Advances in
THE STUDY OF BEHAVIOR VOLUME 22
This Page Intentionally Left Blank
Advances in THE STUDY OF BEHAVIOR E...
20 downloads
993 Views
17MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
Advances in
THE STUDY OF BEHAVIOR VOLUME 22
This Page Intentionally Left Blank
Advances in THE STUDY OF BEHAVIOR Edited by PETERJ. B. SLATER School of Biological und Medical Sciences University of St. Andrews F$e9 Scotland
JAYS. ROSENBLATT Institute of Animal Behavior Rutgers University Newark, New Jersey
CHARLES T. SNOWDON Department of Psychology University of Wisconsin-Madison Madison, Wisconsin
MANFRED MILINSKI Zoologisches Institut Universitat Bern Hinterkappelen Switzerland
VOLUME 22
ACADEMIC PRESS, INC. Harcourt Brace Jovanovich, Publishers San Diego New York Boston London Sydney Tokyo Toronto
This book is printed on acid-free paper. @
Copyright 0 1993 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. 1250 Sixth Avenue, San Diego. California92101 United Kingdom Edifionpublished by
Academic Press Limited 24-28 Oval Road, London NW I 7DX Library of Congress Catalog Number: 64-8031 International Standard Book Number: 0- 12-004522-2 PRINTED IN T H E UNITED STATES OF AMERICA 93 94 95 96 91 98 QW 9 8 1 6 5 . 4 3 2
1
Contents Contributors .............................................. Preface ..................................................
ix xi
Male Aggression and Sexual Coercion of Females in Nonhuman Primates and Other Mammals: Evidence and Theoretical Implications
BARBARA B. SMUTS AND ROBERT W. SMUTS I. Introduction ....................................... 11. Male Aggression and Sexual Coercion in Nonhuman Primates .......................................... 111. Costs to Female Primates of Male Aggression .......... IV. Primate Female Counterstrategies to Male Aggression . . V. Male Aggression against Females in Chimpanzees ...... VI. Male Aggression against Females in Other Mammals . . . . VII. Variation in Male Aggression against Females .......... VIII. Evaluating the Sexual Coercion Hypothesis ............ IX. Implications of Male Sexual Coercion for Sexual Selection Theory ................................... X. Conclusions ....................................... XI. Summary .......................................... References ........................................
1
3 9
II 19
24 31 37 43 49 49 50
Parasites and the Evolution of Host Social Behavior ANDERS PAPE MOLLER, REIJA DUFVA, AND KLAS ALLANDER
I. Introduction ....................................... 11. Group Living and Parasites .......................... 111. Parasites, Reproduction, and Sexual Selection . . . . . . . . . IV. Social Behavior and Parasite-Host Coevolution . . . . . . . . V. Summary .......................................... References ........................................
V
65
73 83 91 93 94
vi
CONTENTS
The Evolution of Behavioral Phenotypes: Lessons Learned from Divergent Spider Populations SUSAN E. RIECHERT I. Introduction
.......................................
11. The Spider System ................................. 111. Fitness-Linked Behavioral Traits .....................
IV. Arizona Riparian Population Deviation from Adaptive Equilibrium ........................................ V. Factors That May Have Limited Adaptation . . . . . . . . . . . VI. Experimental Manipulation of Gene Flow versus Selection .......................................... VII. Discussion and Conclusions ......................... VIII. Summary ........................................... References ........................................
103 I04 I12 119 I23 I28 I30 132 133
Proximate and Developmental Aspects of Antipredator Behavior E. CURIO
I. Introduction ....................................... 11. Causal Aspects of Enemy Recognition . . . . . . . . . . . . . . . . 111. Developmental Aspects of Enemy Recognition . . . . . . . . . IV. Conclusions ....................................... V. Summary .......................................... References ........................................
135 I39 I87 221 225 227
Newborn Lambs and Their Dams: The Interaction That Leads to Sucking MARGARET A. VINCE
I. Introduction ....................................... 11. Pre- and Perinatal Factors That Play a Part in the Ewe and Lamb Relationship after Birth .................... 111. Sensory and Behavioral Factors Involved in the Postnatal Relationship between Ewe and Lamb . . . . . . . . IV. Discussion ........................................ V. Summary .......................................... References ........................................
239 240 244 262 263 264
CONTENTS
vii
The Ontogeny of Social Displays: Form Development. Form Fixation. and Change in Context T . G . G . GROOTHUIS I . General Introduction ................................ I1 . Mechanisms of Form Development . . . . . . . . . . . . . . . . . . . 111. Form Fixation of Display ............................ IV . Change in Context of Display ........................ V . Functional Aspects of Display Development ........... V1 . A Summarizing Scheme ............................. VII . Summary .......................................... References ........................................
lndex
..............................................
Contents of Previous Volumes ...........................
269 273 302 306 310 315 318 319 323 328
This Page Intentionally Left Blank
Contributors
Numbers in parentheses indicate the pages on which the authors’ contributions begin.
KLAS ALLANDER (65). Department of Zoology, Uppsala University, S-75122 Uppsala, Sweden E. CURIO ( 1351, Arbeitsgruppe fur Verhaltensforschung, Fakultat fur Biologie, Rithr-Universitat Bochum, 0 4 6 3 0 Bochum 1, Germany REIJA DUFVA (65), Department of Zoology, Uppsala University, S-751 22 Uppsala, Sweden T. G . G . GROOTHUIS (2691, Zoological Laboratory, University of Groningen, 9750 A A Haren, The Netherlands ANDERS PAPE MOLLER (651, The Galton Laboratory, Department of Genetics and Biometry, University College of London, London, England SUSAN E. RIECHERT (103). Department of Zoology, The University of Tennessee, Knoxville, Tennessee 37996 BARBARA B. SMUTS ( I ) , Departments of Psychology and Anthropology, und Center for Human Growth and Development, University of Michigan, Ann Arbor, Michigan 48109 ROBERT W. SMUTS ( I ) , Ann Arbor, Michigan 48104 MARGARET A. VINCE (239), AFRC Institute of Animal Physiology und Genetics Research, Babraham, Cambridge CB2 4AT,England
ix
This Page Intentionally Left Blank
Preface The aim of Advances remains as it has been since the series began: to serve the increasing number of scientists engaged in the study of animal behavior by presenting their theoretical ideas and research to their colleagues and to those in neighboring fields. We hope that 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. While the recent rise of behavioral ecology and sociobiology has tended to overshadow other areas, scientists studying behavior today range more widely than ever before: from ecologists and evolutionary biologists to geneticists, endocrinologists, pharmacologists, neurobiologists. and developmental psychobiologists, not forgetting the ethologists and comparative psychologists whose prime domain is this subject. It is not our intention to focus narrowly on one or a few of these fields, but rather 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 the publisher of Advances in rhe Srudy of Behavior are committed. We will continue to provide the means to this end by publishing critical reviews, by inviting extended presentations of significant research programs, by encouraging the writing of theoretical syntheses and reformulations of persistent problems, and by highlighting especially penetrating research that introduces important new concepts. The realization of these aims is well illustrated by the spectrum of topics presented in the present volume. With this volume we welcome Dr. Charles T. Snowdon, of the University of Wisconsin at Madison, as an editor. A primatologist with wideranging interests, his expertise will make a valuable contribution to the breadth of our editorial team, and his contribution will do much to maintain the vigor of the series.
xi
This Page Intentionally Left Blank
ADVANCES
IN THE STUDY OF B E H A V I O R . VOL. ?.
Male Aggression and Sexual Coercion of Females in Nonhuman Primates and Other Mammals: Evidence and Theoretical Implications BARBARAB. SMUTS DEPARTMENTS OF PSYCHOLOGY AND ANTHROPOLOGY AND CENTER FOR HUMAN GROWTH AND DEVELOPMENT UNIVERSITY OF MICHIGAN ANN ARBOR, MICHIGAN 48109
ROBERTW. SMUTS ANN ARBOR. MICHIGAN 48104
I.
INTRODUCTION
The single most important difference between the sexes is the difference in their investment in offspring. The general rule is this: females do all of the investing: males do none of it. (Trivers. 1985, p. 207)
Although Trivers' general rule has many exceptions, it accurately identifies the primary source of conflict between the sexes: in most sexual organisms most of the energy and time invested in offspring comes from females. From this basic fact it follows that, for males more than females, reproductive success is limited by the number of matings with fertile partners. For females more than males, on the other hand, reproductive success is limited by the time and effort required to garner and transfer energy to offspring and to protect and care for them (Bateman, 1948; Trivers, 1972). Males therefore are usually more eager than females to mate at any time with any partner who may be fertile, while females are usually more careful than males to choose mates who seem likely to provide good genes, protection, parental care, or resources in addition to gametes (Trivers, 1972; Alexander and Borgia, 1979). Combined with female interest in mate quality, male interest in mate quantity creates a widespread conflict of interest between the sexes (Borgia, 1979; Parker, 1979; Hammerstein and Parker, 1987). The conflict is I Copyright 0 1993 by Academic Press. Inc. All rights of reproduction in any form reserved.
2
BARBARA B. S M U T S A N D ROBERT W . SMUTS
mitigated when males court females by offering them the benefits females want from males, such as food, protection, or help in rearing young. These benefits are typically costly in terms of male time and energy, however, and males may often be able to overcome female reluctance at lower cost to themselves by using force or the threat of force, behavior that we call “sexual coercion.” Theoretical treatments (e.g., Hammerstein and Parker, 1987) indicate that sexual coercion can function as an important selective force influencing the evolution of both male and female behaviors. However, male aggression toward females, including sexual coercion, has rarely been a focus of study, and for the vast majority of animals, including most mammals, quantitative information is unavailable. These limitations severely constrain our ability to determine the evolutionary significance of sexual coercion. This article aims to stimulate research and theorizing about sexual coercion by reviewing the relevant evidence for nonhuman primates and some other mammals in which sexual coercion is especially well documented. Two contrasting goals guide this review. On the one hand, we hope to persuade the reader that sexual coercion is an important phenomenon worthy of further study. On the other, we wish to highlight important gaps in our knowledge of sexual coercion. We have tried to balance these two goals by using limited evidence from a small number of species to generate hypotheses, while emphasizing that, to test these hypotheses, we need much better information from a larger number of species. We begin by describing aggressive male behaviors that appear to function as sexual coercion, the costs that this male aggression imposes on females and young, and the counterstrategies that females employ to reduce these costs. The data that we review for primates and other mammals reveal extensive variation in the form and frequency of male aggression against females, and we propose several hypotheses to help account for this variation. We also consider the kinds of evidence needed to determine whether particular cases of male aggression against females function as sexual coercion. In the final section, we argue that sexual coercion has been underestimated as a significant force in social evolution, and indicate how more attention to intersexual coercion as a form of sexual selection can enhance our understanding of animal societies. OF SEXUAL COERCION THECONCEPT
We define sexual coercion as use by a male of force, or threat of force, that functions to increase the chances that a female will mate with him at a time when she is likely to be fertile, and to decrease the chances that
MALE AGGRESSION AND SEXUAL COERCION
3
she will mate with other males, at some cost to the female. The functional consequences of male sexual coercion distinguish it from other instances of male aggression against females (e.g., in the context of feeding competition) that do not appear to involve manipulation of sexual opportunities. Our definition of sexual coercion as a subset of aggressive male behaviors toward females that is delineated by their function means that sexual coercion is not a purely behavioral concept, but involves a combination of behavioral description and functional explanation. Sexual coercion cannot be identified by observing only the immediate behavior of the aggressor; it is also necessary to observe the subsequent behavior of the aggressor, the target, and even of other individuals. It is not an easy concept to work with, but we believe it is nevertheless useful because it accurately reflects the complexity of agonistic sexual behavior in animals. Toward the end of this article, we consider in some detail how one can test the hypothesis that particular acts of male aggression against females fit the functional definition of sexual coercion given here. We delay this discussion until later because it requires a basic understanding of the wide variety of male aggression toward females that is observed in nature. Thus, we will proceed for the moment on the assumption that sexual coercion does indeed exist, while keeping in mind the need to examine functional consequences before accepting the hypothesis that a particular aggressive act (or set of acts) actually functions as sexual coercion. Our definition also limits sexual coercion to behavior that involves the use or the threat of force. Although males can (and do) manipulate female mating behavior to their own advantage by inflicting other kinds of costs or by withholding benefits, such a broad definition of sexual coercion would encompass so large a part of all interactions between males and females that it would prove useless. 11. MALEAGGRESSION A N D SEXUAL COERCION IN NONHUMAN PRIMATES In what follows, we concentrate on polygynous primates living in groups in which a single male monopolizes matings with two or more females, or multiple males compete for mating opportunities with multiple females. Because polygyny is typically associated with much more intense male-male competition for mates (Clutton-Brock and Harvey, 1976, 1978), these species are expected to show more male sexual coercion than species living in monogamous or polyandrous groups. Reduced sexual coercion is especially likely in monogamous and polyandrous primates, because these species invariably establish long-lasting pair bonds and defend territories
4
BARBARA B. SMUTS AND ROBERT W. SMUTS
against other groups (Goldizen, 1987; Robinson et al., 1987; Robbins Leighton, 1987), minimizing opportunities for contact between oppositesex individuals other than mates. This is in contrast to the situation in many monogamous birds and in most human groups, in which, because of high mobility and/or communal living, mated individuals may frequently encounter opposite-sex individuals other than their mates (Westneat et al., 1990; Rodseth ef al., 1991). Although, on theoretical grounds, sexual coercion is expected to be considerably less common in monogamous and polyandrous nonhuman primates, we do not imply that it is entirely absent in these species, and, indeed, the possible significance of sexual coercion of females by mates (e.g., Goldizen, 1989) or male neighbors during encounters between family groups, or by strange males when they encounter lone females, or by mates when females approach territory boundaries, deserves further attention. We focus on information from wild primates, when it is available, because wild groups are more likely to reflect socioecological conditions that obtained during the species’ evolutionary history, but we also include relevant evidence from provisioned and captive animals. Caution is necessary when such information is used to support an argument related to selection pressures in the wild. However, evidence from captive or provisioned animals can provide useful indications of behavioral potentials not typically shown in the wild, which may nevertheless reflect a species’ evolved capacity to respond adaptively to novel circumstances (R. W. Smuts, 1993). Finally, we conclude our discussion of primates with a special section on chimpanzees. We discuss chimpanzees as a separate “case study” because much more information is available on male aggression against females in this species than in any other primate, and we wish to present this information as a coherent whole. A.
FREQUENCY OF MALEAGGRESSION AGAINST FEMALES AND CONTEXTS OF OCCURRENCE
Male aggression against females is frequently mentioned in passing or briefly described in the literature on wild nonhuman primates, which suggests its widespread occurrence through the Primate order (Tracy and Crawford, 1992). However, few quantitative data are available on male aggression against female nonhuman primates. Smuts (1985) determined rates of male aggression toward anestrous (i.e., pregnant and lactating) females in a troop of wild olive baboons. During daylight hours, the average anestrous adult female was a victim of male aggression five times per week. One-quarter of these episodes involved physical attack, and
MALE AGGRESSION A N D SEXUAL COERCION
5
roughly 1 of every 50 attacks resulted in a serious wound. Put another way, each adult pregnant or lactating female baboon in the troop could expect to receive at least one serious wound from a male every year (Smuts, 1985). The rate at which female mountain gorillas receive aggression from the silverback male is even higher (ranging from 1 to 4.3 times per female per 12-hour day, depending on the group and time period), but, in contrast to baboons, this aggression very rarely leads to injury (Watts, 1992).In some other species, male aggression toward females occurs much less often. Among red howlers, for example, Sekulic (1983a) observed male aggression toward females at a rate of less than 0.04 times per female per day. The contexts in which males show aggression toward females also vary widely, both within and between species. In many species, a significant proportion of male-female agonism occurs during feeding competition (e.g., olive baboons: 20% [Smuts, 19851; mountain gorillas: 5-20% [Watts, 19921; wedge-capped capuchins: 63% [O’Brien, 19911; chimpanzees: about 18% [Goodall, 1986, fig. 12.31). Smuts (1985) found that males were also aggressive toward anestrous females in a wide variety of social situations, including defense of other females and young who were affiliated with the aggressor. Mountain gorilla males and macaque males also frequently direct aggression toward females in order to break up fights between females (Kaplan, 1977; Harcourt, 1979; Bernstein and Ehardt, 1986; Oi, 1990; Watts, 1992). Smuts (1985) also observed young, high-ranking males attacking the close female associates of older, lower ranking rivals, apparently in order to provoke the older males into aggressive confrontations that they were likely to lose. Similarly, de Waal (1982) described how, during a power struggle between captive chimpanzee male allies Nikki and Luit on the one hand, and alpha male Yeroen on the other, Luit and Nikki often attacked one of Yeroen’s female supporters near Yeroen, apparently to test his willingness to protect females against the rivals. These examples indicate that bonds with particular males sometimes make females vulnerable to manipulative aggression by rival males. The examples just given highlight the fact that not all male aggression toward females functions as sexual coercion. However, quantitative data from several species show that male aggression toward females is more likely when the females are in estrus (macaques: Tokuda, 1961; Kurland, 1977; Enomoto, 1981; Fedigan, 1982; Eaton, 1984; Teas, 1984 [but see Ruehlmann et al., 1988, for an exception]; savanna baboons: Hausfater, 1975;chimpanzees: Goodall, 1986; mountain gorillas: Nadler, 1989b).This widespread tendency for males to show more aggression toward potentially fertile females is consistent with the hypothesis that male aggression often functions to increase access to mates.
6
BARBARA B. SMUTS A N D ROBERT W . SMUTS
B. POSSIBLE EXAMPLES OF MALE SEXUAL COERCION The primate literature contains numerous descriptions of behaviors that appear to satisfy our definition of sexual coercion. For example, male rhesus monkeys attack females caught mating or consorting with lower ranking rivals and sometimes injure them severely (Carpenter, 1942; Lindburg, 1983; Manson, 1991). Manson (l991), studying free-ranging, provisioned rhesus monkeys on Cay0 Santiago, reported a significant positive relationship between the frequency with which estrous females associated with lower ranking males and the rate at which they received aggression from high-rdnking males, who were apparently intent on disrupting these mating relationships. Chimpanzees, like male macaques, also tend to attack the female, rather than the lower ranking rival, if the two are caught courting (de Waal, 1982; Goodall, 1986; Hauser, 1990). Male chimpanzees (Goodall, 1986; see also Section V,A), rhesus monkeys (Carpenter, 1942; Lindburg, 1983), Japanese macaques (Enomoto, 1981), and savanna baboons (Hausfater, 1975) also use aggression to try to initiate or maintain consortships with uncooperative females. The most dramatic examples of apparent sexual coercion come from wild orangutans, in which most copulations by subadult males (MacKinnon, 1971; Rodman, 1973; Rijksen, 1978; Galdikas, 1985; Mitani, 1985) and nearly half of all copulations by adult males (Mitani, 1985) occur after the female’s fierce resistance has been overcome through violent restraint. Similar forced copulations have occasionally been observed among wild chimpanzees; in most cases these involved incestuous matings (Goodall, 1986;Nishida, 1990). In a series of studies ofcaptive chimpanzees, lowland gorillas, and orangutans, Nadler (1982, 1988; Nadler and Miller, 1982) found that, when heterosexual pairs were housed alone together, males in all three species used aggression to force females to copulate throughout the estrous cycle. When females were given control over proximity to the male, however, copulations occurred only with female cooperation and only at mid-cycle. These observations indicate that males in all three of these species of apes will employ sexual coercion when the opportunity for females to escape is minimized. Even when a female is not sexually receptive, male aggression may be designed to increase, or maintain, future mating access. A well-known example involves the male hamadryas baboon, who uses coercion to keep the females he gathers around him away from other males at all times. Should one of “his” females stray toward another male, the hamadryas male will instantly threaten the female with an eyebrow flash; if she fails to approach him immediately he will attack her with a neckbite (Kummer, 1968). Male use of aggression to herd mates away from strange males
MALE AGGRESSION A N D SEXUAL COERCION
7
during encounters with other groups has been reported for species in all major primate taxa, including prosimians (M. E. Pereira, personal communication), cercopithecines (Cheney and Seyfarth, 1977; van Noordwijk and van Schaik, 1985; Byrne et ul., 19871, colobines (Stanford, 19911, New World monkeys (Goldizen, 19891, and apes (Nishida and HiraiwaHasegawa, 1987; Sicotte, 1989; Watts, 1991). Our definition of sexual coercion in functional as well as behavioral terms means that it may sometimes be difficult to determine whether a particular behavior qualifies as sexual coercion. For example, in several primates, male ritualized courtship displays incorporate aggressive behaviors that are typically directed against other males (e.g., stalking in rhesus macaques: Manson, 1991; hair erection and bipedal swagger in chimpanzees: Goodall, 1986; charging in gorillas: Nadler, 1989b). The functional significance of “ritualized” aggression during courtship is not well understood; such displays could possibly function to demonstrate a male’s health and vigor and might thereby facilitate female mate choice. Thus, the fact that a male directs aggression toward an estrous female does not in and of itself constitute evidence for sexual coercion (see Section VlII for further discussion of this issue). On the other hand, male aggression that has no obvious sexual significance may nevertheless function to increase female sexual cooperation in the future and thus qualify as aform of sexual coercion. Goodall (l986),for example, notes that 83% of severe male attacks on females that occurred in no obvious context involved cycling females whose sexual swellings had not yet reached the stage of full tumescence associated with ovulation. She suggests that these attacks intimidate the female so that, when she is close to ovulation, she will respond positively to the male’s mating initiatives. Another possible example of sexual coercion involves the frequent cooperative aggression against single females by allied male spider monkeys (black-handed spider monkeys: Fedigan and Baxter, 1984; black spider monkeys: McFarland Symington, 1987). This aggression has not been observed to injure females, but it can be intense; McFarland Symington (1987, p. 153) describes “frenzied chases involving three males and lasting up to 15 minutes.” The functional significance of this aggression remains obscure; although it is directed only at cycling females (McFarland Symington, 1987), it has not been observed as a prelude to copulation (Fedigan and Baxter, 1984; McFarland Symington, 1987). Spider monkeys are among a handful of primates in which males remain in their natal groups and form lifelong bonds with one another, while females transfer to other groups. They are also one of the few polygynous anthropoid primates that show little sexual dimorphism, but females are nevertheless consistently
8
BARBARA B. SMUTS AND ROBERT W. SMUTS
subordinate to males and relations between the sexes are “generally tense” (McFarland Symington, 1987, p. 161). Given the slight sexual dimorphism in these species, it seems reasonable to hypothesize that male dominance over females is a product of aggression by male coalitions and that, by increasing their power over females, cooperating males also increase their ability to gain sexual access to them. Since courtship and copulation have rarely been observed in wild spider monkeys (Fedigan and Baxter, 1984; McFarland Symington, 1987), further evidence is needed to evaluate this hypothesis. Male primates’ use of force to increase sexual access to females can also involve considerably longer-term strategies such as infanticide (Hrdy , 1979). Males from a wide variety of nonhuman primates, including Old and New World monkeys, apes, and prosimians, kill infants sired by other males (Hausfater and Hrdy, 1984; Crockett and Sekulic, 1984; Struhsaker and Leland, 1987; Pereira and Weiss, 1991). Male infanticide occurs most often in species that live in groups with a single breeding male after a strange male aggressively usurps the breeding position and attempts to kill the immature offspring of the previous male (grey langurs: Hrdy, 1977; red howlers: Sekulic, 1983a; Crockett and Sekulic, 1984; mantled howlers: Clarke, 1983; red-tail monkeys: Struhsaker, 1977; blue monkeys: Butynski, 1982). Male infanticide can also occur when immigrant males enter groups with multiple breeding males (baboons: Collins er id., 1984), in association with a change in male status relationships within multimale groups (red colobus: Struhsaker and Leland, 1985), or after a breeding male dies, leaving vulnerable mothers and infants without protection (gorillas: Watts, 1989). Because a return to sexual cycling is inhibited by lactation, death of the infant typically brings the mother into estrus sooner than would occur otherwise. In many instances, the infanticidal male subsequently mates with the female (reviewed by Struhsaker and Leland, 1987). Although the aggression involved in infanticide targets the infant rather than the mother, it is appropriate to view infanticide as a form of sexual coercion for two reasons. First, like other forms of sexual coercion, it involves the use of force to manipulate the female’s sexual state and mating behavior to the male’s advantage; killing the infant is simply a means to this end. Second, like other forms of sexual coercion, it imposes a cost on the female.
C. A NOTE ON TERMINOLOGY In the following section, we discuss the costs to females of male aggression that we hypothesize functions as sexual coercion. However, below and in later sections, we refer to specific behaviors as “male aggression”
MALE AGGRESSION AND SEXUAL COERCION
9
rather than “sexual coercion” because, in most cases, the functional consequences of these behaviors have yet to be demonstrated conclusively. 111. COSTSTO PRIMATE FEMALES OF MALEAGGRESSION
Although the reproductive costs to females of male aggression have seldom been measured, they may often be considerable. Lindburg (1983) saw a top-ranking rhesus male fatally injure his consort partner after she repeatedly approached another male, and B. B. Smuts (personal observation) saw an adult male olive baboon kill an adolescent, estrous female. Rajpurohit and Sommer (1991) reported the death of a grey langur female as a result of wounds inflicted by a male, but the context of the attack was not described. Enomoto (1981) and Manson (1991) reported frequent male aggression toward estrous female macaques (Enomoto: 0.86 times per day per female; Manson: 0.26-0.44 times per day per female). These two studies and Teas’ (1984), study of wild rhesus “temple monkeys” in India agree that male aggression against estrous females often resulted in serious wounds. These results should be regarded with caution, however, because they are from provisioned troops living in crowded conditions, which may increase rates of male aggression and wounding. At Gombe, observers have witnessed numerous brutal attacks by male chimpanzees on females from other communities, and some of these females died from their wounds (Goodall, 1986). Finally, even when females themselves are not severely injured by male attacks, male violence can lead to abortion (baboons: Pereira, 1983), disruption of estrous cycles (chimpanzees: Goodall, 1986; rhesus macaques: J. Manson and s. Perry, personal communication), and perhaps other deleterious, stress-related effects. The reproductive costs of male infanticide are easier to ascertain. Among grey langur troops near Jodhpur, when the previous resident male was replaced by a new male, 40% of infants present at the time of replacement (n = 81 in 12 different troops) and 35% of the infants born shortly thereafter (n = 34) were victims of infanticide (Sommer, 1992). Since male takeovers occurred on average every 26.5 months (Sommer and Rajpurohit, 1989), infanticide is clearly an important source of infant mortality. Among mountain gorillas, at least 37% of infant mortality is due to male infanticide (Watts, 1989). Crockett and Rudran (1987) and Clarke and Glander (1984) give similar estimates (44 and 40%) for red howler monkeys and mantled howlers, respectively. Male infanticide may also be responsible for a significant proportion of infant mortality in chimpanzees
10
BARBARA B. SMUTS AND ROBERT W. SMUTS
(Goodall, 1986; Nishida, 1990; Nishida er al., 19901, baboons (Collins er ul., 1984), and probably a number of other species. Potential costs of infanticide for females are probably even higher, since rates of infanticide are undoubtedly reduced, sometimes substantially, by female counterstrategies (see below). In addition to the obvious costs due to severe injuries or death, male aggression can inflict subtle but perhaps significant costs by constraining female behavior in many ways. For example, male herding in hamadryas baboons sometimes prevents a female from joining her female kin in a different group, thus depriving her of potential allies (Abegglen, 1984). In mountain gorillas, male infanticide constrains female movements between groups. Mothers with young infants must remain in their current group until the infant is older, abandon the infant and transfer without it (as sometimes occurs), or transfer with the infant, which nearly always leads to infanticide (Watts, 1989). When males employ aggression to exact female sexual cooperation, the benefits females derive from free mate choice will be reduced (for discussion of possible benefits of mate choice. see Smuts, 1987a; Small, 1989; Manson, 1991). Manson (1991), for example, found that among rhesus macaques on Cay0 Santiago, estrous females preferentially maintained proximity to lower ranking males, and such proximity-maintaining behavior correlated with higher copulation rates. However, after higher ranking males chased or attacked females in consort with lower ranking males, the females often failed to restore proximity to their previous partners. This suggested to Manson that male aggression disrupted females’ attempts to express their mating preferences. The time and energy involved in maintaining vigilance toward potentially aggressive males may sometimes be costly, although such costs are difficult to measure. Female baboons with young infants consistently avoid proximity to recent male immigrants (the males most likely to commit infanticide; Collins er a / . , 1984; Busse, 1984), female vervets restrain their infants significantly more often in the presence of new males (Fairbanks and McGuire, 1987), and female ring-tailed lemurs carefully monitor the movements of recently immigrated males who are likely to commit infanticide (Pereira and Weiss, 1991). Finally, it is important to note that the costs discussed here occur in spite of whatever female counterstrategies exist to resist or reduce male aggression. In the absence of such counterstrategies, the costs to females of male aggression presumably would often be considerably higher. These “original,” higher costs are the selective forces that promote the evolution of female counterstrategies. In addition, the counterstrategies that females
MALE AGGRESSION A N D SEXUAL COERCION
11
employ to resist or reduce male aggression themselves often involve costs, as indicated below. IV. A.
PRIMATE FEMALE COUNTERSTRATEGIES TO MALE AGGRESSION
FIGHTING BACK
The most obvious first line of defense against male aggression is to fight back. In extreme cases, particularly when protecting vulnerable infants. this is just what female primates tend to do. Mountain gorilla females, for example, usually fight back against male attacks on their infants. However. because male gorillas are twice as large as females, female resistance is usually futile, and resistant females risk severe injury (Watts, 1989). Red howler and grey langur females also attempt to physically thwart infanticidal males, and occasionally wound them, but they are rarely able to prevent infanticide (Hrdy. 1977; Sekulic, 1983a; Crockett and Sekulic, 1984). Similarly, orangutan females struggle free only rarely during forced copulations (Mitani, 1985). In most nonhuman primates in which male aggression toward females has been reported, males are larger than females and dominate them in one-on-one encounters (reviewed by Smuts, 1987b), which limits the effectiveness of retaliatory aggression by single females. A few striking exceptions exist, however. In ring-tailed, crowned, and ruffed lemurs, females consistently win dyadic agonistic encounters with males (Kappeler, 1990; Pereira et a l . , 1990; Kaufman, 1991). Among patas monkeys, individual females often defeat males in one-on-one fights, and males “appear extremely reluctant to use force against females in almost all contexts, presumably because of the threat of female retaliation’’ (Loy, 1989, p. 39). Similarly, in macaques, vervet monkeys, brown capuchins, wedge-capped capuchins, and several other species, individual females sometimes win agonistic encounters against males (e.g., stumptail macaques: Bernstein, 1980; Japanese macaques: Johnson et d., 1982; vervets: Bramblett et al., 1982; brown capuchins: Janson, 1984; wedgecapped capuchins: Robinson 1981; O’Brien, 1991; see Smuts, 1987b. for further details). Since, with the exception of lemurs, males are larger than females in all of these species (and much larger than females in patas monkeys), these observations are puzzling; they are discussed further below. Because of the limited effectiveness in most primates of individual retaliation by females, evolution has favored a variety of other female counterstrategies. These are not trivial, but involve critical aspects of
12
BARBARA B . S M U T S A N D ROBERT W . S M U T S
female life histories, including timing of sex and reproduction, mate choice, choice of group, and the development of social relationships and alliances. Below we give some examples of each.
B. TIMINGOF S E X U A L
ACTIVITY A N D
REPRODUCTION
In grey langurs (Agoramoorthy et d.,1988). gelada baboons (Mori and Dunbar. 1985). and captive hamadryas baboons (Colmenares and Gomendio, 1988),takeover of the unit by a new male induces spontaneous abortions in pregnant females, which has been interpreted as female termination of investment in infants who are likely to be victims of infanticide (Mori and Dunbar, 1985; Sommer, 1987). In addition. in all three species. lactating females confronted with anew male may rapidly return to cycling, shortening lactational amenorrhea (Sigg et d.,1982; Mori and Dunbar. 1985; Colmenares and Gomendio, 1988; Winkler. 1988). In the captive hamadryas group, all six lactating females quickly resumed cycles, regardless of the age of their infants, and one grey langur female resumed cycling only 7 days after giving birth (Winkler. 1988).Thus, the presence of a new male overrode the role that infant suckling normally plays in the control of female reproduction (Colmenares and Gomendio, 1988). Whether rapid return to cycling by lactating females has evolved to prevent infanticide, however, remains an open question. In several wild gelada females (Mori and Dunbar, 1985) and in the single case reported for wild langurs (Winkler, 1988), the infants of the nursing mothers who resumed cycles early were not killed, but in the captive hamadryas group, some were. which led the observers to reject the infanticide hypothesis (Colmenares and Gomendio, 1988). However, the hamadryas data are ambiguous because all four victims of infanticide were killed by a single male described as so aggressive in temperament that he was removed from the colony (Colmenares and Gomendio, 19881, and because of abnormal crowding in captivity. In many primates, pregnant females may solicit copulations when confronted with an unfamiliar male (red colobus: Struhsaker and Leland, 1985; grey langurs: Hrdy, 1977; captive patas: Loy. 1985; gelada baboons: Mori, 1979;redtail monkeys: Cords, 1984; mountain gorillas: Watts, 1989). Hrdy (1977, 1979)first argued that such situation-dependent sexual receptivity may reduce the likelihood of infanticide by confusing paternity. Sommer (1987, 1992) rejects this hypothesis for grey langurs, because the pattern of postconception estrus observed over many years at Jodhpur was virtually the same whether the sire was still resident in the troop or a new male had taken over (in other words, it was not “situationdependent”), and because the presence or absence of copulations with a
MALE AGGRESSION A N D SEXUAL COERCION
13
new male did not affect whether or not the female’s infant was subsequently killed by that male. In contrast, female red colobus monkeys who were pregnant when a male attacked infants in their group mated more frequently and later into their pregnancy than did pregnant females either before or after these attacks. A large fraction of these copulations was with the infanticidal male, who did not attack infants of the pregnant females after they were born (Struhsaker and Leland, 1987, p. 96). Thus postconception estrus may serve different functions in different species. Other species in which infants are vulnerable to infanticide, such as red howlers, fail to show postconception estrus in response to invading males (Sekulic, 1983~). When some infants are killed by infanticidal males, others in the same group often escape harm. At Jodhpur, for example, a substantial proportion of vulnerable langur infants (44%) were not attacked by new males, even though other infants in the troop were killed (Sommer, 1992). The fact that some infants go unharmed raises intriguing questions about the factors that may be responsible for their survival, including, perhaps, presently unidentified female counterstrategies. After red howler females experience invasion and infanticide from immigrant males, they rapidly return to cycling but do not conceive immediately. Crockett and Sekulic (1984) hypothesize that the rapid return to cycling incites male-male competition, hastening resolution of the identity of the new alpha male; similarly, delayed conception may benefit females because it increases the probability that their next infant will be sired by the new alpha male. C. MATE CHOICE In primates that live in multimale troops, females often show preferences for mating with dominant males; these preferences have usually been interpreted in terms of the proven genetic quality of the males (reviewed by Small, 1989). The alternative hypothesis that females choose dominant males in order to reduce harassment of themselves or their infants by other males (Wrangham, 1979; Trivers, 1985) has received little attention and deserves further scrutiny. Manson (1991), for example, showed that when rhesus monkey females consort with high-ranking males, they are attacked significantly less often by rival males than when they consort with lowranking males. Since, as noted above, such attacks can lead to severe injury or even death, mate choice could significantly reduce the costs to females of male aggression. Pereira and Weiss (1991) hypothesize that female ring-tailed lemurs choose to mate with males that indicate superior ability to maintain high rank throughout the subsequent birth season,
14
BARBARA B. S M U T S A N D ROBERT W . S M U T S
because such males will be more effective in protecting infants from infanticide by rival males. Similarly, Pope (1990)and O’Brien (1991)suggest that, among red howlers and wedge-capped capuchins, respectively, females benefit from mating with the alpha male, because he provides the most effective protection against infanticide by other males. Janson (1984,1986) suggests that female brown capuchins benefit from mating with dominant males because the protection they provide enables females and their infants to forage undisturbed at rich food sources. In many Old World primates, female mate choice appears designed to facilitate copulation with a number of different males (reviewed by Smuts, 1987a; Small, 1989). Hrdy (1979) and Wrangham (1980a) suggested that, by mating with many males, a female can confuse paternity and thus reduce the probability of infanticide. This hypothesis predicts that females will be particularly interested in mating with males who are most likely to commit infanticide, namely, males that have recently entered a group, or extragroup males who might later transfer into the group (Hrdy, 1979). Indeed, in a number of primate species, females are sexually attracted to such males (reviewed by Smuts, 1987a; Small, 1989). This has been documented both for species living in multimale troops (e.g., Japanese macaques: Wolfe, 1981; rhesus macaques: Brereton, 1981; Manson, 1991; vervet monkeys: Henzi and Lucas, 1980; savanna baboons: Packer, 1979) and species living in one-male troops (grey langurs: Hrdy, 1977; Mohnot, 1984; Sommer, 1988; blue monkeys: Tsingalia and Rowell. 1984; patas monkeys: Olson, 1985). Sommer (1988) has suggested, in addition, that female grey langurs solicit copulations from male invaders in order to incite male-male competition and induce takeover by the strongest possible male. Sommer’s hypothesis could apply to many other species living in one-male troops in which females copulate enthusiastically with invading males (e.g., redtail monkeys: Cords, 1984; blue monkeys: Tsingalia and Rowell, 1984; patas monkeys: Harding and Olson, 1986). Alternatively, females may induce takeover by the strongest male not by mating, but by inciting male-male competition through other means, such as howling (Sekulic, 1983~). Despite promiscuous tendencies, females in many species living in multimale groups show marked mating preferences for particular male partners (reviewed by Smuts, 1987a). For example, in savanna baboons, females often prefer to mate with males with whom they have developed a long-term, affiliative relationship (Seyfarth, 1978; Smuts, 1983a,b, 1985). Smuts (1985) argued that females form such friendships with males, and prefer them as mates, in exchange for protection by these males against aggression from other males toward themselves and their infants. Indeed, when a male defended a female or her immature offspring against other
MALE AGGRESSION A N D SEXUAL COERCION
15
baboons, in 91% of the cases he was the female’s friend (Smuts. 1985). Females form friendships with both high- and low-ranking males. This reflects the fact that, in olive baboons (unlike macaques; see below), even low-ranking males are useful allies, because they are willing to challenge higher ranking males, especially when they receive agonistic support from other males (Packer, 1977; Strum, 1982; Smuts, 1985; No&, 1990). M. E. Pereira (personal communication) has documented similar special relationships among captive redfronted brown lemurs, which are also characterized by male protection of the female against other males in exchange for enhanced mating opportunities. The significance of such special relationships was highlighted when, after a male transferred from one enclosure to another, he killed the infants of one female and “bonded” with the other, leaving her infants alone (M. E. Pereira. personal communication).
D. CHOICE OF GROUP A female’s choice of which group to live in may be strongly influenced by potential male aggression, particularly in those species in which females commonly transfer. In red colobus monkeys (Marsh, 1979) and grey langurs (Sugiyama, 1967), females sometimes emigrate in response to the presence of a potentially infanticidal male immigrant. In howler monkeys, in contrast, patterns of female emigration seem to be more related to female-female competition than to attempts to avoid infanticide ( Jones, 1980; Crockett and Sekulic, 1984). Since mountain gorilla infants are vulnerable to infanticide by extra group males, mothers will clearly benefit from association with a male who can protect their infants effectively. In two of three cases in which an infant was killed despite the resident male’s presence, the female subsequently deserted the male for another (in the third case, it is not known whether or not she transferred)(Fossey, 1984). Even for females who have not experienced infanticide, evaluation of a potential mate’s ability to protect her infants may be the most important criterion for mate choice (Wrangham, 1979, 1982; Watts, 1983, 1989; Stewart and Harcourt, 1987). In chimpanzees, both females and their infants are vulnerable to severe aggression from males from neighboring communities, particularly when their own community range is shrinking due to intercommunity male-male competition (Goodall, 1986; see below). Consistent with this danger, at Mahale Mountains, when all but one of the adult males of K-group disappeared, K-group females transferred en masse to the neighboring M-group, which contained many adult males (Nishida et al., 1985). However, for the first few years after transfer, most male infants of transferred females were killed by M-group males, even though they were often the infants’
16
BARBARA B. SMUTS AND ROBERT W. SMUTS
likely fathers (Kawanaka, 1981; Nishida and Kawanaka, 1985; Nishida, 1990; Nishida et al., 1990). Observers speculate that the M-group males may have regarded these infants as offspring of K-group males, because the ranging habits of females that had transferred from K-group made their community identity ambiguous (Nishida and Kawanaga, 1985. Nishida, 1990).
E. DEVELOPMENT OF SOCIAL RELATIONSHIPS AND ALLIANCES As indicated above, in savanna baboons, gorillas, and chimpanzees, females choose to associate and mate with males who, in turn, help protect females and infants from aggression by other males. In several species, females who have recently given birth increase the time they spend near male “friends” (savanna baboons: Altmann, 1980; Smuts, 1985; Japanese macaques: Takahata, 1982) or near the probable fathers of their infants (gorillas: Harcourt, 1979; red howlers: Sekulic, 1983b; black spider monkeys: McFarland Symington, 1987;long-tailed macaques: Van Noordwijk and van Schaik, 1988; blue monkeys; Tsingalia and Rowell, 1984). Through these close associations with males, females probably gain protection from potential infanticide. Male-female relationsips in macaques appear to involve mutual protection against males who threaten the established social order-maturing males and male immigrants. Like female savanna baboons, female macaques form long-term, affiliative bonds with particular males who selectively protect their female affiliates and the infants of those females from aggression by other males (Kaufman, 1967; Takahata, 1982; Chapais, 1983b,c). In macaques, however, unlike baboons, females consistently prefer high-rankingmales as associates (Takahata, 1982;Chapais, 1983a,c; Hill, 1990; Manson, 1991). This is consistent with the fact that, in macaques, in contrast to baboons, only high-ranking males can effectively protect females from other males, since aggression directed up the male hierarchy is extremely rare. High-ranking males, in turn, prefer highranking females as associates (Takahata, 1982; Chapais, 1983a,c; Hill, 1990; Manson, 1991),and these females support the males during aggressive competition with other males (Koyama, 1970; Fedigan, 1976; Gouzoules, 1980; Chapais, 1983a,c; de Waal, 1989). This mutual support provides the females with protection against aggression from male immigrants and young natal males (Chapais, 1983a,c; Bernstein and Ehardt. 1986; Oi, 1990), and it helps the resident males to achieve and maintain high rank (Koyama, 1970; Bernstein, 1969; Gouzoules, 1980; Chapais, 1983a,c; de Waal, 1980). Studies of naturalistic, captive groups of vervet monkeys indicate the existence of similar, mutually supportive relationships be-
MALE AGGRESSION A N D SEXUAL COERCION
17
tween high-ranking females and dominant males (Raleigh and McGuire, 1989; Keddy Hector and Raleigh, 1992). In wild brown and wedge-capped capuchins, as well, females preferentially associate and groom with the dominant male, and both females and the dominant male direct aggression toward all subordinate males (Robinson, 1981, 1988; Janson, 1984; O’Brien, 1991). Since infanticide has been observed in capuchins (Valderrama et al., 1990), O’Brien (1991) speculates that a strong association with the dominant male may help females to obtain protection for their infants. Bonds with other females can also prove critical in reducing male aggression toward females and young (Smuts, 1987b; Nadler, 1989a; Strier, 1990). Females form coalitions against males in a wide variety of nonhuman primates, including lemurs; New World monkeys, such as howlers and capuchins; and Old World monkeys, such as macaques, baboons, vervets, patas monkeys, and several colobines (reviewed in Smuts, 1987b; see also Robinson, 1981, 1988; Sekulic, 1983a; Pope, 1990; O’Brien, 1991, for New World monkeys). Female coalitions are especially likely in response to male harassment of females or infants. In many species, females gang up on males when they attack, herd, or frighten other females (rhesus macaques: Bernstein and Ehardt, 1985; Japanese macaques: Watanabe, 1979; pig-tailed macaques: Oi, 1990; olive baboons: B. B. Smuts, personal observation; chacma baboons: Hall, 1962; silver-leaf monkeys: Bernstein, 1968; captive chimpanzees: de Waal, 1982). In common squirrel monkeys (Baldwin, 1968), patas monkeys (Hall, 1967; Loy, 1989), vervets (Andelman, 1985), and captive chimpanzees (de Waal, 1982), several females may turn on a male who solicits sex from an unwilling female. The most frequent context in which females form aggressive coalitions against males involves potential, or actual, threat to an infant (grey langurs: Boggess, 1979; Hrdy, 1977; Jay, 1963; blue monkeys: Butynski, 1982; redtail monkeys: Struhsaker, 1977; vervet monkeys; Lancaster, 1972; patas monkeys: Hall, 1968; rhesus monkeys; Bernstein and Ehardt, 1985; Lindburg, 1971; Japanese macaques; Kurland, 1977; Watanabe, 1979; long-tailed macaques: Chance et ul., 1977; olive baboons: Ransom, 1981; Smuts, 1985; common squirrel monkeys: Baldwin, 1968; red-backed squirrel monkeys: Baldwin and Baldwin, 1972; wedge-capped capuchin monkeys: Valderrama et al., 1990; red howlers: Sekulic, 1983c; Pope, 1990; ring-tailed lemurs: Pereira and Weiss, 1991). In species in which females normally remain in their natal groups, female-female coalitions typically involve close kin and are usually directed against females and juveniles from other matrilines (reviewed by Walters and Seyfarth, 1987). In striking contrast, when the target is an adult male, females often form coalitions with females to whom they are not closely related (rhesus monkeys: Bernstein and Ehardt, 1985; red-
18
BARBARA B. SMUTS A N D ROBERT W. SMUTS
backed squirrel monkeys; Baldwin and Baldwin, 1972;grey langurs: Hrdy , 1977;olive baboons: B. B. Smuts, personal observation; vervets: Cheney, 1983b;bonobos: Kano, 1987; Furuichi, 1989).Such coalitions can mobilize very quickly in response to male aggression, since any females nearby can be recruited (B. B. Smuts, personal observation). This may help to explain why, as noted above, females can sometimes individually dominate males in spite of the females’ smaller size: Males may sometimes defer to individual females because of the ever-present possibility that one female opponent may suddenly become many (cf. Robinson, 1981, 1988, for wedgecapped capuchins). Thus, female cooperation against males may benefit females both in the short-term, by halting male aggression, and in the longterm, by making males more hesitant to harass females or young because of the risks of counterattack by a female mob. How effective are female coalitions in reducing male aggression? In Japanese macaques (Packer and Pusey, 1979), vervet monkeys (Cheney, 1983a,b), and patas monkeys (Hall, 1967), female coalitions can drive males from the troop or prevent them from entering in the first place. Among capuchins, female coalitions probably help to keep non-alpha males peripheral, both socially and spatially (Robinson, I98 1 ; O’Brien, 1991). In wild red colobus monkeys, female-female coalitions have been observed to kill immigrant males (Starin, 1981), and among captive talapoin monkeys, female-female coalitions have also resulted in killing of males (Rowell, 1974). In grey langurs and red howlers, however, female coalitions are not very effective against infanticidal males (Hrdy, 1977; Crockett and Sekulic, 1984). Few data are available to evaluate the effectiveness of female coalitions against males. For instance, no published data indicate whether female coalitionary aggression toward a male reduces the likelihood that he, or male witnesses, will show subsequent aggression toward females or young. Clearly, this topic deserves further attention.
F. FORMOF THE SOCIAL SYSTEM Until this point, we have considered how, given particular features of the social system (e.g., presence of related females; one-male vs. multimale groups), females may develop counter strategies to resist male aggression. Here, we briefly consider how sexual coercion and female strategies to resist it may influence the form of the social system itself. Mountain gorillas provide the clearest evidence that male sexual coercion and female counterstrategies can determine the form the social system takes. In these apes, almost all infants who lose the protection of a mature silverback male (in most cases, because he has recently died) are soon
MALE AGGRESSION AND SEXUAL COERCION
19
killed by other males (Watts, 1989). In contrast, contrary to Fossey's earlier (1984) suggestion, recent data indicate that infants living in a group with a mature silverback are rarely killed (Watts, 1989). These observations provide strong support for Wrangham's hypothesis that infanticide is the selective force responsible for group-living in gorillas ( Wrangham, 1979, 1982, 1987a; Watts, 1983). Because females rely for protection primarily on the silverback male, rather than on other females (Watts, 1989), the gorilla social system is based not on bonds between related females, but on bonds between (usually unrelated) females and the adult male(s) in the group (some gorilla groups have more than one mature male: Harcourt, 1979; Stewart and Harcourt, 1987). Male sexual coercion may also help to explain the distribution of onemale versus multimale polygamous primate groups-a problem that remains unresolved despite numerous attempts to explain it in terms of male competitive strategies (Glutton-Brock et al., 1977; Ridley, 1986). Several people have argued that we also need to consider the effect of female strategies on the number of males in the group (Wrangham, 1980a; van Schaik and van Noordwijk, 1989; Altmann, 1990). Altmann (1990) proposes that the threat of male infanticide may result in the evolution of synchronized female ovulation, which in turn will make it more difficult for one male to control all of the fertile females in his group. This will result in a transformation from one-male to multimale groups (see Section VII for a discussion of why infanticide is generally reduced in multimale groups). V.
MALE AGGRESSION AGAINST FEMALES I N CHIMPANZEES
Chimpanzees have been studied continuously and intensively for more than 25 years at two study sites in Tanzania, Gombe National Park (Goodall, 1986) and Mahale Mountains National Park (Nishida, 1990). Although they have also been studied at other sites in East, Central, and West Africa (Heltne and Marquardt, 19891, these studies have not yet produced detailed information on male aggression against females. Thus, it remains to be seen whether the patterns observed at Gombe and Mahale characterize all chimpanzees or are limited to populations living in particular areas. Chimpanzee males show two main kinds of aggression against females: aggression against potential mates from the same community and aggression against nonestrous females from neighboring communities. Each kind is reviewed in turn.
20 A.
BARBARA B. SMUTS AND ROBERT W. SMUTS
MALE AGGRESSIONTOWARD POTENTIAL MATES
Goodall succinctly summarizes the role of male aggression in chimpanzee sex as follows: “Almost always, unless he is crippled or very old, an adult male can coerce an unwilling female into copulating with him” (1986, p. 481). In chimpanzees, males copulate under three different circumstances (Tutin, 1979; Hasegawa and Hiraiwa-Hasegawa, 1983, 1990):promiscuous, opportunistic mating, which involves frequent copulations with many different males in the group setting; possessive mating, which involves a single male’s attempts to monopolize copulations in spite of the presence of other males; and consortships, in which mating takes place between one male and one female who travel apart from the rest of the community for several days or weeks. Promiscuous, opportunistic mating typically occurs early in the female’s cycle before her swelling reaches maximum tumescence, and is unlikely to result in fertilization. As she nears ovulation, she will typically either participate in a possessive mating relationship (most likely involving the alpha male), or form a consortship. Male aggression against females occurs in all three contexts but especially during consort formation (Goodall, 1986). Among chimpanzees at Gombe, Tanzania, consortships are probably responsible for at least one-third of all conceptions, and they greatly improve a lower ranking male’s chances of fathering offspring (Goodall, 1986). It is thus not surprising that males appear highly motivated to form consortships. In order to do so, they must convince a female to follow them away from other males and to remain with them for at least several days (sometimes as long as 5-6 weeks) until her sexual swelling begins to subside, which indicates that ovulation has occurred. In order to accomplish this end, males employ what Goodall terms “a fair amount of brutality” (1986, p. 453). Males often try to initiate consortships with a female long before her sexual swelling reaches the full size associated with ovulation. The male’s apparent goal is to escape the rest of the group early, before competition for the female becomes too intense, and then to sequester the female through the period of ovulation. Aggression is most common during the early stages of consortship, when the male is trying to lead the female away from other males by traveling away from the core area of the community range (Tutin, 1979; Goodall, 1986). During this time, the female often refuses to follow the male, and she may scream, which sometimes attracts other males. If she is approaching ovulation, a higher ranking male may disrupt the consortship and she can escape her suitor. However, if she is not fully swollen, other males show little interest and she has a harder time escaping.
MALE AGGRESSION AND SEXUAL COERCION
21
Goodall (1986) reports that if the female refuses to accompany the consorting male, he will often use violence to force her to follow him. For example, Evered spent 5 hr leading Winkle north across a valley, away from other males. During these 5 hr, he repeatedly displayed at her aggressively and attacked her six times, twice severely (Tutin [ 1975, 19791 and Goodall [ 19861provide numerous additional vivid examples of male aggression in this context). Once the pair has moved far from the core of the community range, the female becomes more cooperative (probably because she is in an unfamiliar area and relies on the male for protection) and the male becomes more relaxed and tolerant (probably because he has left his mating competition far behind)(Goodall, 1986). Male aggression appears to be quite effective in convincing females to go on consort. This is well illustrated by the case of Jomeo, an adult male who showed the lowest rates of punitive aggression toward consort partners. He was also least successful in forming consortships and was the only adult male who is thought not to have sired any offspring. The significance of male aggression during consort formation may help to explain why males frequently conduct severe, apparently unprovoked attacks on cycling females whose sexual swellings have not reached full tumescence. Goodall (1986) hypothesizes that these attacks function as intimidation designed to increase the chances that the female will submit to the male’s advances in the future. Similarly, she argues that when a female appears to follow a male on consort voluntarily, her lack of resistance does not necessarily indicate willing participation; rather, it may simply reflect previous experiences with male aggression. Along the same lines, the low frequency with which females ignored adult male invitations to copulate (4.1%) may also reflect previous experience with male aggression. When a female did ignore a male’s invitation to copulate, on one out of every five occasions he responded with aggressive displays or chases and she gave in. These hypotheses linking female acquiescence to previous aggression, or to the expectation of future aggression, seem intuitively reasonable but are difficult to test (see Section VIII,B for further discussion).
B. MALEAGGRESSION AGAINST FEMALES FROM OTHERCOMMUNITIES At Gombe (Pusey, 1979; Goodall, 1986) and Mahale Mountains (Nishida, 1979; Nishida and Hiraiwa-Hasegawa, 1985), young, sexually cycling, nulliparous females typically transfer, either temporarily or permanently, to neighboring communities; while there, they mate with community males. Males welcome such females and sometimes even pro-
22
BARBARA B. SMUTS A N D ROBERT W. SMUTS
tect them from hostility by resident females. In dramatic contrast, when chimpanzee males encounter mature, anestrous females from another community, they typically respond with intense, sometimes lethal aggression, as illustrated by the killing of the old female, Madam Bee, at Gombe. The attacked females are not immigrants but are encountered in areas of overlap between the ranges of the two communities or in their own community range during invasions by neighboring males (Bygott, 1972; Goodall et ul., 1979; Goodall, 1986; Nishidaand Hiraiwa-Hasegawa, 1985). From 1975 until 1982, observers at Gombe witnessed 25 encounters between adult males from the habituated community and strange, anestrous mothers from neighboring communities (Goodall, 1986). Nineteen of these encounters were aggressive, involving chases or attacks. Fifteen attacks were observed, and, with one exception, they were extremely severe. Three attacks resulted in the death of the female’s infant. In 10 cases, observers were able to see the victim after the attack. Each time she was bleeding heavily from wounds on the limbs and/or back and, in at least 8 cases, on the face or head; some females may have died of their wounds. Males showed a marked degree of cooperation in this context. All of the attacks involved aggression by more than one male; some involved as many as six males. The males often embraced one another before attacking the female. In one case the males persistently “hunted” (Goodall’s term) a strange female before attacking her, and, in another case the males cooperated to surround the female as they sometimes do when hunting baboons (Goodall, 1986, p. 494). Several similar attacks have also been observed at Mahale Mountains (Nishida and Hiraiwa-Hasegawa, 1985). In two instances involving the same female, observers intervened because they were certain she would be killed (Nishida and Hiraiwa-Hasegawa, 1985). At both Gombe and Mahale, although infants may be killed and even cannibalized during these attacks, observers gained the impression that the males’ aggression was directed primarily at the mother (Goodall, 1986; Nishida and HiraiwaHasegawa, 1985). Several explanations have been proposed to account for aggression toward anestrous females from other communities. Wolf and Schulman (1984) argued that males attack older females because they have low reproductive value, and, if killed, additional habitat becomes available for younger females of higher reproductive value who may eventually mate with the killers. Many of the females attacked by males from other communities were not, however, especially old (Goodall, 1986; Nishida and Hiraiwa-Hasegawa, 19851, so this explanation cannot account for all of the cases. Nishida and Hiraiwa-Hasegawa (1985, p.12) speculated that, by attacking neighboring females who may compete with resident females for
MALE AGGRESSION A N D SEXUAL COERCION
23
food and other resources, the males may “court the favor” of resident females. However, resident females will rarely witness such attacks, since they are relatively uncommon, and typically only one or two resident females are likely to be present (at Gombe, on average, only I .25 resident females were present during attacks on strange females, based on data in Goodall, 1986, Table 17.2). Goodall (1986) provides a third hypothesis, suggesting that repeated brutal attacks on mothers may facilitate recruitment of their daughters to the attacker’s group. In support of this idea, she notes that, at least at Gombe, many daughters retain close bonds with their mothers and remain as residents in their natal groups. If the mother-daughter bond is weakened due to repeated attacks, or destroyed because the mother is killed, the daughters may be more likely to transfer permanently to the neighboring group. Consistent with this explanation, all but one of the (at least) five attacks on the old female, Madam Bee, occurred when the attacking males were recruiting her daughter, Little Bee; during this period, Little Bee transferred to their community. After Madam Bee’s death, her other daughter, Honey Bee, associated with the attackers’ community off and on for 3 years. If Goodall’s explanation is correct, then male aggression toward females from other communities would qualify as a form of sexual coercion (although in this instance the individual who directly suffers the cost of the coercion is not the males’ potential mate, but her mother). Whatever the explanation for the brutal attacks on strange females, they clearly occur regularly at Gombe and Mahale, and thus constitute an important selection pressure influencing the behavior of female chimpanzees. Female chimpanzees forage, often on their own with dependent young, in dispersed, but overlapping, home ranges. Males range more widely and cooperate in defending a community range that encompasses that of several females. As adults, and often after transferring from their natal communities, female chimpanzees become clearly identified with a particular community, i.e., with a particular group of males (Goodall, 1986; Nishida and Hiraiwa-Hasegawa, 1987). Although female dispersion is probably a product of feeding competition (Wrangham, 1975, 1979), the fact that females “belong” to a particular male community, rather than ranging and associating freely regardless of community boundaries, is probably a response to violence by males from neighboring communities. This conclusion is supported by observations from Mahale Mountains indicating that infants of lactating females with ambiguous community identity are especially vulnerable to infanticide by males (Kawanaka, 1981; Nishida and Kawanaka, 1985; Nishida, 1990; Nishida et af., 1990). Thus, among chimpanzees, as among gorillas (see above), male aggression against females appears to have influenced the form the social system takes.
24
BARBARA B. SMUTS AND ROBERT W. SMUTS
VI. MALEAGGRESSIONAGAINST FEMALES IN OTHERMAMMALS Table I summarizes information on male aggression against females in selected mammals. It is not exhaustive, and it is biased toward large, diurnal mammals whose behavior has been studied in the wild. We present the information in Table I to illustrate (a) the fact that male aggression against females and infants occurs in a variety of mammals, (b) the varied contexts and forms of this aggression, (c) the potential costs to females, and (d) the different kinds of counterstrategies that females exhibit. Most of the instances of male aggression toward females were interpreted by the authors as sexual coercion, as defined in this article. In surveying the literature on nonprimate mammals, we encountered few instances of male aggression toward females in nonsexual contexts. A. TYPESOF MALE AGGRESSION
Females in many mammalian species experience both sexual aggression and infanticide by males. Male sexual aggression appears to be most common in gregarious species in which females do not form long-term bonds with a single male (or, as in lions, with a group of allied males), so that females are exposed to a number of males competing for sexual access to them (e.g., fallow deer, bighorn sheep, African elephants, several pinnipeds, bottlenose dolphins). In contrast, females that do form longterm bonds with particular males (wild horses, lions) are usually protected from routine sexual harassment by other males and do not experience sexual aggression from their long-term male associates. These females, however, are vulnerable to infanticide (in lions) or induced abortion (in horses) during male takeovers. Female rodents and farm cats also experience infanticide when they encounter strange males. In species in which estrous females are exposed to several competing males, they are typically chased and herded, and sometimes kicked, pushed, or bitten by males attempting to mount. In some species (such as fallow deer or African elephants), males apparently do not frequently injure females, and the main costs to females of sexual aggression are probably loss of feeding time and energy expended in escape. In other species, sexually aggressive males sometimes severely injure and even kill females (e.g., several pinnipeds). In addition, in their aggressive attempts to gain access to estrous females, males sometimes cause death of infants (e.g., crabeater seals, sea lions, elephant seals). Little information is available on species in which females are solitary. In sea otters (Foote, 1970) and many other mustelids (martens, weasels, skunks, mink) and viverrids (civets, fossas, some mongooses), copulation is accompanied by intersex-
TABLE I MALEAGGRESSIONAGAINST FEMALES A N D INFANTSIN SELECTED NONPRIMATE MAMMALS‘ Species Fallow deer (Dama damn)
Social/mating system Polygynous; dominant males defend territories on leks
Context of male aggression Prolonged chases of fertile females by nonterritorial males
Potential costs to females
Female countentrategies
Energetic costs of avoiding male Remaining in territories of “sexual harrassment”; dominant males provides protection from other males potential wounding by male antlers Rocky Mountain bighorn sheep Romiscuous; multimale. Single males or groups of Potential injury from attacks by Females try to escape chasing (Ouis canodensis) multiiemale groups subordinate males chase blocking males; prevention of and blocking males; if females and push, butt. and female mate choice: unsuccessful, female mates. kick them until they submit to disruption of feeding; perhaps to avoid further copulation. “Blocking”: male restricted movements; attacks forceably sequesters female energetic costs of fleeing and mates with her; prevents her from approaching other males by herding, kicking, and pushing Wild horses and Assateague Polygyny; single-male, Males invade bands and try to Females occasionally suffer bite Wild hones: females kick, turn ponies (Equus caballus) multifemale bands steal femaks by herding. or wounds; in wild horses. 86% away. and run from males Iry to take over band by of females 20 mg dry weight) Between year variability Cumulative trap days
Mean (No. years)
CVh (%)
S E (%)
.35 .85 .80
.28(6) .87(5) .89(4)
91.5 9.3 21.2
21.9 0.8 7.6
NM. Desert grassland; AZ, riparian; and TX, riparian. Only prey encountering sticky traps and between 4 and 25 mm in body length (capture range of mature A. aperra) were included in the analyses. CV, Coefficient of variation computed on arcsine transformed data.
to a limit (Fig. 2a). Beyond this limit, rather than producing additional eggs, A. aperfa incorporates greater quantities of yolk in its eggs (Fig. 2b; Riechert and Tracy, 1975). Because the yolk is what the newly hatched spiderlings subsist on until they are able to obtain their first insect meal, I assume that the quantities of yolk a female is able to provide for her offspring affects offspring survival. Recent work by Fred Singer and myself on male reproductive success indicates that the body mass that A. upertu males are able to achieve as a result of their feeding history is a significant determinant of mating success as well. Males collected at those web sites that afford higher levels of prey weigh far more than males collected from lower quality microhabitats (Fig. 3). At maturity, males give up their sites and search for matings. We have found that the distances males can travel is a function of their body masses (Fig. 4a) and that mating success is also correlated with body mass (Fig. 4b). Note that, as is the case for most spider species, male A. uperta are smaller on average than females and small males not only are frequently chased off the webs by large females but may be killed during courtship as well. (Why are males smaller than females? Females need a larger frame to accommodate egg masses, while we have some evidence that there is a reproductive advantage for males to mature earlier and hence at a smaller body size than females [S. E. Riechert and F. D. Singer, unpublished data]). Active habitat discrimination is evidenced for members of this population (Riechert, 1985),with a preference shown for those few sites (3% of the
106
SUSAN E. RIECHERT
a
Surface 60.0
-
Shade
Sun
-@-
-----
upper lower
C
10.0
0
'
'
'
*
'
'
. .
1000
I
.
2000
Time
60.0 r 50.0 -
0
1000
Time
2000
'
EVOLUTION OF BEHAVIORAL PHENOTYPES
I07
available habitat cover) that afford the foraging time and prey availabilities requisite to survival and reproduction by females in most years (Fig. 5 ; Riechert, 1979, 1981). The high-quality sites in the grassland habitat are holes or depressions that are formed by the scouring effect heavy summer rains have in the area (Riechert, 1976). These protected sites provide the spiders with shade and often have accumulated litter which decreases the thermal radiation a spider receives. There is also less wind damage to webs that are built in these depressions than occurs on the grassland surface itself. The spiders do not cue in on the depressions because of their structural characteristics. Rather, they actively seek shade and, when in shade, move along temperature gradients to settle at their preferred body temperature of 30°C (Riechert, 1985).Agelenopsis aperta also responds to both olfactory and vibratory cues emitted by insects in locating its web sites in the vicinity of such insect attractants as flowering herbs and fecal material (Riechert, 1976, 1985). Thus, individual success in the NM grassland habitat is highly dependent on a spider’s ability to locate a web site in a microhabitat that provides it with sufficient insect prey to survive to reproduction. This population is food limited and the availability of adequate sites is a major selection pressure on individuals. Unlike the AZ riparian population described below, the grassland population receives no measurable predation pressure from foraging birds, the major visual predators on this spider species (Riechert and Hedrick, 1990). B. RIPARIAN(ARIZONA) The second population occupies a riparian (woodland) area in southeastern Arizona. This is characterized by a tree canopy, a grass and leaf litter forest floor, and the presence of a permament, spring-fed stream which serves as an insect reservoir. Prey are thus present in abundant supply (Table I ) and 89% of the habitat cover is appropriate to ensure spider survival to reproduction (Riechert, 1979). Interestingly, the 11% of the
FIG. I . Predicted Agelenopsis aperra body temperatures for the course of a clear day in mid June at (a) a surface site (poor quality) versus (b) a depression site (high quality) in the NM grassland habitat. Two curves reflect spider temperature in full sun and full shade, respectively. Actual spider temperature when out on the web will fall somewhere between these two curves depending on how much sunlight the individual is exposed to. (Note that, in actuality, little shade is available at surface sites.) Spider foraging is permitted only when body temperature falls between the upper and lower temperatures of its activity range, the area demarcated in the figure by the dashed lines. (Data from Riechert and Tracy, 1975.)
SUSAN E. RIECHERT
108
-
900-
a
I-
$
-
000-
t
700-
W
W
E 600-
V
3
a
n
500-
u)
W
s2
400-
300-
CORRELATION COEF F I C I EN T
II J
SIGN. AT P=
m
8 6
4
Geospizo fuliginosa
0
1
L,, 0
2
3
L
5
6
10
1
2
3
L
5
15m
J
Minimum Distance FIG.1. Comparison of effect of tethered live owl and dummy owl (Asio.flummeiis) on minimum distance of mobbing by individuals of two species of Darwin’s finches. The live owl is approached less closely (both species pooled: two-tailed pz = 2 x 10“. Mann Whitney U-test). (Redrawn from Curio, 1969.)
ASPECTS OF ANTIPREDATOR BEHAVIOR
I39
not differ from that given to the live predator (n = 33); if anything, there was a tendency to exceed it (Curio, 1975,Table I). Similarly, the responses given to a live tawny owl fell within the range of responses elicited by both mounts of a tawny owl and even the tiny pygmy owl (Glaucidiurn passerinurn)(Curio, 1975, Table 11). Hence, there is good evidence that dummies can justifiably be viewed as stimuli eliciting the full-blown response to the predators they represent, but the response strength thus obtained should always be calibrated against that elicited by the natural predator. Almost all studies of antipredator behavior employing predator dummies fall short of this control. Milinski (personal communication) has suggested an objection: That the stimulus value of a dummy may equal that of the real thing by chance. Although it is recognized as a weak form of predator, the strangeness of a mount per se may compensate for any stimulus deficiency. However, as will be shown below for a nonpredatory bird, the nature of a mount per se would not appear to intimidate prey birds. The skepticism about the dummy method expressed by Knight and Temple (1986a) appears not quite justified. They base it mainly on the fact that (a) a dummy used in their own study elicited a quantitatiuely different response compared to the real animal; (b) other authors (refs. in Knight and Temple, 1986a) who failed to employ the real animal as a control gave up experimenting with dummies (in one case-ironically-because the author was not sure whether the dummy-elicited response was deficient in any way). Their reason (a) undermines my insistence that one must control for any deficiency of a dummy against the standard of the live predator. Their reason (b) is based on negative evidence and therefore not very helpful. Knight and Temple (1986b, 1988) themselves used predator models without calibrating them against the live predators and without mentioning their previous skepticism. Loughry (1987) found that blacktailed prairie dogs (Cynornys ludouicianus) harassed an unrestrained snake in ways that differed subtly from responses to a tethered one. Most of the differences reflected the different locale of the two types of presentation: the mouth of burrows in natural encounters and above ground in staged encounters. However, differences between categories of animal (male/ female; parenthonparent) were consistent despite the two manners of presentation. The assertion of Loughry (1987) that this result was the first to validate the use of a restrained predator is clearly unwarranted in view of the earlier evidence summarized above. Lorenz (1943) had described in a number of birds a response (“Gespenstreaktion” = “ghost response”) that is elicited by a conspecific of unusual appearance, e.g., an albino. This response resembles the typical antipredator response of the species concerned on a number of counts.
I40
E. CURIO
According to Lorenz’s view, a predator dummy might trigger protective behavior because of its “Gespenst” character. In order to test for this potential effect of predator dummies, one would have to render a familiar object odd by, for example, mounting a specimen of a familiar, nonpredatory species and comparing it to similar-sized genuine predator stimuli. We performed this experiment with Geospiza difficilis, the sharp-beaked ground finch, living syntopically (= in the same habitat) with the extremely common Galapagos dove (Zenaida galupagoensis) on Tower Island (Genovesa). In experimentally naive finches, a stuffed Galapagos dove facing the subject elicits only a fraction of the fear response given to a stuffed forward-facing head of the short-eared owl that also occurs on Tower Island (see Fig. 23, Section 111,C). The dove is even less effective than is a torch, a novel stimulus, though this difference falls short of significance (Curio, 1969). The low stimulus effectiveness of both the dove and the torch rules out an alternative explanation based on novelty. One could submit that the owl surrogate is more effective than the dove because the owl is less common than the dove and, hence, some fear of novelty (Section II,A,3) might have rendered it more effective. This idea, however, can be dismissed since the lamp was even more novel than the owl; as this owl is active during the day, it was certainly familiar to the finches. The fact that the mounted dove released some fear may have been due to the novel circumstances under which it was presented to the finches. The mounted dove experiment thus seems to rule out the Gespenst hypothesis for the efficacy of dummies; the Galapagos dove was totally familiar to the finches, did not elicit a fear response in the wild, and was larger than the owl head. Therefore, being dead or mounted as a specimen per se can hardly trigger antipredator behavior, thus validating in yet another way the approach using predator dummies. The same confidence may not apply to a dead conspecific (Section II,B ,3,b). 2 . Suddenness A sudden movement or a loud noise set forth a variety of alert behaviors like freezing, crouching, precipitous flight, or facing the alerting stimulus for closer examination. Both of these alerting events may signal danger, since most predators rely on surprise, and surprise depends on speed of attack. The exact nature of the response will depend on many details of context, the subject’s history, and on the stimulus itself. The last of these may vary widely but still elicits a response, given some minimum intensity. Given this low stimulus specificity, examining the effect of context appears more rewarding than examining stimulus quality. The startle response, a simple muscle twitch, has been studied in European starlings (Sturnus vulgaris) by Pomeroy and Heppner ( 1977). This
ASPECTS OF ANTIPREDATOR BEHAVIOR
141
response is very quick, averaging 76.4 msec for visual and 80.6 msec for acoustical stimuli. Because of their greater realism, studies of the more complex takeoff response to sudden movements deserve closer attention. Davis (1975) studied socially induced takeoff responses in domestic pigeons, the vertebrate with the highest recorded flicker fusion frequency (15O/sec) (ref. in Frankenberg, 1981). On being alarmed by an electric current through a perching grid an actor pigeon took flight and other observer pigeons followed suit immediately. Response latency to the electric stimulus was 250 msec but there was no measurable time difference between the alarmed actor’s takeoff and that of the observers. The actor’s takeoff was not preceded by any preflight activities whereas spontaneous, noninduced flights were. The more prevalent were these preflight activities the less contagious were spontaneous takeoffs. Although subtler differences, as yet undiscovered, between both types of takeoff may exist, it appears likely that preflight activities suppress socially induced takeoffs of observers. (These inhibitory signals must differ from those that spark off the synchronized takeoffs of pigeon flocks not preceded by any external alerting stimulus.) The impact of the social context on protective takeoff has been studied by Frankenberg (1981) in a more natural setting in European blackbirds (Turdus merufa).Following up my suggestion, he examined experimentally the “alerting others” hypothesis of avian mobbing behavior. This predicts that birds that perceive others mob are alerted to flee from imminent danger, which normally would be the very predator that elicited mobbing from the actor(s). An actor blackbird was induced, in its cage, to mob a stuffed owl that an observer blackbird was prevented from seeing (Fig. 2). The observer in a neighboring cage was allowed to see and hear the actor, and it was additionally exposed to a startle stimulus, a wooden lever that the actor could not see. This lever could be made to rotate suddenly through an arc of 180”,in a movement across the observer’s field of view to the observer’s view, thereby triggering an escape response away from the startle stimulus. Without the actor’s mobbing, the observer took 600 msec to takeoff, whereas with the actor’s mobbing preceding the sudden startle stimulus, response latency fell to 300 msec. This alerting effect of the actor occurred prior to its mobbing the owl, and thus was due to subtle signs of the actor’s incipient alert behavior. The benefit to the observer bird is considerable; with an accipiter hawk attacking at a speed of 65 km/hr, an unwarned blackbird would escape after 12 m of approach, an alerted one would have about 6 m, a meaningful difference. In another experiment by Frankenberg (1981), both the actor and the observer were allowed to perceive the startle stimulus, all other features of the experiment remaining the same (Section III,D,4,a). As a conse-
142
E. CURIO
: :
Front Lid
C
S t i mulus,
Observer l m
FIG.2. Experimental setup for testing the effect on the startle response of an observer blackbird of an actor blackbird’s mobbing that is induced by a little owl (Arhene nocruu) mount. Curtains (C) prevent the actor from seeing the startle stimulus visible to the observer and prevent the observer from spotting the owl. The actor’s mobbing is started by removal of the front lid. (Redrawn from Frankenberg, 1981.)
quence of this, the actor bird was found to show the same decrease of its response latency as did the observer. Hence, a blackbird’s own predator mobbing and/or being alerted facilitates its escape should the necessity arise. This is quite remarkable since the actor might have anticipated an attack from the owl, not the innocuous wooden lever, and yet its escape response was not compromised. Perhaps any conflict between an anticipated attack by the owl and the actual movement of the lever was offset by the bird continually monitoring the source of danger, whereas the observer could only infer a danger from the actor’s behavior. This possibility could be tested by scoring the actor’s latency upon an attack from the owl; one would expect it to decrease still further. There were mutual effects between the two blackbirds, one of which is reminiscent of Davis’s (1975) experiment with pigeons. After the observer took off on seeing the startle stimulus when it was hidden from the actor, the latter stopped mobbing, and after a latency averaging 3.9 sec took off itself. It later resumed mobbing and this induced the same response in the observer. The magnitude of its delay to takeoff contrasts with virtually no delay in the pigeons, perhaps due to the more social nature of the latter species. The impact of the observer’s startle response on the same response in the actor demonstrates unequivocally that a social influence
ASPECTS OF ANTIPREDATOR BEHAVIOR
143
overrode the effect of the unchanging enemy stimulus. The actor’s takeoff latency was about 13 times the observer’s when it responded to the startle stimulus (3.9 vs. 0.3 sec). This cautions against comparing the two takeoff responses directly. A startle stimulus may imply a real danger or an apparent danger. Discrimination between the two takes time, although there is no direct evidence for the tacit assumption that an artificial startle stimulus takes the same response time as does a biologically more meaningful stimulus pattern. If the startle stimulus persists and if, on closer examination, it turns out to involve a predatory threat, a typical startle response will give way to a different class of responses. These consist of predator exploration, mobbing, or attack behavior, depending on the species, past experience, and context (see also Andrew and Clayton, 1979; Clayton and Andrew, 1979). Depending on the time available for identification, one and the same innocuous stimulus may elicit a startle response when appearing suddenly, or be ignored if examined for some minimum period of time. Harmless flying birds like blackbirds or wood pigeons (Columbapalurnbus) triggered an aerial predator call in my silver pheasants (Lophuru n. nycrhemeru) when they suddenly came into view as they crossed the edge of a nearby roof. If flying individuals of the same two species were detected while approaching from a longer distance, giving the pheasants time for identification, those predator calls were produced much less often. The distinct avoidance by the pheasants of house walls may be regarded as a means to keep as large a section of the sky as possible under surveillance and thus forestall the surprise attack of raptors (E. Curio, unpublished). Similarly, carrion crows (Coruus corone) sent black-headed gulls (Larus ridibundus) into precipitous flight when appearing over the rim of a steep slope but were ignored when seen flying for longer (Kruuk, 1964). From these observations springs the hypothesis that greater opportunity for enemy identification is linked to fewer startle responses and, hence, to a lower rate of misidentifications. This idea is supported by yet another line of evidence. The threshold for startle responses is lowered and vigilance perhaps increases when the scope for surprise attack increases. Lohrl (1950b) found that marsh tits (Parus palustris) emitted aerial predator calls in foggy weather more often and to a broader range of flying birds (see also Ferguson, 1987). Furthermore, vigilance for predators increases under several conditions: 1. When the potential victim is closer to a site more liable to attacks (Lendrem, 1983). 2. When the potential victim is at the periphery of a flock, that is, in a
144
E. CURIO
more vulnerable position (Altmann, 1958; Goldman, 1980; Jennings and Evans, 1980; Prins and Iason, 1989). 3. When the potential victim is in a smaller flock (Lazarus, 1972; Bertram, 1980; Caraco et al., 1980; Hoogland, 1979). The detection time for a sudden stimulus is longer in smaller groups (Dimond and Lazarus, 1974: Quelea quelea; Magurran et al., 1985: Phoxinus phoxinus). Despite a decrease of the individual scanning rate with flock size, the corporate vigilance of the flock as a whole increases (Caraco, 1979). 4. When the potential victim has recently perceived a predator nearby (Caraco et al., 1980; Lendrem, 1980, 1984a,b; Poysa, 1987). In whitebrowed sparrow-weavers (Plocepasser mahali), Ferguson (1 987) found the individual scanning rate did not decrease with increasing flock size and that larger flocks attracted predators more often. This greater vulnerability of flocks is hypothesized to cancel the usual inverse relationship between vigilance and flock size. This raises the question why so many other species can afford to lower individual vigilance as flock size increases. Elgar (1989) lists a large number of confounding factors and calls for an experimental approach to the flock size-vigilance relationship. 5 . When the potential victim wears a conspicuous plumage as compared to other times of the year (Lendrem, 1983). 6. When the potential victim has a poor knowledge of its home range. The longer the period for which the home range has not been monitored for predators the more vigilant a bird is during a current episode of monitoring. When patrolling an area of our garden where they had not been for a long time, my silver pheasants were more on the alert, scanned more often, and bunched more closely together, especially the chicks before their independence (E. Curio, unpublished, 8 years of observations). Similarly, a nest-building blue tit (Parus caeruleus) spent more time in scanning for predators after longer absences from the nest box (Lendrem, 1980). The time scales over which this memory window works can apparently vary tremendously. The silver pheasants patrolled parts of their common home range at intervals of many days or even weeks; the blue tit’s nest visits were spaced apart by seconds or minutes. Given that the threshold for a startle response tends to vary as a function of context (see above), one would predict that startle responses should be commoner in less frequently visited parts of the home range. This idea has so far gone untested. 3. Novelty a . General. A stimulus will be called “novel” here if it fulfils two criteria: First, it elicits antipredator responses based on pattern recogni-
ASPECTS OF ANTIPREDATOR BEHAVIOR
145
tion, as opposed to the simple startle stimuli relying on mere conspicuousness (Section Il,A,2). The following qualify as such responses: concealment of self and offspring, mobbing, distraction displays, and attacking the predator. According to Ratner (1967), freezing should be included here too, though I think it is more common in response to startle stimuli. Second, the object is encountered for the first time in an individual’s lifetime. For the fear of novelty to work there must have been “previous encoding of the familiar” (Bronson, 1968). The research reported below fulfils this criterion since the birds under study have in no case been deprived of patterned stimulation (about the value of restricted deprivation see Section 111,A). Sometimes objects fitting the first criterion have been called “strange,” especially if a response could be triggered by some dummy that, at first glance, bore little resemblance to a real predator. However, the term has never been operationally defined and, on closer inspection of the examples that have been quoted, novelty could not be ruled out as a sufficient explanation, and/or failure to match in all respects a true enemy stimulus pattern could not be ruled out either. Here I therefore examine whether either novelty or enemy-specific cues are sufficient for analyzing the information content encoded in the objects triggering antipredator behavior. A discussion of the role of novelty in shaping the development of antipredator responses and the processes underlying it will be deferred to Section III,B and II1,C. b. The Stimulus Effectiveness of Novel Objects. Objects tested for the effect of novelty fall largely into two classes. In order to know whether only predators elicit protective behavior in their prey, novel nonpredatory species have been tested and compared to predator species of the same class of vertebrates. Novel objects should ideally not contain any key stimuli that are known to be features of sympatric enemies. Novelty was ensured by picking similar-sized species that live in allopatry with the species to be tested. Kaspar Hausers deprived of any experience with heterospecifics have been used to test for the effectiveness of any species, both harmful and harmless. This potentially very powerful technique depends on having excellent raising and maintenance skills (e.g., Schleidt, 1964). The other class of stimuli comprise artificial, inanimate objects that are as dissimilar as possible to the range of objects in the natural world of the species to be tested. Such stimuli could be used to test if novelty per se were effective. Studies using both types of stimulus are illustrated in Fig. 3. Of the two studies with captives (b and c), one was validated by comparing the responses of caged birds to nesting ones in the wild (c). As can be seen,
20
0
80
60
LO
100% Q
C
era^ Pigeon
A
m
, 0
,
, 20
,
, LO
,
, 60
,
, 80
,
, 100 %
FIG.3. Relative stimulus value of objects that differ in naturalness and/or predatory threat t o pied flycatcher (a), European blackbird (b), both in Germany, and small ground finch (Geospizafuliginosa),on Santa Cruz Island in the Galapagos (c). Stimulus value is expressed as a percentage of the maximally effective predator stimulus known in experimentally naive birds. Stimulus presentations in b and c involved an initial movement of the dummy. either horizontally or upward above a screen, respectively. In a, dummies were presented stationary in front of the nest hole. All stimulus values within species are significantly different. [Based on data in Curio, 1975 (a). Curio, 1969 (c); and Curio et a / . , 1978a (b).]
ASPECTS OF ANTIPREDATOR BEHAVIOR
147
a stuffed novel bird (Coracias garrulus in a, Philemon corniculatus in b, a long-billed Columba livia var. domestica in c) elicits invariably weaker, though qualitatively identical, responses compared with those to genuine natural predators, that were likewise presented as mounts. Results parallel to c were also obtained with two other island forms of Darwin's finches (Geospiza d$jcilis). All three island forms, including Geospiza fuliginosu from Santa Cruz Island, had not seen Galapagos hawks (Buteo galapagoensis) before, though G . fuliginosa had existed in sympatry with this predator until recently (Curio, 1969). Analogous results, supporting the hypothesis of greater response to the predator, have been obtained with nesting great tits (Parus major) tested with live exemplars of stimuli (two owl species and a sparrow hawk Accipiter nisus [Curio et al., 19831 vs. cockatiel Nymphicus hollandicus [W. Fleuster, unpublished results]. See also Toenhardt, 1935: many German woodland birds; von St. Paul, 1948: Lanius collurio, L . senator; Scaife, 1976a: Gallus gallus var. domesticus; Galloway, 1970: Cinclodes antarcticus; Kobayashi, 1987: Eutamias sibiricus). A puzzling exception is the Falkland thrush (Turdusf. fulcklandii) which treated both a control rubber snake and a stuffed cat with little fear but survived attacks from introduced cats and rats much better than the tussock-bird (Cinclodes) despite the latter's more sophisticated enemy recognition (Galloway, 1970).Perhaps the measure of avoidance used, attendance times at afeeder with a dummy, was inappropriate. Furthermore, for unknown reasons the bullfinches (Pyrrhula pyrrhula) of Kramer and von St. Paul (1951) did not differentiate between raptors and harmless species, although both of these stimulus classes elicited more response than clearly novel inanimate objects. With the exception of the studies in the wild (great tit and other woodland and Falkland Island birds; part of the Eutamias study), all the animals mentioned above were predator-naive when tested. In two cases, the novel harmless bird was larger than the predator species. Therefore body size alone cannot explain the superiority of the latter, but is known to enhance responsivity: Pied flycatchers mob a red-backed shrike significantly less (Fig. 3a) when it is half the normal size (Curio, 1975; see also Geospiza, Section II,A,4,a; and Kramer and von St. Paul, 1951). Likewise, the superiority of the predator compared to novel birds cannot be explained by novelty per se, as the predator was of the same novelty (Fig. 3c and a large fraction of birds yielding the same stimulus values for both stimuli in Fig. 3a), or this was very likely (Fig. 3b). This conclusion touches on the hawk-goose controversy sparked off by Lorenz in Tinbergen (1951), a discussion of which is deferred to Section III,B. In view of the disproportionately large number of harmless bird species as opposed to predatory ones living in sympatry with prey birds, any
148
E. CURIO
control for the efficacy of the raptor stimulus pattern must logically remain open-ended. However, the evidence supporting the hypothesis of a perceptual mechanism encoding rather predator-specific patterns of information is strong. Controlled experiments involving songbirds of different bird taxa (Furnariidae, Muscicapinae, Turdinae, Geospizinae, Paridae, Laniidae, Phasianidae) and a mammal (Sciuridae) have established the superiority of at least one predator stimulus pattern over mere novelty. The basis seems to be some innate releasing mechanism evolved during periods of sympatry with the predator under consideration. An alternative explanation proposing that enemy recognition in these taxa is due to experience with the predator under consideration can be dismissed in a number of cases: these involve tests with animals known to be predator-naive either because they are “natural Kaspar Hausers” living allopatrically from certain predators (Geospizinae, Muscicapinae), or because they have been hand-raised (Laniidae, Muscicapinae, Phasianidae, Sciuridae), as is discussed more fully below (Section III,A,2). The moderate effectiveness of novel harmless birds raises the question of whether all novel objects possess properties apt to elicit protective behavior. To control for the “animal nature” of the novel bird stimuli employed, inanimate objects were presented to experimentally naive birds (Fig. 3b,c). These objects have a lower stimulus value than novel birds, though it is still above zero. To the inanimate objects shown in Fig. 3 can be added the effect of an empty cage on nesting great tits (Curio, 1989). The clearly intermediate nature of novel birds permits us to categorize them as a stimulus class sui generis. In the pied flycatcher various incomplete shrike dummies have a stimulus value of zero and therefore probably qualify as effectively inanimate (e.g., Fig. 3a; Curio, 1975: Fig. 22). Other studies have also found an extremely low stimulus value of novel inanimate objects as compared to the predator but have failed to look for the relative effectiveness of novel harmless species (e.g.; Galloway, 1972; but see Galloway, 1970). The almost total ineffectiveness of these zero value stimuli raises the question of whether they are decoded in a way different from equally novel, harmless but “animate” objects or their live counterparts. A working hypothesis is as follows: For a subject to identify a thing as “bird” it must fulfil minimal requirements in terms of naturalness, for example, have an upright rather than an inverted posture (Fig. 3a), and of surface texture, color pattern, and so on. Thereafter more specific stimulus properties are decoded. This would suggest that birds decode objects in a sequential way: first, a categorization into inanimate versus animate; second, a categorization within each of these classes, for example, bird versus mammal, then, within birds, harmless bird versus raptor, etc. At each
ASPECTS OF ANTIPREDATOR BEHAVIOR
149
level of this sequence of perceptual events different key stimuli would be required (see also Section II,A,6,d). Differences of releasing value among inanimate novel objects may be due to stimulus pattern complexity. A multicolored plastic bottle ranks next to a honeyeater (Fig. 3b), yet a wooden box used to bring either of these two stimuli into view of the subjects and moved the same way ranks lower (Curio et al., 1978a). The findings on novelty have important implications for our understanding of fear responses if one accepts that a potential or real danger activates a motivational system termed fear. In an influential review, Bronson (1968) advocated the view that all fear responses can be related to degrees of novelty. Accordingly, maximum fear is seen as elicited by “absolute novelty.” In a perceptive critique of this novelty theory, Murphy (1978) found the concepts of fear and the related one of exploration ill-defined. She did not question the feasibility of ranking all stimuli along one novelty dimension. However, she hinted at the problem that, for example, the change of the animal from one environment (cage) to another new one inhibited responses that the reciprocal change did not, that is, when the new environment was made to be the familiar and the familiar the new one. The protective responses reported here demonstrate further that the unidimensionality of the novelty theory is untenable. Given the same “absolute” degree of novelty, objects fall into at least two, if not three classes of pattern quality, as measured by response strength. A description of the corresponding stimulus-response contingencies in terms of merely one stimulus dimension would miss most important aspects of protective behavior. Two puzzling aspects of response to novelty have received merely passing attention. First, some European woodland birds have been observed vigorously mobbing nonpredatory objects like a live death’s-head hawk moth (Acherontia atropos) caterpillar (Meinertzhagen, 1959) and models of a scorpion, but not a lizard, on a bird table (Meinertzhagen, 1955). Although not controlled, these observations should be followed up experimentally. They may suggest that, apart from the object categories mentioned above, there are more. There may be cultural transmission of aversion to particular food objects via mobbing (cf. Rothschild and Lane, 1960). Second, Heinroth (1917) and Heinroth and Heinroth (1924-1934) first described the fact that many captive birds respond by panicking when seeing certain colors. More examples came to light later (Koenig, 1951: Merops apiaster; Sauer, 1954: Sylvia communis). A closer analysis of these fear responses would also have to include a consideration of UV patterns. These are now known to play a vital role in the recognition by many birds of plumage patterns, flowers, and fruit (Burkhardt, 1990).
I50
E. CURIO
4. Predator K e y Stimuli: Identification and Properties
a . Localized versus Diffuse Key Stimuli. The plumage pattern of the red-backed shrike male is conspicuous from a long distance. This pattern, not the feathering, is all-important since a plain plastic model is as effective as a stuffed mount in eliciting mobbing by pied flycatchers. Removal of the conspicuous eye stripe (including the equally dark eye) renders the model almost ineffective (Fig. 4a,b). The same happens if, in another experiment, one removes the whole color pattern except for the black bar (Fig. 4c). These changes of the whole pattern demonstrate that both the eye stripe and the “rest” of the shrike’s pattern are localized key stimuli or contain such stimuli. The two pattern components must combine in the receiver to yield the full response. A green or red eye stripe is much less effective than the natural black one (Curio, 1975, Fig. 21). Therefore, it came as a surprise that a reversal of contrast of the eye bar-only pattern (Fig. 4d) lowers the response to about the same level. This seems to indicate that a black-white contrast of either sign is somewhat effective but falls short of the composite natural pattern. In another experiment, it was shown that only a bar aligned with the beak will do: other locations-and another orientation-of the bar are virtually ineffective (Fig. 5 ) . To examine if the eye stripe operates in. an all or none fashion, various shades from jet black to light grey, the color of the head, were tried (Fig. 6). This last model is the same as that in Fig. 4b. As the contrast between the stripe and the rest of head declines, stimulus efficacy falls in parallel. Hence, the eye stripe is a key stimulus that is allowed to vary in intensity, including a concomitant change in response strength. Conversely, if the rest of head is made to vary from white to black with the eye stripe held constant (Ostwald gray level p), there is little if any change in stimulus value as long as there is a minimum of contrast (Curio, 1975, Fig. 25). Taken together, these dummy experiments suggest that almost any deviation from the male shrike’s natural pattern is inferior to it. That the red-backed shrike female also represents an inferior pattern will become the starting point of a new perspective on releasing mechanisms (Section II,A,6). As well as the red-backed shrike, owls provide an example of a localized key stimulus for enemy recognition. Whereas pied flycatchers still recognize an owl whose yellow eyes have been changed to same-colored, convex triangles (Fig. 7 a,b), they fail to do so when the number of eyes has been reduced to one (Fig. 7c). The slight though nonsignificant increase of stimulus value with a further reduction of eye number to zero (Fig. 7d) might be due to the restoration of bilateral symmetry. From these comparisons, it follows that the presence rather than the shape of eyes is
I51
ASPECTS OF ANTIPREDATOR BEHAVIOR
0
50
100 *I.
I
I
I
93
31
P
I n. s I
10
1
d
9
FIG.4. The stimulus effect ( = median value) of the eye stripe of the red-backed \hrike. its presence and the sign of its contrast in the pied flycatcher. Numbers below bars denote sample sizes of experimentally naive birds. Any one of stimulus values b-d is statistically different from a. (Redrawn from Curio. 1975.)
crucial for owl recognition (see also Inglis et al., 1983). The breakdown of recognition due to the elimination of one eye suggests two things: (a) Two eyes act as a configural or Gestalt stimulus. The alternative explanation, that eyes add up by each contributing a certain stimulus value, can be dismissed: the one-eyed owl should then have about half the stimulus value of the intact owl. (b) Like the red-backed shrike's eye stripe, the eyes of the owl combine in the receiver, that is, display coaction with the rest of the owl to produce the full response. In other species, eye color has been alleged to affect responsiveness (Kerlinger and Lehrer, 1982; Inglis et al., 1983), but luminance and/or contrast with the background could have been the relevant cues. In an
I52
E. CURIO 50 I
0 I
61
100 *I.
1
1
P = .904
I 21
C
I 08
I 19
I 20
FIG.5. The stimulus effect of the position and orientation of the black eye stripe of the red-backed shrike on the pied flycatcher. P is based on a two-tailed Mann Whitney U-test. Other conventions are as in Fig. 4. (Redrawn from Curio, 1975.)
iguana, eye size was found to be important (Burger er ul., 1991). More importantly, Coss ( 1978a, 1979) demonstrated in jewelfish (Hemichromis bimaculurus) and in primates (Coss, 1970, 1978b),and Jones (1980)in chickens, that two horizontally oriented facing eyes are superior to one, three, or four eyes or unnatural orientations. This again supports the configurational
153
ASPECTS OF ANTIPREDATOR BEHAVlOR
p-!?
.,‘
0
50
100 m.i
I
I
1
I
61 0
W‘15
Ok h \ 20
c
I
P = ,031 8
FIG.6. The stimulus effect of the shade of the eye stripe p-d (scored after Ostwald) in otherwise unchanged red-backed shrike models. Conventions as in Fig. 4. (Redrawn from Curio, 1975.)
stimulus idea, contrary to the idea of simple stimulus summation (see also Karplus and Algom, 1981). In a most penetrating analysis, Inglis et cil. (1983) found in starlings (Sturnus uufgaris) that two eyes were slightly superior to three eyes only when surrounded by a simple head outline; otherwise, the reverse was true, which is perhaps an artifact. Strangely, changing the orientation of a pair of eyes from horizontal to vertical only slightly reduced their aversiveness. This finding differs from all previous ones, and what Inglis et a f . (1983) call “inconsistencies” among different studies could well be species differences. Despite the functional similarity of their role in relation to the whole enemy Gestalt, the eyes appear to differ markedly from the eye stripe.
154
E. CURIO
I
73
I
I
1
0
50 100 % FIG.7. The stimulus effect of the shape and the number of eyes in a mounted pygmy owl (Claucidictm posscrinum) on pied flycatchers with nestlings. (a) Normal mount: (b) Perspex eyes of natural coloration and near to natural gloss and “corneal” curvature: (c) one eye; (d) both eyes covered with tiny chicken feathers resembling the face color (conventions as in Fig. 4). (Redrawn from Curio. 1975: courtesy of Academic Press. London.)
When presented in isolation, the black bar fails to elicit any response; it does not even have any effect when presented on an oblique wooden dowel the size of a shrike (Curio, 1975; Fig. 22). The owl’s eyes alone have not been tried on the pied flycatcher, but Scaife (l976b) presented a pair of yellow glass eyes to naive domestic chicks and elicited avoidance responses as strong as those given to a stuffed kestrel (Fulco tinnr4nculus). The Gestalt of the eye or eyes may be, due to their intrinsic structuring. a much more powerful aversive stimulus than is the eye stripe (see also Section III,D,3). Similarly, Gallup et al. (1971) found in young chickens that isolated glass eyes prolonged tonic immobility, an antipredator response. From this the authors erroneously concluded that the stimulus effect of simulated eyes is “contextually independent of other facial and/or bodily features of potential predators” (p. 80). All that the study showed is that simulated eyes suf$ce to bring about immobility rather than that there is no coaction with other stimuli which are part of the whole predator. In view of the universality of owl recognition in potential owl victims, the question arises whether the key stimuli used to decode the owl pattern
155
ASPECTS OF ANTIPREDATOR BEHAVIOR
are the same across species (Curio, 1963). The general answer is no, but it deserves qualification. On the Galapagos, where the short-eared owl preys partly on Darwin’s finches, the eyes possess virtually no releasing value when examined on the entire owl (Fig. 8A). Note that this is the way the owl’s eyes were examined in pied flycatchers. If, however, the owl’s head is presented alone, the eyes exert a dramatic effect (Fig. 8B); their elimination renders the head virtually ineffective (see also Smith and Graves, 1978). The finches on Wenman Island are owl-naive but the same effect as in Fig. 8 can be found, though less strongly, on other islands where the owl is a real threat (Curio, 1969). From the experiments on the Galapagos it follows (a) that the role of the owl’s eyes in decoding the owl is not universal, and (b) that, again, the eyes interact with the “rest of the owl” pattern, yet in a way that differs as a direct consequence of (a). For Geospiza, the eyes are a dispensable key stimulus, for Ficedula they are indispensable. The manner of interaction of
1
1
1
I
20 40 60 80 100 % FIG.8. The stimulus value (arithmetic mean) of the eyes depends on the “rest” of the compound stimulus representing an entire owl (A) or its head (B). Results are for the sharpbeaked ground finch (Ceospizo c/ij$ci/is sc.prc,,ir~iorio/is) of Wenman Island/Galapagos. The stimulus value of the eye-less owl is not significantly less than for the unaltered owl. (Redrawn from Curio, 1969; courtesy of Parey, Berlin, Hamburg.)
0
156
E. CURIO
the eyes with the rest of the owl pattern is perhaps different in the two cases. One might propose that the amount of stimulation by a diffuse key stimulus surrounding the eyes of Asio is the key for both an understanding of their different effect when on the owl as opposed to only on its head and their different effects in the two species. The greater the quantity of a hypothetical diffuse stimulus (e.g., the mottled plumage pattern) surrounding the eyes, the less the finch attends to them. Accordingly, key stimuli would compete for attention and the more there is of one the less effective is another. In this view, the eyes would be indispensable in Ficedula because the surface area of the pygmy owl is absolutely smaller than that of Asio and comparable to the latter’s head. To test this idea, different amounts of mottled plumage were presented to a Darwin’s finch species on Genovesa (Fig. 9). Although the models tested also differed in other details from one another, the resulting stimulus values appear to confirm the prediction: The strength of response increases with the amount of a diffuse key stimulus, in this case, the amount of mottled owl pattern seen. This then would support the competition-forattention hypothesis of compound stimuli put forward above. If true, it would furnish another example of stimulus summation in the broad sense.
L
I
1
100 O h FIG. 9. The stimulus effect of three owl (Asio flammeus) dummies differing largely in the amount of mottled plumage visible to sharp-beaked ground finch (Geospiza difJicilis acurirosrris) subjects on Genovesa (Tower Island) Galapagos (from data in Curio, 1969). Eyes were concealed by neck feathers. Stimulus values (arithmetic means) did not differ significantly from dummy on top.
0
50
ASPECTS OF ANTIPREDATOR BEHAVIOR
157
There must be other key stimuli at work (see also Hartley. 1950). The entire, intact enemy Gestalt must encode features above and beyond this size-of-stimulus effect. A pygmy owl, a species not found in the Galapagos, elicits on Genovesa the same response as does the much larger short-eared owl. On Santa Cruz, the pygmy owl is markedly less effective. Even when seeing only the pygmy owl’s back, the finches on Genovesa still respond about as strongly as when seeing its front (Curio, 1969). Although this latter finding would be in line with the minute effect of the owl’s eyes within the intact owl (see Fig. 8) the lack of size dependence in the Genovesa finch, as compared to birds on Santa Cruz. is difficult to explain. Hence, the size-of-stimulus effect does not appear to be a universal explanation of the decoding rule for diffuse key stimuli, but more work is necessary to eliminate possibly confounding features of the owl as a whole. b. Rules of Stimulus Coaction. In no case has the coaction of key stimuli obeyed any simple rule of algebraic summation as was envisaged by the early proponents of the rule of heterogeneous summation. Indeed, such summation seems to be the exception rather than the rule. A neat example is provided by Leong (1969) and Heiligenberg et a / . (1972). but a closer look at one of the two localized key stimuli involved demonstrated the existence of multiplicative processes as well (see below). The analysis of some of the key stimuli involved in the recognition of the perched avian predator follows rules implying multiplicative processes. The eye stripe of the red-backed male shrike yields, when superimposed on the rest of the pattern, the full response. Both stimuli when presented alone on the proper shrike shape are almost ineffective. Hence, their coaction on the natural shrike Gestalt obeys a rule of stimirlus dilation, by which the whole yields more than the sum of its component pattern parts. The same would apply to the owl’s eyes and the rest of the owl (Ficedirla) or that of its head (Geospiza).For a diagrammatic representation see Fig. 10. This effect is so pronounced that it is obvious even without statistical treatment. Yet, Baerends and Drent (1982) contend that the effect is not “proven” rigorously. Unfortunately, they do not give the reasons for their critical stance. They also misquote my paper (Curio, 1975) by calling the opposite effect (see below) response dilation when they mean stimulus compression. By contrast, the opposite type of coaction of key stimuli operates when separate key stimuli are so effective that their combination theoretically exceeds the value of the whole. Therefore a form of stimulus compression must apply (Fig. 10).(The alternative, banal explanation of a “ceiling effect” [= efferent saturation] can be ruled out.) This is the case with parts of the owl and the hawk in Darwin’s finches. In the pied flycatcher it has not been looked for. Both types of coaction can be modeled by various algorithms of
158
E. CURIO
*I.
a
b
C
100 5 L
C m
u
L
In
50 C
0
n
m u
a 0
A
A + B >AB
+ B Lota Iota
4
Hemichromis bimaculatus
2 staring eyes > 3 staring eyes
a. i
5
Heterodon platirhinos
a. i
6
Geospiza sp.
Otus asio D; human with stare > human with eyes averted Dromicus sp. + D
a
7 8
Ceospiza sp. Taeniopygia guttata
Asio D, Buteo D Glaucidium brasilianum
a(D). i a
9
Ficedula hypoleuca
Strix D, Athene, Lanius excubitor Staring eyes D, human face
a. i
10
Callus gallus bankiva
-
I
Comment
Source
Adaptive geographic variation
Seghers (1973. 1974)
Adaptive geographic variation Avoidance of Esox only increases on perceiving it Serves avoidance of piscivores + dominant conspecifics Recovery time from cataleptic response
Tulley and Huntingford (1987a) Jacobsson and Jarvi (1977)
Natural Kaspar Hausers on snake-free islands. Adaptive geographic variation Adaptive geographic variation Presence of pair mate andlor open view of landscape required. Dummies fail Lanius: adaptive geographic variation Recovery time from tonic immobility. Domestic fowl, human-experienced
Curio (I%% E. Curio, unpublished
Coss (1978a) Burghardt and Greene (1988)
Curio ( 1%9) Hoffmann (1979). Lombardi and Curio (1985a) Curio (1975) Gallup et al. (1971. 1972)
(continues
TABLE 1 (Continued)
No. II
Kaspar Hauser species Gallus gallus bankiva
Predator stimuli and ranking where knownb Falco tinnunculus D, staring
eyes D 12
Gallus gallus bankiva
Human
13
Bubo virginianus
14
Aphelocoma coerulescens, Aphelocoma ultramarina Alectoris rufa
I5 16
Neotoma albigula Rattus norvegicus, 5
Pit uophis catenifer Felis catus
Stimulus specificity controlled for by a or i' aD aD
-
r\o 4
Human
laboratory strains 17
Mesocricetus auratus
18
Mesocricetus auratus
19
Peromyscus maniculatus
20
Spermophilus beecheyi
Odor of Mustela putorius, +D Mustela putorius f. furo. + D Dog, odor of dog, +D Mustela, Felis , Coluber , Pituophis Crotalus viridis > Pituophis melanoleucus
Comment Fear increases on stare of eyes following chick. Domestic fowl Avoidance differs among two stocks; modifiable more by reward/habituation in birds of tame stock Response matures earlier in less social A. coerulescens Only if not familiarized with man within 48 hr of hatching
a
a, i
Response most pronounced in least selected strain. Strain differences Freezing, threat, attack, occlusion of burrow Flight into burrow; freezing rarely Adaptive geographic variation Adaptive (?) geographic variation
Source Scaife (1976a.b) Murphy and Duncan (1977, 1978) Cully and Ligon (1976) Csemely et al. (1983, 1984) Richardson (1942) Satinder (1976)
Dieterlen (1959) Dieterlen (1959) Hirsch and Bolles (1980) Owings and Coss (1977);
s
21
Spermophilus beecheyi
Crotalus viridis = Pituophis melanoleucus
22
Canis lupus
Human
23
Damadama
Human
24
Didelphis virginianus
Human (dog)
a
Kaspar Hausers both snakenaive + burrow-naive dealt appropriately with snake in burow but did not discriminate the two species as they do above ground (see no. 20) Only if not familiarized with humans in early life. Independent of parents' tamenessd Only if not familiarized with humans during first 2 days of life. Independent of mother's tameness Feigning death when grabbed: dependent on rearing by mother and/or housing conditions
Coss and Owings (1978)
Woolpy and Ginsburg (l%7)
Gilbert (1968)
Francq (1%9)
Curio (1%3). . .
* D denotes response to dummy only; + D denotes response to dummy and to live predator.
' a, Animate, denotes either live, taxidermic, or other replica of control stimulus. i, Inanimate, denotes man-made control stimulus of about predator size, or simply test environment (the latter when not specified). There are qualitative interindividual differences among littermates in their response to humans which complicate things (Shaker et al., 1977;see also Fox, 1972).
194
E. CURIO
study (no. 4 in Table I), some work on stimulus analysis appears to make it certain that the performance is specific for the targeting predator/ adversary. However, even here some control for preexposure conducive to positive responses would have been necessary (see point 2 below). The stronger effect of the predator as compared to an animate object would appear to guarantee the stimulus specificity looked for. The importance of the discrimination of predators from innocuous species derives from the popular notion that animals have “to learn what not to fear” (refs. in Nice, 1943; Bronson, 1968). This idea is clearly rejected by the evidence for an innate discrimination between the two classes of objects. It is also rejected by evidence showing that avoidance of particular enemies requires learning (Section III,C,3). The idea is less easily falsified in the case of the discrimination between harmless species and novel inanimate objects (see Section III,B,C). 2. The gaze of two staring eyes is one potential key stimulus that is contained in the human face. It is possible that preexposure during rearing, maintenance, or capture by humans could bias responses to staring eyes, where these have been employed as a stimulus. Apart from one study (no. 7 in Table I ) where experimenters covered their faces with a stocking mask whenever handling the subjects, no study controlled for this potential source of bias (but see Coss, 1978a). A similar caution applies to the effectiveness of a mammal’s pelage (e.g., Kramer and von St. Paul, 1951) and the hair of the experimenter’s head. 3. In some cases (nos. 9 and 13), only some of the Kaspar Hausers tested responded as would be expected from the behavior of wild birds. This may be no serious drawback since in the wild the adults tested had a territory and/or a brood, factors that the Kaspar Hausers were deprived of. Therefore the test must be regarded as highly conservative. It is important to note that the maximum response strength attained a level equivalent to that in the wild (Curio, 1975). 4. Experiments in which dummies (D) were used instead of the real enemy must be regarded as conservative, too; dummies provoke weaker responses (Section II,A,l) and, in one case (no. 8), failed to release any: this had led an earlier investigator to conclude that zebra finches lack any owl response (refs. in Lombardi and Curio, 1985a). This latter case is particularly instructive in that it demonstrates the difficulty of replicating results. The owl response of zebra finches obtained earlier (Hoffmann, 1979)could-for unknown reasons-only be replicated by subtly changing the context of testing (see Comment in Table I, no. 8). There are intriguing sensitive period effects that determine whether an object is appropriately responded to (nos. 14,22,23 in Table I). Discussion
ASPECTS OF ANTIPREDATOR BEHAVIOR
195
of these effects would go beyond the scope of this review. Discussion of responses to flying predators is deferred to Section 1II.C because of the long-standing contention that the underlying IRM is relatively unspecific. An ecological correlation between the antipredator behavior of captiveborn individuals and the predatory threat to the populations they came from has similarly been used to infer innate differences among the groups concerned. It could be argued that the correlations observed are due to some interaction between a hypothetical population-specific susceptibility to captivity and the development of the behavior under study. However, this objection seems farfetched since, without exception, the correlations mirror, on a ranking scale, the population-specific threat from predation in the wild in guppies (Poecilia reticulata),sticklebacks, minnows (Seghers, 1973, 1974; Breden and Stoner, 1987; Stoner and Breden, 1988; McPhail, 1969; Huntingford, 1982; Giles and Huntingford, 1984; Levesley and Magurran, 1988; Tulley and Huntingford, 1987b, 1988), voles (Chlethrionomys glareolus; Alder, 1975). deer mice (Peromyscus maniculatus; Hirsch and Bolles, 1980), and California ground squirrels (Spermophilus beecheyi: Owings and Coss, 1977; Coss and Owings, 1985). A possible exception is in sticklebacks, where population-specific escape responses to simulated attacks from birds differed in the details of the orientation among wildcaptured and laboratory-raised fish (Giles, 1984), perhaps suggesting an influence of maintenance conditions. Work on anti-snake behavior of snake-adapted versus snake-nonadapted populations of black-tailed prairie dogs broadly parallels the results on ground squirrels, but experience with snakes could have contributed to the population differences found (Owings and Owings, 1979; Loughry, 1988). Only for guppies, sticklebacks, voles, and deer mice did the experimental protocol ensure that individual experience with the predator(s) could not have caused the observed avoidance. The adaptiveness of the avoidance involved was most directly demonstrated by Hirsch and Bolles (1980; Table I, no. 19). They showed that deer mice that failed to recognize predators which they had not lived with in sympatry succumbed to their attacks whereas animals from an area of historical sympatry survived. That the differences between predated and unpredated populations are largely due to genetic adaptations is corroborated in some cases by accompanying differences in antipredator devices that are morphological (McPhail, 1969; Giles, 1987; Reist. 1983), physiological (Coss, 1985), behavioral (escape response: McPhail, 1969), or behavioral in other contexts (agonistic: Tulley and Huntingford, 1988; courtship and female choice: Breden and Stoner, 1987; Stoner and Breden, 1988; mate choice: McPhail, 1969). As a cautionary point, one has to consider that part of these correlations, if not shown otherwise, may be
I96
E. CURIO
due to individual differences. Individual sticklebacks with a heavy armor and many spines were found to be bolder against their fish predators (Reist, 1983). b. Dissection of the Stimulus Pattern into Parts. A number of rules governing coaction of predator and novel bird stimuli have been elucidated above (Section II,A,4,b). Do these rules operate on first encounter with a predator or do they result from some experience with it? Some suggestive results have been obtained by comparing populations of Darwin’s finches with different opportunities for acquiring information on predators. In Darwin’s finches, the eyes of an owl interact with the “rest” of the pattern surrounding them. While the presence of eyes is immaterial for the whole of the owl Gestalt, they are of paramount importance when only the owl’s head is perceived (Fig. 8). There are two effects: There is stimulus compression in that the eyes are scarcely, if at all, effectual in the intact owl, but have an extraordinarily great stimulus value in the owl’s isolated head; the eyes are “compressed” but this becomes clear only when their surroundings are reduced. The same effect can be found, though less dramatically, on the islands of Genovesa and Santa Cruz (Curio, 1969, Fig. 48). On these islands the finches can gain experience with owls, though only on Santa Cruz are adults much preyed upon; on Genovesa, their nests are raided and some immatures or adults are taken by owls (Grant and Grant, 1980).It remains unknown whether the differential compression of the eyes is a consequence of experience (with owls) that differs among the populations concerned. What can be safely concluded is that the compression effect operates independently of any experience with owls. Similarly, the great efficacy of the head alone and the rump alone (see Fig. 9) cannot depend on individual experience with the predator in question. The Galapagos hawk’s (Buteo galapagoensis) head displays a stimulus value comparable to the owl’s head in Fig. 8 (Curio, 1969).Again, since this is so on three islands where the finches are hawk-naive (Wenman, Genovesa, Santa Cruz), experience with the hawk cannot be a prerequisite. It could be argued that an experience with the owl might be generalized to perception of the hawk and determine the rules governing decoding of the hawk’s parts in relation to the whole. However, this is an unlikely explanation since on Wenman Island the finches are both owl- and hawknaive, and the same efficacy of the hawk’s head applies. Furthermore, the stimulus compression observed on Santa Cruz Island, where the hawk’s head and the hawk’s rump together yield more stimulus value than the whole of the hawk, cannot be contingent on experiencing a hawk before. It is remarkable that stimulus compression in the human infant, as elucidated by its response to facila features, is also independent of seeing the whole. (Bower, 1966).
ASPECTS OF ANTIPREDATOR BEHAVIOR
197
The epigenetic basis of stimulus dilation, as found in the pied flycatcher (Section II,A,4,b), remains unknown at present. c . The Multichannel Organization of Enemy Recognition. Six lines of evidence suggest that, in pied flycatchers, owls, shrikes, and harmless birds enter perceptual channels specific to each of these object classes (Section II,A,6). Furthermore, as shown by the deprivation experiment (Table I, no. 9), owls and shrikes are recognized innately, and a live harmless bird (Nymphicus hollandicus), the animate control for predator specificity of the underlying IRMs, was responded to much less than were the predators (Curio, 1975). This ranking of responses parallels that found in the wild. Hence, the decoding of the three object classes fulfils the criterion of innateness (for a possible qualification of the novel bird response, see Section 111,C). It remains to be explored whether the multichannel organization itself is innate or requires experience with any of the objects classified by it.
B. NOVELTYAS
AN EPIGENETIC DETERMINANT OF ANTIPREDATOR BEHAVIOR
A number of birds have been shown to respond more strongly to genuine predators than to harmless birds and inanimate objects (Fig. 3, Section II,A,3). Since these results were obtained with birds in the wild, one could argue that the stronger response given to the predator was due to the predator being rarer than are harmless birds (Schleidt, 1961a,b).This idea, the “rarity principle” (Curio, 1969),predicts that all potentially dangerous objects are initially of equal stimulus value. A difficulty with this idea is that the harmless birds tested were, by design, entirely novel whereas the raptor could have been encountered previously, though rarely. Harmless birds should then have provoked at least the same level of response, which was not the case. One might, however, save the idea that the response to the two classes of stimuli is due to differential novelty by assuming that the test birds conceptualized a class “harmless bird” by generalizing from previous and frequent experience with harmless birds to the novel harmless bird presented to them experimentally. As an alternative to the differential rarity idea, a harmless bird might be any bird that is a nonpredator. A basis for such a negatively defined stimulus class would be the sequential stimulus decoding suggested above (Section II,A,6,c). In a first step of identification, an object would be classified as “bird.” In a second step, a harmless species would be identified as “nonraptor” by not entering any of the raptor-specific channels. This not very stimulusspecific mechanism would not require differential encounter frequencies, as does that based on rarity, but it might need experience with stimuli characterizing what is a bird.
198
E. CURIO
The idea of differential novelty underlying enemy recognition received partial support from Schleidt's (1961a,b)experiments with predator-naive domestic turkeys (Meleagris gallopavo). He demonstrated that chicks of 14-16 weeks of age responded by aerial predator calls to various equalsized silhouettes, regardless of whether they mimicked a raptor or any meaningless object (Fig. 22). Stimulus efficacy was inversely proportional
5
Circle
Hawk
' R a p t o r ' 'Goose'
Rectangle
FIG.22. Response strength (aerial predator "prrr" calls) given by turkey chicks to various equal-sized (30 cm2) silhouettes moving overhead on test days (abscissa). From day 3 on. points denote mean values from 10 trials each. After day 5 three stimuli that were only shown on days 1 and 2, or were novel. displayed stimulus values equivalent to those on days I through 3. Apparent sizes correspond to a hawk seen 27 m above ground. (From Schleidt, I%lb; courtesy of author and Springer Verlag, Berlin, Heidelberg, New York.)
ASPECTS OF ANTlPREDATOR BEHAVIOR
199
to frequency of encounter, apparently irrespective of the particular shape involved. It is important to note that on first encounter all stimuli were absolutely novel (at least the one seen first). They were not effective because of being relatively novel, that is, as compared to a stored “background” of innocuous stimuli, to which subjects had become habituated as a consequence of frequent encounters. This falsifies the notion that relative novelty is the exclusive mechanism underlying decoding based on habituation (Bronson, 1968; see also Section 11,A,3). Although only a limited number of stimuli were tested after habituation had occurred, the result seems also to be at variance with Kagan’s (1970) popular notion of the “discrepancy principle,” which holds that stimuli that differ only mildly from remembered ones are more effective than strongly discrepant stimuli. Schleidt’s (196la,b) finding was taken to resolve the controversy concerning the short-necked key stimulusfor avian fear responses (see Hirsch et al., 1955; Tinbergen, 1957), for Lorenz’s and Tinbergen’s turkeys and pheasants had seen flying geese more often than the short-necked raptors (Schleidt, 1961b; see also Markgren, 1960, for Anatidae). While the differential encounter rate may be one reason for the raptor specificity of the response in the wild, a stimulus specificity independent of rarity cannot be dismissed. In fact, birds have been observed to gear their aerial predator response to the species-specific stimuli (visual pattern, flight style?) of the raptor species and to the threat it poses (refs. in Curio, 1963). Similarly, American coots (Fulica americana) show highly adaptive responses when discriminatingbetween planes, hawks, and eagles that cannot be explained by either the short- versus long-neck dichotomy or their abundance (Grubb, 1977; see also Markgren, 1960; Muller, 1961; Martin and Melvin, 1964). Even the same raptor species may provoke different behavior. While flying in nonhunting style, a sparrowhawk provokes mobbing from woodland birds, whereas when hunting it induces aerial predator calls (Klump, 1984) or silence (Hartley, 1950). Experiments with predator-naive mallard (Anus platyrhynchos) ducklings seem to support the view of an innate superiority of the hawk versus the goose silhouette (Green et al., 1966; Mueller and Parker, 1980), and thus do not confirm the rarity principle advocated by Schleidt (1961a,b). Unfortunately, the models used, the manner of their presentation (height, speed, path length), and the behavior measures used all differed from those in Schleidt’s turkey work. Green et al. (1966) were not even aware of the standardized methodology of stimulus presentation pioneered by Schleidt. Furthermore, the measures used are difficult to interpret biologically. In one study (Green et al., 1966), the measure used (running) was not validated at all, in the other (Mueller and Parker, 1980), the measure
200
E. CURIO
(uariance of heart rate) was backed up by pointing to the correlation between it and the threat posed by the stimulus leading to it. If taken at face value, the studies falsify, like others (nos. 1, 5, 6 , 7 , 8 , 9 , 11, 15, 17, 18, 20 in Table I), one prediction of the rarity principle, that is, that all potentially dangerous objects are initially of equal stimulus value. Furthermore, different genuine predators may be treated, from the start, differently according to the threat they pose (nos. 3, 17, 18, 20 in Table I; see also Martin and Melvin, 1964). Evidence that permits the rejection of any explanation based on absolute novelty of harmless animals at the same time falsifies an alternative idea on their correct classification as mentioned above: This idea is based on individual experience with other harmless species from which to generalize. For the pied flycatcher, this can be dismissed (see Section III,A,2,c). Whether (peaceful) experience with conspecifics could shape the response to absolutely novel harmless animals (beware of any nonconspecific!) is unknown. For an animal to be classified as relatively novel familiar ones must be remembered. In species-rich habitats, more than 40 species can be remembered by a fish (Ebersole, 1977), and probably by great tits (Fig. 20), which is not surprising in view of the enormous number of visual patterns that pigeons can store (von Fersen and Delius, 1989). In paradise fish, a 1-min exposure to a goldfish effects an habituation for at least 3 months (Csanyi et al., 1989).
C . HABITUATION AS
A
DETERMINANT OF ANTIPREDATOR BEHAVIOR
Two interrelated questions can be asked: (a) Is the lower alarm behavior directed to harmless animals in the wild the result of habituation to them as a consequence of their higher abundance? (b) If yes, do prey animals habituate to predators in the wild as well and to what extent?
I . The Rarity Principle The habituation hypothesis of enemy recognition predicts that nonconspecifics should provoke protective responses as a function of their abundance; because of their rarity, predators should rank highest (Schleidt, 1961a,b; Curio, 1969). To test this, we manipulated abundance of the harmless Galapagos dove by capitalizing on its different abundance on different islands. When presented as a mount to Geospiza finches on Genovesa and Santa Cruz Island, responses were stronger on the latter where the dove no longer exists; on Genovesa it is extremely common, feeding on the ground next to the finches (Fig. 23). Similarly, great tits fear a novel bird more than a familiar one (Fig. 20). Hence, the results
ASPECTS OF ANTIPREDATOR BEHAVIOR
30
r
20 1
Ge n o v e s a
01
Santa C r u r 0
1
2
FIG.23. Testing the “rarity principle” by comparing the fear responses to the harmless Galapagos dove on two different Galapagos islands where it differs in abundance. Finches tested were Geospiza difficilis ( n = 10) on Genovesa, with a high density of doves; and Geospizaf. fuliginosa ( n = 10) on Santa Cruz, with f zero density of doves. 1, 2 denote sequence of stimuli. (1) To make responses on Genovesa measurable, mounts were shown at half the distance ( I m) from that on Santa Cruz.(2) Since the response measure used might reflect different degrees of fear in different species, the dove was compared with a similarsized yet moderately effective owl head. The difference between stimuli is significant on Genovesa. (Modified from Curio, 1969; courtesy of Parey, Berlin, Hamburg.)
tend to support the rarity principle for harmless species. Pertinent questions remain unanswered: Does habituation differ depending on the nature of these species, the sequence of encountering them, or other circumstances? Attempts to mimic under captive conditions the diverse processes of habituation to an array of syntopic species have been notoriously jeopardized by the fact that even effective predator stimuli rapidly lose their efficacy (Hinde, 1954a,b; Curio, 1969). This loss appears maladaptive in view of the vulnerability that would ensue if it occurred in nature. Since
202
E. CURIO
stimulus presentations have usually involved the same position of appearance in relation to the subject, the rapid onset and long-lasting nature of the habituation was rightly thought to be an artifact. As shown for domestic and wild fowl, pied flycatcher, and angel fish (Pterophyllum eimecki), a waned response was renewed by changing the site of presentation of a fear/predator stimulus to which habituation had occurred (Shalter, 1975, 1978a; Schleidt et al., 1983). This suggests that the (relative) “lack of habituation to recurring predators in nature, is due, in part, to everchanging relationships of the predator relative to the inanimate environment” (Shalter, 1978a, p. 1219). Similarly, habituation has also been avoided by shifting the subject (chaffinch) from one aviary to another between presentations (Hinde, 1954a). In all these experiments renewed mean responsiveness attained, at most, the same level as before. (A higher level later on may result from shifting the stimulus from a ground position to an overhead one, thus perhaps involving two different predator channels [Shalter, 19751.) By contrast, in a cultural transmission experiment with European blackbirds, response level was dramatically increased far beyond the initial level (Curio, 1988b). A novel honeyeater (Philemon corniculatus) mount attained a conditioned stimulus value when presented at the place where it had been seen during cultural transmission. When, after one-trial learning, the honeyeater was presented at the opposite side of the aviary, its stimulus value had trebled. This shows that conditioned enemy stimuli can also gain in stimulus value by a change in locale. Despite overwhelming evidence bearing out the importance of patternspecific recognition of enemies in the widest sense (Sections II,A,4; 11,B; III,A), Schleidt et al. (1983) still maintain that the rarity principle is sufficient to protect prey from their predators. This mechanism, however, may fail (see also Seyfarth and Cheney, 1980) where a predator has become commoner than harmless species (examples in Curio, 1969), or where harmless species are as rare as are many predators, a common situation, especially in species-rich faunas; too many false alarms to innocuous creatures would be energetically prohibitive. 2. Differential Habituation to Predators and Harmless Species? Prey could escape the dilemma posed by the rarity principle if they habituated to predators less, or not at all, as compared to harmless species. Though not systematically fested, there is a sizeable amount of circumstantial evidence in favor of such a view (Fig. 24). The lower stimulus value of nonpredatory species has been found in all deprivation experiments controlling for the stimulus specificity of genuine enemy recognition (Table I). The response elicited by the dummy red-backed shrike male does not seem to abate until after 6-8 presentations to pied flycatchers of this and
ASPECTS OF ANTIPREDATOR BEHAVIOR
Anti -
Predator
203
Predator
I
Age of Prey Bird
FIG. 24. Schematic representation of habituation to predators and to harmless birds dependent on encounter number, or time. Ordinate is arbitrary.
other less effective shrike dummies, whereas the response to mounts of the roller and the red-backed shrike female, another novel bird (see Section II,A,6,b), appears to wane immediately after the first encounter (Curio, 1975, Fig. 18). Similarly, pied flycatchers that had their territories adjacent to red-backed shrike pairs, and birds from populations living syntopically with red-backed shrikes displayed the same level of response to the male shrike model as did pairs and populations, respectively, living unmolested by shrikes. Again, the male shrike appears immune to habituation (Curio, 1975; see also Mineka and Keir, 1983). Furthermore, paradise fish habituate to free-moving goldfish more rapidly than they do to pike under the same conditions, despite an initially identical amount of interest (Cshnyi, 1985; see also Buitron, 1983; Magurran and Girling, 1986). In other cases, waning of the response is more rapid for the more potent predator stimulus (Melvin, 1969; Hinde, 1954b), but this may be an artifact due to massed presentations in the same position. Likewise, “transfer of habituation” (Hinde, 1954b), which entails reduction of predator stimulus value as a consequence of exposure to less effective stimuli before it (Hinde, 1954b; Melvin, 1969; Martin and Melvin, 1964), may be such an artifact. If it occurred in nature, this would clearly be detrimental. Significantly, other dummy experiments have failed to confirm the effect (Fig. 15; Curio, 1969). Furthermore, langurs (Presbytis entellus) in India that had become fully habituated to native Indians still exhibited the full range of fear responses to Europeans; among the latter, novel movements, like lifting a pair of binoculars, enhanced these responses still further (Vogel, 1975). (It is unknown whether the initial fear of humans is innate which, if true, would violate the hypothesis underlying Fig. 24.) Transfer of habituation may follow different rules when experiences are widely spaced in ontogeny and/or the objects habituated to are inanimate.
204
E. CURIO
For example, Rubel (1970)found that quail (Coturnix coturnix) feared such objects as adults less when they had already encountered them much earlier. The circumstances under which transfer of habituation to fearsome stimuli takes place need further study. In other cases, habituation to the predator may have been obviated by seeing it hunt, though this has not been demonstrated. In sociable weavers (Philetairus socius),alarm calls given to pygmy falcons (Polihierax sernitorquatus) breeding in their colony do not wane at all (MacLean, 1973). Similarly, H. Lohrl (personal communication) found that passerines did not cease alarm calling to a sparrowhawk that lived in an adjacent wood and was seen daily. Further, a pair of pygmy owls refrained form nesting for 3 years while being housed adjacent to an eagle owl’s aviary. As soon as the pygmy owls were shifted to an aviary out of its sight they bred (W. Scherzinger, personal communication). Some casual hunters, like gulls nesting in colonies of terns, tend to hunt more away from the “host colony” rather than robbing nests therein. Accordingly, terns attack foreign gulls rather than their “guest” gulls. Such observations led McNicholl(l973) to contend that the host terns had become habituated to their guest gulls while they remained responsive to same species gulls roaming over the colony from afar. An alternative view, however, may be that the “domestic” gulls, being less prone to hunt in the colony, were correctly classified as harmless by the would-be victim terns. Hence, the habituation view is not necessarily at variance with the view advocated here according to which predators under natural conditions are immune against habituation. Things may be very different when birds have the chance to avoid the locale of a predator that cannot move about. Thus W. Scherzinger (personal communication)observed passerines each spring, after their return from the south, to mob his caged owls (Strix nebulosa, Surnia ulula), yet only for a couple of days. This differs strikingly from the behavior of resident jays toward a free owl (Section II,B,3,b). Thus, when danger is permanent there may be no habituation to it at all, which clearly falsifies part of the rarity principle and calls for some higher order process of enemy recognition. The difference in response may be due to simple habituation. Or, the birds confronting the caged owls may have assessed correctly that mobbing was of no avail: Whatever the strength of the response, the raptors could not be made to move on. Similarly, sociable weavers gave up showing alarms to snakes, which plundered nests in the breeding colony with impunity, perhaps because they recognized that alarms were futile (MacLean, 1973).Thus, the reasons for response decrement may be varied, but this has not been analyzed in detail. (Even for turkeys, the birds Schleidt studied, it remains to be shown if some more
ASPECTS OF ANTIPREDATOR BEHAVIOR
205
realistic raptor model [with flapping flight, nonlinear flight path, etc.] would be immune to habituation. Significantly, Martin and Melvin [ 19641 found a clear superiority of a flying hawk as opposed to its moving silhouette. The naive bobwhite quails [Colinusuirginianus] used, however, did habituate to the live hawk, perhaps due to the unvarying conditions of its appearance [see also Murray and King, 19731.) The opposite from habituation, an increase of distraction display with an increase of encounter frequency, has been observed by Hudson and Newborn (1990) in red grouse (Lagopus lagopus scoticus) encountering red foxes (Vulpes uulpes). This incremental process, best ascribed to sensitization, undermines even more forcefully than does the lack of habituation the special reaction to predators among all stimulus objects eliciting avoidance responses. In conclusion: There is evidence that harmless animals lose their moderately alarming properties as the prey species becomes familiar with them. The underlying habituation, or some higher order risk assessment, accentuates the preexisting distinction between them and genuine predators, which is due to inherent properties of both object classes rather than to different encounter frequencies. There is little evidence that prey become habituated to genuine predators under natural conditions.
D. LEARNING ABOUT PREDATORS: CONTENT A N D MODES In view of the substantial evidence of innate information underlying avoidance of predators in vertebrates (Table I), some questions become pertinent: (a) If and to what extent are these responses altered by experience that relates to predators? (b) What characteristics of predators are learned? (c) Can harmless species, that is, nonspecialist hunters or brood parasites, become recognized as harmful through experience? (d) What are the modes of and the constraints on learning?
1.
What Is Learned? Many vertebrate prey modify their antipredator responses in ways that are extraordinarily flexibile, thereby supplementing their rigid framework of IRMs and preprogrammed startle reflexes. The evidence for this is often based only on the reasoning that the observed fine-tuning of responses is too unlikely to result form innate “wiring.” a. Acquired Predator Species Characteristics. There is a dearth of information as to which and how many animal species (“types”) can come to elicit the protective behavior of prey. Seyfarth and Cheney (1980) inferred from their observations and playback experiments with vervet monkeys that they discriminate four different classes of predator, as
206
E. CURIO
judged from the fact that they give acoustically different alarms to leopards, martial eagles, pythons, and baboons. On hearing the alarms, vervets respond adaptively in terms of risk avoidance. When compared to adults, infant vervets give alarm calls to a significantly wider variety of species, that is, give many false alarms. Additionally, they display more “wrong” responses. However, their signaling is not arbitrary; they restrict leopard alarms to terrestrial mammals, eagle alarms to birds, snake alarms to snakes, and baboon alarms to baboons. How infants come to “sharpen” their discriminative abilities over the years till adulthood is not known; social transmission is a likely candidate. For example, during alarm playbacks infants look at their mothers and the responses are more adultlike if their mothers are nearby. The gradual sharpening, that is, narrowing down to the species level, or to a particular spatial context (Section II,B, I), may build on four innate channels tuned to the four classes of predators mentioned (a possibility not discussed by Seyfarth and Cheney [ 19801). Without conducting a deprivation experiment, however, the innateness of such channels remains conjectural; in view of the high sociality of vervets, cultural transmission, even of these baseline channels, remains a viable alternative (see Section III,D,4). A taxon-wide generalization from an adverse experience with one of the taxon’s members has been demonstrated in bullfinches(end of Section II,A,6,g). This need not imply a difference in the underlying learning process between birds and primates. The one-trial learning (pursuit) by which that predator fear was brought about in the bullfinches vastly differs from the gradual improvement observed in the vervets. It would be important to know whether bullfinches are able to separate out particular species from their taxon-wide fear of an avian family as a consequence of further experience with them. It is as yet uncertain whether the decoding of innately recognized predators differs fundamentally from that of acquired recognition. Furthermore, the process(es) underlying the ontogenetic increase of caution in confronting dangerous predators (Owings and Coss, 1977; Coss and Owings, 1978) awaits elucidation. b. Acquired Predator Individual Characteristics. There are numerous reports that the individual human (Nicolai, 1950; von Frisch, 1964; Brown, 1970; Drost, 1971; Stinson, 1976, Merritt, 1984; McLean and Rhodes, 1991), cat (Lendrem, 19801, or gull conspecific committed to cannibalism (Tinbergen, 1958; Veen, 19771, provokes protective behavior from birds. A reinforcing event has been identified as handling of a brood or an adult, thus mimicking predation (Nicolai, 1950; von Frisch, 1964; Drost, 1971; Stinson, 1976; Merritt, 1984), or true predation itself (Tinbergen, 1958). A particular person may also be singled out, for example, by parent red-
ASPECTS OF ANTIPREDATOR BEHAVIOR
207
winged blackbirds (Agelaius phoeniceus), if that person alone repeatedly visits the nest such that only the familiar visitor provokes bolder defense responses. This difference is thought to result either from habituation of the fear system to the familiar person, or from positive reinforcement following its moving on after each encounter (Knight and Temple, 1986a; see also Brown, 1970). Whatever the reason, the finding is in line with the idea developed above (Section 111,C,2)that the response to nonpredators (= harmless species that are potentially dangerous) becomes modified as a result of peaceful encounters. As they are not game birds, blackbirds almost certainly lack any innate avoidance of humans. This raises the important question whether only the low-keyed responsiveness to harmless species is a prerequisite that imparts on prey the flexibility reflected by recognition of individual adversaries. If so, there would be no individual labeling of genuine predators that are decoded innately. Whether such predators may still be partitioned into classes following habituation to one class is uncertain. Thus, langurs discriminating between morphs of humans (Section III,C,2) may or may not fear them innately. The details underlying such adversary learning may differ fundamentally from those involved in the acquisition of fear of species or higher categories: First, there is no generalization from the individual concerned to the whole of that individual’s species, whereas bullfinches generalize from one predator species to the whole family of that species (see above). Second, compared to genuine predator recognition, they seem to forget the individual they were conditioned to previously; for example, the same mockingbirds (Mimus polyglottos) that harassed a person one summer failed to do so the next (Merritt, 1984; see also Drost, 1971). As with acquired species recognition (see above), there is much scope for experimentation. c . The Behavior of the Predator. Much mobbing (Marler and Hamilton, 1966) and inspection behavior (Magurran, 1986) vis-a-vis a predator is thought to be information gathering. There is also active testing of whether an attack is imminent, for example, it may be dependent on the thermal condition of an ectothermic predator (Rowe and Owings, 1990). There is a general belief that the details of the behavior of such a predator have to be acquired because a n y IRM, should it exist, would appear to be overtaxed. Though one can sympathize with this view there is no hard evidence to support it. Prey animals apparently discern a predator in hunting mood from one that is not (Meinertzhagen, 1959; refs. in Curio, 1963; Berger et al., 1963; Kruuk, 1972; Potts, 1980; Robinson, 1980; Buitron, 1983). The extraordinary behavior (termed “baiting” by Brown [1971]) by a Syke’s monkey (Cercopithecus mitis) vis-a-vis its most dangerous enemy, an
208
E. CURIO
incubating/broodingcrowned eagle (Stephanoaerus coronatrrs),is difficult to categorize. In a number of episodes, the male monkey bounded back and forth over the eagle many times, even touching the nest on occasions, just keeping out of reach. The eagle was visibly molested by these maneuvers, which are reminiscent of mobbing, and for which a number of functional explanations can be dreamed up. When analyzing from film the preattack posture of a red-tailed hawk (Buteojarnaicensis),Grier (1971) was unable to find any cues that might be utilized by prey to predict an attack. There were no differences compared to the movements preceding any change of perch. As correctly observed by Grier, a prey that is targeted visually prior to an attack may well infer from the movements involved that an attack is imminent. This point could not be settled since the prey used were laboratory mice that were unaware of the distant hawk. In an attempt to mimic the situation of a predator in different moods, Licht (1989) presented guppies with a choice of a satiated and a hungry piscivore. Guppies almost invariably chose the satiated one, that is, then avoided the one that continually attempted to prey on them. Unfortunately, this experiment does not address the problem of how prey recognize the subtle details differentiatingdifferent states of readiness to attack; rather it demonstrates, not surprisingly,that overt attacks elicit avoidance. We still do not know how prey animals come to decipher, to their advantage, different states of the predator. More specifically, the question of whether an animal can become aversively conditioned to another one as a consequence of the latter being treated (stationary) near a vulnerable possession (Section II,B,4) cannot be answered for any species.
2. The Effect of Perceiving the Predator Perceiving the complete releasing situation for a response may change the range of effective stimuli. By “priming,” a previously ineffective stimulus acquires eliciting properties through association with initially effective ones so that the range of stimulus objects is broadened, given some degree of similarity between them (Hinde, 1970). Could such a process safeguard a prey against its predators even if these occur in a situation not readily decoded by an IRM? Shalter (1978b) found that 42% of parent pied flycatchers did not mob a pygmy owl mount which others vociferously harassed. When the nonmobbers were presented with a live pygmy owl, all of them mobbed it. When one day later these same birds were again given the pygmy owl mount, all responded in a way indistinguishable from those that had reacted during its first presentation (median call rate 85 and 79/min, respectively). Thus, the range of stimulus objects had increased, although proba-
ASPECTS OF ANTIPREDATOR BEHAVIOR
209
bly not that of the constituent key stimuli, since a stuffed owl lacks such features as movement that the live owl presents. In other experiments with great tits overlooking a tawny owl mount, one head turn by 90" sufficed to induce incessant mobbing of the mount (E. Curio, unpublished). Shaker's (1978b) priming experiment revealed a still unsettled enigma. During an experiment 15 years earlier, employing the same owl mount, Curio (1975) had found that only 8% of the pied flycatchers, that is, significantly less ( p < .001), had failed to mob. One possible explanation is that the densities of owls in the pied flycatcher's central-west African winter quarters might have diminished between the two experiments. If true, this would suggest species differences in the ability to decode owl stimuli since both man-made and natural Kaspar Hausers of several species have been found to respond to a stationary model owl without previous priming with a live one (Table I). A similar consideration applies to the ability to recognize part of the predator without having had a chance to see a whole live one previously (Section III,A,2,a). A rather sophisticated mechanism, perhaps entailing priming sensu lato, was reported by Owings and Coss (1981) for California ground squirrels. Young animals that have on first encounter innately avoided a rattlesnake are thereafter, but not before, reluctant to flee from a human into their burrow. Burrows are the habitual place of retreat when thus threatened. Rattlesnakes routinely enter squirrel burrows to escape the heat or search for squirrel pups. Perception of the snake has apparently been generalized to encompass places that a snake habitually visits. During an encounter in a burrow squirrels are at a disadvantage as compared to aboveground. The fear of a burrow upon perceiving a snake nearby is reminiscent of the sign-of-predator hypothesis suggested above for birds (Section II,B,3,b), though the underlying mechanisms may, of course, differ. There is a dearth of information on priming in relation to adversary avoidance. An increase of distraction displays by red grouse due to a rise of encounter rate with red foxes has been onerously explained by Hudson and Newborn (1990). They suggest that more foxes mean a higher proportion of young foxes; young foxes in turn mean, because of their inexperience, more reward for distraction displays. It would be worthwhile to follow up this functional interpretation. It predicts that predators less in need of education by their prey should not, by their increase in numbers, lead to more pronounced defense. 3 . Pursuit of the Subject As a result of being pursued, animals may rapidly develop avoidance behavior that forestalls a successful attack by the pursuer (e.g., Benzie, 1965: stickleback). In a thoroughly studied case (Dill, 1974: Bruchydunio
210
E. CURIO
rerio), the changes were found to concern the R M responding to a looming stimulus rather than the motor pattern of flight. Similarly, Darwin's finches (Geospiza sp.), which initially allow a human to approach to a distance of 0.5 m, hop away on the ground at an increasing distance as slow pursuit continues (E. Curio, personal observation). With cats, the degree of avoidance initially present does not effectively protect the more ground-living finches against an attack (A. Kastdalen, personal communication). How quickly they develop avoidance behavior as a consequence of being stalked is not known. As a consequence of relaxed predator pressure, mainly from carnivores, island birds have apparently developed a reduced ability to recognize them innately and to learn from being pursued by them. Of all avian species that have become extinct through introduced mammalian predators, 90% have lived on oceanic islands (Johnson and Stattersfield, 1990). Yet, even game animals may lack any innate fear of humans. American elk (Ceruirs elaphirs) are not wary where humans are rare (Altmann, 1958). Pursuit may take quite subtle forms. In predator-naive domestic chicks, a pair of artificial staring eyes elicits an initial avoidance that is enhanced if the eyes are made to track the subject with their gaze (Scaife, 1976b). Unfortunately, this tracking movement was not controlled for by applying other modes of eye movement excluding the threatening gaze. Moreover, prior to testing, the chicks were not naive in terms of a human face, thus marring this and many other similar experiments; the experimenter may unwittingly engender fear from staring eyes while picking the subject up prior to a test (see Section III,A,2,a). Staring at animals, or lifting binoculars to look at them, is believed to scare them only after they have been persecuted (shot at) by man (Goodwin, 1976: corvids; Booth, 1962: Ceropithecus). Nevertheless, continually fixating a particular subject may engender an unusually high degree of fear of humans. During a 5-day observation period, a focal Javan mannikin (Lonchura leucogasrroides) in a flock of conspecifics became dramatically afraid by merely being looked at, so that it hid in plants or a nest box once the observer resumed her observation of it (B. Krause. personal communication). From the above, one may conclude that the widespread innate fear of a pair of staring eyes (Section II,A,4,a) may be reinforced by the subject merely being uisucilly targeted, without any other overt signs of imminent attack. Targeting a victim, being a prelude to an attack, seems likewise to be decoded innately and apparently reinforces an initially low level of avoidance. Whether this enhanced avoidance results from the stare only, or includes the whole of the adversary, is presently unknown. The mere proximity or approach of a human may scare naturally raised birds. As is well known for captive birds, talking or whistling relieves this
ASPECTS OF ANTIPREDATOR BEHAVIOR
21 1
tension in different species to various degrees. The reason probably is that stalking predators are always silent, and when they vocalize they apparently pose less of a risk. 4.
Cultural Transmission of Enemy Recognition
a . Learning in the Presence of a Predator. A protective response may become enhanced by perceiving either a (potential) predator chase, kill, or hold of a (would-be) victim, or an animal being mobbed. In both cases, learning takes place only if the adversary is present. Pursuit of would-be victims by a predator offers another potential source of information on how to survive an attack both to the pursued and to onlookers. The best knowledge presently available is in regard of the response of animals that perceive others being pursued. How this experience translates into protective action later on remains unknown. Many prey animals, such as fish (refs. in Dill, 1974; Suboski e t a / . , 1990). birds (Davis, 1975), and mammals (Gerkema and Verhulst, 1990), flee into cover when they notice others flee who are knowledgeable about an imminent threat. This knowledge may derive from earlier experience with that threat (refs. in Dill, 1974). In many situations, information on the direction in which to flee as a result of observing informed prey animals is crucial. Depending on the species and on the circumstances, the observers either flee into the nearest cover, even at the risk that they thereby approach the predator (Tinbergen, 1951), or they flee in the direction they see others take. What happens when the social information about the direction where an attack is expected to come from interferes with information on a real attack? In an experiment testing the “alerting others” hypothesis of mobbing, Frankenberg (1981) pitted these two sources of information against each other. An observer blackbird had a free view of an actor who mobbed a little owl mount, but the actor could not see a rod which was visible to the observer (Fig. 25; curtain shown by solid and dashed lines). As a result of the interaction, the observer mobbed as well and, when startled by sudden movement of the rod startle stimulus [SS], took off with a significantly lower latency than normal. Taking off was invariably directed away from the SS that the actor could not see. With a small delay, the actor took off in the same direction, that is, away from the owl. With a slight modification of the setup, the actor was allowed to see the SS as well (Fig. 25; curtain solid line only), in order to test for an effect of the actor’s mobbing on its own escape performance. When the owl and the SS were in the same direction from the cages, both birds took off away from the SS (Fig. 26: Tl). However, when the SS startled the birds into flight from the opposite direction to the owl (T2), the actor still fled from
212
E. CURIO
/4 I I +1I Actor
1 m
FIG. 25. Experimental setup to unravel the information content an observer blackbird extracts from the mobbing of an owl by the actor blackbird. The owl is visible only to the actor; the startle stimulus is visible only to the observer in one experiment (both solid and dashed line of c curtained), and to both birds in another (solid line of c only curtained). (Modified from Frankenberg, 1981.)
Actor
c1
11
Observer
QQ QQ Owl
12
ss c2
00 ss
FIG.26. Spatial arrangement of predator stimulus (owl) and startle stimulus (SS) to test for directional information encoded in the mobbing of an actor blackbird when seeing the owl and in the startle response of an observer blackbird when scared by SS. Arrows denote the direction of first takeoff upon receiving stimulation from the conspecific's action in a neighboring cage; thin lines denote the SD of takeoff directions (for details of setup see Fig. 25). CI, C2 and TI, T2 denote control trials and test trials, respectively. in order from top to bottom. See text for discussion. (Modified from Frankenberg, 1981; courtesy of author and Parey, Berlin, Hamburg.)
ASPECTS OF ANTIPREDATOR BEHAVIOR
213
the owl; the observer, with the owl barred from its view, largely behaved the same but part of this group compromised by jumping toward the SS or sideways. Hence, with conflicting information about a potential attack (T2), the real predator had precedence over the SS although it did not move, and, even more surprisingly, the observer apparently rated the information extracted from the actor’s behavior higher than what it sees itself when the SS moves. Two control experiments (Cl, C2), with the SS being the only disturbance, ensured that the direction of escape observed in experiments T1 and T2 was not in any way biased by the geometry of the cages and their environs. Directional information about the place of the predator is probably transferred by the actor’s mobbing behavior. The actor’s way of calling may provide a clue: The call rate increases both after each takeoff away from the owl and after each landing close to it (Frankenberg, 1981). The direction of monocular fixation may be another clue. In domesticated zebra finches there is social facilitation of mobbing among pair mates only one of which can see an adversary. Performance of the knowledgeable bird successfully conveys information about the adversary’s position so that harassment of both mates is directed at it (Hoffmann, 1979). Acoustic information from a male suffices to induce the behavior in a female separated visually from it (Lombardi and Curio, 1985b). A distressed or killed conspecijic near an enemy has been shown to elicit antipredator behavior from individuals or groups (Section II,B,3). The strength of the behavior surpasses that given to the predator alone. As a consequence of such behavior, experienced individuals tend to respond more strongly or earlier on any future occasion when they have the same or a similar encounter, or they avoid the site where it has occurred previously. Foreshadowed by the anecdotal observations of Lorenz (193I), Conover and Perito (1981), and Conover (1987) demonstrated that starlings and ring-billed gulls (Larus delawarensis) mob more vigorously at a predator (owl, human) after having experienced it hold a live, adult conspecific. In both these cases and similar ones (Csanyi, 1985), the experienced birds had partaken in mobbing assembliesaround the threatened or dead conspecific. Therefore, it remains unknown whether merely perceiving the threat and/or the commotion around it would suffice to bring about conditioning. Strangely, pairing the owl with a dead starling or the species’ distress calls failed to render the owl more effective later on (Conover and Perito, 1981). From an experiment with domestic ducks on “empathic learning” it would appear that the mere sight of a distressed conspecific leads to an avoidance of the place of danger (Klopfer, 1957), though both the latter and the aversive stimulus used (electric current) were unnatural. To see whether animals can extract information about a predator by
214
E. CURIO
perceiving it in conjunction with a killed victim, Kruuk (1976) presented a dead gull next to a stuffed predator to mixed gull colonies (Larus argentarus, L . fuscus). Incidentally, this combination of stimuli, simulating an act of recent predation, scared the gulls more than did the sum of the releasing values of each one of them. This furnishes yet another example of stimulus dilation (Section II,A,4,b), demonstrating its validity not only for key stimuli of one Gestalt (predator) but also for a combination of two Gestalten (predator, prey). The releasing values of the predators used, that is, hedgehog (Erinac-euseuropaeus), stoat (Mustela erminea),and red fox (Vulpes uulpes), increased in that order, thus reflecting their biological significance to the gulls. Having seen a stoat with a dead gull, the gulls responded to the stoat thereafter, at the same site, with increased avoidance, as measured by alighting distance (Fig. 27) and greater flock size. A habituation experiment with the stoat only verified that the increase was due to some learning effect. In order to examine whether this effect was due to conditioning to the site of the encounter or to the circumstances of presentation, a control
FIG.27. Avoidance by herring gulls and lesser black-backed gulls of a stoat, a stoat plus a dead gull, and a stoat during the 5-min presentations 1 to 3 (-), respectively. versus three trials with the stoat alone (---). Avoidance was measured by the mean alighting distance of a flock circling above the stimulus. Significance level relates to the difference between trials 1 and 3 in the first-mentioned experiment. (From data in Kruuk, 1976.)
ASPECTS OF ANTIPREDATOR BEHAVIOR
215
experiment was run with a hedgehog being shown in trials 1 and 3, while the combined stimulus of the stoat and gull was presented in trial 2 (Fig. 28). Since experiencing the combined stimulus left the releasing value for the hedgehog unaltered, the learning observed must have been specific for the stoat. This stimulus specificity may suggest that the two predators involved enter two distinct channels (see Section 11,A,6). Unfortunately, the reciprocal experiment testing for any reinforcing role of the hedgehog in trial 2 has not been performed. The experiments carried out so far do not yet permit us to conclude whether the predator attraction itself is important in this learning process; birds that fly away immediately, rather than stay and circle above the enemy, might show the same kind of learning process. The increase of avoidance behavior due to this type of social learning is probably adaptive in coping with predators that engage in surplus killing. Stoats and foxes and, to a lesser degree, hedgehogs are known to belong to these species; a predator that has killed one prey in a place is likely to kill others there once the opportunity arises (Kruuk, 1976). Learning by observing others interact with a predator lies at the heart
- 6
E
c u C
0
4
0
FIG. 28. Avoidance by gulls of a hedgehog in trials 1 and 3 to test for the stimulus specificity of seeing and avoiding a stoat plus a dead gull inbetween (trial 2). *denotes the releasing value of the stoat in the conditioning experiment in Fig. 27. (From data in Kruuk, 1976.)
216
E. CURIO
of cultural transmission of enemy recognition and has become an area of active research. This mode of learning about adversaries safeguards the prey when an IRM is of no avail, for example, after invasion of an area by a predator. Knowledge about the new danger in those transmitting it to novices may in this case have come about by more conventional modes of learning (e.g., Section III,D,3). The experimental evidence on cultural transmission of enemy recognition in the European blackbird and the rhesus monkey (Macaca mulatta) has been summarized (Curio, 1988b; Mineka and Cook, 1988), so a brief outline will suffice here. The basic method, pioneered by Curio et al. (1978a),consists of permittingan enemynaive learner (Lr) to perceive a conspecific teacher (Tr) interact with a predator that is invisible to the Lr, while the Lr is looking at the conditioning object to be feared; this latter object is, in turn, not visible to the Tr but is closely juxtaposed to the dangerous object the Tr is interacting with. This ensures that the Lr regards the conditioning object as something feared by the Tr. The criteria for selecting a conditioning object were as follows: (a) it had to be novel; (b) it had to be unlike any predator that is recognized innately; (c) it had to be similar in size to a dangerous predator. A jay-sized Australian honeyeater (Ho) fulfilled the criteria reasonably well. A test for novelty ensured that the Ho elicited mobbing at only a low level (Fig. 3), though at a level that was higher than the previous response to a wooden box used to rotate the Ho into view of the Lr (Fig. 29, trial 2). In a third trial (reinforcement), this response was reinforced by pairing the Ho with a conspecific mobbing a little owl, thereby raising the Lr’s response level about threefold. In a final test for learning (trial 4),the Ho proved to be about as effective a stimulus eliciting mobbing as is the owl. Repeated presentation of the Ho without any social reinforcement brought about habituation of the novelty response (trial 2) so that sensitization (pseudoconditioning)to the Ho could be ruled out as an explanation of the transmitted response in the trial 4 test. Although mobbing is sympathetically induced in the Lr while perceiving the Tr mob the position of the Ho, it is dispensible for full transmission to occur (Curio, 1988b). In order to discover the extent to which transmission of the Ho was natural, whether learning would occur as well when a subject had previously encountered a potentially dangerous animal in the absence of a knowledgeable Tr was examined. To this end, the Ho was first rendered ineffective by habituating blackbirds to it, and they were subsequently subjected to the conditioning procedure outlined above. As a result, transmission took place unimpaired, thus raising confidence in the naturalness of the transmission procedure substantially. Encountering an adversary without any education does not blindfold the prey to successful education
ASPECTS OF ANTIPREDATOR BEHAVIOR
217
3 3
-0
aJ .-N
?
1
0 0 C
0
f
p9 aoe in weeks
60
255
234
149
I
n
FIG.4. Changes in form of oblique-like postures in the course of ontogeny of black-headed gulls. Shown are the relative frequencies (left y-axis) of three categories of combinations of different bill and neck positions, together with the duration of the display (right y-axis), for four age classes of young black-headed gulls. Bars with densest stippling: bill downward, neck oblique or vertical; bars with less stippling: bill horizontal, neck vertical; white bars: the normal adult form: neck oblique, bill horizontal or upward or neck vertical, bill upward. The depicted postures represent the most frequently performed form within the form category.
intervals. The long call consists of such a sequence of notes also, and the final parts of the long call in adult birds may be quite similar to the choking call. However, especially at the beginning of the bout, the notes of the former differ from those of the latter in that they are louder, longer lasting, and higher in pitch. Both these call types gradually change in form in the course of ontogeny. Furthermore, these changes support the interpretation that the oblique display emerges from the choking display. Quantitative evidence for this, based on the form criteria mentioned in Section II,b,l,b, can be found elsewhere (Groothuis, 1989b). The course of development of these calls is qualitatively depicted in Fig. 5 . The early-appearing harsh-sounding notes that can already be heard from the egg become more harsh and embedded in the typical sequential structure with short note intervals (Fig. SA-C). From this point on, two developmental pathways can be found: first, toward the adult choking call. Here a gradual change in pitch and duration of the call takes place (Fig. SD),resulting in the call sounding more muffled. This is probably caused by the fact that the bird is now performing the
i"?yPi time (sec)
I
1
1
I
ONTOGENY OF SOCIAL DISPLAYS
283
choking display with closed bill. Second, a change toward the adult long call. This starts at about 3 weeks of age with the insertion of some highpitched notes with a slightly longer duration between the precursors of the choking call (Fig. 5E).Although the essential structure of the call does not change thereafter, changes in the overall tonal quality still take place. In the course of the third month after hatching, the calls sound much clearer and higher in pitch (Fig. 5F), whereas in early spring at the age of 10 months the calls sound extremely hoarse (Fig. 5G). Finally, about 2 months later, the calls are indistinguishable from the normal adult call (Fig. 5H). d . Conclusion. Both the species-specific display postures and accompanying calls emerge gradually in the course of ontogeny. The ontogenetic sequence of changes in forward-like postures and in the pumping movement, the begging display of young birds, which have not been discussed so far, are also depicted in Fig. 1. Based on data concerning the form, the sequence, and the accompanying vocalizations of the different motor patterns, it could be shown that the development of pumping concerns the integration of the alert (in which the chick scans the surroundings with an extended vertical neck when there are disturbances in the environment) and the forward-like postures. Ontogenetic changes in the latter are therefore also reflected by ontogenetic changes in the pumping movement, which thereby becomes an increasingly conspicuous begging display. The course of changes in form of the motor patterns consists of changes in the spatial position of separate elements and the addition of new form elements. Combinations of a specific call and a specific posture become gradually more clear. Furthermore, a display may emerge from building blocks of another display or by an integration of two different motor patterns which have emerged independently at an earlier stage. By these changes, the motor patterns gradually change in form toward a more pronounced display. Finally, in the course of development, transitional forms of display follow each other in a more or less fixed sequence. e . Final Remarks. i. Frequency changes and regression. All four display types gradually increase in frequency in the course of the first 4 weeks
FIG.5. Sonagrams representing the course of emergence of the choking call (A to D) and the long call (A, B, C, E to H) in the black-headed gull. (A) Harsh call day I after hatching, showing irregular note intervals; (B) harsh call day 5, showing the typical regular note intervals of short duration and less well defined frequency spectra; (C) harsh call day 19 in which tonal quality is almost totally absent; (D) typical stifled rhythmic choking call, performed with the bill closed, day 26; (E) long-call-like vocalization, day 30, with high-pitched notes of a slightly longer duration in the beginning of a harsh call bout; (F) long-call-like vocalization week 10, with well-defined frequency spectra; (G) hoarse long call in first spring, similar in frequency spectra with early harsh calls; (H)adult long call.
284
T. G. G. GROOTHUIS
after hatching. Interestingly, the expression of a more complete display in a certain individual during a certain developmental stage often took place in concordance with a higher frequency of performance of that display by that individual. On the other hand, the frequency of performance of certain transitional forms of display decreases considerably with, increasing age of the bird. Moreover, the complete pumping display disappears altogether from the repertoire of older birds (although the incomplete pattern returns in a slightly different form more than a year later in the male’s repertoire as a precopulation display). This leads to the question of whether the neural mechanisms for some displays become reorganized. This seems not to be the case. For example, a gull raised in abnormal social conditions (see below) started to reperform complete pumping frequently when adult. Further, adult birds, when heavily attacked by other birds in a situation where they cannot escape (for example in a small cage and with their wings clipped), reperform crouching in the form typical of young chicks. Thus, the neural coordination mechanisms for juvenile display still seem to be present in adult birds. This is reminiscent of the finding of Bekoff and Kauer (1982) that the neural circuitry for the hatching movement of chicks is still present after hatching. The inhibition of motor patterns that were expressed at an earlier stage of development draws attention to the fact that ontogeny of behavior includes not only progression but also regression. This, of course, must be related to the function the motor patterns fulfill at each particular stage of ontogeny (see Section V). ii. Discontinuities. To some extent, emphasis has been laid on the finding that displays gradually change in form in the course of ontogeny. However, the data presented so far are mean scores of groups of birds and differences were found between individuals in the time span in which the different transitional forms of display were performed. Furthermore, individual birds could suddenly come up with the next transitional display, which then became the main display type for some time, although they had hardly performed this motor pattern before. Thus, a young bird may suddenly jump from one developmental stage to another. Such discontinuities may mark important stages in development and should be analyzed in more detail (see also Plooy [1980] and Plooy and Van de Rijt-Plooy [1989] for the development of behavior in chimpanzees and humans). iii. Integration and Differentiation. In the first half of this century, two contrasting views on motor development were prevalent. One was the idea that, in the course of ontogeny, separate motor units become integrated in larger functional units (Kortlandt, 1940). The other was that the subunits, such as limb coordination, gradually differentiate from an early movement
ONTOGENY OF SOCIAL DISPLAYS
285
pattern that involves the whole body of the animal (Coghill, 1929). However, several authors are now of the opinion that both processes take place in ontogeny (e.g., Plooy, 1980; Fentress and Mcleod, 1987), and this is clearly the case for the development of displays after hatching in the blackheaded gull. For example, in the course of development of the choking display, different motor units, such as extended carpaljoints, a tilted body, and movements of the head, gradually “differentiate” as separate motor units that can be performed together or apart from the other units. Furthermore, the oblique display, both the posture and the long call, seems to differentiate from the choking display, and later in ontogeny,both displays are performed in slightly different contexts (Moynihan, 1955). Integration of two different motor units, the alert and forward-like displays, takes place in the development of the pumping display. However, whether a description of motor development in these two ways really refers to causal processes underlying the development of neural circuitry needs further study. iu. Onrogeny and phylogeny. Displays are thought to have evolved in the course of evolution from intention movements, for example, for locomotion, aggression or escape behavior (e.g., Tinbergen, 1952; Baerends, 1975). Because more conspicuous forms of these intention movements could function as important tools for intraspecific communication, these movements gradually evolved by natural selection to the present conspicuous displays. This process is called ritualization (Daanje, 1951;Tinbergen, 1952;Morris, 1957;Hinde, 1969).Because species recognition is strongly adaptive, closely related species developed different displays, although often with common basic features because of their common ancestor. The gradual development in form of displays in the course of ontogeny in the black-headed gull shows striking similarities with the changes in form that are postulated to have occurred in evolution. As discussed earlier, the display patterns in young gulls become increasingly conspicuous. Furthermore, young gulls frequently perform incomplete display in alternation with intention movements of aggression and escape (Groothuis, 1989~).Moreover, several incomplete or transitional forms of display in young black-headed gulls show great resemblance to adult display of related gull species (Groothuis, 1989~).It is often found in developmental biology that ontogeny “recapitulates” phylogeny. The reason for this is the fact that evolution takes place by modifying already existing mechanisms in the course of the life of an organism. Therefore, phylogeny is modified ontogeny (de Beer, 1940). From this perspective, the data on the ontogeny of social displays presented so far seem to support current ideas about the evolution of these motor patterns.
286
c.
T. G. G. GROOTHUIS
POSSIBLE MECHANISMS OF DISPLAY DEVELOPMENT
In this section, four not mutually exclusive hypotheses are discussed to explain the data presented in Section I1,B. In the next Section (II,D), I discuss experiments aimed at testing these mechanisms. The first three hypotheses have in common that the neural coordination for the complete display is present early in ontogeny. The gradual emergence of the complete display may than be explained by assuming that the expression of the circuitry changes gradually, depending on other factors: 1 . A change in the motor apparatus, such as growth of muscles or efferent neural connections. This is often called maturation, and might explain changes in motor patterns early in life, such as those reported in the development of bill pecking in gulls (Hailman, 1967) or of head-bobbing in lizards (Roggenbruck and Jenssen, 1986). 2. A change in external stimuli. A young animal may have most of its interactions with young of approximately the same,age, and only later in ontogeny with larger adult conspecifics. The latter may provide stronger external stimuli than the former and therefore may trigger the motor pattern more completely. 3. A change in internal stimuli, such as the development of the proper motivational state for aggressive behavior, which triggers the complete display. In his classical study on the ontogeny of social behavior in junglefowl, Kruijt (1964) explained the development of displays in this way by postulating a development of different motivational systems and changes in the relationship among these systems. However, alternatively, the neural coordination mechanisms themselves may not be complete at first and their development may also explain the gradual emergence of the adult displays. Which factors may influence this development? 4. One plausible hypothesis is that the young animal may shape its display on the basis of proprioceptive feedback, resulting from the performance of incomplete display. This kind of feedback may than be matched against some sort of neural template, in which the information about the species-specific form of display is encoded. This mechanism, discussed by Baerends van Roon and Baerends (1979), is similar to that proposed for the development of song in songbirds, although in the case of display postures proprioceptive feedback will be involved while in the case of song development the feedback is auditory. If this hypothesis is correct, the next question to be answered is how the template itself may develop. Two types of external influences could be relevant for this: 4A. The young animal may acquire the information in the template by
ONTOGENY OF SOCIAL DISPLAYS
287
observing display of conspecifics and copying that form of display through imitation. This is clearly the case in song development in several species of songbirds. However, in song development, the song of the tutor (auditory input) is compared with the bird’s own song output (again using auditory input), and therefore only one sensory mode is involved. This is not the case in imitation of postures or movements. In the many cases when an animal cannot see its own performance, visual input (seeing the motor pattern of the tutor) must be compared with proprioceptive feedback. Complex as this seems, we cannot exclude such a complex cognitive mechanism out of hand and there is some evidence that young human babies already have the capacity to imitate facial expressions (Meltzoff and Moore, 1977; Field et al., 1982). Whether in this case the young baby is really shaping its motor output into a new form on the basis of visual input, or whether it is just triggering already existing motor patterns in response to this visual information remains, however, to be seen. 4B. The young animal may also acquire information about the proper from of display on the basis of social reactions of conspecifics to the different forms of display it shows in interactions with them. If conspecifics are able to differentiate between incomplete and complete display, they may react consistently differently to these forms. If reactions to the property complete display are more reinforcing than reactions to incomplete display, the performer may shape this display by operant conditioning. A prerequisite for this is that subtle differences in the form of display are recognized and carry meaningful information for conspecifics. This has been shown to be the case for several gull species (Beer, 1980; Veen, 1980; Groothuis, 1989b, 1992). That the reactions of a conspecific to the display performed may influence the development in form of that display has been shown to be the case for song development in the cowbird (West and King, 1988). These four hypothesis are not mutually exclusive. Maturation of motor systems has been shown to be dependent on the activity of these motor systems (Drachman and Sokoloff, 1968; for reviews, see Oppenheim. 1981, and Prechtl, 1984). Therefore, the presence of the proper motivational state (hypothesis 3) may stimulate the performance of display, which in turn may stimulate maturation (hypothesis 1). (In fact, the term maturation is so strongly associated with autonomous growth of tissue that this term has become obsolete and the application of it should be avoided.) The proper motivational state may also increase the frequency with which the animal matches its display against the template (hypothesis 4), and stimulate the young to initiate social interactions. These then create the proper situation for the animal to imitate the display of the opponent (4A),
288
T. G. G. GROOTHUIS
or to shape its display on the basis of reactions of conspecifics to its own behavior (4B).Furthermore, as a “safeguard,” more than one mechanism may guide the display to its normal adult form. Therefore, in revealing the mechanisms of display development, confirming one hypothesis does not necessarily make testing of the other hypothesis redundant. D. TESTING THE HYPOTHESES 1 . Limitations in the Motor Side
This hypothesis predicts that limitations in the peripheral motor apparatus explain the occurrence of incomplete display postures in the blackheaded gull. However, even I-week-old chicks are capable of performing many motor elements of the complete adult display postures. This can be deduced from two findings. First, many elements of the complete display can be observed in the young chick in various contexts long before they become incorporated in the display. For example, raised carpal joints are shown early in ontogeny during running; the proper bill and neck position for the adult oblique can be seen in chicks during begging, when they peck to the parent’s bill for food. The closed bill and standing instead of crouching, two elements of complete choking, occur very early in ontogeny, long before the emergence of the complete display. Second, complete oblique display has been observed in young gull chicks, although only briefly and rarely. Although the performance of all the proper form elements in the correct combination may be attributed to a coincidence and not to the result of central coordination, all motor elements did appear. Therefore the first hypothesis should be rejected. However, limitations on the motor side of the machinery may be relevant to explain display development very early in ontogeny and may be more pronounced in altricial species than in a species like the black-headed gull. Maturation may explain the development of some,characteristics of the vocalizations. The increase in note duration may be caused by an increase in the volume of the air sacs of the bird. Changes in pitch may be caused by growth of the membranes of the syrinx, the birds’ primary vocal organ. Evidence for the latter follows from the finding that young gulls that has been treated with testosterone performed calls with adult characteristics, while autopsy showed an enlargement of some parts of the syrinx (Groothuis, 1992).
2 . External Stimuli The second hypothesis postulates that the increase in the completeness of display in the course of ontogeny is caused by an increase in strength
ONTOGENY OF SOCIAL DISPLAYS
289
of external stimuli to which the young are exposed. This hypothesis was tested by rearing young chicks by hand and confronting them weekly with the same strong external stimulus: a stuffed adult black-headed gull in breeding plumage and in the oblique posture. The model was mounted at the top of a long stick and was maneuvered in such a way that attack movements of adult intruders on the territory in the field were simulted. This provided the proper context for the agonistic displays. The experimental birds showed essentially the same gradual emergence of complete display as control birds, although display postures with raised carpal joints were slightly more often performed than in agonistic interactions with young of the same age. These results showed that the second hypothesis is unlikely as far as the black-headed gull is concerned. However, in other species, the role of external stimuli in triggering display early in ontogeny is important, such as in the cichlid fish Aequidens rivularus (Groothuis and Ros, in preparation). In this species, the young gather in large schools during the first months of their life and show a gradual increase in frequency of different displays. Fish that are younger than 6 weeks hardly ever show agonistic display and never show sexual displays. However, in confrontations between young of 6 weeks of age, display frequency increased strongly if the participants had been isolated for 2 days before the tests. This suggests that, in a school, young cichlid fish habituate to each other and, consequently, their companions are less effective for triggering display behavior. This is obviously functional, because premature agonistic interactions will disrupt schooling behavior, which is important for reducing the risk of predation. When the stimulus fish was treated with sex hormones, which induced premature appearance of the adult color patterns, the isolated but hormonally untreated opponent even showed sexual displays in confrontations with this stimulus fish. Similar effects of short-lasting social isolation on the increase of display in confrontations with other young early in ontogeny have been reported by Kruijt (1964) for junglefowl. In this species, adultlike although shortlasting crowing postures have been observed in chicks of I week old, reared in social isolation, during confrontation with an imprinting stimulus (a small blue ball) (T. Groothuis, unpublished observation). These examples show that certain uncommon external stimuli can provoke relatively high frequencies of adultlike display in young animals of several species. Unfortunately, the forms of these premature displays have not been analyzed in detail. 3. Development of Motivational States The gradual emergence of a complete display may be attributed to the fact that the motivational systems controlling that display develop
290
T. G. G. GROOTHUIS
gradually or become increasingly activated. In this case (as in the previous case, 2), one would expect the emergence of display to be the result of a decrease in the threshold for the different motor elements to appear, and not the result of increasing coordination between these elements. Indeed, in the black-headed gull, evidence for such an increased coordination is lacking. In each age class, even in chicks, many elements characteristic for a particular adult display occur more often in combination with another element of that display than could be expected if these elements occurred independently from each other (Groothuis, 1989b). Thus, the change in form of the displays does not consist of a development from random combinations of form elements to well-coordinated motor patterns. This conclusion is in accordance with the finding that display in young gulls is not necessarily less stereotyped than display in adults. On the contrary, the course of changes in display consists of a more or less fixed sequence of transitional forms of display, as is shown in Figs. 2 and 4. This course of changes can be adequately described as a change to the performance of more pronounced display or, in other words, from a low-intensity display to a strongly motivated display. The influence of the development of motivational states can be tested by manipulation of these states that control display in the young animal. If the hypothesis is correct, precocious strong activation of the relevant motivational system should induce precocious complete adult display. In order to carry out such an experiment, determination of the motivational background of display in the young animal is necessary. To this end, an ethological black-box analysis of the motivational systems underlying the occurrence in frequency and form of the four displays depicted in Fig. 2 has been undertaken (Groothuis, 1989~).In this analysis, it is assumed that: (a) if external stimuli are held constant, changes in behavior are due to changes in the internal factors controlling that behavior; (b) that motor patterns showing similarities with each other in form, frequency changes, temporal sequences, and context of occurrence share an internal common causal factor. A stuffed adult black-headed gull was used as the standard external stimulus in this study. It was placed at least weekly in the territories of the chicks, inducing a lot of agonistic behavior. Display behavior, as well as other motor patterns such as overt aggression, overt escape, feeding, and locomotion, were compared with each other. This analysis resulted in the following conclusions: Early in ontogeny both choking-like and forward-like display are primarily under the influence of two subsystems that control fear behavior. These are, respectively, a system controlling a tendency to hide or to stay put and that controlling a tendency to escape. The pumping display is influenced by a state of alertness, induced by hunger. The changes in form of the forward- and
ONTOGENY OF SOCIAL DISPLAYS
29 1
choking-like displays, and the emergence of the oblique from the latter display are due to the influence of a motivational system for aggression. This system becomes increasingly activated, simultaneously with the activation of the other motivational systems. If this interpretation is correct, manipulation of the internal variables controlling aggressive behavior should result in a change in the gradual emergence of complete display. In many animal species, including birds, aggressive behavior and agonistic display in adults are known to be under the influence of sex hormones. Interestingly, in gull species in which the sexes are monomorphic both in plumage and in display behavior, both the male and the female have been shown to be sensitive to the male hormone testosterone (laughing gull; Terkel et al., 1976). Furthermore, both sexes show comparable blood levels of testosterone during a considerable part of the breeding period (Western gull; Wingfield et al., 1982). Moreover, juvenile black-headed gulls of both sexes show an increase in blood levels of testosterone from less than 0.1 to 0.5 ng/ml during the period when they persistently show complete display for the first time in their life (Groothuis and Meeuwissen, 1992). These data justify further analysis of the effect of testosterone on display development in young black-headed gulls. Young black-headed gulls were treated with testosterone at a stage of ontogeny before the emergence of complete display. If the neural coordination mechanisms for the complete display are indeed present early in ontogeny, and if their expression can be activated by testosterone, two kinds of results are to be expected: (a) the complete adult display will be expressed early in ontogeny; (b) birds will be able to skip the performance of incomplete display in developing the complete motor pattern. The second finding would imply that incomplete display patterns are not necessary precursors that fulfill a function in the development of the form of complete display. In contrast, incomplete motor patterns can then be taken as merely epiphenomena of a gradually increasing motivational state for agonistic behavior. This would imply that a mechanism by which the bird matches the actual motor output against a template, in order to develop the complete display (the fourth hypothesis), is not a necessary part of the developmental process of display patterns. To overcome possible confounding influences of limitations in the motor apparatus and in the metabolism of testosterone, the first experiment was carried out with gulls that were already 10 weeks of age (Groothuis and Meeuwissen, 1992). At this age, normally reared gulls are able to perform the complete displays. To prevent the experimental animals from gaining experience with the performance of incomplete display before hormonal treatment, all birds, including the controls that were not hormonally treated, were reared in social deprivation (in isolation or in very small
292
T. G. G. GROOTHUIS
groups). The complete lack of the performance of incomplete display before hormonal treatment, due to the lack of adequate social stimulation, was confirmed by observations. Hormonal treatment consisted of subcutaneous implantation of one or two 25-mg pellets of testosterone propionate which released hormone for at least 3 weeks. After hormonal treatment, both aggression and the oblique-,,forward-, and choking-like displays increased rapidly in frequency; the pumping display did not show such an increase. This confirmed the interpretation of the results of the motivation analysis discussed above: the first three are agonistic and the pumping display is part of nonagonistic behavior. The three agonistic displays also showed a strong acceleration in form development. Some of the data on the oblique display are summarized in Fig. 6. The young birds reached adult percentages of complete oblique postures within 5 days after implantation, which is significantly shorter than in normal development. Moreover, in some birds, during numerous observations, incomplete display was hardly observed or not at all. These individuals showed high percentages of complete oblique postures from the first day after implantation onward (Fig. 7). This indicates that young black-headed gulls do not need to practice their motor output in order to develop complete display.
1001 Mcomplete oblique O O ’
c
0 1 2 3 4 5 days after implantation FIG.6. Frequency and form of oblique-like display in approximately 10-week-old naive young black-headed gulls over the course of 1 I days after testosterone implantation in 14 experimental and 7 control birds. Day 0 is the day of implantation. (A) Total frequency and (B) average percentages for 14 birds of oblique-like postures that consisted of complete adult display. Bars are standard errors. (Modified after Groothuis and Meeuwissen, 1992.)
293
ONTOGENY OF SOCIAL DISPLAYS
40 20 N=O
7
18
0
1
2
21
3
34
0
0
3
10
4
24
3 4 5 0 1 2 3 4 5 days after implantation FIG.7. Percentage of oblique-like display that consisted of complete adult oblique postures in two young naive black-headed gulls (A and B) after implantation of testosterone at the age of 10 weeks (day 0). (Modified after Groothuis and Meeuwissen, 1992.)
This result raises the question of how early in ontogeny the neural coordination mechanism for the adult display may be present. The answer is provided by the results of an experiment in which young black-headed gull chicks were treated with testosterone at day 5 after hatching. Essentially the same results were obtained as with the older gulls in the previous experiment. The hormonally treated chicks performed conspicuous longlasting complete oblique, choking, and forward display (Fig. 8), together with overt aggressive pecking. Some chicks showed complete display in percentages characteristic for adults from the very first day after hormonal treatment onward (Fig. 9). In both experiments, the time period during which the long call emerged was also shortened. After 2 weeks of hormonal treatment in the first experiment, the juvenile gulls produced calls which were indistinguishable from the hoarse calls normally produced at the age of 10 months (Fig. 5G). Some of these individuals eventually produced adult calls before the age of 5 months. Moreover, some of the young chicks in the second experiment even performed long calls with characteristics of the high-pitched long calls (Fig. 5F) or hoarse long calls (Fig. 5G),typical for much older birds.
FIG.8. Depiction of complete choking (A), oblique (B),and forward display ( C )in blackheaded gull chicks treated with testosterone on day 5 after hatching. (Modified after Groothuis and Meeuwissen, 1992.)
294
T. G . G. GROOTHUIS
A with head movements
a complete choking
4 5 11/12 0 1 2 3 4 5 11/12 days after implantation FIG.9. Percentage of choking-like display that was performed with head movements, or that consisted of complete choking, in two black-headed gull chicks after implantation of testosterone at the age of 5 days (day 0). (Modified after Groothuis and Meeuwissen, 1992.)
0
1
2
3
Autopsy of some of these birds and of controls revealed that testosterone treatment had modified some parts of the syrinx, the main vocal organ in birds. This indicates that testosterone influences the development of the vocal displays partly by peripheral effects. The chick calls, however, never acquired the duration of the adult calls, supporting the suggestion given earlier that the size of the bird, and especially of its air sacs, may be the limiting factor for note duration. The data presented so far support the hypothesis that complete display is present early in ontogeny. However, it must be stressed that, although the effect of hormonal treatment is spectacular, in many of the experimental birds, especially in the chick group, a gradual change in form and frequency of display could still be noticed, although in an unusually short time period. This opens the possibiltiy that, although practicing motor output according to the matching hypothesis is not indispensable, for at least some individuals, such a,supplementary mechanism may still be of some importance. Alternatively, gradual emergence of complete display in testosterone-treated birds may be due to the fact that biochemical processes, by which the hormone modifies behavior, need time to become effective. These biochemical processes include conversion of testosterone to active metabolites, development of receptor systems, and modification of DNA transcription and protein synthesis. To test the two possibilities, one could either treat chicks still in the egg with testosterone or test the matching hypothesis more directly. The first experiment may give the biochemical processes time to become effective, while the embryo, because it is still in the egg, is prevented from practicing display behavior.
ONTOGENY OF SOCIAL DISPLAYS
295
These experiments are currently being carried out in Haren. The second experiment has already been carried out and is discussed in the following section. It should be stressed here again (see earlier in this section) that, in the black-headed gull, testosterone appears to influence form and frequency of agonistic behavior in both the male and the female, in contrast with many other animal species. This may be related to the fact that gull species are monomorphic both in plumage and behavior: both sexes perform the same displays, except for precopulation begging in which the male performs vertical head bobbing and the female bill tossing. Furthermore, in homosexual pairs of males, I have seen males performing bill tossing and adopting the female copulation posture, and I have seen a female gull performing the male copulation posture. Moreover, in gull species in spring both sexes develop the breeding plumage, while in most other bird species this only occurs in the male. In males of many animal species this breeding plumage and display behavior are under the control of androgens. Although behavioral effects of testosterone in males may be mediated by aromatization to estrogen and conversion to dihydrotestosterone (DHT) in the brain, the breeding plumage, like other peripheral effects of sex hormones, has been shown to be under the control of androgens. The fact that female gulls need testosterone to develop the breeding plumage, and perform the same displays as males, displays which are normally under the control of testosterone in the male sex, makes the sensitivty of female gulls to the male sex hormone testosterone understandable. In this section, the effect of testosterone on the expression of complete display has been interpreted as due to hormonal effects on a behavioral system for agonistic behavior. Although testosterone affected the development of the calls partly via effects on the syrinx, such peripheral effects do not seem the proper explanation for the hormonal effects on the expression of the display postures. As was concluded earlier in this section, the neural mechanisms for the complete adult display postures are present very early in ontogeny, and need only to be activated by the proper internal stimuli. Therefore, an explanation of the effects of testosterone in terms of sensory motor development does not seem appropriate. Furthermore, testosterone not only activated the complete displays, but also overt aggressive behavior. Moreover, the hormone changed the temporal sequences of the behavior patterns (such as the oblique forward sequence) and the context of the displays in relation to external stimuli and locomotion (T. Groothuis, unpublished data). This leads to the interpretation that testosterone not only lowers the threshold for the displays to occur, but also influences a behavior system coordinating agonistic behavior. (For a further discussion of motivational factors influencing display development,
296
T. G. G. GROOTHUIS
and age-related effects of testosterone on motivational effects, see Groothuis, in preparation). 4 . Matching of the Motor Output
To establish the influence of feedback gained from the performance of motor output, by which a young animal may match its display against a template, the feedback itself should preferably be manipulated. In contrast to the case of song development, where one can easily manipulate the relevant feedback by deafening the bird, this is not readily possible in the case of development of postures, where proprioceptive feedback is involved. There are, however, two other ways to test the hypothesis: (a) One may prevent the young animal from practicing relevant motor patterns for a certain period, after which a possible retardation in motor development could be tested; (b) one may manipulate the information present in the template itself, If this template encodes information which is gained by observing display postures of conspecifics, one could provide the young animal with deviating models. If the relevant information is gained by reactions of conspecifics to the display performed, one could expose the bird to abnormal reactions or try to induce deviating display by operant conditioning. These three experiments (testing the influences of practice, imitation, and social reactions) have been carried out with the black-headed gull (Groothuis, 1989b, 1992; Groothuis and van Mulekom, 1991) and are discussed in this sequence below. a . Influence of Practice. Young birds almost exclusively perform display during social interactions. To prevent young gulls from practicing display postures, chicks were reared either in social isolation, or in small groups of 2 to 4 individuals. In the latter situation, chicks treat each other as siblings, and hardly ever interact aggressively. Display development in these birds was tested either at the age of 15 weeks (when normally reared birds have completed display development) or at the age of 10 to 12 months. All birds tested at the age of 15 weeks showed a retardation in display development, showing a mean incidence of oblique postures with the adult bill and neck position of 60% only, against 85% in control birds. The incidence of oblique display with raised carpal joints was 13%, against 80% in the controls. Birds tested at the age of 10 to 12 months showed high interindividual variation in display development (Table I). Forty percent of the birds raised in visual isolation from other birds and 55% of the birds reared in small groups without adult birds showed consistently normal complete display. Thus, imitation is not indispensible for the development of the proper species-specific adult display. However, many of the other birds
ONTOGENY OF SOCIAL DISPLAYS
297
TABLE I
OBLIQUE DISPLAY AND RAISING CONDITION Display
Control
0
Isolated
Small group
35 25
None Fragmentary Deviating Normal
0 0 100
40
0 20 25 55
N=
18
20
20
0
showed no display at all or fragmentary display (deviating display is discussed later). The latter contained characteristics of the incomplete display of much younger birds. Thus, social deprivation may cause a retardation in display development. The retardation in display development occurring in both age classes may be interpreted as support for the idea that birds need to practice the motor patterns in a social context in order to develop normal display. However, indirect evidence opens a more plausible alternative explanation. In the birds raised in social deprivation, the development of the black plumage on the head also appeared to be retarded as compared with the mask of birds reared in large groups. The development of the mask occurs in this gull species every spring and is regulated by testosterone (Groothuis, 1992). Furthermore, the stage of development of the mask in spring in birds housed in large groups is positively correlated with blood levels of testosterone, as measured by radioimmunoassays (R = .78, n = 6, p