of octopods' difficulty in weight perception, tactile learning, and position sense. In detour experiments, an octopus followed a wall by continual tactile contact or fixating the wall continually with one eye. The observation by Fiorito et al. (1990) that 0. vulgaris did not reduce exploration time over successive trials when learning to pull a stopper out of a hole to get a crab supported a limitation on learning position information. Nevertheless, Walker et al. (1970) trained 0. maya to make correct turns in one direction in a T maze when the reward was reentry into the water. Mather (1991b) was similarly able to train 0. rubescens to go to a beacon for a food reward, regardless of the beacon's orientation relative to the start point, which was the home of the octopus. Spatial memory (see Gallistel, 1990) is probably used by the animals to return to their shelters in the ocean. 0. vulgaris returned from foraging trips on average from a heading 30" away from their outward route (Mather, 1991b) (see Fig. 4). The return was probably guided by visual landmarks, as octopuses jetted backward through the water when leaving or approaching home, and displacements resulted in either a return to the outward path (n = 3) or a diagonal return to the home (n = 8). Such spatial memory should be testable in the laboratory, but 0. rubescens failed to demonstrate working memory when tested in an aquatic equivalent of the radial maze devised by Olton and Samuelson (1976; J. Mather, personal observation). The problem may be choice of species or of testing situation, as the pigeons studied by Spetch and Edwards (1988) were similarly unable to work on a radial maze, yet demonstrated spatial memory in an open-room situation. Octopuses tested on orientation to a beacon also showed wide individual variation in response to the open-area testing (Mather, 1991b). D. INDIVIDUALDIFFERENCES

Traditional testing of behavior emphasized similarities rather than differences among individuals of a species. Descriptions of species-specific abilities thus tended have the individual variability factored out. Yet,

344

JENNIFER A. MATHER 7

0

1

2

3

SCALE IN METRES

FIG. 4. Long daytime foraging trips (>2 m) out from the central home of one Ocropus vulgaris in Bermuda over a 10-day period. From Mather (1991b).

individual variation is a conspicuous aspect of octopus behavior, and once the overall pattern is established, individual variation must be taken into account. In standard testing situations, some octopuses attempt to escape from the tank (Boycott and Young, 1957), shoot jets of water at the experimenter (Dews, 1959), or circle the testing arena in jet-propelled swimming ( J. Mather, personal observation), when the experimenter was presenting a standard set of stimuli and other individuals were learning the task in a few trials. Such variability among individuals has only recently been stressed for vertebrates (see Clark and Ehlinger, 1987). Individual variation may be the result of both inherited temperament and experiences during life, as has been suggested for reactivity to stress in monkeys (Suomi, 1991). This may even result in a range of cognitive capacity within a single species that is far larger than cross-species differences, as Povinelli, Rulf, Landau, and Bierschwale (1993) found for self-recognition in a mirror by chimpanzees. With such a range, what is a true assessment of species' capacity? As octopuses were known to show variability in behavior, Mather and Anderson (1993) documented its extent in 0. rubescens using laboratory tests in standard situations of alerting, feeding, and threat by touch. A computer-based multiple discriminant function analysis sorted the re-

COGNITION IN CEPHALOPODS

345

sulting behavior scores and isolated three “personality dimensions,” which could be described as reactivity, activity, and avoidance. Reactivity may be similar to the emotionality found in monkeys by Suomi (1991), and differences on the activity dimension would generate the kinds of differences found by Povinelli et al. (1993) on his tests of self-awareness. These were intriguingly similar to “personality” dimensions found in “higher” vertebrates and humans, although, as the octopuses were not lab reared, the developmental history of such personality differences is not known. Burghardt (1991) found similar large differences in newborn hognose snakes from the same litter, suggesting that the differences between them were inherited. Wells (1958) found newborn Sepia to have a restricted range of prey choice, which expanded considerablywith maturation and brain growth. These individual differences could emerge as a result either of maturation or of experience during a sensitive period. Von Boletzky (1992) concluded that the pelagic environment of the young paralarval phase in nearly all octopods may explain why the limited behavioral repertoire of young juveniles is adaptive. The later benthic phase in the variable and diverse nearshore environment would demand learning and might foster individual differences. Studies of development of behavior would help to elucidate the capacities of paralarval and adult phases. Ironically, large-egged cephalopods with minimal pelagic stages, such as Sepia, Octopus joubini (Mather, 1984), and 0. briarius, are easier to raise in the lab. Thus, answers to the question of how this change in behavior occurs may be a long time in coming. V.

SUMMARY

The information presented here by no means characterizes the ways in which sensory information is “transformed, reduced, elaborated, stored, recovered and used” (Neisser, 1976) by cephalopods. Rather, it gives us a basis on which to make such a characterization. Cephalopods may have considerable information-processingcapacity, and also the ability to make decisions on the basis of visual and/or tactile cues. Such information processing must be used in survival strategies, such as those involved in finding food and avoiding predation. The behavior of open-ocean squid may be most represented by visual communication, avoiding predation, signaling to conspecifics in schools, and finding and dealing with prey using vision. More solitary bottom-living octopods and squid have an added dimension: seeking and making a suitable shelter and returning to it after foraging. The flexibility of movement of these boneless animals may make it impossible for their brain to process all available position

346

JENNlFER A. MATHER

information, and the tremendous ability of their skin to present color and pattern changes offers a challenge for studies of communication. Thus, it is in the context of decision making about feeding, spatial memory for central place foraging, communicating by complex skin signals, use of the physical world, and processing information about their own body position that the phenomena that may indicate cognition in cephalopods must be examined.

Acknowledgments This review was begun during study leave from the University of Lethbridge, and improved as a result of comments from R. Anderson, C. Abramson, and P. J. B. Slater. The author

would like to thank D. Griffin for making these ideas respectable, D. L. Mather for making them legible, and C. T. Snowdon for making them comprehensible.

References Ambrose, R. F. (1982). Shelter utilization by the molluscan cephalopod Octopus bimaculatus. Mar. Ecol. 30,261-272. Ambrose, R. F. (1984). Food preferences, prey availability, and the diet of Octopus bimaculatus Vemll. J . Exp. Mar. Biol. Ecol. 77,29-44. Ambrose, R. F., and Nelson, B. V. (1983). Predation by Octopus vulgaris in the Mediterranean. Mar. Ecol. 4,251-261. Anderson, R. C. (1987). Cephalopods at the Seattle Aquarium. Intl. Zoo. Yrbk. 26,41-48. Arnold, E. N. (1988). Caudal autonomy as a defense. In “Biology of the Reptilia” (C. Gans and R. B. Huey, eds.), pp. 235-273. Liss, New York. Arnold, J. M.,and Arnold, K.0. (1%9). Some aspects of hole-boring predation by Octopus vulgaris. Amer. Zool. 9, 991-996. Aronson, R. B. (1986). Life history and den ecology of Octopus briareus in a marine lake. J . Exp. Mar. Bio. Ecol. 95, 37-56. Barth, F. G. (1985). “Insects and Flowers: The Biology of a Partnership.” Princeton Univ. Press, Princeton. Beck, B. B. (1980). “AnimalTool Behavior: The Use and ManufactureofTools by Animals.” Garland, London. Beer, C. (1992). Conceptual issues in cognitive ethology. In “Advances in the Study of Behavior, Vol. 21” (P. J. B. Slater, J. S. Rosenblatt, C. Beer, and M. Milinski, eds.), pp. 69-1 10. Academic Press, New York. Bierens de Haan, J. A. (1926). Versuche uber den farbensinn und das psychische leben von Octopus vulgaris. Z. Vergl. Psychol. 4, 766-796. Bitterman, M. E. (1975). Critical commentary. In “Invertebrate Learning” ( W . C. Corning, ed.), pp. 139-145. Plenum, New York. Boal, J. (1991). Complex learning in Octopus bimaculoides. Amer. Malacol. Bull. 9,75-80. Boal, J. (1994). A review of simultaneous discrimination as a method for training octopuses. Paper presented at the annual meeting of the Animal Behavior Society, Seattle. Bogdany, F. J. (1978). Linking of learning signals in honeybee orientation. Beh. Ecol. Sociobiol. 3, 323-336.

COGNITION IN CEPHALOPODS

347

Boulet, P. C. (1958). Contribution a I’ttude experimentale de la perception visuelle du mouvement chez la perche et la seiche. Mem. MusCum, Paris XVII, 1-131. Boulet, P. C. (1989). Etude en aquarium des strattgies prkdatrices de la seiche (Sepia oficinalis L.). Bull. Inst. Oceanog. Num. Spec. 15, 141-147. Boycott. B. B. (1954). Learning in Octopus uulgaris and other Cephalopods. Pubbl. Staz. Zool. Nap. 25, 1-27. Boycott, B. B., and Young, J. Z. (1957). Reversal of learned responses in Octopus uulgaris Lamarck. Anim. Behau. 6,45-52. Boyle, P. R. (Ed.). (1983). “Cephalopod Life Cycles I-Species Accounts.” Academic Press, London. Boyle, P. R. (Ed.). (1987). “Cephalopod Life Cycles 11-Comparative Reviews.” Academic Press, London. Boyle, P. R. (1991). “The UFAW Handbook on the Care and Management of Cephalopods in the Laboratory.” United Federation for Animal Welfare, Potter’s Bar, United Kingdom. Budelmann, B. U., and Bleckmann, H. (1988). A lateral line analogue in Cephalopods: Water waves generate microphonic potentials in the epidermal head lines of Sepia and Lolliguncula. J . Comp. Physiol. A . 164, 1-5. Budelmann, B. U., Sachse, M., and Staudigl, M. (1987). The angular acceleration receptor system of the statocyst of Octopus uulgaris: Morphometry, ultrastructure, and neuronal and synaptic organization. Phil. Trans. R. Soc. Lond. B 315, 305-343. Bullock, T. H. (1993). How are more complex brains different? Brain Behau. Euol. 41, 88-96. Bullock, T., and Budelmann, B. (1991). Sensory evoked potentials in unanaesthetized unrestrained cuttlefish: A new preparation for brain physiology in cephalopods. J . Comp. Physi01. 168, 141-150. Burghardt, G . M.(1991). Cognitive ethology and critical anthropomorphism: A snake with two heads and hog-nose snakes that play dead. I n “Cognitive Ethology: The Minds of Other Animals” (C. A. Ristau, ed.), pp. 53-90. Erlbaum, Hillsdale. New Jersey. Cheney, D. L., and Seyfarth, R. M. (1990). “How Monkeys See the World: Inside the Mind of Another Species.” Univ. of Chicago Press, Chicago. Cigliano, J. A. (1993). Dominance and den use in Octopus bimaculoides. Anim. Behau. 46, 677-684. Clark, A. B., and Ehlinger, T. J. (1987). Pattern and adaptation in individual behavioral differences. In “Perspectives in Ethology” (P. P. G. Bateson and P. H. Klopfer, eds.), pp. 1-47. Plenum, New York. Clark, D. L., and Uetz, G. W. (1990). Video image recognition by the jumping spider Maeuia inclemens (Aranae: Salticidae). Anim. Behau. 40, 884-890. Crancher, P., King, M.G., Bennett, A., and Montgomery, R. B. (1972). Conditioning of a free operant in Octopus cyaneus Gray. J . Exp. Anal. Behau. 17, 359-362. Davis, H., and Bradford, S. A. (1986). Counting behavior by rats in a simulated natural environment. Ethology 73, 265-280. Dawkins, M. S. (1993). “Through Our Eyes Only?” Freeman, New York. Deutsch, J. A. (1960). The plexiform zone and shape recognition in Octopus. Nature (London) 185,443-445. Dews, P. B. (1959). Some observations on an operant in the octopus. J . Exp. Anal. Behau. 2, 57-63. Dilly, N., Nixon, M..and Packard, A. (1964). Forces exerted by Octopus vulgaris. Pubbl. Staz. Zool. Nap. 34, 86-97. Dubas, F., Hanlon, R. T., Ferguson, G. P., and Pinsker, H. M. (1986). Localization and stimulation of chromatophore motoneurons in the brain of the squid, Lolliguncula breuis. J . EXP.Bi01. U1, 1-25.

348

JENNIFER A. MATHER

Duval, P., Chichery, M. P., and Chichery, R. (1984). Prey capture by the cuttlefish (Sepia oficinalis L.): An experimental study of two strategies. Behau. Proc. 9, 13-21. Dyer, F. C. (1994). Understanding the constraints on spatial cognition in insects. Paper presented at the annual meeting of the Animal Behaviour Society, Seattle. Fentress, J. C. (1992). The covalent animal: On bonds and their boundaries in behavioral research. In “The Inevitable Bond: Examining Scientist-Animal Interactions” (H. Davis and D. Balfour, eds.), pp. 44-71. Cambridge, Univ. Press, Cambridge. Fiorito, G., and Scotto, P. (1992). Observational learning in Octopus uulgaris. Science 254, 545-547. Fiorito, G., von Planta, C., and Scotto, P. (1990). Problem solvingability of Octopus uulgaris Lamarck (Mollusca, Cephalopoda). Behau. Neur. Biol. 53, 217-230. Foyle, T. P., and O’Dor, R. K. (1988). Predatory strategies of squid (Illex illecebrosus) attacking small and large fish. Mar. Behau. Physiol. W, 155-168. Franks, N., Wilby, A., Silverman, B., and Tofts, C. (1992). Self-organizingnest construction in ants: Sophisticated building by blind bulldozing. Anim. Behau. 44, 357-375. Galef, Jr., B. G. (1976). Social transmission of acquired behavior: A discussion of tradition and social learning in vertebrates. In “Advances in the Study of Behavior, Vol. 6” ( J . S. Rosenblatt, R. A. Hinde, E. Shaw, and C. Beer, eds.) pp. 77-100. Academic Press, New York. Gallistel, C. R. (1990). “The Organization of Learning.” MIT Press, Cambridge. Ghiretti, F. (1960). Toxicity of Octopus saliva against crustacea. Ann. N.Y. Acad. Sci. 90, 726-741. Gould, J. L., and Marler, P. (1984). Ethology and the natural history of learning. In “The Biology of Learning” (P. Marler and H.S. Terrace, eds.), pp. 47-74. Springer, Berlin. Graziadei, P. (1964). Muscle receptors in Cephalopods. Proc. R. SOC.Lond. B. 161,392-402. Graziadei, P. (1971). The nervous system of the arms. In “The Anatomy of the Nervous System of Octopus uulgaris” (J. Z . Young, ed.), pp. 45-62. Clarendon Press, Oxford. Griffin, D. R. (1993). “Animal Minds.” UNv. of Chicago Press, Chicago. Hanlon, R. T. (1982). The functional organization of chromatophores and irridescent cells in the body patterning of Loligo plei (Cephalopoda: Myopsida). Malacologia 23, 89-119. Hanlon, R. T. (1987). Mariculture. In “Cephalopod Life Cycles 11” (P. R. Boyle, ed.), pp. 291-305. Academic Press, London. Hanlon, R. T. (1988). Behavioral and body patterning characters useful in taxonomy and field identification of Cephalopods. Malacologia 29, 247-264. Hanlon, R. T., and Messenger, J. B. (1988). Adaptive coloration in young cuttlefish (Sepia oficinalis L.): The morphology and development of body patterns and their relation to behaviour. Phil. Trans. R. SOC. Lond. 320, 437-487. Hanlon, R. T., and Messenger, J. B. (1995). “Cephalopod Behaviour.” Cambridge University Press, Cambridge. Harlow, H. F. (1949). The formation of learning sets. Psych. Reu. 56, 51-65. Hartwick, E. B., Ambrose, R. F., and Robinson, S. M. C. (1984). Den utilization and the movement of tagged Octopus dojeini. Mar. Behau. Physiol. 11, 95-110. Hartwick, E. B., Thorarinsson, G., and Tulloch, L. (1978). Methods of attack by Octopus dojeini (Wiilker) on captured bivalve and gastropod prey. Mar. Behau. Physiol. 5 , 193-200. Haussknecht, T.. and Kuenzler, P. (1990). An experimental study of the building behaviour sequence of a shell-breeding cichlid-fish from Lake Tanganyika (Lamprologus ocellatus). Behaviour 116, 127-142. Herrnstein, R. J. (1990). Levels of stimulus control: A functional approach. Cognition 37, 133-136.

COGNITION IN CEPHALOPODS

349

Hockett, C. F. (1960). Logical considerations in the study of animal communication. I n “Animal Sounds and Communication” (W. E. Lanyon and W. N. Tavogla, eds.), pp. 392-430. Amer. Inst. Biol. Sci., Washington, D.C. Hubel, D. H., and Wiesel, T. N. (1%2). Receptive fields, binocular interact and functional architecture in the cat’s visual cortex. J . Physiol. 160, 106-154. Humphrey, N. K. (1976). The social function of intellect. In “Growing Points in Ethology” (P. P. G. Bateson and R. A. Hinde, eds.), pp. 303-317. Cambridge Univ. Press, Cambridge. Iribarne, 0. 0. (1990). Use of shelter by the small Patagonian octopus Octopus tehuelchus: Availability, selection and effects on fecundity. Mar. Ecol. 66, 251-258. Kier, W. M., and Smith, A. M. (1990). The morphology and mechanics of octopus suckers. Biol. Bull. 178, 126-136. Lederman, S. J., Browse, R. A., and Klatzky, R. L. (1988). Haptic processing of spatially distributed information. Percep. Psychophys. 44,222-232. Mackintosh, M. J., and Mackintosh, J. (1%3).Reversallearning in Octopus vulgaris Lamarck with and without irrelevant cues. Q . J. Exp. Psych. 15, 236-242. Mackintosh, M. J., and Mackintosh, J. (1964a). Performance of Octopus over a series of reversals of a simultaneous discrimination. Anim. Behau. U,321-324. Mackintosh, M. J., and Mackintosh, J. (1964b). The effect of overtraining on a non-reversal shift in Octopus. J . Genet. Psychol. 106, 373-377. Maldonaldo, H. (1963). The visual attack learning system in Octopus vulgaris. J . Theoret. Biol. 5, 470-488. Mansfield, R. (1974). Neural basis of orientation perception in primate vision. Science 186, 1133-1 135. Mason, W. A. (1984). Learning: Physiology and behavior. Verh. Dtsch. Zool. Ges. 77, 45-56. Mather, J. A. (1980). Social organization and use of space by Octopus joubini in a seminatural situation. Bull. Mar. Sci. 30, 848-857. Mather, J. A. (1982a). Choice and competition: Their effects on occupancy of shell homes by Octopus joubini. Mar. Behav. Physiol. 8, 285-293. Mather, J. A. (1982b). Factors affecting the spatial distribution of natural populations of Octopus joubini Robson. Anim. Behav. 30, 1166-1 170. Mather, J. A. (1984). Development of the behaviour of Octopus joubini Robson. Vie Mileu 34, 17-20. Mather, J. A. (1986). Sand digging ofjuvenile Sepia officinalis:Assessment of a cephalopod mollusc’s “fixed” behavior pattern. J . Comp. Psychol. 100, 315-320. Mather, J . A. (1991a). Foraging, feeding and prey remains in middens ofjuvenile Octopus vulgaris. J . Zool. (Lond.) 224, 27-39. Mather, J. A. (1991b). Navigation by spatial memory and use of landmarks in octopuses. J . Comp. Physiol. A . 168, 491-497. Mather, J. A. (1992). Interaction ofjuvenile Octopus vulgaris ‘with scavenging and territorial fishes. Mar. Behav. Physiol. 19, 175-182. Mather, J. A. (1994). “Home” choice and modification by juvenile Octopus vulgaris: Specialized intelligence and tool use? J . Zool. (Lond.) 233, 359-368. Mather, J. A., and Anderson, R. C. (1993). “Personalities” of octopuses (Octopus rubescens). J . Comp. Psychol. 107, 336-340. Mather, J. A., and Anderson, R. C. (1994). Octopuses and clams: The packagingproblern. Paper presented at the CIAC Conference “The behaviour and natural history of Cephalopods.” Vico Equense, Italy. Mather, J. A., and Anderson, R. C. Species-typical patterns of octopus drilling in bivalve shells (manuscript in preparation).

350

JENNIFER A. MATHER

Mather, J. A., and Mather, D. L. (1994). Skin colours and patterns of juvenile Octopus vulgaris in Bermuda. Vie Mileu. 44, 267-272. Mather, J. A,, and O’Dor. R. K. (1984). Spatial organization of schools of the squid Illex illecebrosus. Mar. Eehau. Physiol. 10,259-271. Mather, J . A,, and O’Dor, R. K. (1991). Foraging strategies and predation risk shape the natural history of juvenile Octopus uulgaris. Bull. Mar. Sci. 49, 256-269. Mather, J . A., Resler. S., and Cosgrove, J. (1985). Activity and movement of Octopus dofleini. Mar. Eehau. Physiol. 11, 301-314. Menzel, R. (1985). Learning in honey bees in an ecological and behavioral context. In “Experimental Behavioral Ecology” (B. Holldobler and M. Lindauer, eds.), pp. SS74. Fisher-Verlag, Stuttgart. Messenger, J. B. (I%@. The visual attack of the cuttlefish, Sepia oflcinalis. Anim. Eehau. 16,342-357. Messenger, J. B. (1973a). Monocular discrimination of mirror images in Octopus. Pubbl. Staz. Zool. Nap. 36, 103-1 11. Messenger, J. B. (1973b). Learning performance and brain structure: A study in development. Brain Res. 58, 519-523. Messenger, J. B., and Sanders, G. D. (1972). Visual preference and two-cue discrimination learning in Octopus. Anim. Eehau. 20,580-585. Messenger, J. B., Wilson, A. P., and Hedge, A. (1973). Some evidence for colour-blindness in Octopus. J . Exp. Eiol. 59, 77-94. Moody, M. F., and Pamss, J. R. (1960). The discrimination of polarized light by Octopus. Nature (London) 186,839-840. Moynihan, M. (1975). Conservatism of displays and comparable stereotyped patterns among Cephalopods. In “Function and Evolution in Behavior” ( G . Baerends, C. Beer, and A. Manning, eds.), pp. 276-292. Clarendon, Oxford. Moynihan, M. (1985). “Communication and Noncommunication by Cephalopods.” Indiana Univ. Press, Bloomington. Moynihan, M., and Rodaniche, A. F. (1982). The behavior and natural history of the Caribbean reef squid, Sepioreuthis sepioidea. In “Advances in Ethology,” Vol. 25, pp. 1-150. Verlag Paul Parey, Berlin. Muntz, W. R. A. (1961). Interocular transfer in Octopus vulgaris. J . Cornp. Physiol. Psych. 54,49-55. Muntz, W. R. A. (1963). lntraretinal transfer and the function of the optic lobes in Octopus. Q. J . EXP.Psych. 15, 116-124. Muntz, W. R. A. (1970). An experiment on shape discrimination and signal detection in octopus. Q. J . Exp. Psych. 2 2 , 8 2 4 0 . Muntz, W. R. A., and Gwyther, J. (1988). Visual acuity in Octopus pallidus and Octopus australis. J . Exp. Eiol. W, 119-129. Neisser, U. (1976). “Cognitive Psychology.” Appleton-Century-Crofts, New York. O’Dor. R. K., and Webber, D. M. (1986). The constraints on Cephalopods: Why squid aren’t fish. Can. J . Zool. 64, 1591-1605. O’Dor, R. K., and Wells, M. J. (1987). Energy and nutrient flow in cephalopods. I n “Cephalopod Life Cycles 11” (P. R. Boyle, ed.), pp. 109-133. Academic Press, London. Olton, D. S., and Samuelson, R. J. (1976). Remembrances of places passed: Spatial memory in rats. J . Exp. Psychol. Anim. Eeh. Proc. 2, 97-116. Packard, A. (1972). Cephalopods and fish: The limits of convergence. Biol. Rev. 47,241-307. Packard. A. (1988a). Visual tactics and evolutionary strategies. In “Second International Cephalopod Symposium: Cephalopods Present and Past” (J. Wiedmann and J. Kullman, eds.), pp. 89-103. Schweizerbart’sche Verlagsbuchhandlung, Stuttgart.

COGNITION IN CEPHALOPODS

351

Packard, A. (1988b). The skin of cephalopods (coleoids): General and special adaptations. In “The Mollusca 11-Form and Function” (E. R. Trueman and M. R. Clarke, eds.), pp. 37-67. Academic Press, San Diego. Packard, A., and Sanders, G. (1971). Body patterns of Octopus uulgaris and maturation of the response to disturbance. Anim. Behau. 19, 780-790. Papini, M. R., and Bitterman, M. E. (1991). Appetitive conditioning in Octopus cyanea. J . Comp. Psychol. 105, 107-114. Parriss, J. R. (1963). Interference in learning and lesions in the visual system of Octopus uulgaris. Behaviour 21, 233-243. Pepperburg, 1. M. (1994). Numerical competence in an African gray parrot (Psittacus erirhacus). J . Comp. Psychol. 108, 36-44. Piaget, J. (1962). “Play, Dreams and Imitation in Childhood.” Norton, New York (originally published in 1945). Povinelli, D. J.. Rulf, A. B., Landau, K. R., and Bierschwale, D. T. (1993). Self-recognition in chimpanzees (Pan troglodytes): Distribution, ontogeny and pattern of emergence. J . Comp. Psychol. lM, 347-372. Remy, M., and Watanabe, S. (1993). Two eyes and one world: Studies of interocular and intraocular transfer in birds. I n “Vision, Brain, and Behavior in Birds” (H. P. Ziegler and H.-J. Bischof, eds.), pp. 333-350. MIT Press, Cambridge. Ristau, C. (1991). Aspects of the cognitive ethology of an injured-feigning bird, the piping plover. I n “Cognitive Ethology: The Minds of Other Animals” (C. A. Ristau, ed.), pp. 91-126. Erlbaum, Hillsdale, New Jersey. Robinson, M. H. (1990). Predator-prey interactions, informational complexity, and the origins of intelligence. I n “Insect Defenses” (D. L. Evans and J. 0. Schmidt, eds.), pp. 129-149. State University of New York Press, Albany. Ronacher, B. (1992). Influence of unrewarded stimuli on the classification of visual patterns by honey bees. Ethology 92,205-216. Ross, D. M. (1970). Protection of hermit crabs (Dardanus spp.) from octopus by commensal sea anemones (Calliacrus spp.). Nature (London) 230,401-402. Rowell, C. H. F. (1966). Activity of interneurons in the arm of Ocropus in response to tactile stimulation. J . Exp. B i d . 44, 589-605. Sanders, F. K.. and Young, J. Z. (1940). Learning and other functions of the higher nervous centre of Sepia. J . Neurophysiol. 3, 501-526. Sanders, G. D. (1975).The Cephalopods. In “Invertebrate Learning, Vol. 3” (W. C. Corning, ed.), pp. 1-101. Plenum, New York. Saunders, W. B., and Landsman, N. H. (Eds.). (1988). “Nautilus: The Biology and Paleobiology of a Living Fossil.” Plenum, New York. Schiffman, H. R. (1990). “Sensation and Perception: An Integrated Approach.” Wiley, New York. Spetch, M. L., and Edwards, C. A. (1988). Pigeons’, Columbia liuia, use of global and local cues for spatial memory. Anim. Behau. 36,293-3%. Stephens, D. W., and Krebs, J. R. (1986). “Foraging Theory.” Princeton Univ. Press, Princeton. Suomi, S. J. (1991). Uptight and laid-back monkeys: Individual differences in the response to social changes. I n “Plasticity of Development” (S. E. Brauth, W. S. Hall, and R. J. Dooling, eds.), pp. 27-56. MIT Press, Cambridge. Sutherland, N. S. (1957). Visual discrimination of orientation and shape by the octopus. Br. J . Psychol. 40, 55-71. Sutherland, N. S. (1960). Theories of shape discrimination in Octopus. Nature (London) 186,840-844.

352

JENNIFER A. MATHER

Sutherland, N. S. (1%3). Shape discrimination and receptive fields. Nature (London) 197, 1 18- 122. Sutherland, N. S., Mackintosh, N. J., and Mackintosh, J. (1%3). Simultaneous discrimination training of octopus and transfer of training along a continuum. J. Comp. Psychol. 56, 150-156. Sutherland, N. S . , Mackintosh, N. J., and Mackintosh, J. (1%5). Shape and size discrimination in Octopus: The effects of pretraining along different dimensions. J. Genetic Psychol. 106, 1-10. Ten Cate, J. (1928).Contribution a I’innervation des ventouses chez Octopus vulgaris. Arch. Neerl. Physiol. Homme Anim. W, 407-422. Thomas, R.T. (1980).Evolution of intelligence: An approach to its assessment. Brain Behav. Evol. 17, 454-472. Von Boletzky. S. (1992).Evolutionary aspects of development, life style and reproductive mode in incirrate octopods (Mollusca, Cephalopoda). Rev. Suisse Zool. 99, 755-770. Von Boletzky, S.,and Von Boletzky, M. V. (1970).Das eingraben in sand bei Sepiola und Sepietta. Rev. Suisse Zool. 77,536-548. von Frisch, K. (1967).“The Dance Language and Orientation of Bees.” Harvard Univ. Press, Cambridge. Walker, J. J., Longo, N., and Bitterman, M. E. (1970).The octopus in the laboratory. Handling, maintenance, training. Beh. Res. Meth. & Instru. 2, 15-18. Watanabe, S., Lea, S. E. G., and Dittrich, W. H. (1993).What can we learnfromexperiments on pigeon concept discrimination? In “Vision, Brain, and Behavior in Birds” (H. P. Ziegler and H.-J. Bischof, eds.), pp. 351-376. MIT Press, Cambridge. Wells, M. J. (1958).Factors affecting reactions to Mysis by newly hatched Sepia. Behaviour l3,%-111. Wells, M. J. (1961).Weight discrimination by Octopus. J. Exp. Biol. 38, 127-133. Wells, M. J. (1964a). Tactile discrimination of surface curvature and shape by the octopus. J . Exp. Biol. 41,433-445. Wells, M. J. (1964b). Detour experiments with octopuses. J. Exp. Bio. 41, 621-642. Wells, M. J. (1%7). Short-term learning and interocular transfer in detour experiments with octopuses. J. Exp. Biol. 47, 393-408. Wells, M. J. (1978).“Octopus: Physiology and Behaviour of an Advanced Invertebrate.’’ Chapman and Hall, London. Wells, M. J. (1990).Oxygen extraction and jet propulsion in cephalopods. Can. J. Zool. 68,815-824. Wells, M. J., and Wells, J. (1957).The function of the brain of Octopus in tactile discrimination. J. Exp. Biol. 34, 378-393. Wells, M. J., and Young, J. Z. (1966). Lateral interaction and transfer in the tactile memory of the octopus. J. Exp. Biol. 45, 383-400. Wells, M. J., and Young, J. 2.(1970).Stimulus generalization in the tactile system of Octopus. J . Neurobiol. 2, 31-46. Wodinsky, J. (1%9). Penetration of the shell and feeding on gastropods by Octopus. A m . ZOO^. 9,997-1010. Wodinsky, J. (1973).Mechanism of hole boring in Octopus vulgaris. J. Gen. Psychol. 88, 179-183. Wodinsky, J. (1978).Feeding behaviour of broody female Octopus vulgaris. Anirn. Behau. 26,803-813. Yarnall, J. (l%9). Aspects of the behaviour of Octopus cyanea Gray. Anim. Behau. 17, 747-754. Yoerg, S. I., and Kamil, A. (1991). Integratingcognitive ethology with cognitive psychology.

COGNITION IN CEPHALOPODS

353

I n "Cognitive Ethology: The Minds of Other Animals" (C. A. Ristau, ed.), pp. 273-289.

Erlbaum, Hillsdale, New Jersey. Young, J. Z. (1965). The organization of a memory system. Proc. R. Soc. B 163,285-320. Young, J. Z. (1971). "The Anatomy of the Nervous System of Octopus uulgaris." Clarendon, Oxford. Young, J. Z. (1991). Computation in the learning system of Cephalopods. Biol. Bull. 180, 200-208. Zhang, S.W., and Srinivasan, M. V.(1994). Prior experience enhances pattern discrimination in insect vision. Nature (London) 368, 330-332.

This Page Intentionally Left Blank

Index

A Aardwolves, 212 Acrosome, 302 Adaptation, sex differences, 48-49 Aequidens pulcher, 159 African wild dog (Lycaon pictus), 203 group hunting, 219, 220 home range, 204 intraspecific fights, 218 phylogenetic regression, 234 reproduction breeding group, 207, 224 litter size, 223 reproductive suppression, 209-21 1 Agelaius phoeniceus (red-winged blackbird), 24 Aggression, threat display, 272, 277-278 AGL, see Anogenital licking Agonostomus monticola, 160 Ailurids, 204. 235 Ailuropodids, 204 Alloparental care, carnivores, 223-224 Allozyme divergence, guppies, 179-180 Androgens, fetal, maternal behavior, 57-59,93 Androstenedione, masculinized genitalia, 64 Anogenital licking, reproductive development, 78-89,93-97, 104 Antipredator behavior birds, communal roosts, 3 carnivores, 222-223 guppies, 161-163 phenotypic plasticity, 177 schooling, 161-162, 167, 174, 176, 185-187, 189 Ants, information transfer, 35-36 Apis spp., information transfer, 36 Arctic fox, 215

Arctic ground squirrel (Spermophilus SPP.), 177 Ardeids, information transfer, 29 Aromatase, masculinized genitalia, 64 Astyanax bimaculatus, 159 Auditory information processing, cephalopods, 320, 327-329 Auditory stimuli, reproductive behavior, 74

B Badgers home range, 204, 217, 218 reproduction, 224 off-territory copulation, 205 reproductive suppression, 209, 210, 225 social breeding, 215 Bambara spp., reproductive anatomy, 300 Banded mongoose (Mungos mungo) intraspecific fighting, 218 phylogenetic regression, 237 reproduction breeding group, 208, 227 reproductive suppression, 210, 225 sociality, 216, 222,224 Bank swallow, information transfer, 28 Barn swallow (Hirundo rustica), information transfer, 28 Bat-eared fox (Octocyon megalotis), 207, 234 Bats, information transfer, 26-27 Bears, 205, 235 Bees information transfer, 36-37 vision, 320 Beetles, reproductive anatomy, 300 Behavior learned, see Learned behavior 355

INDEX

Behavioral ecology, 156, 184-194 Biased mutation, 295 Binturong (Nandinia binotata), 203, 237 Biological diversity, 155-156 guppies, evolution in the wild, 172-177, 181-183 Biparental care, carnivores, 206, 223 Bird roosts, see Communal roosts Birds biparental care, 223 colonies, 26 cooperative breeding, 225 group living, 2 habitat saturation, 217 information center hypothesis, 1-39 information transfer, colonial feeding, 27-29, 30-35 reproductive anatomy, 300 reproductive behavior, 49-5 I , 66, 68, 74 seabirds, 26 songbirds, 49-50 Bivalves, predation by octopus, 342 Blackbacked jackals, 207, 219, 221 Black-billed gull (Larus bulleri), information transfer, 32 Blackbirds feeding, 3, 24 information transfer, 34-35 Black-headed gull (Larus ridibundus), information transfer, 32 Black rat ( R a m s r a m s ) , pine cone feeding, 119-152 Black-tipped mongoose (Herpestes sanguineus) phylogenetic regression, 237 sociality, 205,216 Black vulture (Coragyps atratus), information transfer, 33, 34 Blanford's fox (Vulpes cana), 205, 215 Bluegill sunfish (Lepornis rnacrochirus), mating, 191 Brain, octopus, 319-320, 325, 330-332 Brambling (Fringilla rnontifringilla) food scarcity, 24 Breeding communal, 207 cooperative, 225 multiple mating, 291-307 Breeding group size, carnivores, 203-208, 213, 242-243

Brewer's blackbird (Euphagus cyanocephalus), group feeding, 3 Brown bear (Ursus arcros), 205, 235 Brownheaded cowbird (Molorhrus ater), food scarcity, 24 Brown hyena, 205, 215, 224 Bulbospongiosus muscles, 97

C Carnponotus spp., information transfer, 35 Canids phylogenetic regression, 234 reproduction breeding temtory, 218 litter size, 223 reproductive suppression, 2 10 social breeding, 205-207 Canis aureus (golden jackal) breeding, 204, 207 grouping, 216, 219 home range, 204 phylogenetic regression, 234 Canis latrans (coyote) breeding group, 207 group hunting, 219 phylogenetic regression, 234 Canis rnesornelas (silverbacked jackal) breeding group, 207, 224 phylogenetic regression, 234 Canis sirnensis (Ethiopian wolf) breeding group, 205, 207, 212, 224 intraspecific fighting, 218, 222 phylogenetic regression, 234 Carduelis chloris (greenfinch), information transfer, 33 Carnivores, 203-204 breeding, 204-208, 213-225 evolution, 213-228 phylogenetic regression analysis, 204, 228-247 reproductive suppression, 208-2 12, 225-228 social organization, 204-212 Canion crow (Cornus corone corone) group foraging, 23 information transfer, 32 Cathartes aura (turkey vulture), information transfer, 34

INDEX

Cephalopods cognition, 317-346 communication, 335-338, 346 nervous system, 319 Chaffinch (Fringilla coelebs), food scarcity, 24 Characins, guppy predator, 159, 188 Cheetahs, 218, 221, 222 Chickens, communication, 271 Chimpanzees learned feeding technique, 149 mating and anatomy, 300-302 metacommunication, 275 relationships, 271 Cichlids, guppy predator, 159, 185 Ciconiiformes, colonies, 26 Circadian rhythm, intrauterine, 56 Citral, 70-71 Clams, predation by octopus, 342 Cliff swallow (Hirundo pyrrhonofa) colony size, 25 information transfer, 28 Cognition, cephalopods, 317-346 Coho salmon (Oncorhynchus kisufch), mating, 191 Coleoid cephalopods, 319, 322 Coloniality colony size, 24-25 cooperation, 15-21 group foraging, 7-14, 22-23 information center hypothesis, 4-39 information transfer, 7, 10-1 1, 15, 20-39 waiter-searcher model, 13 information transfer, 25-26 birds, 27-29, 30-35 mammals, 26-27, 29-30 social hymenopterans, 35-40 reproductive success, 25 synchronous departure, 22, 26 Color display, guppies, 164-166 Common grackle (Quiscalus quiscala), food scarcity, 24 Common tern (Sterna hirundo), information transfer, 28-29 Communal breeding, 207 Communal roosts, 1-4, 24 information center hypothesis, 4-39 Communication, 264, 271; see also Information transfer cephalopods, 335-338, 346

357

coregulation, 272-276 emotions, 271-272 framing, 273-276, 279 metacornmunication, 275 mother-infant relationships, 259-286 Conditioning experiments, octopus, 333 Consensual reframing, 279 Cooperation information center, 5-7, 38-39 models, 15-21 reciprocal, 5-7, 15-16, 38-39 Coooperative breeding, birds, 225 Copulation birds, 49-50 extra-pair copulation, 205 guppies, 164-166, 170-172, 181, 182 multiple mating, 291-307 off-territory copulation, 205, 212 rats intromission, 94-95 juvenile play, 93 maternal anogenital licking, 94-97, 103-105 maternal stress affecting, 59-60 neural components, 100 restricted rearing, 74 sex differences, 49-5 1 restricted rearing, 72-75 Coragyps atratus (black vulture), information transfer, 33, 34 Coregulation, communication, 272-276 Corticosteroids, maternal, prenatal stress, 60 Corvus C o r a (raven), information transfer, 31-32 Corvus corone cornix (hooded crow), information transfer, 3 1 Corvus corone corone (carrion crow) group foraging, 23 information transfer, 32 Cotton-top tamarin (Saguinus oedipus), play behavior, 73 Courtship, risk and, 163-165 Cowbirds, 24, 271 Coyote (Canis larrans) breeding group, 207 group hunting, 219 phylogenetic regression, 234 Cremaster nucleus, 98 Crenicichla alta, guppy predator, 159-163, 166-170, 173, 175, 178, 185

358

INDEX

Crocuta crocuta (spotted hyena) group hunting, 219-222 juvenile play behavior, 92 masculinized genitalia, 64 phylogenetic regression, 236 reproductive success, 224 reproductive suppression, 208, 209 social organization, 214 Crossarchus spp. (cusimanse), 208, 216 Crossbill (Loxia curuirostra), pine cone feeding, 119 Cross-fostering, 66, 152 Crows group foraging, 23 information transfer, 31-32 Cultural transmission feeding methods chimpanzees, 149 mice, 147 rats, 119-152 Cuon alpinus (dhole) breeding group, 207 group hunting, 221-222 phylogenetic regression, 234 Cusimanse (Crossarchus spp.), 208, 216 Cuttlefish (Sepia spp.) brain, 330 concealment by burying, 340 predation habits, 331-332, 345 skeleton, 320 skin color changes, 323, 337 Cypress tree, pine cone feeding by black rats, 145-146

D Dance language, bees, 36-37 Deer, differential rearing, 78 Defense harem polygyny, 207 Dhole (Cuon alpinus) breeding group, 207 group hunting, 221-222 phylogenetic regression, 234 Diet selection model, birds, 24 Differential rearing, reproductive development, 75-82 Dimethyl disulfide, 71 Dingoes, 209, 234 Display of colors, guppies, 164-166

Diversity, see Biological diversity DLN, see Dorsolateral nucleus Dodecyl propionate, 82-83 Dogs wild dogs, 203 breeding group, 207-21 1, 224 group hunting, 219, 220 home range, 204-205 intraspecific fights, 218 litter size, 223 phylogenetic regression, 234 Dorsolateral nucleus (DLN), 97 Drosophila melanogaster, 297-298 Ducks, reproductive stimuli, 74 Dwarf mongoose (Helogale paruula) phylogenetic regression, 237 reproduction breeding group, 207, 208 reproductive suppression, 210-212 sociality, 216, 222, 224

E Egrets, information transfer, 29 Eleotris pisonis, guppy predator, 160 Emotions, communication, 27 1-272 Enhydra lutra (sea otter), 225, 236 Episodic metacommunication, 275 Equifinality, 52-53 Estrogen, reproductive development, 90 Ethiopian wolf (Canis simensis) breeding group, 205, 207, 212, 224 intraspecific fighting, 218, 222 phylogenetic regression, 234 Ethology, 155-156 Euphagus cyanocephalus (Brewer’s blackbird), group feeding, 3 Eurasian badger, 204, 205, 209, 215 European minnow (Phoxinus phoxinus), antipredator responses, 177 Evening bat (Nycticeius humeralis), information center, 26 Evolution biodiversity, 155-156 carnivores, reproductive suppression, 213-225, 225-228 guppies, 172-177, 181-183 Exclusive territorial polygyny, 207

359

INDEX

Extra-pair copulation, 205 Eye, cephalopods, 319-320

F Feeding birds colony-based group foraging, 8-1 1 diet selection model, 24 information center hypothesis, 4-39 learned behavior chimpanzees, 149 mice, 147 rats, 119-152 rats, pine cone feeding, 119-152 Felids diet, 204 mating habits, 206, 208 phylogenetic regression, 234-235 Females birds, mounting and lordosis, 49-51 carnivores, reproductive suppression, 208-209 fish, foraging habits, 169-170 mammals, reproductive development, 47-105 multiple mating, 291-307 primates, mounting, 73 rats differential rearing, 75-82 juvenile play behavior, 73, 91-94 maternal anogenital licking, 78-89, 93-94, 97 Fennec fox, 208 Ferrets, heterosexual stimuli, 71 Fertilization efficiency, 295-298, 302 Fetal development, reproductive, 55-64,93 Finches food scarcity, 24 information transfer, 33 Fish guppies, 156-158, 179 behavioral diversification, 156, 184-194 copulation, 164-166, 170-172, 181, I82 evolution, 172- 177, 181-183 genetic divergence, 178-184

male coloration, 165-166, 172-174 predation, 159- 177 speciation, 156, 183-184, 191-192 sticklebacks, 162, 177, 191-192 Fissiped carnivores, 203 Fitness bias model, 225-228 Flash expansion, guppies, 161 Food patches fish, defense, 167, 185-186 information center hypothesis, 4, 9-12, 23 renewal, 216 Foot-clasp mount, primates, 73 Foraging group foraging, 22-25 carnivores, 205-206 colony-based, 7-14 information center hypothesis, 22-25 synchronous departure, 22, 26 guppies, sex differences, 169-170 information center hypothesis, 4-39 models, 15-21 octopus, 325, 343 Fountain maneuver, guppies, 161 Foxes home range, 204 hunting, 218, 222 phylogenetic regression, 234 reproductive suppression, 209, 21 I , 212 social breeding, 205, 207, 215, 216, 224 Framing, 273-276, 279 Freshwater prawns (Macrobrachium spp.), guppy predator, 160, 161, 163, 166 Fringilla coelebs (chaffinch), food scarcity, 24 Fringilla montifringilla (brambling), food scarcity, 24 Fulmars, group feeding, 2 G

Gambusia afinis (mosquitofish), 169 Gasterosteus aculeatus (three-spined stickleback) antipredator activities, 162 evolution, 177 population differentiation, 191-192 Genetics biased mutation, 295

360

INDEX

divergence of populations, 178-184, 188-194 maternal effects, 54-55 multiple mating, 291-293, 297, 307 reproductive suppression, 2 I 1-2 12 sex differences, 48-49 Genitalia chimpanzee, 300-302, 304 penis, variation, 97-98, 304 primates, inspection in newborn, 76-77 rats, penile function, 97-98 spotted hyena, masculinized, 64 Gerbils reproductive development fetal androgens, 57-58 maternal anogenital licking, 81-82 uterine position, 61, 89-90 Gestation period reproductive development, 104-105 fetal testosterone, 57-59, 60-63, 88 maternal stress, 59-60, 89, 94 sexual divergence, 63-64 uterine position, 60-63, 89-90 Gobies, guppy predator, 160 Gobiomorus dormitor, guppy predator, 160 Golden jackal (Canis aureus) breeding, 204, 207 grouping, 216, 219 home range, 204 phylogenetic regression, 234 Gonadal steroids reproductive development, 54, 88-90, 105 androgens, fetal, 57-59, 93 heterosexual stimuli, 71-72 maternal stress, 59-60 testosterone, fetal, 57-59, 105 uterine position, 60-63, 89-90 Gonopodial thrusting, guppies, 164-165 Grackles, food scarcity, 24 Granger causation, 227 Great grey mongoose (Herpestes ichneumon), 216 Greenfinch (Carduelis chloris), information transfer, 33 Grey wolf, 205 Griffon vulture (Gyps fulvus), group living, 2 Group foraging, 22-25 carnivores, 205-206 colony-based, 7- 14

models, 15-21, 38-39 synchronous departure, 22, 26 waiter-searcher model, 13 Group hunting, carnivores, 219-220 Group living, birds, 2 Gulls, information transfer, 32 Guppy (Poecilia reticulata), 156-158, 179 behavioral diversification, 156, 184-194 copulation, 164-166, 170-172, 181, 182 evolution, 172- 177, 181- 183 genetic divergence, 178-184 male coloration, 165-166, 172-174 predation, 159-177 speciation, 156, 183-184, 191-192 Gyps fulvus (griffon vulture), group living, 2

H Habitat saturation, 217 Hamsters reproductive anatomy, 300 reproductive behavior, 68-69 reproductive development, 63-64 Haptics, 328 Helogale parvula (dwarf mongoose) phylogenetic regression, 237 reproduction breeding group, 207, 208 reproductive suppression, 210-212 sociality, 216, 222, 224 Herons, information transfer, 29 Herpestes ichneumon (great grey mongoose), 216 Herpestes sanguineus (black-tipped mongoose) phylogenetic regression, 237 sociality, 205, 216 Herpestids home range, 204 phylogenetic regression, 237 sociality, 204-205, 222 Heterochrony, 54, 63 Heterosexual stimuli, 71-72 Hirundo pyrrhonota (cliff swallow) colony size, 25 information transfer, 28 Hirundo rustica (barn swallow), information transfer, 28 Home range, 204, 206, 228

36 1

INDEX

Honeybees, information transfer, 36-37 Hooded crow (Cornus corone cornix), information transfer, 3 1 Hoplias rnalabaricus, guppy predator, 159, 188 Hormones, see also specific hormones reproductive development, 54, 88-90 androgens, fetal, 57-59,93 heterosexual stimuli, 71-72 maternal stress, 59-60 testosterone, fetal, 57-59, 105 uterine position, 60-63, 89-90 reproductive suppression, 208-209 Human infant, development and relationships, 259-286 Hunting, carnivores, 219-222 Hyaenids, 205, 208, 238 Hyenas breeding group, 208 hunting, 219-221 masculinized genitalia, 64 phylogenetic regression, 238 play behavior, 92 sociality, 214, 215, 224 Hymenoptera spp., information transfer, 35-40 I Ichneumia albicauda (white-tailed mongoose) phylogenetic regression, 237 sociality, 205, 208, 216 Illex spp., 341 Imprinting, 66 Information center, 1-39 colony-based group foraging, 7- 14, 2 1-25 cooperation, 15-21 models, 15-21 Information processing, cephalopods, 317-346 Information transfer, see also Communication coloniality, 7, 10-1 I, 15, 20-39 colonies, 25-26 birds, 27-29, 30-35 insects, 35-37 mammals, 26-27, 29-30 social hymenopterans, 35-40

Insects information transfer, 35-37 reproductive anatomy, 300 sperm precedence, 296 Insulting, communication as part of, 272 Intrauterine environment, reproductive development in mammals, 55-64 Intromission chimpanzees, 300-302 guppies, 164 rats, 94-95, 104 Ischiocavernosus muscles, 97 Isolation, sexual development, 74

J Jackals breeding group, 207 home range, 204 phylogenetic regression, 234 sociality, 216, 224 Jaguar (Punthera onca), 222 Jerusalem pine, pine cone feeding by black rats, 119-152

K Kinkajou (Potos Pauus), 204, 235 Kin selection model, group foraging, 18-21, 38-39 Kissing, communication as part of, 272, 273, 276

L Large maned wolf,207 Larus bulleri (black-billed gull), information transfer, 32 Larus ridibundus (black-headed gull), information transfer, 32 Learned behavior cephalopods, 317-346 chimpanzees, 149 mice, 147 rats, 119-152 Learning, 330-332, 335 Leopard (Panihera pardus), 222

INDEX

Lepomis macrochirus (bluegill sunfish), mating, 191 Leprorhorux acervorurn, information transfer, 35 Licking, see Anogenital licking Lions, 225 group hunting, 219-222 intraspecific fighting, 218 reproductive suppression, 210-212 Little brown bat (Myoris lucijiugus), information center, 27 Local enhancement, 2-5, 16 Loligo plei, 338 Lolliguncula, 330 Long-Evans rats copulatory performance, 104 differential rearing, 78 maternal anogenital licking, 78-82, 101. 104 Lordosis, 49-51 Loxia curvirostra (crossbill), pine cone feeding, 119 Lutra lurra (otter), 215, 225, 236 Lycaon picrus (African wild dog), 203 group hunting, 219, 220 home range, 204 intraspecific fights, 218 phylogenetic regression, 234 reproduction breeding group, 207, 224 litter size, 223 reproductive suppression, 209-21 1

M Macaques reproductive development differential rearing, 77 fetal androgens, 57 Macrobrachiurn crenularurn, guppy predator, 161 Macrobrachiurn spp. (freshwater prawns), guppy predator, 160, 161, 163, 166 Males birds, mounting and lordosis, 49-51 carnivores, reproductive suppression, 209 fish, foraging habits, 169-170 guppies, coloration, 165-166, 172-174 mammals, reproductive development, 47-105

multiple mating, 29 1-307 primates, foot-clasp mount, 73 rats differential rearing, 75-82 juvenile play behavior, 73, 91-94 maternal anogenital licking, 78-89, 93-97, 104 Mammals information transfer, 26-27, 29-30 paternal, 53 reproductive anatomy, 300, 302 reproductive development anogenital licking, 78-89, 93-97, 104 maternal stimulation, 91-102 offspring reproductive success, 103-104 restricted rearing, 72-75 sex-related biases, 75-91, 105 sexual divergence, 63-64 social stimuli, 66-72 sperm precedence, 2% Masculinized genitalia, spotted hyena, 64 Mate choice fish, 166, 170, 181 rats, olfactory stimuli, 66-72 Maternal contributions feeding behaviors, learned chimpanzees, 149 mice, 147 rats, 119-152 genetics, 54-55 mother-infant relationship, relational perspective, 260-286 reproductive development, rats, 65, 104-105 anogenital licking, 78-89,93-97, 104 maternal stimulation, 91-102 offspring reproductive success, 102-104 postnatal, 65-102 prenatal, 55-64 sex-related biases, 75-91, 105 social stimuli, 66-72 restricted rearing, 72-75 Maternal stress, reproductive development, 59-60, 89.94 Mating, see Reproductive development; Sexual behavior 6-MBOA. 58 Meadow vole (Microtus pennsylvanicus), reproductive behavior, 52, 53

363

INDEX

Medroxyprogesterone acetate (MPA), reproductive development, 65, 88-90 Meerkat (Suricafasuricaffa) phylogenetic regression, 237 sociality, 208, 218, 222 Melipona spp., information transfer, 36 Mesocricetus auratus (Syrian hamster), reproductive behavior, 69 Mesocricetus brandti (Turkish hamster), reproductive behavior, 69 Metacommunication, 275 Mice learned feeding technique, 147 maternal anogenital licking, 82 reproductive behavior, 67 reproductive development, 59-61 sperm characteristics, 297, 303 Microtus ochrogaster (prairie vole), reproductive behavior, 52, 53, 58 Microtus pennsyluanicus (meadow vole), reproductive behavior, 52, 53 Milk, hormones in, 65 Minnows antipredator responses, 177 inspection behavior, 162 Molluscs, cephalopods, 317-346 Molothrus ater (brownheaded cowbird), food scarcity, 24 Mongolian gerbils reproductive development fetal androgens, 57-58 maternal anogenital licking, 81-82 uterine position, 61 Mongooses intraspecific fighting, 218 phylogenetic regression, 237 reproduction breeding group, 208, 227 reproductive suppression, 210-2 12, 225 sociality, 205, 208, 216, 222, 224 Monkeys differential rearing, 77 juvenile play behavior, 92-93 reproductive development fetal androgens, 57 restricted rearing, 72-75 Mosquitofish (Gambusia afJinis), 169 Motacilla alba (white wagtail), food scarcity, 24

Mother-infant relationship, relational perspective, 260-286 Motor neurons, rats, reproductive development, 97-98, 104 Mounting primates, foot-clasp mount, 73 rats, 49-51, 73, 92 juvenile play behavior, 92 maternal anogenital licking, 94-97 MPA, see Medroxyprogesterone acetate Mullet, guppy predator, 160 Multimale polygyny, 207 Multiple mating, 291-307 Multiple paternity, carnivores, 21 1-212 Mungos mungo (banded mongoose) intraspecific fighting, 218 phylogenetic regression, 237 reproduction breeding group, 208,227 reproductive suppression, 210, 225 sociality, 216, 222, 224 Mussels, predation by octopus, 342 Mustela erminea (stoat), 204, 236 Mustelids breeding group, 206 home range, 204,206 phylogenetic regression, 236 as predator, 223 Myofis lucfugus (little brown bat), information center, 27 Mytilus spp., predation by octopus, 342

N Nandinia binotata (binturong), 203, 237 Natural selection, see Selection Nautilus, 319 Nervous system, cephalopods, 319 Neurons, rats, reproductive development, 97-101 No-cost model, group foraging, 15, 20 Nonsocial breeding, 206 carnivores, phylogenetic regression analysis, 239-242 Norway rat (Rattus noruegicus) information transfer, 29-30 reproductive development, 61, 63 Nutcracking behavior, learned, 149 Nycticeius humeralis (evening bat), information center, 26

364

INDEX

0

Observational learning, 335 Octocyon megalotis (bat-eared fox), 207, 234 octopus anatomy, 319-322 brain, 319-320, 325, 330-332 cognition, 317, 322-346 conditioned learning, 333 predation, 341-343 reproduction, 322, 334, 342 Octopus bimacularus, 341 Octopus braireus, 338, 345 Octopus dofleini, 342 Octopus joubini, 323, 345 arm, 321 shelter, 338-339 Octopus maya, 343 Octopus rubescens, 342-344 Octopus vulgaris, 317 cognition, 324, 326 feeding, 342-343 reproduction, 334 shelter, 339 skin color, 338 Oddity, 333 Odor hypothesis, food foraging, 36-37 Offspring, see also Parental care carnivores, offspring dispersal and retention, 217-219 mammals, reproductive success, 103-104 Off-territory copulation, 205, 212 Olfactory stimuli, reproductive behavior, 66-72, 82-93, 100, 101 Oncorhynchus kisutch (Coho salmon), mating, 191 Operant conditioning, octopus, 333 Optic lobe, octopus, 319-320, 325, 330 Osprey (Pandion haliaetus), information transfer, 27-28 Otter (Lurra lutra), 215, 225, 236 P Pandion haliaetus (osprey), information transfer, 27-28 Panthera onca (jaguar), 222

Panthera pardus (leopard), 222 Panrhera rigris (tiger) breeding, 207 feeding habits, 215, 220, 222 phylogenetic regression, 235 Parasitism, group foraging, 15 Parental care, see also Offspring alloparental care, 223-224 biparental care, 206, 223 birds, 223 carnivores, 223-224 chimpanzees, feeding behavior, 149 cross-fostering, 66, 152 differential rearing, 75-82 human, mother-infant relationships, 259-286 mammals, 53, 72-75 mice, feeding behavior, 147 paternal care, 53 primates, play-mothering, 92 rats feeding behavior, 119-152 reproductive development, 47- 105 restricted rearing, 72-75 Paria guppies, 163, 166 Paternal mammals, 53 Peking ducks, reproductive behavior, stimuli, 74 Penis anatomy and function, 300-304 chimpanzees, 300-304 rats, maternal anogenital licking, 97-98 Phenotypic plasticity, 177 Pheromones, information transfer, 35-36 Phoxinus phoxinus (European minnow), antipredator responses, 177 Phylogenetic regression analysis, 204, 228-247 Pied wagtails, information transfer, 32-33 Pigs, reproductive development, 71, 90 Pigtail macaque, reproductive development, 57, 77 Pine cone feeding, black rats, learned behavior, 119-152 Placenta, sex differences, 56-57 Play behavior communication during, 274, 279-283 sex differences, 73, 91-94 Play-mothering, primates, 92

365

INDEX

Poecilia reticulata (guppy), 156-158, 179 behavioral diversification, 156, 184-194 copulation, 164-166, 170-172, 181, 182 evolution, 172- 177. 181-183 genetic divergence, 178-184 predation, 159-177 speciation, 156, 183-184, 191-192 Polyandry, 205, 207 Polygyny, 205, 207 Potos f l a w s (kinkajou), 204, 235 Prairie vole (Microtus ochrogaster), reproductive behavior, 52, 53, 58 Prawns, guppy predator, 160, 161, 163, I66 Predation of birds, 3, 9-1 1 of carnivores, 222-223 of cephalopods, 338 by cuttlefish, 331-332, 341 of guppies, 159-161, 167 evolution, 172- 179, 180- I84 sex ratios, 167-169 sexual harassment, 170-172 by octopus, 341-343 phenotypic plasticity, 177 Prenatal maternal contribution, see Gestation period Prenatal stress, reproductive development, 59-60, 89, 93 Preputial gland, anogenital licking, 82-88 Present posture, primates, 73 Primates differential rearing, 76-77 juvenile play behavior, 92-93 metacommunication, 275 mother-infant relationship, 27 1 present posture, 73 reproductive anatomy, 76-77, 300-302, 304 reproductive development, 57, 76-77 restricted rearing, 72-73 social transmission, feeding methods, 149 Probing behavior, 69-70 Procyonids, 205, 208, 235-236 Progestins, reproductive development, 65, 88-90 Protathaca spp., predation by octopus, 342 Pseudohermaphroditism, 77

Q Quelea quelea (red-billed quelea), food scarcity, 24 Quiscalus quiscala (common grackle), food scarcity, 24

R Rabbits, 297 Rats differential rearing, 78-82 feeding behavior, learned, 119-152 information transfer, 29-30 juvenile play behavior, 92-94 maternal anogenital licking, 78-89, 93-94, 98-101, 104 pine cone feeding, 119-152 reproductive behavior, 69-72 reproductive development, 63-64, 104-105

isolation, 74 maternal contributions, 49, 56-57 maternal stress, 59-60 nerves, 97-101 olfactory stimulation, 66-72 uterine position, 61, 63 R a m s noruegicus (Norway rat) information transfer, 29-30 reproductive development, 61, 63 Rattus rattus (black rat), pine cone feeding, 119-152 Raven (Cowus corax), information transfer, 31-32 Rearing patterns, see also Parental care differential rearing, 75-82 isolation, 74 restricted rearing, 75-82 Reciprocal cooperation, 5-7, 15-16, 38-39 Recruitment center model, group foraging, 18, 21, 38, 39 Red-billed quelea (Quelea quelea) food scarcity, 24 information transfer, 30-3 1 Red deer, differential rearing, 78 Red fox (Vulpes vulpes) home range, 204 hunting, 222 phylogenetic regression, 234

366 reproductive suppression, 209, 21 1, 212 social breeding, 205, 207, 215, 216, 224 Red-winged blackbird (Agelaius phoeniceus), 24 Relational metacommunication, 275 Relational perspective, relationships, 259-260 Relationships coregulation, 277-286 human infant, development, 259-286 Reproductive behavior, see Sexual behavior Reproductive development mammals, 104-105 anogenital licking, 78-89, 93-97, 104 maternal stimulation, 66-72, 91-102 offspring reproductive success, 103-104 postnatal maternal contribution, 65-102 prenatal maternal contribution, 55-64 restricted rearing, 72-75 sex-related biases, 75-91, 105 sexual divergence, 63-64 Reproductive suppression carnivores, 208-212, 225-228 hormones, 208-209 phylogenetic regression analysis, 243-244 Resource availability, carnivore sociality, 214-216 Resource dispersion hypothesis (RDH), 21 5-216 Restricted rearing, reproductive development, 72-75 Reversal learning, 333 Rhesus monkeys mother-infant relationship, 271 reproductive development fetal androgens, 57 juvenile play behavior, 92-93 rearing patterns, 72-75, 77 Risk, courtship and, 163-165 Riuulus hartii, guppy predator, 160-161, 166-169, 173, 174 Rodents differential rearing, 78-82 information transfer, 29-30 juvenile play behavior, 92-94 learned feeding behavior, 119-152

INDEX

pine cone feeding, 119-152 reproductive development, 63-64 differential rearing, 78-82 maternal anogenital licking, 78-89, 93-94, 98-101. 104 maternal contributions, 47-105 maternal stress, 59-60 nerves, 97-101 olfactory stimulation, 66-72 uterine position, 61-63 Roosts. see Communal roosts

S Saguinus oedipus (cotton-top tamarin), play behavior, 73 Salmon, mating, 191 Sandwich tern (Sterna sanduicensis), information transfer, 28-29 Scent-marking behavior, 58, 204 Schooling guppies antipredator techniques, 161- 162, 167, 174, 176, 185-187 genetics, 189 Seabirds, colonies, 26 Sea otter (Enhydra lutra), 225, 236 Selection coloniality, 20-21 genetic diversity, 190 Sensorimotor development, human, 264-270 Sepia spp. (cuttlefish) brain, 330 concealment by burying, 340 predation habits, 331-332, 345 skeleton, 320 skin color changes, 323, 337 Sepioteuthis sepioidea, 323, 335, 336 Sex differences, 49-51 adaptation, 48-49 fish, foraging habits, 169-170 mammals, reproductive development, 47- 105 primates foot-clasp mount, 73 juvenile play behavior, 73, 91-94 rats, reproductive development, 47-105

INDEX

Sex ratio carnivores, 218 guppies, predation, 167-169 rats, maternal influence, 103 Sexual behavior birds, 49-51 carnivores, 208-212, 225-228 chimpanzees, 300-302 guppies, 164-166, 170-172, 181, 182 genetic diversity, 190-191 mounting, 49-51, 92, 94-97 multiple mating, 291-307 off-territory mating, 205, 212 rats copulation, 72-75, 93-95 intromission, 94-95 juvenile play, 93 maternal contributions to reproductive development, 47-105 reproductive suppression, 208-212, 225-228 restricted rearing, 72-75 Sexual dimorphism guppies, 157 rats, reproductive development, 97-101 Sexual harassment, guppies, 170-172, 190 Sexual imprinting, 66 Shelters, octopus, 338-339 Sigmoid posture, guppies, 164-165 Signaling, 12 Silverbacked jackal (Canis mesomelas) breeding group, 207, 224 phylogenetic regression, 234 Skeletal system, cephalopods, 320 Skin color, squid, 323, 336-338, 346 Slight unfamiliarity principle, 71-72 SNB, see Spinal nucleus of the bulbocavernosus Sneaky mating, fish, 164-166, 170-172, 181, 190-191 Social breeding carnivores, 206, 213-225 phylogenetic regression analysis, 239-242 Social hymenopterians, information transfer, 35-40 Sociality carnivores alloparental care, 223-224 competition at kills, 220-221

367

group hunting, 219-220 group vigilance, 222-223 offspring, dispersal and retention, 2 17-2 19 phylogenetic regression analysis, 228-247 predator defense, 222-223 resource availability and renewal, 213-216 Social organization, carnivores, 204-212, 228-229 Social relationships, human infant, development, 259-286 Social restriction studies, reproductive development of mammals, 72-75 Social stimuli, reproductive development, 66-72 Social transmission feeding methods chimpanzees, 149 mice, 147 rats, 119-152 Software, phylogenetic regression analysis, 23 1-232 Solenopsis spp., information transfer, colonial feeding, 35 Songbirds, reproductive behavior, sex differences, 49-50 Spatial ability, sex differences, 52 Speciation, 156, 183-184, 191-192 Speech, coregulation, 273 Spermophilus spp.(Arctic ground squirrel), antipredator responses, 177 Sperm precedence, 2% Sperm sexual selection hypothesis, 293-307 Spinal cord, sexually dimorphic nuclei, 98 Spinal nucleus of the bulbocavernosus (SNB), 97 Spotted hyena (Crocuta crocuta) group hunting, 219-222 juvenile play behavior, 92 masculinized genitalia, 64 phylogenetic regression, 236 reproductive success, 224 reproductive suppression, 208, 209 social organization, 214 Squid, 317, 322-323, 335-337, 345 skin color, 323, 336-338, 346 Squirrels, antipredator responses, 177

368

INDEX

Starlings, food scarcity, 24 Sterna hirundo (common tern), information transfer, 28-29 Sterna sanduicensis (sandwich tern), information transfer, 28-29 Stickleback (Gasterosreus aculeatus) antipredator activities, 162 evolution, 177 population differentiation, 191- 192 Stimuli reproductive behavior, 104-105 anogenital licking, 78-89, 93-97, 104 auditory, 74 differential rearing, 75-82 heterosexual, 71 imprinting, 66 neural development, 98 olfactory, 66-72, 82-88, 93, 100, 101 social, 66 tactile, 81, 100 Stoat (Musrela erminea), 204, 236 Stress, prenatal, reproductive development, 59-60, 89,93 Stumptail macaque, reproductive development, 57 Sunfish, mating, 191 Suricara suricarra (meerkat), 208, 218, 222, 237 Swallows, colony size, 25 Sycidium, guppy predator, 160 Synchronous departure, group foraging, 22, 26 Syrian hamster (Mesocricerus auratus), reproductive behavior, 69

T Tactile information processing, cephalopods, 320 Tactile stimuli, reproductive behavior, 8, 100 Tamarins, play behavior, 73 Tandem running, 35 Tapes spp., predation by octopus, 342 Teasing, communication as part of, 272 Terns colonies, 26 information transfer, 28-29

Testosterone copulatory behavior, 94 fetal maternal behavior, 57-59 maternal stress, 59-60 uterine position, 60-63, 88 sexual dimorphism, 97-98 Threat display, as form of communication, 272, 277-278 Three-spined stickleback (Gasterosreus aculeatus) antipredator activities, 162 evolution, 177 population differentiation, 191-192 Tiger (Panthera tigris) breeding, 207 feeding habits, 215, 220, 222 phylogenetic regression, 235 Tool use chimpanzees nut cracking, 149 octopus, 339-340 Tribolium casraneum. 2% Trigona spp., information transfer, 36 Trivers-Willard hypothesis, 103 Turkey vulture (Cathartes aura). information transfer, 34 Turkish hamster (Mesocricerus brandri), reproductive behavior, 69 Two-handed strategy model, group foraging, 17, 20-21, 38, 39

U Unavoidable parasite model, group foraging, 15, 20 Ungulates, 297 Ursids, 205, 235 Ursus arcros (brown bear), 205, 235 Uterine position, gestation, reproductive development, 60-63, 89-90

V Visual information processing, cephalopods, 320, 324-327 Visual system, cephalopods, 3 19-320 Vivenids, 204, 237 Voles, reproductive behavior, 52, 53

369

INDEX

Vulpes cana (Blanford’s fox), 205, 215 Vulpes vulpes (red fox) home range, 204 hunting, 222 phylogenetic regression, 234 reproductive suppression, 209, 21 1, 212 social breeding, 205, 207, 215, 216, 224 Vultures group living, 2 information transfer, 33-34

W Wagtails food scarcity, 24 information transfer, 32-33 Waiter-searcher model, 13 Weasels, 203 Weaverbirds, information transfer, 30-3 I White-tailed mongoose (Ichneumia albicauda)

phylogenetic regression, 237 sociality, 205, 208, 216

White wagtail (Motacilla a h ) , food scarcity, 24 Wild dogs, 203 group hunting, 219, 220 home range, 204-205 intraspecific fights, 218 phylogenetic regression, 234 reproduction breeding group, 207-21 1, 224 litter size, 223 Wistar rats, 79 Wolves breeding group, 205, 207, 212, 224 intraspecific fighting, 222 phylogenetic regression, 234 reproductive suppression, 209, 210 Wood rats, 78

Y Yellow-headed blackbirds (Xunthocephalus xanrhocephnlus), information transfer, 34

Contents of Previous Volumes Volume 14 Group Mating in the Domestic Rat as a Context for Sexual Selection: Consequences for the Analysis of Sexual Behavior and Neuroendocdne Responses MARTHA K. MCCLINTOCK Plasticity and Adaptive Radiation of Dermapteran Behavior: Results and Perspectives MICHEL VANCASSEL Social Organization of Raiding and Emigrations in Army Ants HOWARD TOPOFF Learning and Cognition in the Everyday Life of Human Infants HANUS PAPOUSEK AND MECHTHILD PAPOUSEK Ethology and Ecology of Sleep in Monkeys and Apes JAMES R. ANDERSON

Volume 15 Sex Differences in Social Play: The Socialization of Sex Roles MICHAEL J. MEANEY, JANE STEWART, AND WILLIAM W. BEA'ITY

Vocal Meet Signaling: A Comparative Approach KLAUS R. SCHERER A Response-Competition Model Designed to Account for the Aversion to Feed on Conspecific Flesh W. J. CARR AND DARLENE F. KENNEDY

Volume 16 Sensory Organization of Alimentary Behavior in the Kitten K. V. SHULEIKINA-TURPAEVA Individual Odors among Mammals: Origins and Functions ZULEYMA TANG HALPIN The Physiology and Ecology of Puberty Modulation by Primer Pheromones JOHN G. VANDENBERGH AND DAVID M. COPPOLA Relationships between Social Organization and Behavioral Endocrinology in a Monogamous Mammal C. SUE CARTER, LOWELL L. GETZ, AND MARTHA COHEN-PARSONS Lateralization of Learning Chicks L. J. ROGERS

On the Functions of Play and Its Role in Behavioral Development PAUL MARTIN AND T. M. CAR0

Circannual Rhythms in the Control of Avian Migrations EBERHARD GWINNER

Sensory Factors in the Behavioral Ontogeny of Altricial Birds S. N. KHAYUTIN

The Economics of Fleeing from Predators R. C. YDENBERG AND L. M. DILL

Food Storage by Birds and Mammals DAVID F. SHERRY

Social Ecology and Behavior of Coyotes MARC BEKOFF AND MICHAEL C. WELLS

370

CONTENTS OF PREVIOUS VOLUMES

37 1

Volume 17

P. V. ALATALO AND A. LUNDBERG

Receptive Competencies of LanguageTrained Animals LOUIS M. HERMAN

Kin Recognition: Problems, Prospects, and the Evolution of Discrimination Systems C. J. BARNARD

Self-Generated Experience and the Development of Lateralized Neurobehavioral Organization in Infants GEORGE F. MICHEL

Maternal Responsiveness in Humans: Emotional, Cognitive, and Biological Factors CARL M. CORTER AND ALISON S. FLEMING

Behavioral Ecology: Theory into Practice NEIL B. METCALFE AND PAT MONAGHAN

The Evolution of Courtship Behavior in Newts and Salamanders T. R. HALLIDAY

The Dwarf Mongoose: A Study of Behavior and Social Structure in Relation to Ecology in a Small, Social Carnivore 0. ANNE E. RASA

Ethopharmacology: A Biological Approach to the Study of Drug-Induced Changes in Behavior A. K. DIXON, H. U. FISCH, AND K. H. MCALLISTER

Ontogenetic Development of Behavior: The Cricket Visual World RAYJOND CAMPAN, GUY BEUGNON, AND MICHEL LAMBIN

Volume 18 Song Learning in Zebra Finches (Taeniopygia guttata): Progress and Prospects PETER J. B. SLATER, LUCY A. EALES, AND N. S. CLAYTON

Behavioral Aspects of Sperm Competition in Birds T. R. BIRDHEAD Neural Mechanisms of Perception and Motor Control in a Weakly Electric Fish WALTER HEILIGENBERG Behavioral Adaptations of Aquatic Life in Insects: An Example ANN CLOAREC The Cicadian Organization of Behavior: Timekeeping in the Tsetse Fly, A Model System JOHN BRADY

Additive and Interactive Effects of Genotype and Maternal Environment PIERRE L. ROUBERTOUX, MARIKA NOSTEN-BERTRAND, AND MICHELE CARLIER Mode Selection and Mode Switching in Foraging Animals GENE S. HELFMAN Cricket Neuroethology: Neuronal Basis of Intraspecific Acoustic Communication FRANZ HUBER Some Cognitive Capacities of an African Grey Parrot (Psitracus erirhacus) IRENE MAXINE PEPPERBERG

Volume 20 Social Behavior and Organization in the Macropodoidea PETER J. JARMAN

Volume 19

The t Complex: A Story of Genes, Behavior, and Population SARAH LENINGTON

Polyterritorial Polygyny in the Pied Flycatcher

The Ergonomics of Worker Behavior in Social Hymenoptera PAUL SCHMID-HEMPEL

372

CONTENTS OF PREVIOUS VOLUMES

“Microsmatic Humans” Revisited: The Generation and Perception of Chemical Signals BENOIST SCHAAL AND RICHARD H. PORTER Lekking in Birds and Mammals: Behavioral and Evolutionary Issues R. HAVEN WILEY

Volume 21 Primate Social Relationships: Their Determinants and Consequences ERIC B. KEVERNE The Role of Parasites in Sexual Selection: Current Evidence and Future Directions MARLENE ZUK Conceptual Issues in Cognitive Ethology COLIN BEER Responses in Warning Coloration in Avian Predators W. SCHULER AND T. J. ROPER Analysis and Interpretation of Orb Spider Exploration and Web-Building Behavior FRITZ VOLLRATH Motor Aspects of Masculine Sexual Behavior in Rats and Rabbits GABRIELA MORAL1 AND CARLOS BEYER On the Nature and Evolution of Imitation in the Animal Kingdom: Reappraisal of a Century of Research A. WHITEN AND R. HAM

Volume 22 Male Aggression and Sexual Coercion of Females in Nonhuman Primates and Other Mammals: Evidence and Theoretical Implications BARBARA B. SMUTS AND ROBERT W. SMUTS

Parasites and the Evolution of Host Social Behavior ANDERS PAPE MOLLER, REIJA DUFVA, AND KLAS ALLANDER The Evolution of Behavioral Phenotypes: Lessons Learned from Divergent Spider Populations SUSAN E. RIECHERT Proximate and Developmental Aspects of Antipredator Behavior E. CURIO Newborn Lambs and Their Dams: The Interaction That Leads to Sucking MARGARET A. VINCE The Ontogeny of Social Displays: Form Development, Form Fixation, and Change in Context T. G. GROOTHUIS

Volume 23 Sneakers, Satellites, and Helpers: Parasitic and Cooperative Behavior in Fish Reproduction MICHAEL TABORSKY Behavioral Ecology and Levels of Selection: Dissolving the Group Selection Controversy LEE ALAN DUGATKIN AND HUDSON KERN REEVE Genetic Correlations and the Control of Behavior, Exemplified by Aggressiveness in Sticklebacks THE0 C. M. BAKKER Territorial Behavior: Testing the Assumptions JUDY STAMPS Communication Behavior and Sensory Mechanisms in Weakly Electric Fishes BERND KRAMER

This Page Intentionally Left Blank

I S B N 0-12-004524-9