Advances in the Study of Behavior, Volume 23 (Advances in the Study of Behavior)

Advances in

THE STUDY OF BEHAVIOR VOLUME 23

This Page Intentionally Left Blank

Advances in THE STUDY OF BEHAVIOR Edited by

PETERJ. B. SLATER School of Biological and Medical Sciences University of St. Andre1t.s F f e , Scotland

JAY S. ROSENBLATT Institute of Animal Behavior Rirtgers University Nekwrk, New Jersey

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

MANFREDMILINSKI Zoologisches Institirt Universitat Bern Hinterkappelen Switzerland

VOLUME 23

ACADEMIC PRESS A Division of Harcourt Brace & Company

San Diego New York Boston London Sydney Tokyo Toronto

This book is printed on acid-free paper. @

Copyright 0 1994 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. 525 B Street, Suite 1900, San Diego, California 92101-449s United Kingdom Edition published by Academic Press Limited 24-28 Oval Road. London NWl 7DX

International Standard Serial Number: 0065-3454 International Standard Book Number: 0- 12-oO4S23-0 PRINTED IN THE UNITED STATES OF AMERICA

94 95 96 91 98 99 QW 9 8 7 6 5 4 3 2 1

Contents

ContrihutorJ ............................................................................. Prefuce ...................................................................................

ix xi

Sneakers, Satellites, and Helpers: Parasitic and Cooperative Behavior in Fish Reproduction MICHAEL TABORSKY

I. Introduction ............................................................ 11. Reproductive Competition ......................................... 111. Associations between Reproductive Competitors

and Cooperation ...................................................... IV. General Chapter Discussion ....................................... V. Summary ............................................................... References ..............................................................

1 1

47 71 79 81

Behavioral Ecology and Levels of Selection: Dissolving the Group Selection Controversy LEE ALAN DUGATKIN AND HUDSON KERN REEVE

I. Introduction ............................................................ 11. Historical Perspective ............................................... 111. Some Issues Surrounding the Controversy .................... IV. Alternative Explanations for the Evolution of Behavior: Analogies and Examples ............................................ V. Conclusion .............................................................. VI. Summary ................................................................ References ..............................................................

102 102 107 119 129 130 130

Genetic Correlations and the Control of Behavior, Exemplified by Aggressiveness in Sticklebacks T H E 0 C. M. BAKKER I. Genetic Correlations as a Tool in Studying the Control of Behavior .................................................. V

I35

vi

CONTENTS

I1. I11. IV . V.

Why Study Stickleback Aggression? ............................ Life Cycle and Aggressive Behavior of Sticklebacks ....... Choice of the Breeding Design .................................... Pros and Cons of Estimating Correlations from Selection Designs ..................................................... VI . Genetic Correlations and the Causation of Aggressive Behavior: Double Selection Experiments ...................... VII . Concluding Remarks ................................................. VIII . Summary ................................................................ References ..............................................................

138 139 i40 141 147 164 165 166

Territorial Behavior: Testing the Assumptions JUDY STAMPS I . Introduction ............................................................ I1. Territory Function. Habitat Selection. and Assessment ... I11. The Function of Territorial Behavior ............................ IV . Future Directions ..................................................... V . Summary ................................................................ References ..............................................................

173 174 204 217 222 223

Communication Behavior and Sensory Mechanisms in Weakly Electric Fishes BERND KRAMER Introduction ............................................................ Weakly Electric Fishes .............................................. Pulse and Wave Fishes .............................................. The Interdischarge Interval Code in the Mormyridae ....... Electrical Signaling in the Courtship and Spawning of a Mormyrid Fish ........................................................ VI . Individual Discrimination in a Mormyrid Fish ................ VII . Constancy of the Mormyrid EOD Waveform in a Variable Environment by Impedance Matching .............. VIII . Electrical Signaling in Gymnotiform Pulse Species .......... IX . Electrical Signaling in Gymnotiform Wave Species ......... X . The “So What?” Question ......................................... I. I1. I11. IV . V.

233 233 234 236 238 242 249 252 254 262

CONTENTS

vii

XI . Summary ................................................................ References ..............................................................

263 264

Index ..................................................................... Contents of Previous Volumes ....................................

271 281

This Page Intentionally Left Blank

Contributors

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

T H E 0 C. M. BAKKER ( 1 3 9 , Abteilung Verhaltensokologie, Zoologisches Institut, Universitat Bern, CH-3032 Hinterkappelen, Switzerland LEE ALAN DUGATKIN (101), Center for Evolutionary Ecology, T. H. Morgan School of Biological Science, University of Kentucky, Lexington, Kentucky 40506

BERND KRAMER (233). Zoofogisches Institut der Uniuersitat, 0-93040 Regensburg, Germany HUDSON KERN REEVE (lo]), Museum of Comparative Harvurd Uniuersity, Cambridge, Massachusetts 02138

Zoology,

JUDY STAMPS (173), Section of Evolution and Ecology, University of California at Davis, Dnuis, California 95616 MICHAEL TABORSKY ( 11, Konrad Lorenz-Institut fur Vergleichende, Verhaltensforschung, A-1160 Vienna, Austria

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 who are engaged in the study of animal behavior by presenting their theoretical ideas and research to their colleagues and to those in neighboring fields. We 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 our intention not to focus narrowly on one or a few of these fields but to publish articles covering the best behavioral work from a broad spectrum. The skills and concepts of scientists in such diverse fields necessarily differ, making the task of developing cooperation and communication among them a difficult one. But it is one that is of great importance and one to which the editors and publisher of Advances in the Strtdy 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 dealt with in the present volume. While all the chapters are written by researchers whose work concentrates on lower vertebrates, they demonstrate the power of such studies to shed light on diverse matters of fundamental importance to the study of behavior: the functional significance of different breeding strategies, the level at which natural selection acts, methods of teasing apart the genetic control of behavior, the assumptions underlying models of territoriality, and finally, signaling systems and the sensory mechanisms on which they depend. xi

This Page Intentionally Left Blank

ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 23

Sneakers, Satellites, and Helpers: Parasitic and Cooperative Behavior in Fish Reproduction MICHAELTABORSKY KONRAD LORENZ-INSTITUT FUR VERGLEICHENDE VERHALTENSFORSCHUNG A- I I 60 VIENNA, AUSTRIA

I. INTRODUCTION Organisms compete for various resources in the course of sexual reproduction. First, there is intrasexual competition for obtaining mates. Then, there is the need to exclude reproductive competitors who might displace or affect the individual’s own gametes (e.g., sperm competition, egg dumping). There is competition for sites that are used to raise progeny and/or that will optimally support them, and it may be highly advantageous to monopolize the resources that are essential for offspring survival and development. In this chapter I review our current knowledge of the ways in which fish compete at these different levels. I further describe how competition for resources may lead to cooperative behavior, even between the competitors themselves. Finally, I draw attention to the model character of fish social systems and suggest crucial directions for future research.

11. REPRODUCTIVE COMPETITION

I begin with a description of different levels of reproductive competition among males. Group spawning appears to be a mating pattern with little competition between males, but this impression may result from our ignorance of the subtleties involved in this mating pattern and in its reproductive consequences. The competitive character of male behavior that serves the purpose of gaining access to females is much more obvious when it is coupled with some sort of resource or mate monopolization. The attempt to monopolize resources or females to obtain fertilizations (i.e., the “bourgeois” tactic) may not always be the best choice for a male. He may be 1

Copyright 0 1994 by Academic Press, Inc. All rights of reproduction in any form reserved.

2

MICHAEL TABORSKY

better off choosing alternative tactics when he is not in the position to compete successfully, for example, because of a weak resource holding potential (sensu Parker, 1974), or when there are “cheaper” ways to obtain fertilizations, that is, when the costibenefit ratio of the alternative tactic will fall below that of the bourgeois tactic (see Rubenstein. 1980: Dunbar, 1982; Arak, 1984; Magurran, 1986; for general discussions). Basically, the monopolization of mates may be overcome in two alternative ways. Males may either obtain partners from others who have already invested in their acquisition, or they may directly try to obtain parasitic fertilizations, that is, release sperm when a female spawns with another male. The first route is chosen by males taking over a nest, mating site, or breeding hole from its owner who already invested in behaviors like nest building, preparation of a spawning surface, site advertisement, or defense (nest takeover). A specific version of this tactic is to take charge of the nest for only a limited spawning period and then leave the broodcare to the previous nest owner (piracy). Males may also try to steal females within other males’ territories or intercept females who are on their way to a spawning site that is monopolized by another male (female thefr and interception). The second alternative route to obtain fertilizations differs from the first one in that parasitic males do not attempt to get exclusive access to a female, that is, monopolize her for some period of time. but rather shed sperm while a bourgeois competitor spawns (sperm competition). This “simultaneous parasitic fertilization” tactic is very widespread in fish and I will give an overview of its taxonomic distribution (see Table I). I summarize the information on the different types of males participating in kleptogamic fertilizations, from bourgeois territory neighbors to males behaviorally and morphologically specialized for this type of mating (e.g., female mimics). In live-bearers, fertilization stealing often involves coercive copulations that may also be at the expense of males investing in courtship to attract females. I compare bourgeois and parasitic males with regard to their relative abundances, costs (e.g., behavioral, morphological, and gonadal effort), reproductive success, and origin, that is, to what extent their tactics are genetically or phenotypically determined. I then discuss how females behave toward bourgeois and parasitic males, and review female reproductive competition and parasitic behavior (e.g., egg dumping). A.

GROUPSPAWNING

Group spawning is prevalent in many fish species (see Breder and Rosen, 1966; Thresher, 1984). In many surgeonfish (Acanthuridae), for example,

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

3

thousands of individuals aggregate for spawning (e.g., Robertson et al., 1979; Robertson, 1983; Colin and Clavijo, 1988). Often, there is both pair and group spawning among members of the same species (e.g., in Salmonidae, Cameron, 1940; Mullidae, Colin and Clavijo, 1978, 1988; Sparidae, Smith, 1986; Labridae, Randall and Randall, 1963; Reinboth, 1973; Moyer and Shepard, 1975; Warner et al., 1975; Meyer, 1977; Warner and Robertson, 1978; Pottle et al., 1981; Wernerus and Tessari, 1991; Scaridae, Randall and Randall, 1963; Choat and Robertson, 1975; Robertson and Warner, 1978). In spite of the fact that male competition for fertilizations or sperm competition is intense in these “explosive breeding assemblages” (sensu Emlen and Oring, 1977), no specific behavioral mechanisms have been reported that may give individuals performing them higher fertilization probabilities. However, even where it looks as if gametes are released by many fish simultaneously and without regular pattern, there may well be organized pair spawning (e.g., Brawn, 1961). This leaves ample scope for intrasexual male parasitism (see the following), but there are as yet no data on this in group spawners.

B. COMPETITION FOR ACCESS TO FEMALES Pair spawning will be discussed in the following sections. It involves at least a short-term monopolization of a mate. The effort of males may be in defense of a place, shelter, or nest, in modification of the substrate (e.g., by digging, cleaning, nest building), in courtship, and in broodcare. There are various ways in which the exclusion shown to them can be overcome by competitors to parasitize this effort and/or the success of territorial males. 1 . Temporary and Permanent Nest Takeover for Spawning

Males may save effort by temporarily taking over nests, holes, or other structures that have been obtained and/or prepared by territorial males for the purpose of spawning and/or rearing offspring. In the cyprinid Margariscus margarita “adolescent,” nonterritorial males may spawn within the territories of adult male conspecifics (Langlois, 1929). Similar observations were made in desert pupfish (Kodric-Brown, 1977) and in the wrasses Bodianus diplotaenia, Halichoeres maculipinna (Robertson and Hoffman, 1977), and Symphodus ocellatus; in the latter species there was spawning by both “satellites” and “sneakers” (Taborsky et al., 1987). In arctic graylings (Thymallus arcticus: Salmonidae), subdominant males spawn within territories when their owners are distracted by other activities (Beauchamp, 1990).

4

MICHAEL TABORSKY

Nest takeovers have been described in creek chub (Semotilus atromacularus: Cyprinidae), in which nonnesting males try to occupy nests of other males temporarily in order to attract females to clasp them for spawning. Males stay and “watch” at the margin of nests and take over when the nesting male is engaged in agonistic interactions away from the nest (Ross, 1977). Toward the end of the season, when few females mate, many of the nests are taken over by males that are smaller than the previous owners. Brightly colored, territorial Pseudocrenilabrus philander males (Cichlidae) take over nests of opportunistically courting, semiterritorial males by expelling them from their spawning pits, and spawn with the females that had been attracted by these semiterritorial males (Chan, 1987). In bluegill sunfish (Lepomis macrochirus: Centrarchidae), larger males often displace smaller nest owners after vigorous, often prolonged fighting (Dominey, 1981). Longer-lasting or even permanent nest takeovers also occur in the Mediterranean wrasse S . ocellarus (Labridae). Territorial males of this species build complex nests of algae. Fiedler (1964) observed nest takeovers in this species and in S. mediterraneus. In a population of the former species off Corsica, more than a quarter of the nests studied were taken over by males in nuptial coloration that had built their own nests before (Taborsky et al., 1987). Two-thirds of these takeover males only fed on the eggs contained in the acquired nests, but the other third courted there and most of these also spawned successfully. When compared to building a nest by oneself, a nest takeover reduced the interval between the completion of one nest and the first spawning in the next by 3.7 days on average, which is more than a third of the average length of a whole nest cycle. Usually, takeover males had been immediate neighbors of the individuals that were ousted. In 3 out of 24 cases the previous owner regained his nest at a later stage (Taborsky et al., 1987). tn the river bullhead (Corm gobio: Cottidae; Bisazza and Marconato, 1988) and in the freshwater goby Padogobius martensi (Gobiidae; Bisazza, et al., 1989a) and Pomatoschistus minutus (Magnhagen and Kvarnemo, 1983), large males displace smaller spawning or guarding males to spawn themselves in the acquired nest sites. Hastings (1988) demonstrated experimentally the importance of relative male size in the competition for already occupied spawning shelters in angel blennies (Corafliocetus angelica: Chaenopsidae). Large male greenbreast darters (Etheostomajordani: Percidae) may displace smaller males which guard a female on the spawning ground by lying on top of her (Orr and Ramsey, 1990). At least in the river bullhead such displacements may be a beneficial tactic, as females prefer to spawn with males that already guard eggs (Marconato and Bisazza, 1986; Bisazza and Marconato, 1988; see Section III,C,3).

PARASITIC A N D COOPERATIVE BEHAVIOR I N FISH REPRODUCTION

5

2. Pirucy

Nonnesting males of the Mediterranean wrasse Symphodus tinca (Labridae) may spawn within the nests of territory owners when the latter have “spawning breaks” (i.e., rest between series of spawnings), which occupy a large proportion of the spawning period (Lejeune, 1985). Occasionally, very large males may take over a nest from an owner and spawn there for up to two and a half days, much as in the cases described earlier for the closely related S. ocellatus and the bluegill sunfish. In S. tinca, however, the original nest owners remain at these nests and continue to guard them after the “pirates” have left (van den Berghe, 1988). Pirates seem to have less success than nest owners. Combining the information given by van den Berghe (1988) with his unpublished data (personal communication), pirates seem to average only one-tenth of the spawnings of nesting males. Contrary to this, van den Berghe (1988) believed that they obtained similar spawning rates to nest owners, but this was based on an erroneous comparison of spawning rates measured over different time periods, and without allowing for the fact that in 67% of the observed cases of piracy the pirated nests were abandoned by their owners before the pirates’ eggs could hatch. Despite this, at least some of the nest owners’ own eggs could have hatched in these cases because they had been laid at an earlier stage in the nesting cycle. Why then do the largest males in a population adopt a greatly inferior spawning tactic? First, pirates may build their own nests at a different stage of the breeding season (van den Berghe, 1988). Second, in comparing tactics we must examine the possible alternatives for an individual at any given time. Pirates may compensate for their low spawning rate by saving the time and risk associated with guarding and nest building, as well as by feeding in the takeover nest while in charge of it (see van den Berghe, 1988). In 2 out of 88 observed nests of territorial S. ocellatus, an expelled nest

owner regained his former nest at a later stage to continue broodcare and guarding (Taborsky et al., 1987). Probably, this takeover reversal resulted from an aggressive expulsion of the intruder (i.e., not from his spontaneous abandonment of the nest) and hence this temporary, parasitic nest occupancy should not be viewed as a behavioral “tactic” (i.e., “piracy”). In the tesselated darter (Etheostoma olmstedi: Percidae), Constantz (1985) observed that “fathers” may cruise and search for ripe females and for other nests once they have spawned in their own shelter. They may, “upon encountering consort pairs, attempt to displace courting males” (p. 176). It is not stated, however, if a displaced previous owner will ever regain and guard his shelter afterwards.

6

MICHAEL TABORSKY

3. “Interception” and “Female Theft” In the American angelfish Holacanthuspasser (Pomacanthidae), smaller males occasionally interfere with courting males and may spawn with a female in the territories of the latter. This happens primarily when two to three females visit the territory of a large male simultaneously. This female theft is only very rarely successful (Moyer et al., 1983). It has also been observed in wrasses (Labridae: Thaiassoma bifasciatum, Reinboth, 1973; Clepticusparrae, Warner and Robertson, 1978; Cirrhilabrus temminckii, Bell, 1983). Peripheral males interrupted spawning harem owners in the hawkfish Cirrhitichthys falco (Cirrhitidae) and spawned occasionally with harem females (Donaldson, 1987). Courting males of the cyprinid Zacco temmincki may be attacked at or in a spawning redd by a male competitor. This leads most often to the courting males’ loss of the females they were going to spawn with (Katano, 1990). Similarly, females of the pupfish Cyprinidon uariegatus (Raney et al., 1953), C . macularis (Barlow, I961), and C .pecosensis (Kodric-Brown, 1977) and of the Mediterranean wrasse Symphodus tinca (van den Berghe, 1988) may be intercepted when they are ready to spawn at or around territories. They may subsequently follow the intercepting males and spawn with them outside a territory. In S. tinca this interception by nonnesting males yields apparently very little success, however, as the untended eggs produced by this spawning mode have minute chances of survival (Lejeune, 1985; Wernerus, 1989). Interception of females on their way to a territory has also been observed by groups of “initial phase males” (i.e., small males that do not have the specific color pattern of territory owners) of the tropical wrasse Thalassoma bifasciatum (Warner et al., 1975; Warner and Robertson, 1978) and the parrotfish Sparisoma radians (Robertson and Warner, 1978). This may lead to group spawning. Small parasitic or large neighboring territorial fish were observed to intercept females that are ready to spawn in the wrasse Symphodus melanocercus (Lejeune, 1985), in parrotfishes (Scaridae, Robertson and Warner, 1978), in Chaetodon capistratus (Chaetodontidae, Neudecker and Lobel, 1982), and in tesselated darters (E. olmstedi, Percidae, Constantz, 1985). In an experimental situation, large Padogobius martensi (Gobiidae) nest-males courted females that were spawning in the nests of smaller males and sometimes got the females to follow them into their own nests, where they continued to spawn with the interlopers (Bisazza et al., 1989a). In the field, two P . martensi males “in aggressive livery” were occasionally found together in a nest with a spawning female or freshly spawned eggs (Marconato et al., 1989). Sexually mature “bachelor” males of Canthigaster uaientini (Tetrao-

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

7

dontidae), a harem species, live either alone in home ranges on the periphery of social groups or as non-site-attached wanderers. They have no access to females of the harem to which they belong, but they may spawn in another male’s territory when its owner courts another female at the far end of the territory. Less than 3% of spawnings observed by Gladstone (1987b) involved such bachelor males. When territorial males were removed, bachelor males took over their territories (Gladstone, 1987a). Within the territory of a male, nonterritorial male honey gouramis (Colisa chuna) may clasp a female after she has spawned with the territory owner, and quiver, apparently releasing sperm. It is uncertain if the females release eggs on these occasions, but the sperm of these intruders may at least fertilize eggs that had been spawned before (Janzow, 1982).

C. SPERMCOMPETITION So far I have discussed the competition of males for the opportunity to spawn. This involves the parasitism of the effort of other males by obtaining access to females that had been attracted to them or to structures provided by them. I now turn to a type of competition that involves the participation of more than one male in a spawning. I focus on cases that are asymmetric with regard to effort, that is, cases in which the reproductive effort of one male is exploited by others. In such cases of simultaneous spawning of a female with more than one male, sperm competition adds to the costs borne by the parasitized male that result from the surreptitious use of his reproductive effort (e.g., courtship, defense, broodcare; see the previous section). Figure 1 shows an example of a species with both types of male reproductive parasitism, resulting from competition for access to females and from sperm competition. I . Fertilization Stealing by Territorial Neighbors In several fish species, males may leave their territories temporarily and try to steal fertilizations when neighboring males spawn. This was observed in various sticklebacks (Gasterosteidae; three-spined sticklebacks, Gasterosteus aculeatus, van den Assem, 1967; Li and Owings, 1978a,b; Sargent and Gebler, 1980; four-spined sticklebacks, Apelres quadrueus, Rowland, 1979; Wootton, 1984, p. 142, mentions three more stickleback species). The cuckolding males change from their bright color pattern, which reveals their sex and territorial status, to a drab, femalelike coloration before they sneak into the territory of a neighbor. There they may either prevent females from entering the nest to spawn by lying across it or in its entrance, or they may follow the female through the nest and fertilize the freshly laid eggs before the resident male can do so. In these

Expenditure and RI8k

Sarnl-T Yale

increarad Energy Expenditure and Rlok

Low Energy Expenditure and Decreased Rlak

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

9

cases, the nest owner might only obtain such fertilizations as are achieved by the sperm he had released earlier, for example, in previous spawnings. Nest intrusions and fertilization stealing by neighbors also occur in suckers (Catostomidae; Moxostoma duquesnei, Bowman, 1970), sunfish (Centrarchidae; Lepomis macrochirus, Avila, 1976; Gross, 1982; L. megalotis, Keenleyside, 1972; Bietz, 1980,1981;Dupuis and Keenleyside, 1988; Jennings and Philipp, 1992), cichlids (Cichlidae; Sarotherodon grahami, Albrecht, 1968; ffaplochromis(Astatotilapia)burtoni, Fernald and Hirata, 1977; Pseudocrenilabrus philander, Chan, 1987), damselfish (Pomacentridae; Abudefduf saxatilis, Chromis multilineata, Albrecht, 1969; Chromis cyanea, De Boer, 1981), parrot fish (Scaridae; Sparisoma radians, Robertson and Warner, 1978), three species of surgeonfish (Acanthuridae; Ctenochaetus striatus, Zebrasoma scopas, and Z . veliferum; Robertson, 1983), and in Tripterygion tripteronotus (Tripterygidae; Wirtz, 1978). Jennings and Philipp (1992) showed that cuckoldry by neighbors in longear sunfish reduces the reproductive success of colonial males to a level below that of solitary males. Small and less attractive males even seem to specialize in stealing fertilizations in neighbor’s nests. 2. Fertilization Stealing in Simultaneous Hermaphrodites A parasitic, simultaneous release of sperm is also widespread in simultaneous hermaphrodites (e.g., Fischer, 1986). In Serranus fasciatus, hermaphroditic members of a large male’s harem may try to steal fertilizations when this male is spawning with another harem member, despite the fact that these individuals usually take the female role when spawning with the owner of the harem (Petersen, 1987). In S. Tortugarum apart from the behavioral adaptations of this intraharem reproductive parasitism in this bass, this is probably the reason why a large proportion of the gonad mass of hermaphrodites is assigned to the production of sperm (ca. 25%; Fischer, 1986). 3. Alternative Mating Tactics of Different Types of Mules Commonly, competitively inferior male fish parasitize territorial, often brightly colored or morphologically distinct male conspecifics. Various FIG. 1. Schematic representation of male reproductive options in the African cichlid Pseudocrenilabrus philander. There are three reproductive tactics in this species and the frequency of these options depends on male size and competitive pressure. Individual males may switch between tactics. The costs and benefits as indicated in this graph only illustrate the order of magnitude and should not be interpreted literally, because of problems with quantitative measurements (e.g., all eggs spawned when parasitic intrusions occurred were attributed to the success of sneakers). Reproduced from Chan (1987).

10

MICHAEL TABORSKY

terms have been used to describe this behavior, and these are often descriptive expressions of how these males try to participate in spawning (e.g., “sneaking” for stickleback males secretively approaching a nest to fertilize eggs that have just been spawned there, van den Assem, 1967; or “streaking” for wrasse males that rush into a male’s territory to join its owner and his mate just as they are shedding gametes into the water, Warner et al., 1975).The terms used for the males performing such parasitic behavior are even more diverse. They have been called “sneakers” (e.g.. van den Assem, 1967;Taborsky et al., 1987; Hutchings and Myers, 1988), “sneaky males” (Rowland, 1979), “streakers” (e.g., Warner et al., 1975; Maekawa and Onozato, 1986), “scroungers” (Barnard, 1984), “cuckolders” (e.g., Gross, 1984), “machos furtivos” (furtive males; Santos, 1985). “outsider der Befruchtung” (outsiders of fertilization; Soljan, 1930b. 1931),“pseudofemales” (e.g., Morris, 1952), “transvestite males” (e.g.. Dipper, 1981). “stunted males” (e.g., Shute, et af.,1982). “hiders” (Hutchings and Myers, 1988), “accessory males” (e.g., Winn, 1958a; McCart, 1%9: Hillden, 1981), “Beimannchen” (by-males; Fiedler, 1964), “supernumerary males” (e.g., Ribbink, 1975), “small outlier males” (Keenleyside and Dupuis, 1988), “interference spawning males” (Colin and Bell, 1991), “Type I1 males” (e.g., Bass, 1992),or “satellites” (e.g., Dipper, 1981: Lejeune, 1985; Katano, 1992). I focus my discussion on the functional aspects of this phenomenon. The most important distinction between reproductive tactics in this respect is on the basis of effort. As with any parasitic relationship there are individuals investing in some structure, either morphological, physiological, or behavioral, and others exploiting this investment to obtains access to a limited resource (e.g., Barnard, 1984). I use the term “bourgeois” for a male of the former (i.e., investing)type, in line with the nomenclature of the game theoretic treatment of this problem (e.g., Maynard Smith, 1982).A bourgeois individual behaves in a certain way as the owner of a resource (e.g.. a female that is ready to spawn), but it may also behave very differently to usurp such a resource if it is “owned” by another individual (e.g., another male that has successfully put effort into its procurement). I generally call the alternative tactic “parasitic.” Parasitic spawning is defined as “simultaneous” when the parasite tries to steal fertilizations by participating in the spawning of a pair. Other functional and synonymous terms for male reproductive parasitism that I may use are “kleptogamy” (Barnard, 1984) or “kleptogyny” (Turner, 1986a). I have found published accounts of simultaneous parasitic spawning (SPS) for 123 species belonging to 24 different fish families, ranging from salmon to midshipman. These are listed in Table I. This list, though fairly comprehensive, is certainly not complete. There is little literature

PARASITIC AND COOPERATIVE BEHAVIOR I N FISH REPRODUCTION

II

TABLE I MALESIMULTANEOUS REPRODUCTIVE PARASITISM Species"

Salmonidae Salmo salar

S . henshawi S . trutta Salmo gairdneri Oncorhynchus nerka 0 . keta 0 . kisutch 0 . gorbuscha Salvelinus fontinalis S . alpinus S . malma miyabei Thymallus arcticus Cyprinidae Notropis cornutus Semotilus corporalis Zacco temmincki Rhodeus amarus Catostomidae Catostomus commersonii Hypentelium nigricans Moxostoma erythrurum M . duquesnei M. valenciennesi Mochokidae Synodontis Multipunctatus Gasterosteidae Pungitius pungitius Gasterosteus aculeatus

G . inconstans G . wheatlandi Apeltes quadracus

References

Orton et af. (1938); Jones and King (1950b, 1952a,b); Jones (1959); Myers, Hutchings (1987); Hutchings and Myers (1988); Jordan and Youngson (1992) Smith (1941) Jones and Ball (1954) Hartman (1969) Hanson and Smith (1967); McCart (1969); Chebanov et al. (1983); Foote and Larkin (1988); Foote, 1990 Schroder and Duker (1979); Schroder (1981, 1982) Gross (1985) Wicket (1959); Heard (1972); Chebanov (1980)b; Keenleyside and Dupuis (1988); Noltie (1989) Smith (1941) Jonsson and Hindar (1982); Sigurjonsdottir and Gunnarsson (1989) Maekawa (1983); Maekawa and Hino (1986, 1990); Maekawa and Onozato (1986) Kratt and Smith (1980) Reighard (1943y Ross and Reed (1978); Ross (1983) Katano (1983, 1990, 1992) Heschl (1989) Reighard (1920) Reighard (1920) Reighard (1920)b; Kwak and Skelly, (1992) Bowman (1970) Jenkins and Jenkins ( 1980)b Schrader (1993) Morris (1952) Morris (1952); van den Assem (1967); Li and Owings (1978a); Sargent and Gebler (198od; Sargent (1982); Wootton (1984); Goldschmidt and Bakker (1990); Goldschmidt et al. (1992); Rico et al. (1992) Wootton (1984) Wootton (1984) Rowland (1979) (continued)

12

MICHAEL TABORSKY

TABLE I (Continued) Species"

References

Macrorhamphosidae Macrorhamphosus scolopax

Oliveira et al. (1993)

Cyprhdontidae Cyprinodon variegutus C. rnacularius C'. pecosensis C . nevadensis Aphanius fasciatus

Raney et al. (1953) Barlow (1961); Matsui, unpublished, in KodricBrown (1981) Kodric-Brown (1977, 1981, 1986) Soltz (1974) Marconato (1982)

Poeciliida& Poeciliopsis occidentaiis Poecilia sphenops P . reticula ta

P. lalipinna Xiphophorus nigrensis Gamhusia afinis G. holbrooki

Constantz (1975) Parzefall (1979) Baerends et a / . (1955); Liley (1966); Farr (1980a,b); Endler (1983, 1987);Luyten and Liley (1985); Farr et al., (1986); Kodric-Brown (1992); Reynolds et al. (1993) Woodhead and Armstrong (1985); Travis and Woodward (1989) Zimmerer (1982); Zimmerer and Kallrnann (1989); Ryan and Causey (1989) Hughes (1985) Bisazza et al. (1989b)

Serranidae Serranirs scribu Hypoplectrus nigricans S . tortugarum S . baldwini S. fasciatus

Reinboth (19621, P. Lejeune (personal communication)' Fischer (1980) Fischer (1984, 1986) Petersen and Fischer (1986) Petersen (1987, 1990)

Centrarchidae Lepomis gibbosus L . microlophus L. macrochirus

L . megalotis

Miller (1963); Gross (1979) Gerald ( 1970)' Gerald (1970)'; Gross (1979, 1982); Gross and Charnov (1980); Dominey (1980, 1981) L. m.peltastes: Keenleyside (1972); Bietz (1980); Dupuis and Keenleyside (1988); L. m. megalotis; Jennings and Philipp (1992a,b)

Percidae Etheostoma caeruleum E. spectabile E . nigrurn E . exile E . olmstedi

Reeves (1907); Winn (1958a) Winn (1958b) Winn (1958a) Winn (1958a) Constantz ( I 979) (continued)

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

13

TABLE I (Continued) Species* E. perlongum E. jordani Hadropterus maculatus Percina caprodes Sparidae Chrysophrys auratus Cheimerius nufar Chaetodontidae Chaetodon nippon C. multicinctus Cichlidae Sarotherodon niloticus S. alcalicus S . grahami Pseudocrenilabrus philander P. multicolor Acarichtys heckelii Haplochromis burtoni Cyrtocara eucinostomus Lamprologus brichardif L. callipterus Oreochromis mossambicus Pseudosimochromis curvifrons Simochromis diagramma Nannacara sp. Pelvicachromis pulcher Polycentridae Poly centrus schomburgkii Pomacentridae Abudefduf saxatilis Chromis multilineata C. cyanea C. atrilobata C. dispilus Labridae Symphodus ocellatus

References Shute et al. (1982)b Orr and Ramsey (1990) Petrovicz (1938) Winn (1958a) Smith (1986) Garratt (1991)d Susuki et al. (1980) Lobe1 (1989) Heinrich (1967) Albrecht (1968) Albrecht (1968)d

Ribbink(1975);Chan(1987); ChanandRibbink(l990) W. Mrowka (personal communication)c Cichocki (1976)e Fernald and Hirata (1977)d McKaye (1983) Taborsky (1984a, 1985a) Sato (1988, 1991) Baerends and Baerands van Roon (1950); Turner (1986b) Kuwamura (1987) T. Sat0 (unpublished), in Kuwamura (1987) Romer (1993) Martin and Taborsky (1993) Barlow (1967) Albrecht (1969) Albrecht (1969) De Boer (1981) P. Wirtz (personal communication) M. J. Kingsford (personal communication)c Soljan (1930a,b); Fiedler (1964); Taborsky (1984b, 1985b); Lejeune (1985); Warner and Lejeune (1985); Michel et al. (1987);Taborsky et al. (1987); Wernerus e f al. (1987); van den Berghe et al. (1989); Wernerus (1989) (continued)

14

MICHAEL TABORSKY

TABLE I (Continued) Species” S. roissali S. tinca S . mediterraneus S. melops S. cinereus

S. melanocercus S. rostratus

Thalassoma h a r e T. bifasciatum T. cupid0 T . lucasanum T. pavo T. quinqueuittatum Tautoga onitis Halichoeres bivittatus H . maculipinna Tautogolaburs adspersus Ctenolabrus rupestris Pseudolabrus celidotus Coris julis Centrolabrus exoletus Anampses twistii Gomphosus varius

References Soljan (1931); Fiedler (1964); Helas et al. (1982a); Lejeune (1985); Warner and Lejeune (1985); Michel e f al. (1987) Fiedler (1964); Helas et al. (1982b); Lejeune (1985); Warner and Lejeune (1985); Michel et al. (1987); van den Berghe et al. (1989); Wernerus (1989) Fiedler (1964); Lejeune (1985) Dipper and Pullin (1979)b;Dipper (1981)b Michel and Voss (1982); Lejeune (1985); Michel et al. (1987) Lejeune (1985); Warner and Lejeune (1985); Wernerus et al. (1987); Wernerus (1989) Lejeune (1985); Michel et al. (1987) Robertson and Choat (1974) Warner et al. (1975); Warner and Robertson (1978); Warner and Hoffman (1980a,b) Meyer (1977) Warner and Hoffman (1980a); Warner (1982) Michel et a/. (1987); Wernerus (1989) Colin and Bell (1991) Olla et al. (1977) Warner and Robertson (1978) Warner and Robertson (1978); Thresher (1979) Pottle and Green (1979a,b); Pottle et al. (1981) Hillden (1981, 1984a,b) Jones (1981) Lejeune (1982, 1985, 1987); Michel et al. (1987) Michel et at’. (1987) Colin and Bell (1991) Colin and Bell (1991)

Searidae

Scarus croicensis S. velula S. globius S . psittacus S . schlegeli S. sordidus

Sparisoma radians Calotomus spinidens Leptoscarus vaigiensis Acaanthuridae Ctenochaetus striatus Zebrasoma scopas Z. ueliferum

Warner and Downs (1977); Robertson and Warner ( 1978) Clavijo ( 1983) Colin and Bell (1991) Colin and Bell (1991) Colin and Bell (1991) Colin and Bell (1991) Robertson and Warner (1978) Robertson et al. (1982) Robertson et al. ( 1982)b Robertson (1983) Robertson ( 1983)d Robertson ( 1983)d (continued)

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

15

TABLE I (Continued) Species“ Gobiidae Coryphopterus nicholsi Pomatoschistus spp. P . microps Blenniidae Parablennius sanguinolentus Tripterygiidae Tripterygion tripteronotus T. delaisi T. melanurus Axoclinus carminalis Belontiidae Colisa chuna Ostraciidae Lactoria diaphana Batrachoididae Porichthys notatus

References Cole ( 1982)b Miller (1984)b Magnhagen (1992) Santos (1985); Santos and Almada (1988) Wirtz (1978); Mohr (1986); de Jonge and Videler ( 1989) Wirtz (1978); De Jonge and Videler (1989) Mohr (1986) Thresher (1984) Janzow (1982)

Moyer (1984) Brantley and Bass (1991); Bass (1992)

Species for which simultaneous parasitic spawning (i.e., “sneaking,” “streaking,” “kleptogamy”. . ., see text) has been documented. This table was compiled in collaboration with P. Wirtz, whose generous permission to use his files added nearly 15% of species included in this list. Reproductive parasitism not directly observed, but very likely. Unpublished information, communicated through Peter Wirtz (Univ. Madeira, P-9OOO Funchal). Only territorial (i.e., bourgeois) males were observed to parasitize fertilizations. Cited in Gross (1984). f I do not follow the taxonomic nomenclature suggested by Colombe and Allgayer (1985) for Tanganyika cichlids. In livebearers, reproductive parasitism is not simultaneous.

specifically dealing with parasitic spawning: most accounts were obtained from papers dealing with quite different aspects of fish biology. Without doubt, many examples have escaped my attention. The list should suffice, however, to demonstrate that kleptogamy is an extremely widespread phenomenon; it might even be viewed as “the rule rather than the exception.” This compilation of existing evidence may hopefully encourage observers of this phenomenon to publish their evidence so that a future update of this part of the review could be much more representative. It is obvious from Table I that some fish families are represented by a great number of species (e.g., wrasses: 21 species; cichlids: 14 species;

16

MICHAEL TABORSKY

salmonids: 12 species), whereas others are either absent or only sparsely represented. The most important reason for this pattern is simply a difference in our knowledge about reproductive behavior of these different taxonomic groups. However, this is certainly not the only reason for variation between families. Damselfish, for example, are a well-studied group, but I found accounts of parasitic spawning in only five species of this group. It is nevertheless too early to draw conclusions from the taxonomic distribution of parasitic spawning shown in Table I. Our knowledge of reproductive behavior is too sporadic at present, especially with regard to parasitic spawning, which is a behavior that has often evolved to be extremely quick and cryptic. I proceed in this chapter by (i) introducing the phenomenon of simultaneous parasitic spawning with some examples from the most-studied fish family in this respect, the Salmonidae; (ii) demonstrating specific adaptations that are linked with alternative mating tactics; (iii) comparing bourgeois and parasitic mating strategies: (iv) discussing the success of parasitic mating practices; (v) reviewing the present knowledge on the life histories of parasitic males; and (vi) emphasizing the role of females. Table 11 contains a list of examples on which my discussion of alternative mating tactics is based, in abbreviated and comprehensive form. 4 . Alternative Mating Tactics in Salmon and Char: Some Case Studies

Kleptogamy is best understood in salmonids, partly because they have been intensively studied owing to their commercial importance (see Jones, 1959; Keenleyside, 1979; Hutchings and Myers, 1988). Reproductive competition in the genera Salmo, Oncorhynchus, and Salvelinus may lead to group spawning. with several males spawning with a single female and a dominance hierarchy that is strongly size dependent. Alternatively, large males defend the nesting territories of females and smaller surplus males dart in to steal fertilizations when the pair is spawning (e.g., Jones, 1959; Noltie, 1989; Sigurjonsdottir and Gunnarsson, 1989). In anadromous populations, these smaller males may either be anadromous as well and within the age range of dominant males, but in poorer condition (Noltie, 1989), or they may have spent a much shorter period in the ocean than other males (i.e., "jacks"; e.g., Hanson and Smith, 1967; Gross, 1984), or even be stream resident, much younger and smaller than the migratory territorial males (i.e., parr; e.g., Maekawa, 1983; Maekawa and Hino, 1986). In Oncorhynchus, several types of accessory males may be present in one population. The smallest, nonmigratory males may then wait in close proximity to the spawning pair for a chance to participate in fertilization (see Keenleyside, 1979; Keenleyside and Dupuis, 1988). Alternatively,

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

17

TABLE I1 SUMMARY OF SPECIFIC FEATURES CHARACTERIZING MALE PARASITIC IN THE LISTEDSPECIES REPRODUCTION Family

Female mimicry Salmonidae

Species

References

Oncorhynchus gorbuscha

Keenleyside and Dupuis (1988); Noltie (1989) Pungitius pungitius Moms (1952) Gasterosteidae van den Assem (1967) Gasterosteus aculeatus Cyprinodontidae Cyprinodon pecosensis Kodric-Brown (1986) Centrarchidae Lepomis macrochirus Dominey (1980) Percidae Etheostoma caeruleum Reeves (1907) E. olmstedi Constant2 (1979) Shute et al. (1982) E. perlonglam Cyrtocara eucinostomus McKaye (1983) Cichlidae Pseudocrenilabrus philander Chan (1987) Lamprologus callipterus Sato (1988, 1991) Polycentrus schomburgkii Barlow (1967) Polycentridae Thalassoma lunare Robertson and Choat (1974) Labridae T. bisfasciafum Warner and Robertson (1978) T. lucasanum Warner (1982) Michel et al. (1987); T. pavo Wernerus (1989) Symphodus ocellatus Soljan (1930b); Fiedler (1964); Taborsky (1984b); Lejeune (1985); Michel et al. (1987); but see Taborsky et al. (1987) Dipper and Pullin (1979); S. melops Dipper (1981)" Lejeune (1985)b S . cinereus and S . tinca S. mediterraneus, S. roissali, Lejeune (1985);Michel et al. (1987) and Coris julis Thresher (1979) Halichoeres maculipinna Pseudolabrus celidotus Jones (1981) Scaridae Several species Choat and Robertson (1975)" Scarus croicensis Robertson and Warner (1978) Clavijo (1983) S. vetula Tripterygidae Tripterygion tripteronotus Wirtz (1978) Majority of males parasitic Cyprinidae Semotilus corporafis Ross (1983) Poeciliidae Gambusia holbrooki Bisazza et al. (1989b) Centrarchidae Lepomis macrochirus Gross (1982) Thalassoma lucasanum Warner and Hoffman (1980a) Labridae (continued)

18

MICHAEL TABORSKY

TABLE I1 (Continued) Family

Species

References

Symphodus ocellatus

Warner and Lejeune (1985); Taborsky et al. (1987) S . roissali and S. tinca Warner and Lejeune (1985) Relatively little reproductive effort of bourgeois males to be parasitized upon Catostomidae Catostomus commersonii, Reighard (1920) Moxostoma duquesnei, and aureolum M . volenciennesi Jenkins and Jenkins (1980) Mochokidae Syndontis multipunctatus Schrader ( 1993) Percidae Etheostomu cueruleurn Reeves (1907): Winn (1958a,b) E. exile Winn (1958a) E. spectabile Winn (1958b) Hadropterus maculatus Petravicz (1938) Percina cuprodes Winn (1958a) Pseudocrenilabrus philander Ribbink (1975); Chan (1987) Cichlidae Pseudosimmochromis Kuwamura (1987) curuifrOns Labridae Thalassoma spp. Warner and Robertson (1978); Warner and Hoffman (1980b); Warner (1982) Parasitic males may also eat eggs Salmonidae Salvelinus malma mivabei Maekawa and Hino (1990) Gasterosteidae Pungitius pungitius Morris (1952) Ctchlidae Pseudocrenilabrus philander Ribbink (1971); but see Chan ( 1987) Cyrtocaru eucinostomus McKaye ( 1983) Lamprologits brichardi Taborsky (1984a, 1985a) Gobiidae Coryphopterus nicholsi Cole ( 1982) Testes relatively larger in parasitic than in bourgeois males Centrarchidae Lepomis mucrochirus Dominey (1980); Gross and Charnov (1980); Gross ( 1982) Lepomis megalotis Jennings and Philipp (1992a) Labridae Thalassoma lunare Robertson and Choat (1974) Warner and Robertson Halichoeres bivittatus, H . (1978) maculipinna, and H. pictits Warner and Lejeune (1985) Symphodus roissali S. ocellatus Warner and Lejeune (1985); own data (see fig. 3) Scaridae Several species Choat and Robertson (1975)’ Scarus croicensis Robertson and Warner ( 1978) (contmrwd)

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

19

TABLE I1 (Continued) Family

Species

References

Sparisoma radians

Robertson and Warner (1978) Gobiidae Coryphopterus nicholsi Cole (1982)d Blenniidae Parablennius sanguinolentus Santos and Almada (1988) Tripterygidae Tripterygion tripteronotus Finck (1985); Mohr (1986); de Jonge and Videler ( 1989) T. delaisi Finck (1985); Mohr (1986) Porichthys notatus Bass and Andersen (1991); Batrachoididae Bass (1992) Species with information on rates of male reproductive parasitism Cyprinodontidae Cyprinodon nevadansis Soltz (1974) Poeciliidae Poecilia occidentalis Constantz (1975) P. reticulata Endler (1987) Xiphophorus nigrensis Zimmerer and Kallmann (1989) Centrarchidae Lepomis macrochirus Gross (1982) Percidae Etheostoma caeruleum Reeves (1907) Cichlidae Pseudocrenilabrusphilander Chan (1987) Pelvicachromis pulcher Martin and Taborsky (1993) Labridae Thalassoma bifasciaturn Warner et al. (1975) T. pauo Wernerus (1989) Coris julis Lejeune (1985, 1987) Symphodus ocellarus, S. Lejeune (1985); Warner and tinca, and S . melanocercus Lejeune (1985); Wernerus ( 1989) S. roissali Lejeune (1985); Warner and Lejeune (1985) S. cinereus Lejeune (1985) Success dependent on proximity at spawning Salmonidae Oncorhynchus keta Schroder and Duker (1979); Schroder (1981) Cyprinidae Semotilus corporalis Ross and Reed (1978) Catostomidae Catostomus commersonii Reighard (1920) and Moxostoma aureolum Success of male reproductive parasitism proved Salmonidae Oncorhynchus keta Schroder and Duker (1979); Schroder (1981, 1982) 0. nerka Chebanov et al. (1983) Salvelinus malma Maekawa and Onozato ( 1986) Salmo salar Hutchings and Myers (1988); Jordan and Youngson (1992) (continued)

20

MICHAEL TABORSKY

TABLE I1 (Continued) ~~~

Family

~

Species

Gasterosteidae Poeciliidae

Gasterosteus aculeatus Xiphophorus nigrensis

Centrarchidae Cichlidae

Lepomis macrochirus Lamprologus brichardi

Pelvicachromis pulcher Interspecific male reproductive patasitism Cichlidae Lamprologus brichardi and Juliodochromis ornatus Labridae Pseudolabrus fucicola and P. celidotus Cirrhilabrus temminckii and C . cyanopleura Genetic predisposition of reproductive tactic Salmonidae Oncorhynchus kisutch Poeciliidae Poeciliopsis occidentalis Xiphophorus nigrensis

Centrarchidae

Lepornis macrochirus

Cichlidae

Pelvicachromis pulcher

~~

~

References Rico et al. (1992) Zimmerer and Kallmann (1989); Ryan e f al. (1990, 1992) Gross and Dueck (1989) M. Taborsky (unpublished data) (see text) Martin and Taborsky (1993) M. Taborsky (unpublished data) Ayling (1980)

Moyer (1981); Bell (1983) Iwamoto et al. (1983) Constantz (1975) Zimmerer and Kallmann (1989); Ryan et al. (1990, 1992) Dominey (1980); Gross (1982) (see text) Martin and Taborsky (1993)

Conditional realization of reproductive tactic

Salmonidae

Catostomidae Gasterosteidae Cyprinodontidae Poeciliidae

Cichlidae

Polycentridae

Oncorhynchus gorbuscha 0. nerka Salvelinus alpinus

Noltie (1989) Foote (1990) Sigurjonsdottir and Gunnarson (1989) Beauchamp (1990) Reighard (1920)

Thymallus arcticus Catostomus commersonii and Moxostoma aureolum Pungitius pungitius Moms (1952) Cyprinodon pecosensis Kodric-Brown (1981, 1986) Matsui (unpublished), cited C. macularius in Kodric-Brown (1981) Poeciliopsis occidentalis Constantz (1975) Poecilia latipinna Farr et al. (1986) Xiphophorus nigrensis Zimmerer and Kallmann (1989) Taborsky (1984a, 1985a) Larnprologus brichardi Pseudocrenilabrus philander Chan (1987) Polycentrus schomburgkii Barlow (1967) (continued)

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

21

TABLE I1 (Continued) ~

Family

Species

Labridae

Thalassoma lucasanum

Gobiidae Tripterygidae

T. bifasciatum Symphodus melanocercus Coryphopterus nicholsi Tripterygion tripteronotus

~~

References Warner and Hoffman (1980a); Warner (1982) Warner (1982) Wernerus (1989) Cole (1982) Wirtz (1978); Mohr (1986); de Jonge and Videler ( 1989)

Female choice of bourgeois males Salmonidae Salmo salar Poeciliidae Gambusia afinis Xiphophorus nigrensis

Jones (1959) Hughes (1 985y Zimmerer and Kallmann (1989); Ryan et al. (1990) Serranidae Serranus fasciatus Petersen (1987) Percidae Etheostoma caeruleum Reeves (1907) Cichlidae Pseudocrenilabrus philander Chan (1987) Peluicachromis pulcher E. Martin (personal communication) Pomacentridae Chromis cyanea De Boer (1981) Labridae Thalassoma bifasciatum Warner et al. (1975); Warner and Hoffman (1980b) van den Berghe et al. (1989); Symphodus ocellatus Wernerus (1989); own data (see text) S. tinca van den Berghe et al. (1989); Wernerus (1989) Wirtz (1978) Tripterygidae Tripterygion tripteronotus Interspecific egg dumping Lepisosteidae with Lepisosteus osseus at Goff (1984) Centrarchidae Micropterus dolomieui Cyprinidae Nocomis cornutus, N. Reighard (1943) rubellus, and Campostoma anomalum at N . micropogon Notropis lutipinnis and N. Wallin (1989, 1992) spp. at N . leptocephalus Cyprinidae with Pungtungia herzi at Baba et al. (1990) Serranidae Siniperca kawamebari Cyprinidae with Notemigonus crysoleucas at Can- (1946) Centrarchidae Lepomis punctatus Notemigonus crysoleucas at Kramer and Smith (1960) Micropterus salmoides Notropis umbratilis at Hunter and Hasler (1965)f Lepomis cyanellus (continued)

22

MICHAEL TABORSKY

TABLE I1 (Continued) Family

Catostomidae with Centrarc hidae Mochokidae with Cichlidae

Species

References

Notropis urdens at Lepornis Steele (1978)R; Steele and Pearson ( 198 I ) megalotis Erimyzon sucetta at M . Carr (1942) sulmoides Synodontis multipunctatus at Sato (1986); Schrader (1993) cichlid mouthbrooders

Fertilization stealing not directly observed. but very likely. Only few small males may resemble the female color pattern. ‘ Kleptogamy not mentioned. Small and probably parasitic males are also weakly territorial. That is, choice of Inrgr males. I Authors give information on 22 examples of interspecific nest utilization in cyprinids. Author gives information on 17 examples of interspecific nest utilization in cyprinids.

they may wait downstream of a pair in a linear dominance hierarchy and dart into the nest during oviposition to release sperm (Hutchings and Myers, 1988). Gross (1984) described how in coho salmon (0. kisutch) these smallest males would hide at some distance from a territorial male, but would still reach similar distances to females when releasing sperm as do large. “fighting” males. Medium-sized, anadromous salmon males may defend places near a nest (Gross, 1984),whereas those that are nearly as large as dominant males wait adjacent to the spawning pair and acquire matings through fighting (Hutchings and Myers, 1988). We may ponder over the reproductive success of parasitic males. As early as 1836. Shaw demonstrated that sperm of male salmon p a n (i.e., young, stream resident males) is capable of fertilizing eggs (see also Kazakov, 1981). Since then it has been repeatedly demonstrated that eggs fertilized by them produce viable offspring( Jones and King, 1950a;Thorpe and Morgan, 1980). Jones and King (1952a,b) sterilized large males and observed that parasitically spawning male parr shed sperm. More recently, the proportion of young sired by large and parasitic males, respectively, has been studied with the help of electrophoretic analyses of genetically polymorphic enzyme systems. In Oncorhynchus ketu, siqgle subordinate males were found to fertilize approximately one-quarter of the eggs deposited by a female when spawning in competition with a large male (weight ratio 0.75 : 1). Two subordinate males fertilized 47% of the eggs deposited into a single nest (Schroder, 1981). In one experiment with three male and one female 0. nerka. Chebanov et al. (1983) demonstrated that the two subordinate, parasitic males together sired 10% of the offspring. In

PARASITIC AND COOPERATIVE BEHAVIOR I N FISH REPRODUCTION

23

Salvelinus malma miyabei, Maekawa and Onozato (1986) maintained that nearly 17% of the eggs were fertilized by a subordinate male when experimentally placed with a spawning pair. However, when only cases with “apparent sperm release” (see Table I11 of Maekawa and Onozato, 1986) are considered and the median is calculated instead of the arithmetic mean, which seems more appropriate, only 7% of the eggs were on average fertilized by the small, kleptogamic males. Hutchings and Myers (1988) measured the proportion of eggs fertilized by varying numbers of Salmo salar parr that competed with dominant, anadromous males. Single male parr fertilized only about 5% of the eggs in a nest, but when 20 parr were simultaneously shedding sperm with one anadromous male, nearly a quarter of the eggs deposited by a female were fertilized by these subordinate males. The progeny of stream resident male Atlantic salmon pan- develop faster than those of sea-run males that have themselves matured at a later stage (Thorpe and Morgan, 1978, 1980). The age of first spawning is heritable in this salmon (Thorpe e f al., 1983; see also Schaffer and Elson, 1975), which may result in a predisposition of the reproductive tactics of males. Glebe et al. (1978) inferred from their (unpublished) data that there are both genetic and environmental components to the expression of precocious sexual maturity in this species. Bailey et al. (1980) found evidence for important maternal (i.e., nongenetic) and environmental effects of developmental characters. Lundqvist and Fridberg (1982) also demonstrated a strong environmental influence on the ontogeny of Salmo salar, and hence on the expression of male reproductive behavior. Fastergrowing males become precocious in this species (Dalley et al., 1983; see also Alm, 1959; Schiefer, 1971). In Oncorhynchus kisutch, there is a genetic component to the probability that a male will mature at 3 years of age and develop a “hooknose,” which is a weapon in intrasexual conflicts, as opposed to maturing at 2 years and remaining small (Iwamoto et al., 1983). The two different reproductive tactics exhibited by these “jack” and “hooknose” males were suggested to be about equally successful and are maintained by disruptive selection, as medium-sized males do not obtain good spawning positions neither when fighting nor when trying to steal fertilizations (Gross, 1984, 1985). Disruptive selection may also operate in Oncorh-ynchus nerka, in which Foote and Larkin (1988) observed that anadromous and stream resident forms mated assortatively and preferentially with members of the same form. Only if nonanadromous males could not find matching females did they try to steal fertilizations by approaching pairs of anadromous fish. In Oncorhynchus gorbuscha and Salvelinus alpinus, on the contrary, it is rather conditional whether a male guards or tries to

24

MICHAEL TABORSKY

TABLE 111 SYSTEMS THATARE CHARACTERIZED BY SUMMARY OF REPRODUCTIVE ASSOCIAT~ONS, COOPERATIVE BEHAVIOR, OR ALLOPARENTAL CARE Family

Species

References

Satellites stay near defended sites, not explicitly tolerated Sdmonidae Oncorhynchus nerka McCart (1970) cited in Keenleyside (1979, p. 104) Cyprinidae Sernotilus corporalis (small Ross and Reed (1978); Ross males) (1983) Zacco ternmincke Katano (1992) Centrarchidae Lepomis rnacrochirus Dominey (1981); Gross (1982) (female mimics) Cichlidae Sarotherodon alcalicus Albrecht (1968) Satellites tolerated by dominant males Cyprinidae Semotilus corporalis Ross and Reed (1978); Ross ( 1983) Notropis leptocephalus Wallin (1989) Mochokidae Synodontis multipunctatus Schrader (1993)" Cyprinodontidae Cyprinodon macularis Barlow (1961) C . pecosensis Kodric-Brown (1977, 1981, 1986) Cichlidae Apistogramma borelli Burchard (1965)" Kuwamura ( 1986)' Tropheus irsacae Eretrnodus cyanostictus Kuwamura ( 1986)' Lamprologus furcifer Yanagisawa (1987) L . callipterus (female Sat0 (1988) mimics) Telmatochrornis temporalis Mboko (1989) Peluicachromis pulcher Martin and Taborsky (1993)" Embiotocidae Micrometrus rninimus Warner and Harlan (1982) Pomacen t ridae Amphiprion akallopisos Fricke (1979) Labridae Symphodus ocellatits Soljan (1930a); Fiedler (1964); Taborsky (1984b, 1985b); Warner and Lejeune (1985); Taborsky et al. (1987) S. roissali Soljan (1931); Fielder (1964); Lejeune (1985) S . tinca Lejeune (1985) Halichoeres maculipinna Thresher (1979) Coris julis Lejeune (1985) Blenniidae Parablennius Santos (1985); Santos and Almada (1988) sanguinolentus Tetraodontidae Sikkel (1990) Canthigaster rostrata Ostraciidae Lactoria fornasini Moyer (1979) Joint defense of spawning territory Cichlidae Sarotherodon alcalicus A1brec ht ( 1968) (continued)

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

25

TABLE I11 (Continued) Family Pomacentridae Labridae

Blenniidae

Species Pelvicachromis pulcher Amphiprion akallopisos Symphodus ocellatus

S . roissali S . tinca Halichoeres maculipinna Para blennius sanguinolentus

Joint nest building Cyprinidae Nocomis micropogon Notropis leptocephalus Joint courtship Catostomidae Moxostoma carinatum Etheostoma blennioides Percidae Joint spawning Salvelinus namaycush Salmonidae Notropis lutipinnis Cyprinidae Catostomidae Catostomus commersonii Hypentelium nigricans Moxostoma aureolum M. duquesnei M . carinatum M. macrolepidotum M . erythrurum

M. valenciennesi Erimyzon oblongus Joint broodcare (intra- and interspecific) Cyprinidae Hybopsis biguttata with Notropis cornutus Cichlidae Cichlasoma citrinellum Etroplus surantensis Leptotilapia inuine

Bagridae with Cichlidae

Tilapia rendalli Bagrus meridionalis with Copadichromis

References Martin and Taborsky (1993)" Fricke (1979) Fiedler (1964); Taborsky (1984b, 1985b); Warner and Lejeune (1985); Taborsky et al. (1987) Lejeune (1985) Lejeune (1985) Thresher (1979) Santos (1985, 1986); Santos and Almada (1988) Reighard (1943) Wallin (1989) Hackney et al. (1967) Fahy (1954) Royce (195 1) Wallin (1989) Reighard (1920) Raney and Lachner (1946) Reighard (1920) Bowman (1970) Hackney et al. (1967); Hackney (1993, cited in Jenkins (1970, p. 245) Jenkins (1970) Jenkins (1970); Kwak and Skelly (1992) Jenkins (1970); Jenkins and Jenkins (1980) Page and Johnston (1990) Hankinson ( 1920)c McKaye and McKaye (1977) Ward and Wyman (1975, 1977) P. V. Loiselle (unpublished)", cited in McKaye and McKaye (1977) Ribbink et al. (1981) McKaye (1985); McKaye et al. (1992) (continued)

26

MICHAEL TABORSKY

TABLE 111 (Continued) Family

Species

References

pleurostigmoides Ctenophaqnx pictus, and Rhamphochromis s p . Cottidae Hemilepidotus hemilepidotus Alloparenhl care: (a) intraspecific adoptions Cichlidae Apistogramma trijksciati4tn A. horellii Tilapia rendalli (?) T. ma ria e I

Chrornidotilapia gitetitheri Pelvicachromis prrlcher Herotilopia rnultispinosa Cichlasoma citrinelliim C. longimanics C. nicaraguense C. nigrofasciaticm Neetroplus netnotopics Etropllrs macirlutiis Perissodus microlepis

DeMartini and Patten (1979) Burchard (1965)" Lorenzen (1989): Dieke (1993) Burchard (1967) Burchard (1967); Sjolander ( 1972) Sjolander (1972) Sjolander (1972): E. Martin (personal communication)" Baylis (1974)" McKaye and McKaye (1977) McKaye and McKaye (1977) McKaye and McKaye (1977) Wisenden and Keenleyside ( 1992) McKaye and McKaye (1977) G . W . Barlow (unpublished)," cited in McKaye and McKaye (1977) Yanagisawa and Nshombo (1983); Yanagisawa (1985a) Yanagisawa (1985b. 1986) Thresher (1985)

Xenotilapia juvipitrtiis Acanthochromis po/yacanihw Ailoparental care: (b) mixed-species broods Bagridae with Bugrus meridionalis cares McKaye and Oliver (1980); Cic hlidae for Copadichromis McKaye (1985) pleurostigmoides, Ctenophuqnx pictiis, and Rhamphochromis s p . Tilupia rendalli (?) and T. Cichlidae Burchard (1 967) mciriae T. mariae and T. ;illii Sjolander (1972) Cichlasoina citrinelliim McKaye and McKaye (1977) cares for Neetropliis nematopus C . longimanics cares for C. McKaye and McKaye (1977) citritrellutn Pomacentridae

(continued)

PARASITIC AND COOPERATIVE BEHAVIOR I N FISH REPRODUCTION

27

TABLE I11 (Continued) ~~

Family

~

Species

References

C. nicaraguense cares for C. McKaye and McKaye (1977) longimanus N. nematopus cares for C. McKaye and McKaye (1977) citrinellum 12 Haplochromis spp. and Ribbink (1977); Ribbink et al. Serranochromis robustus (1980) care for fry of 15 diff. species Lamprologus elongatus Yanagisawa and Nshombo cares for Perissodus (1983) microlepis Cichlidae with 10 mouthbrooding spp. care Brichard (1979); Sato (1986); Mochokidae for Synodontis Schrader (1993) multipunctatus Alloparental care: (c) pure heterospecific broods Esocidae with Esox niger cares for Shoemaker (1947) Centrarchidae Lepomis gibbosus Cichlidae Cichlasoma nicaraguense McKaye (1977) cares for C. dovii Alloparental care: (d) nest takeovers Cyprinidae Pimephales promelas Unger and Sargent (1988)" Percidae Etheostoma olmstedi Constanz (1979, 1985) Pomacentridae Amphiprion clarkii Yanagisawa and Ochi (1986) Labridae Symphodus ocellatus Taborsky et al. (1987) Gobiidae Padogobius martensi Bisazza et al. (1989a) Hexagrammidae Ophiodon elongatus Jewel1 (1968) Cottidae Hemilepidotus DeMartini and Patten (1979) hemilepidotus Cottus gobio Bisazza and Marconato (1988)" Harpagiferidae Harpagifer bispinis Daniels (1978, 1979) Alloparental care: (e) egg stealing Gasterosteidae Gasterosteus aculeatus van den Assem (1967); Wootton (1971); Li and Owings (1978b)"; Sargent and Gebler ( 1980)" Cichlidae Pseudocrenilabrus Mrowka (1987b)" multicolor Alloparental care: (f) broodcare helpers Cichlidae Lamprologus brichardi Kalas (1976)"; Taborsky and Limberger (1981); Taborsky (1984a, 1985a); Hert (1985)"; Taborsky et al. 1986"; von Siemens (1990)" (continued)

28

MICHAEL TABORSKY

TABLE I11 (Continued) Family

Species

References

L . pulcher

Taborsky and Limberger

L. savoryid

Taborsky and Limberger (1981); Kondo (1986); Abe (1987) Kalas (1976y; Taborsky and Limberger (1981)” Taborsky and Limberger

(1981)” Julidochromis ornatus

J. regani

(1981)“ J. marlieri

Belontiidae a

Betta brownorumd B . persephoned

Taborsky and Limberger (1981); Yamagishi (1988) Witte and Schmidt (1992)u Witte and Schmidt (1992)”

Aquarium observations only.

* Sex of conspecifics that are tolerated within temtones is unclear. Division of labor: H . bigutfota builds nest and N. cornutus guards it. As yet only cooperative defense of breeding territory observed.

steal fertilizations (Noltie, 1989; Sigurjonsdottir and Gunnarson, 1989). Essential conditions include relative male size and conditions, and male density. 5 . Female Mimicy

Kleptogamic males often resemble females in their appearance. These males have been called “pseudofemales” (e.g., Morris, 1952), “female mimics” (e.g., Dominey, 19801, or “transvestite males” (Dipper, 1981). As early as 1907, Reeves observed that bright, territorial male darters sometimes mistake small, drab males for females. Subsequently, mimetic resemblance of kleptogamic males to females has been observed in more than 30 species belonging to 10 different fish families (see Table 11). These mimetic males may resemble females morphologically (including color: e.g., Kodric-Brown, 1986), behaviorally (e.g., Constantz, 1979), or both (e.g., Warner and Robertson, 1978). They may be the same age as territory owners (e.g., Morris, 1952; Dominey, 1980; and in Trivers, 198.5; but see Gross, 1982) or younger (e.g., Wirtz, 1978; Mohr, 1986), and they may be “initial phase” individuals (see earlier) that may later change into “terminal colour phase” in sex-changing wrasses (e.g., Jones, 1981; Warner, 1982). They may roam about in small schools or loose aggregations with females (e.g., Robertson and Choat, 19741, or stay in the vicinity of nests (e.g., Keenleyside and Dupuis, 1988) o r even within bright males’

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

29

territories as unrecognized reproductive parasites (e.g., Thresher, 1979). The feature they all have in common is that they “scrounge by deception” (Barnard, 1984). In most reported cases of female mimicry it is unclear, however, whether bourgeois males really mistake parasitic males or females. In the ocellated wrasse, Symphodus ocellatus, the resemblance between parasitic males and females has been stressed repeatedly (see Table I1 for references). But detailed behavioral observations revealed that nest owners usually treat females and parasitic males very differently, with regard to both, behavioral qualities and quantities (Taborsky et al., 1987). 6 . Alternative Tactics When Fertilization Is Internal

In species with internal fertilization, kleptogamy is sequential instead of simultaneous and not so much a matter of escaping recognition by a dominant, bourgeois male. Rather, small- or medium-sized males mate more or less forcefully instead of courting females, as the largest males of a population do (e.g., Liley, 1966; Farr, 1980a; Hughes, 1985;Heinrich and Schroder, 1986; Ryan and Causey, 1989). In Gambusia affinis and G . holbrooki, the vast majority of copulations may even result from males forcibly inseminating females (Bisazza et al., 1989b), and the majority of females of a South Carolina population of mosquito fish had been multiply inseminated (Chesser et al., 1984). Table I contains more examples of poeciliid fish with alternative tactics, that is, courting and forced copulations (see also Constantz, 1984, for a discussion of sperm competition in poeciliids). 7. A Comparison between “Bourgeois” and Parasitic Males a. Numbers. Often brightly colored, aggressive individuals constitute the vast majority of reproductively active males fe.g., Albrecht, 1969). In other systems, however, males specialized in parasitic spawning may make up a much larger proportion of reproductive individuals than the more conspicuous males that monopolize some resources (e.g., Ross, 1983). In a population of bluegill sunfish, for example, 85% of the males parasitized the effort of the 15% of males defending territories and providing parental care (Gross, 1982). With regard to number, parasitic males could then be viewed as the primary reproductive form in these species, even though the occurrence of kleptogamy relies on the existence of some individuals whose effort can be parasitized upon. b. Costs. Male effort may be behavioral, morphological, and/or physiological. Behaviorally, there is a wide range of possibilities for expenditure on mate recruitment and paternal care. Bourgeois males may defend a territory, spawning place, or “nest,” invest in courtship, build or dig to

30

MICHAEL TABORSKY

prepare a spawning site, and care for eggs, larvae, and young. These activities may increase the risk of predation because of conspicuous behavior andlor reduced vigilance. On the other hand, there is only little effort by bourgeois males that can be exploited by kleptogamic males in the mochokid catfish Synodontis rniiltipunctatrrs (Schrader, 1993), in some Percidae (e.g., Winn, 1958a,b), cichlids (e.g., Kuwamura, 1987), and wrasses (e.g., Warner and Hoffman, 1980b). In Catostomidae, for example, often pairs of males spawn with a single female (see the following), but there may be additional males trying to get as close to the female as possible and interfere with the spawning trio (Reighard, 1920), thereby presumably “attempting to sneak fertilizations” (Page and Johnston, 1990). The superior position of the two males adjoining the female on either side of her at spawning may be only a matter of the sequence of making contact with a ripe female, and not represent an expensive investment. Male effort may also be morphological, for example, involving a change of color or specific body structure. When males of Mediterranean wrasses become reproductively territorial they show a bright color pattern (Fiedler, 1964; Michel et al., 1987). Such morphological changes are probably associated with physiological costs (see Frischknecht, 1993), and the increased conspicuousness will probably increase the risks of predation, as exemplified in three-spined sticklebacks (Semler, 1971 ; Moodie, 1972) and guppies (e.g., Endler, 1980). A bright nuptial coloration of bourgeois males is also known from other wrasses (e.g., Robertson and Hoffman, 1977; Warner and Robertson, 1978; Colin and Bell, 1991), and from other fish taxa, like shiners and sunfish (Steele, 19781, darters (Petravicz, 1938), cichlids (Voss, 19801, damselfish (Thresher and Moyer, 1983), parrotfish (Colin and Bell, 1991), Hexagrammidae (DeMartini, 1985), and Triperygidae (Wirtz, 1978). Dichromatism in marine fish was reviewed by Thresher (1984). Other temporal features developing toward spawning include morphological structures like the kypes and humps in salmon (e.g., Gross, 1985; Keenleyside and Dupuis, 1988), bright nuptial humps in blennies (e.g., on the head of male Mediterranean Salaria p a w , Fishelson, 1963, and my own observations), and breeding tubercles, for example, in suckers (“pearl organs”, Reighard, 1920) and cyprinids (Wedekind, 1992). Internal morphological and physiological changes of reproductive males that occur in connection with sound production have been found in the plainfin midshipman (Porichrhys notatus; e.g., Bass, 1992). The physiological costs incurred by bourgeois males may be expressed, for example, simply by a difference in growth from other conspecifics. In the Mediterranean wrasse Symphodiis ocellafzrs, for example, a proportion of males refrain from reproduction in a given season (Taborsky et al.,

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

31

1987) and thereby grow during that period while all reproductive males, territorial or parasitic, stagnate in size (Fig. 2). These nonreproductives are very probably the territorial males in subsequent seasons (M. Taborsky, unpublished evidence). In the closely related S . rinca, van den Berghe (1992) showed that nesting males had four times greater costs than nonnesting males during a reproductive season, as measured by weight changes.

x

X

x X Y

-.. 8

x

T8

Sat

Sn

NR

FIG.2. The change of weights of different types of males during two separate spawning seasons (1982 and 1983) in the Mediterranean wrasse Symphodus ocellatus (location: STARESO, Calvi, Corsica). T-male, territorial (bourgeois) male; Sat, satellite; Sn, sneaker (both parasitizing the reproductive effort of T-males by simultaneously spawning with them); NR, nonreproductive males, which do not show any reproductive activities in a specific year (Le., season). Each dot or cross represents one individual. Medians are marked with a horizontal dash.

32

MICHAEL TABORSKY

In longear sunfish. kleptogamic males have higher gonad/body weight ratios and slower somatic growth rates than bourgeois males. The costs of generating all these behavioral or morphological structures may be parasitized upon by kleptogamic males, which usurp the effort of bourgeois males and their attractiveness for females and fertilize (or sometimes eat: see Table 11) a proportion of the eggs spawned by these conspecifics. Parasites may, however, have considerable costs themselves, as exemplified in S. ocellatus. In this species, as a result of their reproductive activities, parasitic males have a similar reduction in growth as that of bourgeois, territorial males (see Fig. 2). The reproductive costs of parasitic males are even more prominent when it comes to gonadal investment. They cannot, obviously, usurp the gonadal effort of conspecific males. Rather, they should put their own effort primarily into the production of sperm, and hence also into large and prolific testes. A higher gonadibody weight ratio is therefore expected in parasitic than in bourgeois males. This is exactly what is found in S. ocellarus (Fig. 3) and in other labroid fish with both types of males, bourgeois and kleptogamic (e.g., Robertson and Choat, 1974; Choat and Robertson 1975; Warner and Downs, 1977; Robertson and Warner, 1978; Warner and Robertson, 1978; Warner and Lejeune, 1985; see also Table 11). It has also been demonstrated in the North American bluegill sunfish, in which “female mimics” have a gonadjbody weight ratio more than twice that of territorial males (Dominey, 1980), and the smaller “sneakers” even exceed the parental male ratio by fourfold (Gross, 1982). See Table I1 for examples from other fish families. The behavioral costs of kleptogamic as compared to bourgeois males are probably low. The only effort they share with the latter is the behavior immediately leading to fertilizations. Apart from that, they need to obtain a good position to interfere in spawning and they may need to interact aggressively with other parasitic males and submissively with bourgeois males. There are very few data with which to compare these costs with the costs of monopolizing males. Time expenditure has been shown to be higher in territorial than in sneaker males in S. ocellatus, which means that rhe latter spend more than twice the time feeding than do the territory owners (Taborsky et a / . , 1987). Energetically, however, there does not seem to be that much difference between sneakers and territorial males, (see earlier; Fig. 2). I do not know of any published data that allow a quantitative comparison of the predation risk of bourgeois and kleptogamic male fish. However, the risk of being killed by larger conspecifics may be considerable for small males aiming to share in reproduction. Of 49 yearling male chinook salmon (Oncorhynchus tshawytscha) found dead on a spawning ground,

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

33

I

__

Td

Sat

Sn

NR

FIG. 3 . Relative gonad mass of male and female Symphodus ocellatus during the reproductive season of 1983. Location, symbols, and abbreviations as in Fig. 1.

21 had been killed by adult conspecifics, probably by males (Gebhards, 1 960). c . Success. Success rates of males may be measured at different levels. The simplest way is to determine the proportions of spawnings that are parasitized. This has been done for 13 species belonging to five different families (Table 11, live bearers excluded). For a proper comparison, the number of fertilization attempts should be measured for both types of males on an individual basis. In the cichlid Pseudocrenilabrus philander, Chan (1987) found that nearly 10% of spawnings involved attempted fertilizations by parasitic males, and he suggested that the fertilization success of these males was only 6.35% of that of territory owners. He did not take account of the competition of sperm of temtorial and parasitic males but assigned all eggs laid while a parasite was present and trying to fertilize them to this male. Most information on rates of fertilization attempts by bourgeois and

34

MICHAEL TABORSKY

parasitic males exists in wrasses. In Corisjulis, about 30% of terminal phase male spawnings were interfered with by initial phase parasites (average number of simultaneously spawning parasites was 1.8; Lejeune, 1987). On an individual basis, terminal phase, territorial males spawned nearly 8 times more often than parasites did. Lejeune (1985) observed rates of attempted fertilizations by territorial and kleptogamic males in six Mediterranean species. I calculated from his figures that, in Symphodus cinereus and S . melanocercus, on an individual basis territorial males spawned about 10 times more often than parasitic males tried to steal fertilizations. Surprisingly, Warner and Lejeune (1985), who observed the same population of S . melanocercus at the same location and time, recorded only one parasitic fertilization attempt out of 269 observed spawnings. In S . roissali, Lejeune’s (1985) measure gave a ratio of 1 fertilization attempt per parasite to 10 attempts per territory owner when all males were considered, that is, regardless of whether they were reproductively active during the observation period or not. When only sexually active males around the nest are considered, however, the fertilization attempts of parasitic and territorial males occurred at a ratio of 1 : 3. In S . ocellatus this ratio was about 1 : 6, whereas in S. tinca the ratio depended on the size of kleptogamic males. Small parasitic males made three times fewer fertilization attempts than territory owners, whereas medium-sized parasitic males made on average nearly three times more. These fertilization attempts of medium-sized or small males in the nests of territory owners were either interferences with nest owners’ spawnings or separate spawnings. To estimate male success when more than one male is involved in spawning, each fertilization attempt of a male has been divided by the total number of males that have participated in a specific spawning (termed “pair spawning equivalents” by Warner et al., 1975). If one territorial and two parasitic males are involved, for example, each of them is assigned one-third fertilization. This is perhaps not a very good estimate of fertilization success because of differences between the males with regard to position, timing, and the amount (and perhaps quality) of released sperm, but it is probably still closer to the truth than if simultaneous, multiple fertilization attempts were disregarded or assigned to one participant only. Unfortunately, there is no reliability analysis available to check such estimates with the true proportions of fertilized eggs. If this method is applied to the species discussed in the foregoing, there is some discrepancy with data based on the pure rates of fertilization attempts, but also between data sets from different studies on the same species. In S . roissafi, one can estimate from the data of Warner and Lejeune (1985) that parasitic males had about 19% of the fertilization success of territory owners; in S . ocellatus this estimate would be about

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

35

12%. Wernerus (1989), however, measured 25.2 fertilizations per hour (i.e., “pair spawning equivalents”) in nest owners and 9.7 in parasites, that is, an estimated fertilization rate in the latter of 38.5% when compared to nest owners. In another data set he reports 41.3 estimated fertilizations per hour for territorial males and 5.7 for parasites, that is, a fertilization rate of 13.8% when compared to territorial nest owners. Van den Berghe et al. (1989) also provide two sets of data on pair spawning equivalents of territorial and parasitic males, one of which cannot be further analyzed from published evidence as the time interval for which the territorial male success rate was given is unclear. The other data set gives an estimate of 35.9% success of parasitic males compared to that of territorial males. In S. tinca the situation is also somewhat unclear. Warner and Lejeune (1985) documented only a single interference of a parasitic male in 109 observed spawnings. In sharp contrast to this, van den Berghe et al. (1989) and Wernerus (1989) both found that in the same population 74% of spawnings involved “peripheral males” (i.e., purely parasitic males and satellites), and their data suggest a fertilization rate of 1.5 per hour for territorial males and 5.4 per hour for reproductive parasites. This discrepancy from the observations of Warner and Lejeune (1985) may be partly due to the fact that van den Berghe et al. (1989) and Wernerus (1989) did not separate simultaneous parasitic spawnings from occasions when peripheral males spawned with a female in the nest of a territory owner without participation of the latter. To summarize these data on Mediterranean wrasses, reproductive parasites always achieved “pair spawning equivalents” that were within the range of about 10 to 40% of those of territorial males. Only in S. tinca do nonterritorial males seem to have higher fertilization rates than nest owners. This is, however, only partly due to simultaneous parasitic spawning. Why do males of this species show nesting behavior in the first place? Lejeune’s (1985) data suggested that eggs spawned outside nests have only minute survival chances. Because a large part of the attempted fertilizations, especially of medium-sized, nonnesting males, occur outside nests (81%, Warner and Lejeune, 1985), the large nest males may still fare as well or even better than medium-sized and small males, despite their considerably fewer ‘‘pair spawning equivalents.” A better estimate of the reproductive success of bourgeois and parasitic males would be possible if position effects of simultaneously spawning males could be accounted for. Position seems to be important in suckers (Bowman, 1970) and fallfish (Ross and Reed, 1978), in which the territories or nests of dominant males serve as spawning sites. In communal spawning acts, parasitic males that stay in a waiting position are always peripheral to the more dominant territory owners. But also in species with very rapid

36

MICHAEL TABORSKY

spawning acts, the distances between eggs at spawning and the positions of bourgeois and parasitic males trying to fertilize them may greatly differ from each other, as may the timing of sperm release (e.g., in S . ocellatus, my own observations). Schroder (1981) demonstrated by paternity analyses that in Oncorhynchus keta, male mating success was directly related to female proximity during spawning. Single parasites fertilized on average a quarter of the eggs deposited by a female when spawning in competition with a large, dominant male (see Section II,C,4). On the basis of these data, Gross (1985) estimated reproductive success of kleptogamic and bourgeois (i.e., dominant) males in Oncorhynchus kisutch. He showed that the best option for gaining proximity to spawning females differs between males of different sizes. Small males did best by simultaneous parasitic spawning (“sneaking”), and large ones by fighting for position. Surprisingly few data exist simply showing that parasitic males do sire offspring. A first hint may be obtained by artificial fertilization experiments with sperm of parasitic males (e.g., Jones and King, 1950a). Van den Assem (1967) showed that eggs had been fertilized and developed normally in the nests of three-spined sticklebacks even when only a parasitic male had passed through after the spawning female, and not the nest owner. The reproductive success of parasitic fertilizations can only be proved unequivocally, however, by comparing genetic patterns between offspring and their potential parents. Paternity analyses have been done by analyzing genetically polymorphic protein markers with electrophoretic techniques in four species of salmonids (see Table 11). Hutchings and Myers (1988) concluded from an interspecific comparison of these results that the weight ratio of dominant and parasitic males is probably important for the proportion of eggs fertilized by them. The smaller male fertilized between 0 and 46% of the eggs when only one parasite competed with a bourgeois male during spawning. The weight ratios between them varied from 2 to 75%. Testis mass would probably be an even better correlate of relative fertilization success of simultaneously spawning males, but this has not been analyzed yet. A size-related difference in male reproductive tactics allowed estimation of relative male success in Xiphophorus nigrensis (Zimmerer and Kallmann, 1989; Ryan et al., 1990). The technique of genetic fingerprinting was used in a study on three-spined sticklebacks. Rico et al. (1992) showed that in one nest 5 out of 10 fry were not sired by the nest owner, in another nest it was 1 fry out of 10. In total, 3.5% of 170 examined fry resulted probably from parasitic fertilizations (cf. also Gross and Dueck, 1989, for a study of bluegill sunfish). Genetic markers may also be more obviously expressed, as in the form of color patterns. In the West African cichlid Pelvicachromis pulcher

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

37

there are two male color morphs. These color patterns are expressed relatively early in ontogeny. Martin and Taborsky (1993) found that male offspring sired by the males of one morph always belonged to that morph, whereas males of the other morph produced male offspring of both types. By combining a harem owner of one type with satellites of the other they found that the relative reproductive success of the most dominant satellite male was on average nearly 30% of that of the territory owner, whereas the figures for beta and gamma satellites were about 15 and 5%, respectively. Overall, the seasonal net reproductive success of harem owners was on average seven times higher than that of their satellites, mainly because the latter were completely excluded from spawning with the alpha female of the harem. Simultaneous parasitic spawning may also occur between members of different species (e.g., Bell, 1983). When viable offspring are produced on these occasions, species-specificfeatures can serve as genetic markers. In a laboratory study of Lake Tanganyika cichlids, I combined specimens of Lamprologus brichardi and Julidochromis ornatus in one tank. When a pair of J . ornatus spawned, males of the other species fertilized a proportion of the eggs, thereby proving that they can successfully sire offspring by simultaneous parasitic spawning (see Fig. 4; the F1 generation was fertile). d. Origin. In principle, there are two possible origins for parasitic male spawning. At these two extremes, the expression of this reproductive tactic may be purely phenotypic, or it may result from an unmodifiable genetic disposition (see Austad, 1984).

FIG. 4. A hybrid (middle) resulting from the simultaneous parasitic spawning of a Lamprologus brichardi (left) male with a pair of Julidochrornis ornatus (right). This is a proof of successful reproductive parasitism.

38

MlCHAEL TABORSKY

There is evidence for a genetic predisposition, albeit perhaps not unmodifiable, toward alternative reproductive strategies in salmonids. In an artificial breeding experiment with coho salmon (Oncorhynchus h u t c h ) , Iwamot0 et ul. (1983) found that eggs fertilized by male parasites produced a significantly higher proportion of parasitic male offspring than did those sired by large, “hooknose” males. Gross (198.5)suggested, from the spatial distribution of males during spawning, that disruptive selection would stabilize the existence of small and large reproductive males by favoring their respective reproductive tactics, if performed by the “right-sized” males. Large males obtain proximity to females at spawning mainly by fighting for position, whereas small males accomplish this by simultaneous parasitic spawning (“sneaking”). Males of intermediate sizes are at a disadvantage, However, there is another, potentially important reason for the prevalence of large and small males. Spawning in coho salmon is seasonal, and small and large males are recruited from different age cohorts. The small “jack” males stay only one “season” (5-8 months) at sea, and the large hooknose males stay for two “seasons” (17-20 months). This fact alone could explain the bimodal size distribution of reproductive male salmon as schematically depicted by Gross (1984). In other words, there i s no age cohort from which to draw intermediate males. The different reproductive behaviors of these males may then be viewed as an adaptation to (i.e., a consequence of) the size-dependent opportunities to get close to a spawning female. Gross (198.5) estimated a similar lifetime reproductive success for the two male types by using a combination of differential ocean survivorship, reproductively active time at the breeding grounds, and mating success as derived from the different malelfemale distances during spawning. A different way of viewing the origin of reproductive parasitism is by looking at whether a male tactic is fixed for life or conditional on circumstances (Dominey, 1984). These two possibilities exist independently of the degree of genetic influence on the form and expression of reproductive behavior. When males remain small and parasitic for life this may primarily result from a genetic disposition, or from an environmental feature that may, for example, set the stage in their early ontogeny. On the other hand, males changing from parasitic to bourgeois reproduction may act purely in a conditional manner or be under a strong genetic influence with regard to the expression of their reproductive tactic. Most likely this behavior will derive from some interaction of genetic and environmental influences, It may be assumed, however, that in species in which males change their tactic the genetic influence is not as strong as it might be in species with fixed, lifelong male reproductive tactics. Male reproduction in the gila topminnow Poeciliopsis occidentalis may

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

39

serve to illustrate the two different levels of analysis. Large males of this species often take a conspicuous, black color pattern and become territorial, court females, and show longer bouts of gonopodial thrusting. Small males retain their cryptic color pattern and try to fertilize females without courtship and within the territories of dominant males (Constantz, 1975). If large males are experimentally removed, small males may instantaneously change color and start to defend territories and court females. This demonstrates clearly that their reproductive behavior is conditional. The probability of showing one or the other tactic, however, is strongly size dependent, and these fish cease to grow upon reaching sexual maturity. Size is known to have a strong, genetic component in the Poeciliidae (see Ryan et al., 1990). Therefore, the behavior of males is not genetically fixed but dependent on conditions (i.e., relative male size and the existence of competitors), but there is probably a pronounced genetic influence on these conditions. A similar system has been demonstrated in Xiphophorus nigrensis (Zimmerer and Kallmann, 1989; Ryan et al., 1990, 1992). In bluegill sunfish (Lepomis macrochirus), Dominey (1980) found that small parasitic males, which he termed “female mimics,” and large nesting males were both 6 years of age on average, suggesting that the two behaviors were pure strategies fixed for life. Gross and Charnov (1980) and Gross (1982), however, found that the corresponding kleptogamous males, which they termed “satellites,” were on average only 4 years of age in another North American population, whereas most parental males were more than twice as old. This discrepancy may have been caused by either a difference in populations or in methodology (Dominey used the rings in otoliths for an age estimate, Gross used those in scales). Regardless of these different results, Gross (1982) also suggested that the small, parasitic males in his population, which started to reproduce as “sneakers” at an estimated age as low as 1 or 2 years, were not transitional stages toward the bourgeois tactic later in life, but members of a different lifetime reproductive strategy. This suggestion was based on the analysis of scale growth patterns of parasitic and bourgeois males (see also Jennings and Philipp, 1992a, for a similar suggestion in long ear sunfish). A critical test of this suggestion would be a comparison between the growth patterns of large (bourgeois) and small (kleptogamic) males during their early years of life, that is, when males of the latter type should have reproduced already while those of the former had presumably refrained from reproduction. On the basis of Gross’s data (1982, Table 6) I compared the growth increments of males belonging to the 7- to 10-year class (i.e., all being potential or real bourgeois males) with those of 3- to 5-year-old parasitic males (as judged by Gross from their gonadal states), during

40

MICHAEL TABORSKY

years 2. 3, and 4 of their lives. The year in which they were finally caught was excluded from this analysis as the date of capture would have influenced the measurable growth increment in that year. There were 24 possible comparisons between age cohorts, of which 17 revealed significant growth differences (t-tests; thep chosen was 0.001 because of multiple analyses). This strongly supports Gross’s conclusion that the males of this species follow reproductive strategies that are fixed for life, at least from the moment when the males have become sexually mature. It remains unclear whether genes or ontogeny, or both, decide the reproductive fate of a male. Gross and Charnov (1980) and Gross (1982) concluded from intrusion frequencies and the proportions of parasitic males in seven populations at Lake Opinicon that the sum of all parasites fertilized as many eggs as all parental males did (all eggs spawned during “successful” intrusions were ascribed to the parasites, however!). This was regarded as evidence that the tactics had evolved as mixed evolutionarily stable strategies (Gross, 1982, 1984, 1991). A similar system exists in the Mediterranean wrasse Symphodus ocellatus. There are two types of parasites, which were called “sneakers” and “satellites” by Warner and Lejeune (1985) and Taborsky et af. (1987), and larger, parental males whose reproductive effort is parasitized upon. There are also males of a fourth type that do not participate in reproduction in a given season and probably become nest-building, bourgeois males in future years (Taborsky et al., 1987). Soljan (1930b) assumed from the growth pattern of scales that the point when these males are born in the season determines whether they will later be “outsiders of fertilization” or nestbuilders. The early-born males, which have extended growth already before the first winter, reproduce early next season (i.e., when they are about 1 year of age) by simultaneous parasitic spawning. They remain parasites for life (i.e.. also for their second reproductive season). The males that are born late in the season grow little before the first winter but grow for a long period of time after this first winter and before they start to reproduce. In their second year they are all nestbuilders. Combined with our long-term field information, it seems likely that these bourgeois males are nonreproductive when 1 year old, but start to reproduce right away as bourgeois males in their second year. If Soljan’s interpretation of scale growth patterns was right, “birthdate” decides in the males of this species which reproductive strategy they follow for life. In the West African cichlid Peluicachromis pufcher, there are two male color morphs. “Yellow males” always breed as pair males (i.e., bourgeois) and “red males” may either become pair or harem males (i.e., bourgeois) or reproduce as satellites, which are tolerated as male helpers within the territories of harem owners. This means that only males of the latter

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

41

morph become reproductive parasites. The color morphs are fixed for life and their expression is subject to a strong, genetic predisposition (Martin and Taborsky, 1993). All of these cases illustrate that a male reproductive strategy may be fixed for life, regardless of the extent to which its causes are genetic or environmental. In the majority of known cases, however, the reproductive role of males is conditional, that is, males may take up bourgeois or kelptogamic tactics depending on the circumstances (see Table I1 for a list of 20 examples from 10 fish families). These circumstances may be either relative size (e.g., Poecifia fatipinna, Farr et al., 1986; Saluelinus alpinus, Sigurjonsdottir and Gunnarson, 1989; Oncorhynchus nerka, Foote, 1990; Tripterygion tripteronotus, de Jonge and Videler, 1989), male condition (e.g., Oncorhynchus gorbuscha, Noltie, 1989), the intensity of intrasexual competition (e.g., Cyprinodon pecosensis and C . macularius, Kodric-Brown, 1981, 1986; Symphodus melanocercus, as demonstrated by removal experiments, Wernerus, 1989), prior residence (e.g., Oncorhynhus nerka, Foote, 1990), or the ontogenetic stage of a male (e.g., Tripterygion tripteronotus, Wirtz, 1978; Mohr, 1986; Thalassoma lucasanum and T . bifasciatum, Warner and Hoffman, 1980a; Warner, 1982; Lamproologus brichardi, Taborsky, 1985a). In some species males may switch back and forth between bourgeois and parasitic tactics (e.g., Pseudocrenilabrus philander, Chan, 1987; see Fig. 5; Pofycentrus schomburgkii, Barlow, 1967). Often, the choice of tactic and/or its success appears to be frequency dependent (see Gadgil, 1972; Maynard Smith, 1982), although conclusive evidence is missing. 8. Female Choice of Males with Different Reproductive Tactics In many species females seem to prefer bourgeois males. Atlantic salmon females try to chase away parasitic males (Jones, 1959). Female Xiphophorus nigrensis prefer large courting males (Zimmerer and Kallmann, 1989; Ryan et al., 1990). Thalassoma bifasciatum females prefer large males in specific territories and are increasingly reluctant to spawn when potential male parasites are nearby (Warner et al., 1975; Warner and Hoffman, 1980b). In Pseudocrenilabrus philander (Chan, 1987), Chromis cyanea (De Boer, 1981), and Symphodus ocellatus and S . tinca (Taborsky et al., 1987; van den Berghe et al., 1989; Wernerus, 1989), females often leave the nest when parasitic intrusions occur, even though in S . ocellatus they prefer to spawn in nests where satellites are present (but, evidently, with the bourgeois nest owners; Taborsky, 1985b, 1987). Spawning Tripterygion tripteronotus females attack parasitic males (Wirtz, 1978). In S. ocellatus, bourgeois nest males and kleptogamic sneakers and satellites all approach and interact with females that are ready to spawn

FIG. 5 . Schematic description of mating activities in a laboratory lek of Pseudocrenilabrus philander. Peripherally placed fishes with dark markings are territorial (bourgeois) males. A group of females (pale fishes) is in the central region. Fishes 1and 4 are semiterritorial (mostly parasitic) males, whereas 5 to 8 are nonterritorial (purely parasitic spawners). In nest A, a semiterritorial male is just caught in parasitic spawning at a territorial male’s nest. At nest B, parasitic male 5 is about to intrude and join the spawning pair for parasitic sperm release. The spawning in nest C is interrupted by the intrusion of an egg-stealing female while the territory owner attempts to ward off other potential intruders. The semiterritorial male 4 has adopted bright colors and courts a female while the nearby tenitory owners (above) are engaged in fighting. Reproduced from Chan (1987).

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

43

and approach a nest. This behavior looks as if females are being herded to the nest by these males (Fig. 6). We followed 30 females in spawning phase for an average period of 25 min each to check their reactions to male approaches. The frequencies with which females were approached did not differ between the male types. However, female reaction did. An approach by a territory owner increased the likelihood that she would enter a nest, which often led to spawning, whereas an approach by a parasitic sneaker or satellite had exactly the opposite effect; females were then more likely to leave (Fig. 7; Taborsky, 1987). This is remarkable as the behavior exhibited by these males looks exactly the same. Yet in the case of nest owners it has the effect of herding females, whereas in the case of the other males it results in female expulsion. The latter was also described qualitatively by Wernerus (1989). Van den Berghe et al. (1989) and Wernerus (1989) removed some of the parasitic sneakers from the vicinity of nests and found a five- to eightfold increase of female spawning rates in these nests. Equivalent removals at S. tinca nests gave similar results. Van den Berghe et al. (1989) suggested that Symphodus females chose mates based on age, defensive ability, or size as an indicator of their genetic quality. Involvement in matings by peripheral males did not show obvious costs to females in assumed fertilization rates, egg mortalities, or the quality of subsequent parental care. D. PARASITIC BEHAVIOR OF FEMALES Prezygotic investment is generally higher in females than in males, which limits the potential reproductive rate in the former (Clutton-Brock and Vincent, 1991). Therefore, males compete for access to female

FIG. 6. “Contact following,” a behavior that male Symphodus ocellatus (black) perform toward conspecific females (white) in the vicinity of a nest (stippled circle). The figure shows a sequence of positions of one contact following event that was derived from film frames of footage taken in the field. The male is behind and above the female when showing this behavior and may touch her at times, as if he would herd her into the nest. After Taborsky et al. (1987).

44

MICHAEL TABORSKY

Approaching P P are contacted by no

Sat

1

Sn

contact

1

-

w p = 0.01

p < 0.001

FIG. 7. Reactions of female Symphodus ocellatus who approached a nest of a territorial male to being contacted by conspecific males (see Fig. 6) of different types. Abbreviations as in Fig. 2. Each unit is derived from a 25-min behavioral protocol made in the field (location: see Fig. 2) of an individual female that was ready to spawn. The bars mark the number of recordings in which the focal females reacted to these approaches more often by visiting the male’s nest (above zero) or by leaving the area (below zero). The right bar shows how often females completed their nest approaches as compared to Ieaving the area without visiting the nest, when they were not approached by a male; this is intended to serve as a control. Each female was only recorded once.

gametes and not vice versa (e.g., Trivers, 1985). This means that males can parasitize each other’s effort to obtain access to these gametes (see the previous sections) whereas females cannot, because even if there is competition for access to mates among females, this does not involve investment that could be parasitized by others. Females may, however,

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

45

parasitize post-zygotic effort of conspecifics when there is female broodcare. The latter is generally rare among fish, although it is common in a few families (e.g., Breder and Rosen, 1966; Baylis, 1981). Intraspecific brood parasitism is hard to detect (Anderson, 1984; MacWhirter, 1989). Therefore, only recently has evidence begun to accumulate on this phenomenon, for example, in birds (e.g., Moller, 1987; Pinxten et al., 1991; Weigmann and Lamprecht, 1991). In fish, an intraspecific mixing of broods that might be viewed as parasitic has been observed in cichlid maternal mouthbrooders (e.g., Ribbink et al., 1980; Yanagisawa, 1985a, 1986). It seems, however, that this is not a specific female adaptation to save parental effort. There is no evidence for female egg dumping in these cases. Rather, free-swimming fry are taken into the mouth of a parent and spat into a school of fry guarded by other conspecifics (“farming out”; Yanagisawa, 1985a,b, 1986). This was observed in Xenotifupiaflauipinnis when both presumed parents were guarding fry, and in Perrissodus microfepis only when one partner had been experimentally removed. Remarkably, the only two cases in which the sex was known of the pair member successfully farming out parts of its brood involved females (Yanagisawa, 1985a). Female egg dumping in fish has been documented, however, on an interspecific scale. It appears to be common in cyprinids. The nests of Nocomis micropogon, for example, are used as spawning sites by three other species (see Table 11). This interspecific egg dumping occurred during nearly all observation periods and at all nests with spawning activity observed by Reighard (1943). Parasitic “associates” were obviously attracted by building activities of the host species, and up to 200 fish were simultaneously present at an active nest. Abandoned nests were taken over and guarded by members of these egg-dumping species. This “insurance” for a successful completion of broodcare (i.e., defense of eggs) might be an ultimate advantage for the host species. The potential costs for hosts include competition for oxygen among the eggs in a nest, possible cannibalism occurring during the turmoil at spawning, and the chances of hybridization (eggs of the host might be fertilized by sperm of another instead of their own species). There are other cyprinids using nests of cyprinid hosts for spawning (see Wallin, 1989). Yellowfin shiners (Notropis lutipinnis) failed to reproduce in the absence of bluehead chub (Nocomis feptocephalus) nests, into which they usually dump their eggs. Conspecific eggs constituted on average only 3% of all eggs found in bluehead chub nests (Wallin, 1992). Cyprinids use also nests of sunfish for spawning [see Hunter and Hasler (1965) and Steele (1978) for references to egg-dumping Cyprinidae]. Sunfish host species inlcude Lepomis punctatus (Carr, 1946), L. cyanelfus

a (Hunter and Hasler, 1963, L . megalotis (Steele, 1978), and Micropterus sainzoides (Kramer and Smith, 1960). The latter species may also defend eggs and young of the sucker Erimyzon sucetta (Carr, 1942). In that case, the host fry survived better in nests containing fry of the other species ( p <: 0.02, Fisher Exact Probability Test, calculated from data of Carr, 1942; see also McKaye, 1981). Similarly, M . salmoides was found to care for eggs and fry of the garpike Lepisosteus osseus, and nests containing young of both species were more successful than those containing the host species only (Goff, 1984). In contrast to this, egg dumping of the cyprinid Pungtitngia herzi at the freshwater perch Siniperca kawamebari reduced the reproductive success of the host species by 35%. The reason was that another cyprinid, Zacco temmincki, exploited the confusion caused by the spawning P . herzi and robbed its eggs (Baba et al., 1990). A predisposition for egg dumping in cyprinids is probably the habit of pelagically spawning demersal eggs, that is, the eggs spawned in the water column sink passively into the nest below (McKaye, 1981). A mochokid catfish from Lake Tanganyika, Synodontis multipunctatus, was found to parasitize mouthbrooding cichlids (Brichard, 1979; Finley, 1984; Colditz, 1986; Sato, 1986), and the good correlation between the sizes of host and parasite fry suggested that they are of equal age, that is, that the transfer of offspring from parasite to host occurred at spawning (Sato, 1987). This was confirmed by aquarium observations (Staats, 1988; Schrader, 1993). Members of ten different cichlid species were found to brood catfish eggs or young, and Sat0 (1986)found eight broods that consisted only of parasite offspring. The reason for this is probably intrabuccal predation, as aquarium observations revealed that catfish fry consume host fry. In the case of the catfish/cichlid interaction it appears to be clear that the costs of interspecific care for the host by far outweigh any potential benefits. This is not as clear in the spawning associations of cyprinids. The host species may either suffer from competition of their eggs and fry with those of their associates (see the foregoing), or they might somehow benefit from them, for example, by the predator dilution effect, or neither benefit nor suffer. Phrased differently, the relationship may be parasitic, mutualistic, or commensalic. Apart from the two cases in which Micropterits salmoides cared for Erimyzon sucetta and for Lepisosteus osseus, both of which seem to be mutualistic, and the indirect damage of egg dumping Pungtungia herzi to their host Siniperca kawarnebari, the effect of these interspecific associations on the fitness of the host species has not yet been studied. A mixing of broods may also occur when offspring leaving their parent move to a neighboring, guarded school of fry on their own (Ribbink et a / . . 1980), or through kidnapping (McKaye and McKaye, 1977; McKaye,

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

47

1981), or expulsion of parental guards by other, stronger neighbors (Lewis, 1980). This also leads often to an interspecific mixing of broods (e.g., McKaye, 1977,1981; McKaye and McKaye, 1977;Ribbink, 1977; Ribbink et al., 1980), which is apparently not parasitic behavior in the form of a parent trying to save parental effort by passing its young to the care of a stepparent. It will therefore be treated in a separate section together with other examples of interspecific brood mixing. 111. ASSOCIATIONS BETWEEN REPRODUCTIVE COMPETITORS AND COOPERATION

Cooperation might be seen as the opposite of competition. In my view it is instead another form of selfish behavior where individuals attempt to improve access to resources (Taborsky, 1987; see also Harcourt, 1987). Therefore, we should not expect a clear-cut difference between “purely competitive” (i.e., parasitic) and cooperative (i.e., mutualistic) behavior. I trust this will become clear in this section, which aims at summarizing our current knowledge on cooperative behavior in fish reproduction (see Table 111, p. 24). I should stress that I am not concerned with nonreproductive forms of cooperation in this review (e.g., Milinski, 1987; Milinski et al., 1990; see Pitcher, 1992, for a review). I start with a discussion of cases in which males known as “satellite males” associate with bourgeois reproductive competitors. These males are always competitively inferior, but in many cases they are tolerated at or near the bourgeois males’ defended area. In some cases there is joint defense, nest building or courtship by the different males. There is nonaggressive, joint spawning that in certain species occurs without exception (Catostomidue). Within the context of raising offspring, conflict is often less pronounced than in the competition for producing them. This may result from the lower benefits of parasitic behavior the later it is performed in the succession of efforts bearing upon attempts to reproduce. Also, interactions often concern related individuals within the context of broodcare, which also lowers the payoff of parasitic behavior. Cooperative behavior therefore seems more prominent in broodcare systems. For example, parents of different broods may either jointly defend their offspring or one may care for the young of the other. In extreme cases individuals of one species may care for the brood of another even without having young of their own. Such alloparental care may also result from nest takeovers, when the second fish continues to care for the brood of the first. The most advanced forms of intraspecific cooperation are found in systems that are characterized

48

MICHAEL TABORSKY

by higher than average degrees of relatedness. Young of previous broods remain with their parents and help to raise subsequent offspring (broodcare helpers). In the discussion of cooperative associations I focus on the possible costs and benefits to the participants whenever there are data available. We shall see, however, that there is often ample room for speculation given the lack of data. I use the term satellite simply to refer to a spatial relationship of an inferior to a bourgeois male. It does not hint at the role of this male or the type of interactions with the dominant owner of a territory or resource. Time is not included in this definition, but the association should last for some minimum proportion of the reproductive period of these males, to separate this type of association from purely kleptogamic events; for example, when a male parasite enters the territory just to shed sperm when a female is spawning he will not be termed satellite. The term helper is used operationally for an individual participating in some effort and does not necessarily presume a benefit for the receiver of this help (e.g., parent or brood; see Taborsky, 1984a). Cooperation is used in a broad sense and includes behaviors performed by two or more individuals that appear to serve a common purpose. Their behavior does not need to be coordinated.

A.

SATELLITE MALES

1 . Males That Are Not Explicitly Tolerated at a Defended Site

Males may associate with a defended site of a bourgeois male but remain outside or at the margin of the latter’s territory. In Oncorhynchus nerka (Salrnonidae), one to eight satellite males stay in the vicinity of a large male’s nest. Each of them defends its own position. When the dominant nest owner disappears, the biggest satellite takes his position. The satellites parasitize the nest owner’s reproductive effort by kleptogamy (McCart, 1970).Small males of the African cichlid Sarotherodon alcalicus defend small pits (i.e., shallow depressions) at the edge of large pits owned by large conspecific males (Albrecht, 1968). There they spawn with very small females. In fallfish minnows (Sernotilus corporalis), bluegill sunfish (Lepomis rnacrochirus), and the Mediterranean ocellated wrasse ( S . ocellatus)there are two types of alternative reproductive behaviors. The small reproductive parasites in fallfish minnows (Ross and Reed, 1978; Ross, 1983) and oceilated wrasses (i.e., the males termed “sneakers”, Taborsky et al., 1987) stay at a nest for some period of time without being tolerated by the nest owner, similarly to the satellite males in dark chub (Zacco

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTlON

49

temmincki, Katano, 1992). They participate parasitically in spawnings in the nest. In bluegill sunfish, female mimics are only partly expelled. It has been suggested that they are either not recognized as males (i.e., the mimicry is effective) or expulsion may be too expensive for the nesting males (Dominey, 1981). In the hawkfish Cirrhitichthys falco, “sneaker” males lived on the periphery of a harem, but it is unclear whether they were tolerated by its owner (Donaldson, 1987). Satellites also exist in other, nonreproductive stiuations. In the blueheaded wrasse (Thalassoma bifasciatum),for example, subordinate males often stay very close to a bright male’s stationary place or shelter. They are frequently chased by the latter, but often manage to remain in their vicinity and probably benefit from using the shelters of these bourgeois males (Reinboth, 1973).

2 . Males That Are Tolerated by Bourgeois Males With regards to functional explanations, cases in which satellite males are accepted to some extent at bourgeois male’s territories are more interesting than those described in the previous section. Small, parasitic males may be ignored by territory owners, as in the dwarf surfperch Micrometrus minimus (Warner and Harlan, 1982), or they may dwell above or at the boundary of defended sites, as found in the pupfish Cyprinodon pecosensis (Kodric-Brown, 1977, 1981, 1986). In the latter species these brightly colored satellite males are frequently attacked and pursued but still manage to stay at the edge of a territory. They mate primarily with small females, whereas larger females spawn preferentially with territory owners. Often, satellites are tolerated amid territories. In the cichlids Tropheus irsacae and Eretmodus cyanostictus, the sexes of these tolerated individuals are unknown (Kuwamura, 1986); in Lamprologus callipterus, these individuals are males with female color patterns (female mimics; Sato, 1988). The same is true for the wrasse Coris julis, in which these satellites are sometimes even courted by the owners of the territories (Lejeune, 1985). Satellite males of the blenny Parablennius sanguinolentus stay permanently within the territories of large, bourgeois males (Santos, 1985; Santos and Almada, 1988). They court females and share in territory defense (see the following), and steal fertilizations. All of this also applies to the cichlid Peluicachromis pulcher. If there is more than one satellite per territory in this species, there is a strongly size-related dominance hierarchy between them (Martin and Taborsky, 1993). Subordinate males of the anemone fish Amphiprion akallopisos are tolerated by a breeding pair at their host anemone. Fricke (1979) suggested from behavioral observations and histological analyses of gonads that these males are psycholog-

50

MICHAEL TABORSKY

ically castrated, perhaps through high levels of stress (see also Reyer et al., 1986). and hence are not capable of stealing fertilizations. In harem species, several small males may stay within the territories of harem owners (Halichocres macutipinna, Thresher, 1979; Canthigaster rostrata, Sikkel, 1990; Lactoria fornasini, Moyer, 1979). The reproductive role of these satellites is unclear. In L . fornasini, harem owners often attack their satellites, which respond with appeasement behavior (Moyer, 1979). Small males of the pupfish Cyprinodon maciifaris defend subterritories within the territories of large males (Barlow, 1961). They are frequently pursued by the latter but return quickly to their defended places. In Apistogramma borellii, males of female size pair up with females within the territories of large males and also defend subterritories (Burchard, 1965). In the cyprinids Sernotilus corporulis and Notropis leptocephalus, nestmales tolerate other males close to their nests (Ross and Reed, 1978; Ross. 1983; Wallin, 1989). This is similar to the Mediterranean wrasses Symphodus ocellatirs (Soljan, 1930a;Taborsky, 1984b, 1985b; Warner and Lejeune, 1985; Taborsky et al., 1987). S . roissali (Soljan, 1931; Fiedler, 1964: Lejeune, 1985), and S . rinca (Lejeune, 1985). There is a hierarchy between satellites if there are several at a nest ( S . ocellatus, Taborsky et al., 1987; S . tinca, Lejeune, 1985). S . ocellatus satellites behave submissively toward territory owners and they are explicitly tolerated by them. This was demonstrated by a comparison of the behavior of nest owners between encounters with satellites and with other parasitic males (called sneakers; Taborsky et a f . ,1987).Once a satellite is accepted by a territorial male he will usually stay at his nest until the end of the spawning activity. Lejeune (1985) documented a similar constancy of residence by satellites in the closely related S. roissali. In most of the cases described here, satellite males parasitize the effort of bourgeois males by stealing fertilizations. Why are they tolerated? Why is there a range in degrees of tolerance? Expulsion may be simply not possible, as suggested to be the case in Lumprologus Jurcijer (Yanagisawa, 1987). Or attempting to keep these parasites at a distance may be more costly than accepting the loss of a proportion of fertilizations to them (Kodric-Brown, 1977). Ross (1983) suggested that in fallfish tolerance of satellites might be better for nest owners than chasing them, as the latter behavior would interrupt spawning activity. A special benefit from the tolerance of satellite males by territory owners may be found in Arnphipriori akallopisos. In this protandric, sexchanging species, young males may serve as replacement partners when a pair member disappears (Fricke, 1979), that is, tolerance is an insurance strategy to keep potential partners available. A fourth possible benefit for territory owners of the presence of satellites

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

51

is that they might increase the attractiveness of a spawning site to females (Kodric-Brown, 1977; Ross and Reed, 1978). Data from an experimental field study of the ocellated wrasse (Symphodus ocellatus) suggested that territory owners would greatly benefit from the presence of satellites, because females preferred to spawn in nests at which satellites were present (Taborsky, 1985b, 1987). They used satellite males as a cue to assess the probability that their eggs will be tended until hatching (M. Taborsky and P. Wirtz, unpublished data). An attractive function of accessory males was also hypothesized for bluegill sunfish (Lepomis macrochirus; Dominey , 1981). However, alternative explanations to these favored by the various researchers cannot be ruled out completely in any of the foregoing examples. Codbenefit analyses aiming to obtain conclusive evidence on the payoffs of satellites, and especially of bourgeois males, are still a challenge for the future. Bourgeois males may also benefit from the behavioral effort of satellite males, for example, from territory defense, nest building, courtship, or broodcare activities. These possibilities will be treated in the next section. B. COOPERATIVE REPRODUCTIVE BEHAVIOR 1 . Joint Defense‘

Small pupfish (Cyprinodon macularis) satellites tolerated in territories of large males defend their ranges, and hence the common territory as well, against intruding conspecifics (Barlow, 1961). In a functional sense this is similar to the situation in the cichlid Sarotherodon alcalicus, in which the small males surrounding the pits of large reproductive males expel large, roaming males and attack neighboring territory owners (Albrecht, 1968). Harem owners of the West African cichlid Pelvicachromis pulcher often host between one and three satellites permanently in their year-round, all-purpose territories. These satellite males defend the common harem range, both intra- and interspecifically, and they put more effort into this defense behavior than the harem owners themselves do (Martin and Taborsky, 1993). Harems of the wrasse Halichoeres maculipinna may also contain several satellite males that join in territory defense against neighbors and subordinate male competitors (Thresher, 1979). Subdominant male anemone fish (Amphiprion akallopisos) defend the pair’s terriI “Joint” is used here to classify behavior shared by two or more individuals. It does not assume any coordination between the participants.

52

MICHAEL TABORSKY

tory in which they reside against strange conspecifics, predators, and predators of the host anemone (Fncke, 1979). Satellite males of the Mediterranean wrasses Symphodus ocellatus (Fiedler, 1964; Taborsky. 1984b, 1985b; Warner and Lejeune, 1985; Taborsky et d . , 1987), S . roissali (Lejeune, 19851, and S. tinca (Lejeune, 1985)attack conspecific male parasites that try to steal fertilizations when the nest owner is spawning. In the ocellated wrasse ( S . ocellatus), satellites exhibit even higher defense frequencies against these reproductive competitors than territory owners (Taborsky et al., 1987). In none of these species do satellites attack neighboring temtory owners. Such defense against larger, bourgeois neighbors is shown by satellites of the blenny Parablennius sanguinolentus. When frequencies of this behavior are combined with attack rates on smaller competitors, satellites of this species also surpass temtory owners with regard to intrasexual defense frequencies (Santos, 1986). On the ultimate level, the defense effort of satellite males may be purely selfish, that is, only satellites themselves benefit from excluding reproductive competitors because they participate parasitically in spawnings of the territory owner. The latter would in that case tolerate their satellites for other reasons, but not because of a net benefit derived from the satellites’ defense effort. Alternatively, the fact that satellites invest in territory defense may functionally result from a reciprocal relationship with the bourgeois male, that is, satellites are tolerated and their fertilization stealing is accepted to an extent by the dominant male, because they ward off a host of other, purely parasitic reproductive competitors. This could be termed paying for staying (see Taborsky, 1984a, 1985a).A series of removal experiments performed in the field showed that in Symphodus ocellatus, when satellites were present at a nest bourgeois nest owners neither saved defense effort nor experienced reduced rates of parasitized spawnings. Therefore, these accessory males acted in a purely selfish manner (M. Taborsky and P. Wirtz, unpublished data).

2 . Joint Nest Building, Courtship, and Spawning Cyprinid males may cooperate in nest building. Accessory males of Nocomis leptocephalus, which are usually smaller than dominant nest builders, occasionally share in nest building. Their nest building appears to have on average very little effect, however (Reighard, 1943). In the bluehead chub (Notropis leptocephalus), several males may contribute to nest construction. These nest-building associations may last for considerable periods of time. Wallin (1989)observed two large, individually recognizable males who jointly constructed five different nests in succession. I do not know of any studies that test whether these associations are mutualistic or parasitic.

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

53

Males of the northern greenside darter, Etheostoma blennioides, were reported to court females jointly. Fahy (1954) observed in a tank that two males courted a female partly alternating and partly synchronized before the larger one of them spawned with her. In the field, groups of males were found to court a female, one at a time, without any aggressive behavior between these males. Males of the sucker Moxostoma carinaturn jointly court a female at a site prepared by one of the two participants. The other male duplicates the nuptial dance of the nest builder before both of them spawn with the female in unison (see the following; Hackney et al., 1967; Hackney, unpublished, from Jenkins, 1970, p. 245; see also Page and Johnston, 1990). In several temperate freshwater fishes two or more males may spawn jointly, without obvious aggression between them. Often, these males do not differ in their roles, that is, there is no distinction possible between a bourgeois and a satellite (or parasitic) tactic. For example, two or more males of the lake trout, Salvelinus namaycush, court and spawn with a female simultaneously. There may be up to seven males and three females spawning in unison (Royce, 1951). Six males or more may cluster around a female while spawning in the yellowfin shiner (Notropis lutipinnis; Wallin, 1989). Some species of suckers (Catostomidae) spawn either facultatively or usually in trios, involving two males and a single female (Jenkins and Jenkins, 1980; Page and Johnston, 1990). There may be accessory males participating in these spawnings. In a large number of suckers, perhaps in the majority of species, spawning occurs only in trios (see Table I11 for a listing of species and references; especially Reighard, 1920; Jenkins and Jenkins, 1980; Page and Johnston, 1990). Occasionally, they are joined by additional males that may cause an interruption of the spawning act (Reighard, 1920). The two male spawning partners adjoin the female on either side and press against her flanks. This formation is stabilized by breeding tubercles or “pearl organs” (Reighard, 1920) that roughen the body surface of males. Spawning is usually simultaneous by all three members of a trio and has been suggested to be more “efficient” than in pairs of one male and one female only (Kwak and Skelly, 1992). The river redhorse (Moxostoma carinatum) differs from other species because in this sucker a male constructs a redd (i.e., spawning site) and displays there in front of a female. He is then joined and followed in motion by a second male (Hackney et al., 1967; Hackney, unpublished, from Jenkins, 1970 p. 245). The female takes position between the two males for spawning. The ultimate reason for trio spawning in suckers has so far not been studied. Obviously, there is sperm competition for fertilization of the eggs between the simultaneously spawning males. There are three conventional arguments to explain the mutual tolerance of competing males in this

54

MICHAEL TABORSKY

situation. The participants may (i) be closely related to each other (kinship advantage). (ii) take turns with mutually aiding each other (reciprocal altruism), or (iii) behave cooperatively because one partner forces the other to do so (manipulation). All explanations have in common that the benefit over cost ratio of a male participating in joint spawning exceeds that of a male trying t o monopolize a female. There is no evidence that any of these hypotheses explains the observed behavior satisfactorily. It is unlikely that partners in male duos are closer related to each other than the population average, because suckers do not appear to remain localized between their own egg and reproductive stages. There is some migratory behavior before spawning, and young and adult stages often do not share the same parts of a river (Jenkins, 1970). It is very improbable that brothers, for example, would stay together for years from their hatching until spawning and during all of these migratory movements. A scenario in which reciprocity was responsible for the observed behavior kould require that the same two partners meet repeatedly on successive spawning occasions. This cannot be totally excluded given our present state of knowledge, but observations of some species suggest that males associate with different partners for successive spawning events (Reighard, 1920). The third explanation, manipulation, is unlikely because no agonistic behavior has been observed in most of the reported cases. (iv) Another possibility to explain this case of apparent cooperation would be if males were forced to spawn jointly because the simultaneous pressure of two individuals is necessary on the flanks of a female for her to release eggs, or simply to induce spawning. Kwak and Skelly (1992) regarded trio spawning as “more efficient” than pair spawning. This shifts the question from male to female biology. An advantage from increased genetic variability and/or fertilization certainty could perhaps cause the evolution of a habit that allows females to spawn only with at least two male partner? at a time. (v) A fifth hypothesis would imply that joint spawning in suckers is rather a case of parasitism than of cooperation. The first male, which has just obtained a female and presses on to her flank, may simply be incapable of preventing a second male from doing the same on her opposite side. If this holds, more aggression might then be expected tc occur generally between reproductive males at the spawning area. (vi) The last possibility I would like to discuss includes a simultaneous benefit for both male participants. If a female would release, say, 100 eggs when spawning with one male only, but two males could literally press 1000 eggs out of her oviduct when spawning with her simultaneously, each male would increase its fertilization success fivefold by cooperating

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTlON

55

with a competitor (assuming, for ease of argument, fertilizations are on average equally shared between the two male partners). As with the other hypotheses presented to explain trio spawning of suckers, there is no evidence yet that this joint manipulation of females by males occurs. The evolutionary background of joint spawning in suckers is one of the most puzzling riddles in the reproductive behavior of fishes. It might have implications for our understanding of cooperative processes in animal behavior in general. Future studies, preferably performed on facultative trio spawners, should reveal the ultimate and proximate causes of this social phenomenon.

3. Joint Broodcare Cooperative behavior in fish reproduction is not limited to conspecific associations that have the purpose of obtaining mates or fertilizations. It also occurs between parents tending eggs or young (see Keenleyside, 1991). In this section I discuss cases of communal care, in which different parents jointly raise and protect their respective offspring. Intraspecific communal care has been documented for cichlids. The green chromide (Etroplus suratensis) may occasionally exhibit joint care of large schools of young (Ward and Wyman, 1975). Three out of 28 parent/offspring units (1 1%) observed in the field contained more than two adults, ranging from four to six (Ward and Wyman, 1977). Presumably, these were the parents of mixed schools of their offspring, although evidence is missing. The young “glanced and micronipped” mucus from these adults, which is a typical form of provisioning by parents in this genus (Ward and Barlow, 1967). McKaye and McKaye (1977) observed three pairs of the Midas cichlid (Cichlasoma citrinellum)jointly defending a large school of young, which split into three parts when threatened. A large school of young Tilapia rendalli that consisted of two size classes was observed to be guarded by two pairs of adults (Ribbink er al., 1981). In sculpins communal egg guarding, involving mostly one primary female and up to four secondary males, was observed in Hemilepidotus hemilepidotus (DeMartini and Patten, 1979). It has been suggested that the secondary guardian males have probably not spawned in the guarded nest, but may use the clutch as a “courtship dummy” to attract ripe females (see Section III,C,3). More than 50% of the broods guarded by the Lake Malawi catfish Bagrus meridionalis contained cichlid young (see Table 111 for the list of species). In more than half of these cases there were cichlid adults guarding these mixed broods of young as well. McKaye (1985) never observed young of the two Crytocara species involved outside mixed cichlid/catfish groups. As with the intraspecific situation in Etroplus suratensis (see

56

MICHAEL TABORSKY

preceding), cichlid young fed from the body surface of the guarding catfish, but were forced by the latter to the periphery of the school (McKaye et al., 1992). In cyprinids, Hankinson (1920) observed cooperative broodcare of Hybopsis biguttata with Notropis cornutus. They showed some division of labor, with the former species building the nest and the latter guarding it. Joint broodcare by more than one parental unit appears to be relatively rare. This is perhaps not surprising when viewed in the light of cost/ benefit ratios of the participating adults. Guarding young involves costs, with regard to both predation risk and time (i.e., subsequent reproduction is postponed). If the offspring are additionally defended by other parental adults, the benefits of leaving the care completely to these alloparents will probably often outweigh the costs of increased predation on the young after desertion. Codbenefit ratios may be asymmetric between the two parties involved, as was suggested by McKaye’s observations (1985). When adult catfish and cichlids jointly care for a mixed school of young, the cichlid parents may desert without great costs, because the catfish are easily capable of defending the offspring alone. When the catfish parents were experimentally removed, however, the entire school of young was eaten by predators in seven out of nine experiments (i.e., all trials with young < 6 cm), within a period of 15 min (McKaye, 1985). This asymmetry may be the reason why nearly half of the mixed broods of cichlids and catfish only had catfish guards, that is, the cichlid parents may have deserted in these cases. This case study nicely illustrates the likely reason why the occurrence of joint broodcare appears to be relatively rare when compared to the widespread phenomenon of brood mixing. It pays to abandon one’s offspring if another individual provides care anyway. The same argument has been used to explain uniparental care, in the context of task sharing between male and female parents (see Maynard Smith, 1977). The care of mixed broods by one set of parents will be discussed in the subsequent sections.

C. ALLOPARENTAL CARE

I . lntruspecific Adoptions In many cichlid species, groups of young that are guarded by parentai adults may belong to different size/age classes. In substrate brooding, permanently territorial species, this may be a consequence of prolonged filial philopatry, which may result in helper systems (see the following). In the majority of cases, however, it is more likely that this intraspecific

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

57

mixing of broods results from adoptions of strange young. Adoptions can be easily induced experimentally, in both the aquarium and field (e.g., Greenberg, 1963; Sjolander, 1972; Noakes and Barlow, 1973; McKaye and McKaye, 1977; Carlisle, 1985; Wisenden and Keenleyside, 1992), with conspecifics and with young of other species (e.g., Noble and Curtis, 1939; Collins and Braddock, 1962; Myrberg, 1964; Mrowka, 1987a). They also occur on their own, that is, without experimental manipulation (e.g., Burchard, 1965; Baylis, 1974; McKaye and McKaye, 1977; Mrowka, 1987b). The chances of witnessing adoptions in an undisturbed situation are very low. Therefore, the occurrence of adoptions is usually deduced from the fact that broods contain different size classes, or because broods increase in size (McKaye and McKaye, 1977). Table I11 lists species in which intraspecific adoptions have been directly observed or can be safely assumed to occur from these indirect cues. There are several possible reasons for the occurrence of adoptions. Unfortunately, no cost/benefit analyses have been performed yet to study the ultimate (i.e., evolutionary) reasons. We may expect different mechanisms of offspring transfer between broodcaring adults, depending on whether it is advantageous to donors or stepparents, or to both. Information on who initiates the transfer of offspring may provide some hint as to who will ultimately benefit from it (although nonadaptive “mistakes” and “accidents” are potential alternatives). Kidnapping has been observed in the Midas cichlid, Cichlasoma citrinellum (McKaye and McKaye, 1977), in the orange chromide, Etroplus maculatus (G. W. Barlow, unpublished observations, cited in McKaye and McKaye (1977), in Apistogramma borellii (Dieke, 1993), and in Pseudocrenilabrus multicolor (Mrowka, 1987). The latter is a maternal mouthbrooder, and kidnapping refers here to an interference of strange females at spawning and take up of eggs by these interlopers. The stealing of eggs or young may have a positive dilution effect on the kidnapper’s own young when predation on offspring occurs (McKaye and McKaye, 1977). Wisenden and Keenleyside (1992) suggested that this potential antipredation function of brood adoptions may be the reason why in many species only young of equal or smaller sizes than their own offspring are accepted, as predator efficiency is negatively correlated with prey size. Farming out of broods is a term used for a behavior by which parents actively transfer young to strange broodcaring adults (with reference to “egg dumping,” this behavior might be called young dumping). This was observed in the Tanganyika cichlid Perissodus microlepis (Yanagisawa, 1985a). It only happened when one parent was left with the brood, either because its partner has been experimentally removed or had disappeared

58

MICHAEL TABORSKY

for unknown reasons. Yanagisawa argued that the reduced chances of raising a brood alone would make farming out very profitable for single parents. This may also hold for convict cichlids. When the males of broodcaring pairs of Cichlasoma nigrofasciatirm were experimentally removed by Wisenden and Keenleyside (1992), the transfer probability of the brood to other pairs with young increased. In the African mouthbrooder Xenotilapin fiauipinnis, farming out was observed when both parents were still in charge of the brood (Yanagisawa, 1985b, 1986).Generally, as the transfer of young is initiated by donors, it is likely that they are gaining more from this behavior than the stepparents do, in exact contrast to the previously discussed phenomenon of kidnapping. As a consequence, donors should be expected to transfer their offspring to parents caring for young that are still smaller than their own, for the same reasons as Wisenden and Keenleyside (1992) hypothesized that kidnappers should preferably steal offspring that are smaller than their own. There are two other possibilities for how young of different parents may coalesce. When brood-tending adults meet, their young may join the wrong school (family conflux). This is especially likely to happen in species lacking stable territories. It may result in reciprocal adoption, in the displacement of parents (Wisenden and Keenleyside, 1992). or in the collection of all young with the most aggressive parents (Baylis, 1974). Alternatively, young may independently join a guarded brood, after separation from their own parents (independent offspring inclusion). Large offspring of the damselfish Acanthochromis polyacanthus are expelled at some stage by their parents. They may then join a neighboring school of young that is still guarded by adults (Thresher. 1985).The latter appear incapable of separating the two groups of young and expelling the interlopers. Both family conflux and independent offspring inclusion may be beneficial, costly, or neutral to stepparents. It is not possible to assess the likely payoff from the form of this behavior alone; measurements of offspring survival are required to unravel this phenomenon.

2. Mixed-Species Broods and Care for Pure Heterospecifc Broods Frequently, when there are mixed broods containing young of different species, these are guarded by one or two adults of only one of the species involved (see Table 111 for a list of species). For example, all broods guarded by adult Midas cichlids (Cichlnsomn citrinellurn) that were older than 5 weeks of age contained young of Neetroplus nematopus in addition to their own offspring (McKaye and McKaye, 1977).Some guarded broods of N . nematopirs also contained Midas cichlid young. Ribbink (1977) observed that mouthbrooding adults of three predatory Lake Malawi cichlids contained young of Haplochrornis chrysonotus. A closer look revealed

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

59

that members of 12 Haplochromis species and of Serranochromis robustus regularly cared for young of 15 different species, in addition to their own young (Ribbink et al., 1980). The authors concluded from their observations that in Lake Malawi “all species which show well developed parental behavior may have foreign fry mix[ed] with their broods on occasion.” Guarded broods of the predatory, substrate-brooding Tanganyika cichlid Lamprologus elongatus may contain young Perissodus microlepis. Yanagisawa and Nshombo (1983) suggested that these heterospecific young join the host broods on their own. In a later study it was shown that the young of this species are occasionally “farmed out” by their parents (see foregoing), which may be another cause for this interspecific brood mixing. There are intriguing associations between catfish and cichlids in the great East African lakes. Schools of young guarded by the bagrid catfish Bagrus meridionalis often contain young cichlids (McKaye and Oliver, 1980; see preceding), and in nearly half of these cases the mixed broods were guarded by adult catfishes only. The proximate cause for brood mixing was probably active release of cichlid offspring into bagrid broods by their mothers (McKaye, 1985; McKaye et al., 1992), The geometry of these mixed schools is influenced by the aggressive behavior of the catfish parents against cichlid young at the center of the school. Attack frequencies of predators on catfish young were seven times greater in pure broods than in those that were mixed with cichlid offspring (McKaye et al., 1992). Mouthbrooding adults of ten cichlid species were found to tend young mochokid catfish in their mouths, together with their own offspring (Brichard, 1979; Sato, 1986; Schrader, 1993). This is due to egg dumping by female catfishes when the cichlids spawned (see foregoing). Costs and benefits of interspecific brood mixing are probably similar to those involved with intraspecific adoptions. Donors of young will probably benefit from the relief of broodcare. It remains to be studied whether this advantage outweighs the potential costs of reduced survival of young when they are tended by stepparents. Often, they will even gain increased survival probabilities by this transfer to guarding adults of another species, as was suggested by the cichlid/catfish association in Lake Malawi (McKaye, 1985). Cichlid young which are guarded by catfish may even benefit nutritionally because they feed from the body surface of their foster parents (McKaye et al., 1992). Hosts may gain from the inclusion of foreign fry, as demonstrated by the increased offspring survival of the sunfish Micropterus salmoides when broods contained young Erimyzon sucerta (see Section II,D), and of Micropterus dolomieui when the brood was mixed with Lepidosteus osseus (Goff, 1984). Catfish young received seven times fewer attacks when cichlid young were present (McKaye et al., 1992). The effect of foreign young on the survival of hosts could also

60

MICHAEL TABORSKY

be neutral, or disadvantageous as demonstrated by the inclusion of catfish fry in the mouths of broodcaring cichlids. Sat0 (1986) found eight mouthbrooding cichlid individuals that tended only catfish young. This was probably due to catfish fry predating cichlid fry within the mouths of the latters’ mothers (see Section 11,D). Care of pure heterospecific broods was also demonstrated for chain pickerel (&ox niger). Shoemaker (1947) found this species to defend eggs and young of pumpkinseed sunfish (Lepomis gibbosus). He assumed that this apparent interspecific altruism would benefit the performing pickerels because they would thereby get access to the predators of the beneficiaries’ young (i.e., golden shiners, Notemigonirs crysoleucas), upon which they prey. In a Nicaraguan crater lake adult Cichlasoma nicaraguense males were observed to share in the brood defense of C . dovii parents (McKaye, 1977). Four broods guarded by these heterospecific alloparents suffered less mortality than six control broods guarded only by conspecific adults. McKaye (1977, 1979; see also Coyne and Sohn, 1978) suggested an evolutionary scenario to explain this apparent altruism in which the alloparents would take advantage of a population increase of the beneficiaries because the latter prey on the former’s main competitors. This would need to involve group selection, however, as individuals refraining from behaving altruistically would gain (from the alloparental care of other altruists) without paying the costs of alloparental care. The population ecological conditions under which this group selection scenario could work as proposed are very restricted (see Barton and Clark, 1990), and it remains to be shown that these conditions prevail in the case described. 3. Nest Takeovers and Egg Stealing’

In at least eight fish families, nest takeovers occur in combination with subsequent care of the eggs that were already contained in these nests (i.e., probably fertilized by the previous owners; see Table 111 for a list of species). In some species this was observed in the natural, undisturbed situation (Etheostoma olmstedi, Constantz, 1979, 1985; Amphiprion clarkii, Yanagisawa and Ochi, 1986; Symphodirs ocellatus, Taborsky et al., 1987). or it was inferred from indirect evidence (Padogobius martensi, Bisazza et al., 1989a: Cottus gobio, Bisazza and Marconato, 1988). It could be induced by experimental removal of nest owners in Amphiprion clarkii (Yanagisawa and Ochi. 1986), in Ophiodon elongatus (Jewell, 1968). in Hemilepidotus hemilepidotus (DeMartini and Patten, 1973), and in Harpagifer bispinis (Daniels, 1978. 1979). In the last species, clutches

’

The term “nest’‘ is used here in its widest sense for any structure or shelter that serves for spawning and broodcare.

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

61

were originally cared for by females, but after the latter had been removed their place was taken invariably by males who continued to guard and care for the eggs, although with much less effort. Usually, large males displace small nest owners and take over their nests (Pimephalespromelas, Unger and Sargent, 1988; S . ocellatus, Taborsky et al., 1987; Padogobius martensi, Bisazza et al., 1989a; Cottus gobio, Bisazza and Marconato, 1988). In E. olmstedi, however, dominant males frequently abandon the breeding holes in which they had spawned and search for new ones. Subdominant males may then take over their nests and care for the eggs, which includes cleaning and guarding (Constantz, 1979). The eggs get so hard within a day that they cannot be consumed by the new nest owners (Constantz, 1985). The abandonment of clutches by dominant males in order to leave them to the care of smaller conspecifics is reminiscent of the “piracy” tactic observed by the largest males in S . tinca (see Section II,B,2). Why would males compete to take over the nests of others and care for the eggs fertilized by strange conspecifics? Basically, there are two possible benefits to this behavior, predation dilution and mate acquisition. A necessary prerequisite for both hypotheses is that males taking over a nest will subsequently spawn there. This is generally the case with the examples mentioned (see also Section II,B,l), with the exception of the Antartic plunder fish Harpagifer bispinis. The predation dilution hypothesis would predict that the survival probability of eggs fertilized by takeover males will increase from spreading the predation risk to a larger mass of eggs. To my knowledge, this has not yet been demonstrated by empirical data. The second hypothesis proposes that the presence of eggs, or the performance of broodcare behavior, would help to obtain females. In A . clarkii, for example, males taking over an anemone with a female pair up with this new mate and can subsequently spawn with her (Yanagisawa and Ochi, 1986), potentially until the end of her reproductive life. In this case, caring for the acquired brood instead of eating it may be a way of “paying for staying.” This would hold if females were more likely to reject males that were not willing to care for their clutches (Amphiprion females are larger than males and dominant), which remains to be shown. Among species with nest takeovers and alloparental broodcare, an attraction effect of the presence of eggs in a nest was demonstrated in the tessellated darter (E. olmstedi), in the river bullhead (Cottus gobio), and in fathead minnows (Pimephales promelas). In the darter, females seem to prefer nests with several eggs to those with none when deciding where to spawn, but they apparently avoid nests containing large masses of eggs (Constantz, 1985). The latter fact may be the reason why dominant males abandon their breeding shelter after some time to search for new ones

62

MICHAEL TABORSKY

with more bare surface to which eggs can be attached. The likelihood of leaving should additionally depend on the probability of subsequent takeover of the abandoned nest by an inferior male who will continue the broodcare. Female river bullheads chose males depending on the latter's sizes and/or egg presence in their nests (Marconato and Bisazza, 1986; Bisazza and Marconato, 1988). Female fathead minnows preferred to spawn with males who had eggs in their nests, when eggs were randomly assigned to nesting males (Unger and Sargent, 1988). This explains the strong male preference for nest sites with eggs in this species. In an experiment, males prefwred to take over nest sites with eggs that were already guarded by another male even when unguarded empty nests were available. However. they provided less care for adopted clutches than for their own ones (Sargent, 1989). Female preference for males that already guzrd eggs was also demonstrated in other species (e.g., Sikkel, 1988, 1989, and references therein), for example, in three-spined sticklebacks (Ridley and Rechten, 1981). Males of this species also show alloparental care, but instead of taking over a nest they steal eggs from neighbors and deposit them in their own nests (van den Assem, 1967; Wootton, 1971; Li and Owings, 1978b; Sargent and Gebler, 1980). The ultimate reason why females prefer males with eggs may be predation dilution, an increased broodcare motivation of males with large numbers of eggs, or a signal function of male quality (either in the sense of broodcare capabilities or in the sense of selfreinforcing feedback mechanisms acting on female preferences via indirect selection and genetic correlation; see Kirkpatrick, 1982, 1987; Bradbury and Gibson, 1983: Lande, 1987).These hypotheses are not mutually exclusive. Positive correlations of paternal effort and offspring survival with egg number have been repeatedly demonstrated in fishes (e.g., Pressley, 1981; Coleman et a / . , 1985; Sargent, 1988). It appears to be especially important for females to provide males with enough eggs to make their broodcare profitable. If the egg number remains below a critical size, clutches are frequently cannibalized by the nest owners and/or the nests are abandoned (e.g., also in three-spined sticklebacks, van den Assem, 1967; I compiled a list of 21 fish species belonging to 9 families in which small clutches or few eggs are abandoned; M. Taborsky, unpublished). Depending on certain brood size and time cost variables, this may be a decision for optimizing male reproductive success (Taborsky, 1 9 8 5 ~see ; also ten Cate and Taborsky. 1992, for an example with birds). In conclusion, egg raiding in sticklebacks may be a courtship strategy to attract females. as hypothesized by Rohwer (1978) and suggested by experimental results (Ridley and Rechten, 1981). Experiments of Jamieson and Colgan ( 1 989) challenged this conclusion, however, suggesting that

PARASITIC AND COOPERATIVE BEHAVIOR I N FISH REPRODUCTION

63

the presence of eggs would rather work via a “priming effect” on the courtship behavior of males. If this is its only functional significance, egg raiding would be nothing more than self-deception. Another study found, however, that the female preference for nests with eggs is at least partly a consequence of the eggs themselves (Goldschmidt et al., 1993). Nonadaptive hypotheses for egg stealing and subsequent alloparental care have also been proposed to explain such behavior in mouthbrooding Pseudocrenilabrus multicolor females. Mrowka (1987b) suggested that mistaken identity or motivational constraints could be responsible, but he also considered predation dilution and partial cannibalism (of the stolen eggs) as possible adaptive explanations. 4 . Broodcare Helpers

Some cichlids endemic to Lake Tanganyika show levels of cooperative behavior not previously suspected in fishes (Harcourt, 1988;Taborsky and Taborsky, 1993). In three Lamprologus and three Julidochromis species, families have been found that consist of the members of a pair and young of various ages that all share the same territory and the duties of defense, maintenance, and broodcare (Taborsky and Limberger, 1981; see Table 111). All species are monomorphic substrate brooders inhabiting the rocky sublittoral zone of the lake. In Lamprologus brichardi, an average of 7 to 8 young (7.5 5 1.42; 2 2 SE; N = 60 families) of up to four different size classes and a pair share a common shelter site in which eggs and larvae are tended (Fig. 8a). There are male and female helpers, and the largest helpers of most families are sexually mature. They leave their territories to become aggregation members before they are big enough to take over a breeding territory on their own, that is, there is a gap in body size between the largest helpers and the smallest breeders (Taborsky and Limberger, 1981).Aquarium experiments revealed that this transition from family to aggregation is caused by the expulsion of helpers by pair members, and not by an independent decision of the helpers (see the following). In the field, the sex ratio of mature helpers was skewed toward females (2 : 1, female :male), whereas the sex ratio of aggregation members of the same size range was exactly reverse (1 :2, female :male; Taborsky, 1984a). There is not as much information from the field on the other species with helpers. We collected field data on two other species (see Taborsky and Limberger, 1981), near Magara (Burundi), and I briefly describe their social structure and behavior. Julidochromis marlieri is a monogamous species breeding in narrow clefts, and we monitored the composition of 14 families. They consisted of a pair and usually of smaller members of up to four different size classes (Fig. 8b). The larger of these extrapair family members share in intra- and interspecific territory defense, and

x

MICHAEL TABORSKY

Lamprologus bnchardi

Julidochromis marlieri

b

0

1

2

3

4

"

0

no. offspringsize classes/farnily

1

2

3

4

no. offspring size classeslfamily

total no. families 5 '3 1 helper

11 1

4

11 0

4

1 1

-

Lo

2 helpers

E l < 1.5

2.4

3.4

4.4

-5 4

3

3 helpers

1

3

-

-

4

3

1

1

2

3

4

helper size classes

5.4 2 5.5

offspring size classes (cm) 1. (a) Top graph: number of families of the cooperatively breeding cichlid Lamprologus brichardi in which either no young or offspring of up to four different size classes were c ~ m s x l t ~ n o f i m m c l x pn ,% rm r o n t 111 ;n the a,,..u,rr.avvua,, ,OC"' &I.%,

fielrl Iy,.a-.X, t h n r e p""".."., nmhihlw

I."."

r\&n;natprl "L'b.'LUL'"

fcnm i n U . .

I.Y...

cniiivilpnt uyu.<.."...I

nnmhcr L.U.L."I.

of successive broods of the relevant territory owners; total sample size was 60 families). Bottom graph: average numbers of young simultaneously present in L . brichardi families in the field, separated for size. Only families having offspring of the relevant size classes were included in each bar. (b) Top graph: same information as in 8 (a), but for the cooperatively breeding cichlid Julidochromis marlieri (field data: total sample size was 13 families out of 14, in which offspring size classes were clearly distinct). Bottom graph: numbers of J . marlieri families in which either one, two, or three and more helpers were simultaneously present. separated for different relative size classes of helpers, 1 being the largest and 4 the smallest size class.

we know from aquarium observations that they take part in territory maintenance and direct broodcare as well, just as the helpers of L. brichardi do. There was usually only one helper of the largest size class per family (one exception had two, see Fig. 8b), which measured between about 4.5and 6.5 cm and could be approximately I year ofage, as estimated

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

65

from growth rates measured in the aquarium. It was often hard to estimate the actual offspring number of a family, especially of the smaller size classes, as small young remained hidden in their domestic cleft (the distance that young J. marlieri move from their shelter correlates positively with their size, as it does in L . brichardi). The average number of offspring, regardless of size, of the nine families that could be counted satisfactorily was 12.6 ? 8.2 ;( ? 2 SE). Yamagishi (1988) observed large helpers in five out of seven families. They were responsible for approximately one quarter of all interspecific territory defense of the associated families. Lamprologus savoryi is another species with helpers. In the field, we observed dense harems in which females defended very small subterritories, which were often shared by smaller fish of one, two, or rarely three different size classes. These family members joined in intra- and interspecific territory defense. They were particularly aggressive toward members of other subtemtories. It is not yet known whether they alsojoin in direct broodcare (e.g., egg cleaning, fanning, mouthing of larvae). A similar group structure was also reported for another population of this species by Kondo (1986) and Abe (1987). Helping in the other three Tanganyika cichlids listed in Table I11 has oniy been observed in the laboratory, where the behavior and social structure of L . pulcher resemble those of L . brichardi in every detail. The helping behavior of extrapair family members of J. ornafus and J. regani is very similar to that of J. marlieri. From the comparison of the results of field and laboratory observations in the two species for which this was possible (L. brichardi and J . marlieri) it seems very likely that the three species for which helping has as yet only been observed in the aquarium will behave similarly in the wild. Two Bornean fighting fish species, Betta brownorum and B . persephone, have been recently observed to have helpers. Older offspring tend to stay in parental territories and take part in defense against small interspecific intruders (Witte and Schmidt, 1992, aquarium observations only). There is a differential sharing of tasks between members of L . brichardi families (Taborsky et al., 1986). When different competitors and predators were experimentally introduced into family territories, males attacked large, heterospecifics more than their partners did. Both pair members exhibited more aggressive behavior against large intruders than helpers did, whereas the latter specialized in territory maintenance and direct broodcare. When small fishes entered the territory, the helpers showed more aggression against heterospecific intruders than female breeders did, and more displays against conspecific intruders than male breeders did. Among helpers, the large ones spent more time attacking intruders, while small ones performed more territory maintenance and broodcare. This

66

MICHAEL TABORSKY

specializing in different tasks was dependent not only on the type and size of intruders, but also on the stage of the breeding cycle. In the natural situation, results were similar to those of laboratory experiments with regard to the devotion of helpers to interspecific defense, and that of breeders to attacking conspecifics. Also, large helpers showed more territory defense than small ones did, and even more than the male territory owners (Taborsky and Limberger, 1981). See Limberger (1983) for additional data on task sharing in this species between members of a pair or harem. For an understanding of the evolutionary background of these helper systems we first need to know the relationship of helpers to pair members and to the eggs/fry produced in their territory. In L. brichardi, field and aquarium observations revealed that helpers are earlier offspring that greh up in the territories that they defend. They do not move between territories, but the average relatedness between them and the eggs and fry they tend (i.e.. their beneficiaries) declines with their age as a result of the natural replacement of breeders (Taborsky and Limberger, 1980). Helpers stay when one or both parents are replaced by strangers (e.g., because of mortality) and continue to share in the cleaning and fanning of eggs and larvae, removing sand from the breeding hole, removing snails (i.e.. egg predators), and defending the territory and breeding hole against conspecific and interspecific competitors and predators. The degree of relatedness ( r ) between the oldest helpers and eggs they care for was calculated to be approximately 0.25 (Taborsky and Limberger, 1981). The costs and benefits to helpers and pair members were studied by a combination of field observations and laboratory experiments (Taborsky, 1984a). As a measure of helpers’ costs. their growth rates were compared with those of family-independent aggregation members. The possible benefits that were tested ranged from those accruing to both helpers and breeders (mutualism) to those favoring helpers at the expense of breeders (parasitism). They included the advantage of rearing close kin, getting experience in broodcare, increasing the chances of territory takeover, increased survival probabilities in a protected territory, and parasitism of the breeders’ reproduction and cannibalism of eggs and larvae. Helpers grow more slowly than aggregation members of comparable sizes. They are heavier, however, than the latter, which suggests that they accumulate reserves while being protected in a safe territory so that they can pass the subsequent aggregation phase very quickly. The latter is probably a risky period (see the following), and the demonstrated size) weight relationship implies that it is a growth phase that primarily serves to attain sufficient size for the successful conquest of a breeding site. A comparison of growth between helpers differing in their hierarchical status

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

67

and same size territorial controls that were breeding themselves suggested that the delayed growth of helpers is due not only to their behavioral effort but also to their subordinate position within families (Taborsky, 1984a). Helpers did not affect egg or larvae survival in lab experiments, regardless of the presence of predators and competitors, nor did they influence the length of time intervals between subsequent clutches or the growth of pair members whom they assisted. Still, the effort females saved when having between one and three broodcare helpers increased their reproductive success because they were able to produce larger clutches (Taborsky, 1984a). This may select for a genetic disposition toward helping because of the relatedness between helpers and beneficiaries (see foregoing). Field observations revealed that pair members with helpers spent less time in their territories and more in nearby aggregations, in which they feed, than pair members lacking large helpers (Taborsky, 1984a). This may additionally raise the effect of helpers on parental fecundity and hence promote the evolution of helping behavior via natural selection. The importance of gaining broodcare experience was tested by a comparison of helpers and inexperienced controls during their own first broods. Experiments revealed that former helpers and naive controls did not differ in broodcare patterns, nor in clutch sizes, breeding intervals, growth rates, or relative and absolute breeding success. This holds for males and females alike (Taborsky, 1984a). Territory inheritance is another potential benefit to helpers that turned out to have no significancefor the cost/benefit ratios of helpers. This was demonstrated in part by a series of field experiments in which we removed one or both pair members of the 10 families containing the largest helpers out of our total sample of 60 families, to see whether these sexually mature helpers might take over the vacant position. This never happened, and instead it was always taken by a bigger aggregation member. In most of these cases the helpers stayed with the new breeders (Taborsky, 1984a). In the natural habitat of L . brichardiall suitable shelter sites are occupied by fish of various species. Helping could simply be viewed as paying the price for being allowed to stay in a territory defended by larger and more able hosts, and having permanent access to a shelter site. Field observations revealed that the predation pressure on helpers and aggregation members drops sharply when they reach a size of 4-4.5 cm, because their main predator, L . elongatus, is not able to cope with prey above that size. Laboratory experiments showed that the survival probabilities of helpers that are below that size are indeed greatly increased by living in a safe territory and by parental attacks on the predators, even though nonhelper controls had access to shelter sites in this setup that they would

68

MICHAEL TABORSKY

not have in the natural situation (Taborsky, 1984a). This explains the size distribution of family and aggregation members as found in the field (Taborsky and Limberger, 1981, Fig. 3), with a sudden change of the majority of fishes from family to aggregation status when 4-4.5 cm long. Helpers may also benefit from parasitizing the reproductive effort of the pair they stay with. They may eat eggs and larvae instead of tending them, or share in reproduction by simultaneous parasitic spawning (male helpers) and egg dumping (female helpers). Both types of parasitism are shown by large L. brichardi helpers (Taborsky, 1985a). Various experimental analyses suggested, however, that egg cannibalism and reproductive parasitism are probably only of secondary importance for the net balance of helpers and pair members, especially when helpers are below ca. 3.5 cm long and immature. The potential costs incurred to pair members by their helpers are very important, though, with regard to the time and mode of detachment of helpers from their families. When given the chance either to stay in a family as helper or to leave for an aggregation or even for their own breeding territory, a chance helpers would rarely, if ever, get in the natural situation, there was an unequivocal preference for staying (result of two experimental series, see Taborsky, 1985a). Separation from home territories was instead caused by the expulsion of helpers by breeders. This was confined to large helpers, that is, potential reproductive parasties, and to periods of little or no competitive pressure, that is, when the need for helpers was low. Helpers were reaccepted by pair members when the competitive pressure on the territory was experimentally increased by the introduction of conspecifics or heterospecifics, in 11 out of 13 experiments. It was only their own former helpers who were tolerated again and not strange fishes of the same size (Taborsky, 1985a). This points to the capability of breeders to recognize their helpers individually, which was experimentally proven by Hert 1985). Large helpers must decide whether to continue behaving cooperatively or try to cheat the breeders they stay with by cannibalism of their offspring and/or parasitic participation in reproduction. If caught cheating or during periods of increased cheating probability, that is, at spawning and during egg and larvae stages, parental aggression toward helpers and the latter’s chances to be expelled increase (see Taborsky, 1985a, Table I and p. 61, first paragraph). The payoff from these alternatives mainly depends on the potential costs of being expelled. As the predation risk suddenly drops after reaching a size of 4 cm (see earlier), the probability that helpers cheat should be expected to increase greatly at that point. This is exactly what happens. I developed a graphical model to find the optimal solution for helpers caught in the dilemma of having to choose whether they should continue cooperation or start cheating (Fig. 9). It shows that with regard

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

69

1.(

0.5

0

J 1

2

3

4

5

size of helper (cm)

FIG.9. The hypothetical payoff that Lamprologus brichardi helpers receive by helping compared to that for parasitizing the breeders’ reproduction. Abscissa: helpers’ sizes. Ordinate: relative benefit, except for variable a, for which the ordinate is the relative broodcare ability of helpers (1 = perfect ability). b = benefit derived from the effects of helping ( = increased fecundity of breeders times expected degree of relatedness between helpers and the breeders’ offspring). The function is assumed to be directly proportional to r, which declines with helpers’ age (see Taborsky and Limberger, 1981). a times b = benefit from helping mutiplied by actual ability to help at that size. c = benefit from reproductive parasitism, limited by attainment of maturity, which is size or age dependent, and by the costs derived from the production of gonads and germ cells. The latter prevent the curve from reaching 0.5. p = potential maximum payoff of helpers at their optimum, that is, through helping or cheating, whichever is better. Reproduced from Taborsky (1985a).

to immediate fitness consequences helpers should start cheating right after the predation risk has dropped. This suggests that sexual maturity should be reached at this size, which was confirmed by dissection and a check on the gonads of 20 helpers in the field. Thirteen out of 14 helpers 2 4 cm long could be sexed unequivocally: the gonads of one individual (4.4 cm long) were not yet clearly developed. Only 2 out of 6 helpers between 3 and 4 cm long had well-developed gonads already (both were males). The onset of maturity in helpers may hence be phenotypically controlled by the behavior of breeders (i.e., through dominance and punishment by expulsion when helpers try to parasitize their reproduction; see Taborsky, 1985a).

70

MICHAEL TABORSKY

A *strongphenotypic control of the behavior of L . brichardi helpers by that of breeders was also demonstrated by von Siemens (1990). Results of a series of experiments showed that the submissive status of helpers is the most important factor in causing helpers to clean eggs instead of eating them. When allowed to become dominant, helpers switched to cannibalistic behavior. They could be converted to become egg cleaners again, when combined with dominant conspecifics. The behavior of the dominant fishes directed toward eggs also influenced the probability of egg cannibalism of helpers. Egg cleaning by the dominant individuals increased the likelihood that helpers cleaned eggs as well, and cannibalism by dominant individuals caused helpers also to feed on eggs (von Siemens, 1990). Further experiments revealed that the influence of an egg-cleaning, dominant individual that acted as a “model” was size dependent; large potential helpers = 4.4 cm) cannibalized eggs regardless of the model’s behavior (Ladich and Taborsky, 1991). This also points to the potential costs that breeders may suffer from large helpers. The relationship between helpers and breeders in L . brichardi is strongly dependent on age and size of the helpers. Small helpers are highly related to the breeders that they assist. Their tolerance in the territory by breeders dnd their helping behavior is probably due to kin selection. Helpers support close kin and parental tolerance of helpers in their territories may be viewed as prolonged broodcare, which adds to the benefit of receiving help. When helpers have grown up the breeder/helper relationship is characterized by a high degree of reciprocity. The breeders tolerate their helpers at the risk of reproductive parasitism and accept the cost of space competition (see Taborsky, 1985a, p. 62). if the helpers pay by sharing in the breeders’ duties. The helpers will invest in the breeders’ offspring at the cost of reduced growth in order to be protected by the breeders’ terntonality and by access to a shelter site. The less important that these potential benefits become for helpers. because they have grown beyond the size range of prey that their major predators can handle, the more important the helpers’ costs will become in comparison, which should increase their propensity to cheat. This will in turn raise the costs for breeders to tolerate helpers, which finally leads to the expulsion of helpers, especially when their help is dispensable, that is, in low-competition situations. Three findings strongly support the conclusion that helpers are paying for staying. (i) Young stay and help indiscriminately when one or both breeders are replaced (Taborsky and Limberger, 1981), despite their presumed capacity to recognize individuals (Hert, 1985). (ii) Helpers stay in the territory as long as they can, even when their alternative options are experimentally improved far beyond those found in the natural situation;

(x

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

71

this emphasizes the essential importance of a safe territory. (iii) Helpers above a certain size are tolerated by breeders only when they are really needed (Taborsky, 1985a). IV. GENERALCHAPTERDISCUSSION This chapter deals with the different forms of competition that characterize fish reproduction and is limited to the intrasexual level, that is, competition and cooperation between pair members are not considered. The main topics discussed are competition for (i) access to mates (Sections II,B, III,A, III,B,l, III,B,2, 111,C,3), (ii) the production of zygotes (Sections II,C, III,A, III,B,l, III,B,2), and (iii) the effort that is put into raising offspring (Sections II,D, III,B,3, 111,C). Competition may lead to the parasitic exploitation of investment by individuals employing alternative tactics or, at the other end of the scale, to the seemingly altruistic behavior of individuals supporting others without obvious direct reproductive benefit to themselves. However, these examples could be viewed as extremes of a continuum of social interactions found within the context of reproductive competition. Most of the interactions discussed occur between males, which is partly due to the differences between the sexes in prezygotic investment and reproductive rate (see Section 11,D). Yet this may be a biased impression that is caused by conventions in research philosophy, and more could perhaps be gained by looking into female strategies (see, e.g., van den Berghe et af., 1989; Warner, 1990; Ahnesjo et af., 1993). On the level of competition for mates, there are various ways in which males may improve their chances of being selected by females. These include behavioral and morphological features (e.g., courtship, size, color), which are all subject to intersexual selection through the action of female choice. Another way to improve access to mates is by defense, that is, the exclusion of competitors either directly from mates or from spawning sites. Both types of effort may be exploited by male competitors, who temporarily or permanently take over nests, or intercept and steal females that spawn with another male. These social parasites may either belong to the same type of male as the parasitized individuals, for example, territorial neighbors that intrude to recruit females, or they may be competitively inferior males that exploit opportunities without engaging in aggressive interaction with their victims. Competing males may also combine their effort and jointly defend a spawning territory, or jointly court or build nests. These associations may partly regulate the competition among participants and improve their position against other competitors. Male intrasexual competition for the production of zygotes (i.e., sperm

72

MICHAEL TABORSKY

competition; e.g., Smith, 1984; Parker, 1990b; Birkhead and Mgller, 1991) is extremely widespread in fishes (see Table I). This may be partly because fertilization in fishes is usually external, which gives males ample opportunities to attempt to fertilize eggs simirltaneously with other competitors. The timing of fertilization is exactly predictable for these males (i.e., immediately after egg release), which is in contrast to most systems with internal fertilization. It might be argued that the possibility of flooding the spawning site with sperm before a female spawns there could bias the chances of fertilization toward nest owners. This would curtail to some extent the argument that the potentially similar opportunities for bourgeois and parasitic males are a prime mechanism in promoting male reproductive parasitism in fishes. However, the chances are limited that fertilization success can be manipulated by such anticipated sperm release, because of (i) the dilution and dispersal of sperm in water, (ii) the short functional life span of sperm (see Childers, 1967), and (iii) the superior access to eggs of sperm released at spawning. In addition to the predictability of egg release, external fertilization makes it difficult for male fishes to monopolize access to fertilizable eggs. The wide distribution in fishes of simultaneous sperm release by more than one male at spawning is probably a consequence of both factors, perspicuous timing of fertilization and the restricted potential to monopolize access to unfertilized eggs (this latter feature limits the ability of the satellite threshold model to explain fish alternative mating systems; Waltz, 1982). There may be similar costs and similar fertilization probabilities for all participants of simultaneous spawning events, as suggested by group spawning patterns. Fertilization could follow a “fair raffle,” with the success of a given male approximating the proportion of his sperm in the total pool of sperm available (Parker, 1990a).The processes of competition and fertilization involved in these group spawning systems, however, are poorly understood. Most often, there are males investing in some way or another to obtain preferential access to fertilizable eggs. This may be accomplished by morphological structures (e.g., humps, color) and behavior (e.g., courtship) influencing female choice (see earlier), or the provision and defense of spawning or breeding sites. Often, this effort is exploited by other bourgeois males who take the occasion to steal fertilizations in it neighbor’s territory. Of even greater interest is the fact that often competitors that are specialized in alternative mating patterns parasitize bourgeois males. Again, this specialization may be morphological (e.g., size, color) or behavioral (e.g., sneaking toward a spawning site or streaking toward a spawning female) and often serves to camouflage the parasitic individual from the bourgeois male (see the long list of species with female

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

73

mimics in Table 11). There are cases with only a minor proportion of kleptogamic males and others in which the majority of males are parasitic. This raises the question of frequency dependence in this interaction of producing and scrounging strategies (see Ayala and Campbell, 1974; Barnard and Sibly, 1981; Barnard, 1984; Ryan e? al., 1992). How are they maintained within a population? In bluegill sunfishes (Lepomis macrochirus), the proportion of males who take the parasitic pathway early in life (i.e., when 2 years of age) corresponds roughly to the estimated proportion of eggs fertilized by all parasitic males of the population (Gross, 1982). This would suggest that lifetime reproductive payoff for both types of males, bourgeois and parasitic, is at an equilibrium (Gross and Charnov, 1980; see also Gross, 1991) and represents an evolutionarily stable mixture of strategies (Maynard Smith, 1982; but see Ryan et al., 1992). Dominey (1980) found a different pattern of age-related reproductive tactics in another population, however, and the actual fertilization success of the respective males has not been measured in this species (see, however, Gross and Dueck, 1989). Gross (1982) attributed the fertilization of all eggs spawned to these parasites that intruded at spawning. This probably resulted in a substantial overestimate of the success of parasitic individuals (see Section II,C,7,c), which would weaken the former argument. Aside from the need for a more reliable measure of male success rates (see Shuster, 1989), a demonstration of equal morph fitnesses does not necessarily imply that frequencydependent mating success is the mechanism by which the equilibrium of fitnesses is established. Additional evidence is needed on the fitness effects of changes in morph frequency (Ryan e? al., 1990), and it will be most rewarding to unravel the nature of the developmental switch, genetic or otherwise, in such systems (see Maynard Smith, 1982) to find the conditions controlling them. A similar scenario was derived from data on coho salmon (Oncorhynchus kisutch). Small parasitic and large bourgeois males were suggested to have equal lifetime reproductive successes, and negatively frequencydependent disruptive selection was assumed to be responsible for the stability of these alternative strategies (Gross, 1984, 1985). Here, the information used lo estimate relative fertilization success of males was distance from females during spawning. This was combined with information on mortality during the ocean phase and on the time period that males stay on the breeding grounds to render relative lifetime fitness estimates. Obviously, there are a number of critical assumptions involved with this approach, and the conclusions drawn may be premature (see also Section II,C,7,d for an alternative explanation of the existence of two distinctly different size classes of reproductive males). A study on

74

MICHAEL TABORSKY

Atlantic salmon (Salmo solar) suggested that there might be multiple evolutionarily stable equilibria between the proportions of males maturing at different ages, which is closely related to different male reproductive strategies (Myers, 1986). Here also, good information on fertilization success of different male types is lacking, therefore it remains unclear whether the bourgeois and parasitic tactics are at an equilibrium. These studies on sunfishes and salmon illustrate, however, what kind of data would be needed to solve the question of why more than one reproductive tactic so often exists in male fishes. There is an alternative way t o explain the occurrence of parasitic males within this evolutionary framework. Depending on genetic quality and developmental history, there is often a tremendous variation of male features related to resource holding potential (e.g.. body size, Taborsky et a / . , 1987). This may preclude the smaller individuals from competing successfully for resources with better equipped conspecifics. They have to make the best of a bad job (e.g., Dawkins. 1980; Arak, 1984; Waltz and Wolf, 1984; Koprowski. 1993). which may still lead to evolutionarily stable strategies within each phenotype (Parker. 1982). At present. information on most species is insufficient to test whether the occurrence of alternative mating strategies is a matter of equal payoffs for all participants, or whether certain participants are limited to suboptimal solutions (Ryan et a f . , 1992). I would suggest, however, that systems that include lifelong, fixed strategies represent mixed evolutionarily stable states, whereas when the male tactic depends on circumstances (e.g., ontogenetic stages), it is more likely that some tactic is the result of males making the best of a bad job. The first case (fixed strategies) is likely to apply only to a small number of species, whereas in the majority of examples studied so far the reproductive tactics chosen by males seem to depend greatly on conditions (see Table 11, and Section II,C,7,d; for more extensive discussions see Rubenstein, 1980: Arak. 1984; Caro and Bateson, 1986). As has been shown for other groups of animals (e.g., Arak, 1983), female fishes apparently prefer t o mate with bourgeois males. This may exert additional pressure on parasitic males to conceal their presence and potential to attempt fertilizations. for example. by hiding, sneaking, or female mimicry. In the Mediterranean ocellated wrasse, however, females appear to be attracted to nests by the presence of satellite males, despite the fact that they avoid spawning with them (see Section II,C,8). Satellite males were also assumed to influence female preference in pupfish, in bluegill sunfish, and in fallfish minnows (Kodric-Brown, 1977; Dominey, 1981; Ross. 1983). This appears to be a specific variation of “female copying” (see Bradbury and Gibson, 1983; Dugatkin, 1992). In the wrasse,

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

75

females may gain information about the likelihood that the nest will be cared for until hatching. In many other species, females can assess this probability from the presence and number of eggs, but in the thick algae mats provided by bourgeois males of the ocellated wrasse this seems nearly impossible. Satellite males are a reliable cue for the presence of eggs, as they only associate with a nest when females have spawned there already. This interaction between bourgeois male success and satellite behavior illustrates how an initially parasitic system may be stabilized by a substantial degree of reciprocity (i.e., the development toward tolerance of a satellite, see Section III,A,2) via the action of female choice. At first glance this strict preference of female fishes for spawning only with a bourgeois male may be surprising, in the light of recent evidence that females of other groups of animals often choose to mate with several mates in succession (e.g., Birkhead and Mgller, 1991). Potential advantages of multiple inseminations are an increase in the genetic diversity of offspring and a higher fertilization probability. However, a characteristic feature of the spawning patterns of many fish species is that females do not release all their eggs at a time, but rather apportion them between different nests. This allows for the fertilization of their eggs by a number of different bourgeois males, resulting in greater genetic variability of their offspring. At the same time they may choose among these bourgeois males for specific characters related to heritable genetic quality and/or parental abilities. Also, it may benefit females to provide broodcaring males with a high certainty of paternity to provoke their full parental commitment (Knowlton and Greenwell, 1984), which would select against spawning with parasitic males. However, in some species the need to increase fertilization probabilities may override the foregoing factors. For example, species spawning in swift streams may be subject to this problem. The stunning spawning patterns of suckers (e.g., unconditional trio spawning) may have evolved to increase female reproductive success by improving fertilization probabilities. Another male strategy in the competition for the production of zygotes is to raise sperm production. This investment cannot be parasitized upon by other males. Gonadal investment may result in larger testis size and/ or in higher rates of sperm production. To my knowledge, only the first of these two possibilities has been treated to date in any detail (see Fig. 3). Usually, parasitic males produce larger gonads than bourgeois males, when testis mass is related to body mass (see Table 11). In the plainfin midshipman (Porichthys notatus), for example, the gonad/body weight ratio of parasitic males exceeds that of bourgeois males by a factor of 9 (Bass and Andersen, 1991). These examples illustrate the importance of sperm competition and can be viewed as resulting from an evolutionary

76

MICHAEL TABORSKY

arms race between reproductive competitors. Such competition may also lead to differences between species, that is, the gonads of bourgeois males of species with parasitic spawning are larger than those of males belonging to species without alternative spawning mechanisms (e.g., Mohr, 1986). The huge gonadal investment especially of parasitic males may greatly influence their codbenefit ratios and should be considered in future studies on the relative payoffs of alternative male reproductive patterns. Frequently, satellite males are tolerated to some extent by bourgeois male conspecifics. These cases appear to be halfway on the scale between purely parasitic intrasexual competition and cooperative associations. There are various transitions between systems in which satellite males lurk at the edge of a temtory and those in which they are accepted right in its center, in or at nests. These satellites are virtually always competitors for fertilization, therefore it needs to be explained why they are tolerated. Their expulsion may be either physically not possible for the bourgeois male, or too expensive in comparison to the advantage derived from this effort, or there may be some reciprocal benefit from their presence and/ or behavior. This benefit may be related to their mere presence, for example, in anemonefish as an insurance to have a replacement partner if the animal‘s own mate disappears, or in the ocellated wrasse to attract females (see Section III,A,2 for a discussion of these possibilities). Or there may be cooperative behavior between bourgeois and satellite males. This does not necessarily mean that males benefit from each others’ behavioral effort, as was revealed by field experiments in the ocellated wrasse (Section I K B , 1). Joint spawning in suckers might result from mutual benefits of male competitors and is an intriguing behavior with regard to the transition between conflict and cooperation (see the discussion in Section III,B,2). It is beneficial for parents to adopt behaviors that reduce the costs of broodcare (again, I do not consider intru-pair conflict here). This may cause behavioral strategies that differ from those that result from the competition for mates or fertilizations, partly because the broodcare situation mainly involves interactions between relatives, that is, parasitic behavior is not as profitable (see discussion of this argument in Section 111). Additionally, broodcare in fishes is often “sharable” (Wittenberger, 1979) or “nondepreciable” (Clutton-Brock, 1991), as it merely amounts to propagule protection; for example, the inclusion of strange offspring into a brood does not necessarily cause extra costs to parents. We have seen, however, that there is still a fair amount of conflict inherent even in the most seemingly altruistic associations between conspecifics, especially as there may be reproductive competition between the participants on top of joint benefits from cooperative behavior (Section 111,C). Adoptions of eggs or young, for example, may be beneficial to donors

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

77

and stepparents (mutualism), or to one party only (commensalism), sometimes at the expense of the other (parasitism; see the discussion in Sections III,C,l to 111,C,3). They may occur between members of different species, which makes them more obvious, or between conspecifics, which has long been unnoticed in natural situations and is most likely still greatly underestimated because of detection problems (see Anderson, 1984). The benefits involved with alloparental care are clear for the recipients (if there are any), that is, a liberation from broodcare duties and hence an increase in the residual reproductive value of these fish. The potential benefits to alloparents are more varied, ranging from predation dilution of their own offspring to mate attraction (see also Rohwer, 1986, for a discussion of a mate-attracting function of adoptions in birds). The benefits to cichlid broodcare helpers are closely linked to patterns of relatedness within groups (families) and may greatly change with age. The codbenefit analysis of an African cichlid with broodcare helpers showed that predation pressure is the key factor in causing the system to switch from kinshipbased cooperation to a reciprocal association in which helpers are paying for being allowed to stay in the territory and being protected by its owners (Section 111,C,4). This example may illustrate ageneral difference between fish alloparental care systems and those of other taxa, for example, insects, birds, or mammals. In fish broodcare, protection from predation is probably the factor of greatest importance, whereas in the other groups the provision of food appears to be at least as fundamental (Clutton-Brock, 1991; see also Heinsohn, 1991). Why are there so many different reproductive tactics in fishes? Fishes are unique in their physiological and morphological plasticity. Most species show indeterminate growth, that is, there is great variation in size among reproductive competitors. Fishes are faced with the decision to reproduce or grow; they could invest in current reproduction or delay it and invest in growth for future reproductive benefits. Because of lifelong growth this problem is often not confined to a brief period in early ontogeny but is a repeated or permanent dilemma. In addition, sex determination is flexible, ranging from gonochorism to sequential or simultaneous hermaphroditism. And because fertilization is external it is often impossible for males to monopolize access to females or even a single fertilization. All of this predestines fishes to a high level of flexibility in their reproductive behavior and may explain the wealth of alternative reproductive behaviors in this group as compared to the other (in this respect) most heavily studied groups, that is, birds and mammals, and perhpas even insects. The amazing behavioral flexibility involved in fish reproduction is perhaps best illustrated by species with three or more male reproductive strategies. These examples belong to a variety of different fish families,

78

MICHAEL TABORSKY

that is, pupfishes (Cvprinodon pecosensis, Kodric-Brown, 1977, 1981, 198h), sunfishes (Lepomis macrochirus, Dominey, 1981 ; Gross, 1982), cichlids (Pseudocrenilabrus pilander, Chan. 1987; Chan and Ribbink , 1990; Lamprologus callipterus, Sato, 1988;Peivicachromis pulcher, Martin and Taborsky, 1993), wrasses (Symphodus roissafi, Soljan, 1931; Fiedler, 1964; Lejeune, 1985; S . tinca, Lejeune, 1985: Warner and Lejeune, 1985; Wernerus, 1989; S . oceUatus, Taborsky et af., 1987), and blennies (Parablennius sanguinolentus, Santos, 1985, 1986; Santos and Almada, 1988). The wide taxonomic distribution of highly variable male reproductive behavior might point to the importance of ecological factors and/or of very general biological features of fishes (see earlier). Phylogeny appears to be of great importance, too, as it sets the stage for many characteristics of reproductive competition. The great potential for broodcare in cichlids, for example, leads to many different social phenomena that are closely related to the investment in offspring. A proper comparative analysis of alternative mating patterns in fishes would provide valuable insight into the importance of phylogenetic as opposed to ecological factors in leading to different reproductive strategies (see Harvey and Pagel, 1991). Fishes are an exquisite group in which to study ultimate and proximate causes of social behavior as a result of their behavioral variability. Other characteristics make this group additionally suitable for studies on sociality (Taborsky. 1987), for example, there are enormous practical advantages. Many species, including those with social structures matching the most complex organizations known among nonhuman vertebrates, are of small size and have small home ranges. They are easy to handle and experiment with, both in the field and in the laboratory. Because of very moderate spatial requirements they can be observed and manipulated under seminatural conditions while showing their full behavioral repertoire and acting in a way that is indistinguishable from the natural situation in many respects (see Taborsky, 1984a, for a discussion). Many have a fast brood succession and short generation times. The fertilization process is usually very obvious, which allows for crucial behavioral observations on male and female roles and on alternative mating tactics. 1 would suggest that if comparable effort had been put into studies on the social systems of fish as has, for example, been devoted to birds, we would currently have H more complete understanding of functions and mechanisms of complex sexual and social behavior. Much of our information on reproductive behavior of fishes is still superficial. There is need for more detailed studies of specific model systems that are characterized by great complexity and plasticity of behavioral tactics. In order to understand the evolutionary background of these

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

79

systems, we need to address questions concerning, for example, the occurrence and prevention of reciprocity and defection in cooperative systems, and how this might be stabilized by natural selection, and the lifetime reproductive success of individuals performing alternative reproductive strategies could be measured to reveal the evolutionary background of these systems (see, e.g., Shuster and Wade, 1991). Future research should certainly concentrate on the “most interesting” fish families with regard to social behavior, asjudged from our present knowledge; but new, exploratory studies on less understood groups should also be encouraged as this may well lead to the discovery of novel reproductive strategies and provide a better perspective on the systems with which we are already familiar.

V. SUMMARY This chapter reviews our current knowledge of competition in fish reproduction (excluding conflict between members of a pair). The species within this taxonomic group exhibit an impressive range of reproductive tactics. The types of competitive interactions observed range from overt conflict and sexual parasitism at one extreme to cooperation and mutualism at the other. Most of the examples of competition cited occur between males. Males can exploit the effort of other males to attempt to gain exclusive access to females. This occurs through parasitic behavior such as nest takeover, piracy, interception, or female theft. Another form of competition for the production of zygotes involves the more or less simultaneous release of sperm of different males at spawning (sperm competition). This occurs between neighboring bourgeois males (i.e., males that have invested in structures to improve their access to mates, such as territories, nest sites, body coloration, or other morphological structures), but also between bourgeois and parasitic males. The latter may also show morphological, physiological, and behavioral specializations. Simultaneous parasitic spawning (SPS) has been described for 123 fish species belonging to 24 families. Frequently, males that are specialized in parasitic spawning resemble females in size, color, and behavior (female mimics). Their testis/ body mass ratios are frequently higher than those of bourgeois males, and it has been demonstrated that their fertilization success may depend on relative proximity at spawning. In some species, the majority of males specialize in parasitic reproductive tactics. In some species a genetic predisposition was found for the reproductive tactics of males, whereas in many others these tactics are conditionally expressed. Females apparently

80

MICHAEL TABORSKY

prefer to spawn with bourgeois males in most cases, but they may use the presence of subordinate males as a cue for broodcare quality. Competition may lead to associations. Often, subordinate satellite males associate with bourgeois males for some time during the reproductive period. Tolerance of these satellites might be a matter of either defense economy or advantages to bourgeois males derived from satellite presence and/or behavior. Such associations have been observed during the defense of spawning territories, courtship, or nest building. There may even be joint spawning that does not involve any aggressive behavior between participants, as is the case in unconditional trio spawning in some suckers. In a functional sense, this appears to differ greatly from “simultaneous parasitic spawning. Within the context of broodcare, competition may result in the “donation” of offspring to the care of others (e.g.. egg dumping, young dumping) or the usurpation of others’ offspring (e.g., kidnapping). Joint broodcare and aUoparental care may result from various forms of brood mixing (e.g., f a y conflux, independent offspring inclusion) that occur on both intraand interspecific levels, and from nest takeovers or egg stealing. The costs and benefits to donors and stepparents vary greatly, ranging from parasitic through commensal, to mutualistic conditions. Possible benefits to stepparents include a predation dilution effect favoring their own offspring, and increased mate attraction. Females often choose to spawn with males who already have eggs to care for, because this apparently serves as a cue for an increased survival probability of their own eggs in these nests. A special form of cooperative behavior is exemplified by broodcare helpers, which may share in all parental duties and in territory defense and maintenance. Their costs from this behavior include reduced growth, and their benefits result from raising close kin, from being protected by breeders in a safe territory, and from chances to parasitize the reproductive effort of the territory owners. The association of breeders and helpers in an African cichlid changes from a cooperative system among close kin to a reciprocal situation in which helpers are paying for staying, that is, they may parasitize the breeders’ reproductive effort and are only tolerated by breeders when their help is needed. This demonstrates that the nature of an interaction can vary with conditions between forms of mutualism and parasitism. Possible reasons for the great variability of reproductive strategies of fishes are discussed. I argue that morphological and physiological characteristics, such as indeterminate growth, external fertilization, and the versatile mechanisms of sex determination, create the potential for this wealth of solutions to reproductive success. Different strategies have been found to result from behavioral plasticity or, at least partly, from genetic ”

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

81

variability. Several reproductive strategies may exist even within single species. It is argued that this variability, combined with practical advantages, renders fishes a model group for studies of the ultimate and proximate causes of social behavior. Acknowledgments

I thank Peter Wirtz for very generous help with literature, Chris Carbone, Manfred Milinski, Peter Slater, and Peter Wirtz for correcting earlier versions of the manuscript, Barbara Taborsky and especially Chris Carbone for numerous discussions of the topic and for preparing a graph, Brigitte Rauschl for help with the collection of literature, Ingrid Seitter and Barbara Taborsky for help with typing the references, and the editors for their patience. I was supported by a grant from the Austrian Fonds zur Forderung der wissenschaflichen Forschung (P-7394 BIO) while preparing this study.

References

Abe, T. (1987). Population structure of Lamprologus sauoryi. In “Ecological and Limnological Study on Lake Tanganyika and Its Adjacent Regions. IV.” (H. Kawanabe and M. Nagoshi, eds.), pp. 36-37. Kyoto University, Kyoto, Japan. Ahnesjo, I., Vincent, A., Alatalo, R., Halliday, T., and Sutherland, W. .I. (1993). The role of females influencing mating patterns. Behau. Ecol. 4, 187-189. Albrecht, H. (1968). Freiwasserbeobachtungen an Tilapien (Pisces, Cichlidae) in Ostafrika. Z. Tierpsychol. 25, 377-394. Albrecht, H. (1969). Behaviour of four species of Atlantic damselfishes from Columbia, South America (Abudefdufsaxatilis, A . taurus, Chromis multilineata, C . cyanea; Pisces: Pomacentridae). Z . Tierpsychol. 26, 662-676. A h , G. (1959). Connection between maturity, size and age in fishes. Rep. Inst. Freshwater Res. Drottningholm 40, 5-145. Anderson, M. (1984). Brood parasitism within species. In “Producers and Scroungers” (C. J. Barnard, ed.), pp. 195-228. Croom Helm, London. Arak, A. (1983). Male-male competition and mate choice in anuran amphibians. I n “Mate Choice” (P. Bateson, ed.), pp. 181-210. Cambridge Univ. Press, Cambridge. Arak, A. (1984). Sneaky breeders. I n “Producers and Scroungers” (C. J. Barnard, ed.), pp. 154-194. Croom Helm, London. Austad, S. N. (1984). A classification of alternative reproductive behaviors and methods of field-testing ESS models. Am. 2001.24, 309-319. Avila, V. L. (1978). A field study of nesting behaviour of male bluegill sunfish (Lepomis macrochirus Rafinesque). Am. Midl. Nut. %(I), 195-206. Ayala, F. J., and Campbell, C. A. (1974). Frequency-dependent selection. Annu. Reu. Ecol. Syst. 5, 115-138. Ayling, A. M. (1980). Hybridization in the genus Pseudolabrus (Labridae). Copeia, 1980, 176-180. Baba, R., Nagata, Y., and Yamagishi, S. (1990). Brood parasitism and egg robbing among three freshwater fishes. Anim. Behau. 40(4), 776-778. Baerends, G. P., Brouwer, R., and Waterbolk, H. T. (1955). Ethological studies on Lebistes reficuLatus (Peters). Behaviour 8, 249-334.

82

MICHAEL. TABORSKY

Baerends, G. P.. and Baerends van Roon. J. M. (1950). An introduction to the study of the ethology of cichlid fishes. Behuuiorrr (Strppl.) 1, 1-242. Bailey, J. K . . Saunders. R. L.. and Buzeta. M . I. (1980). Influence of parental smolt age and sea age on growth and smolting of hatchery reared Atlantic salmon ( S d m o sulur). C a n . J . Fish. Aqrrut. Sci. 37, 1379-1386. Barlow, G. W. (1961). Social behavior of the desert pupfish. Cyprinodon mucrr/urirr.s, in the field and in the aquarium. A m . Midl. N u t . 65, 339-359. Barlow, G. W. { 1967). Social behavior of a South American leaf fish. Polycentrrrs schomburgkii. with an account of recurring pseudofemale behavior. Am. Midl. N u t . 78, 215-234. Barnard. C. J . , ed. (1984). “Producers and Scroungers.” Croom Helm. London. Barnard, C . J.. and Sibly. R. M. (1981). Producers and scroungers: A general model and its applicability to captive flocks of house sparrows. Anim. Behau. 29, 543-550. Barton. N.. and Clark. A. (1990). Population structure and process in evolution. In ”Population Biology” ( K . Wohrmann and S. K . Jain. eds.). pp. 115-173. Springer-Verlag, Berlin. Bass, A . (1992). Dimorphic male brains and alternative reproductive tactics in a vocalizing fish. Trend3 Nertrosci. 15, 139-145. Bass, A . H.. and Andersen. K. (1991). Inter- and intrasexual dimorphisms in vocal control system of a teleost fish: motor axon number and site. Bruin, Behau. Euol. 37,204-214. Baylis. J . R. t 1974). The behavior and ecology of Herotilapin multispinosu (Teleostei. Cichlidae). Z. Tierpsycho/. 34, 115-146. Baylis. J. R. (1981). The evolution of parental care in fishes. with reference to Darwin’s rule of male sexual selection. Environ. B i d . Fishes 6 , 223-251. Beauchamp. D. A. (1990). Movements, habitat use, and spawning strategies of arctic grayling in a subalpine lake tributary. Norrhhsest Sci. 64, 195-207. Bell, L. J. (1983). Aspects of the reproductive biology of the wrasse, Cirr/zilubru.stemminckii, at Miyake-Jima, Japan. Jpn. J . Ichfhyol. 30, 158-167. Bietz. B. F. (1980). Adaptive significance of territorial aggregation by male longear sunfish. Am. Zoo/. u)(4). 826. Bietz. B. F. (1981). Habitat availability. social attraction and nest distribution patterns in longear sunfish. iLrpomis megalotis peltastes). Enuiron. Biol. Fish. 6, 193-200. Birkhead, T.. and Mdler. A . P. (1991). ”Sperm Competition in Birds.“ Academic Press. L.ondon. Bisazza, A , . and Marconato. A. (1988). Female mate choice. male-male competition and parental care in the river bullhead, Cotfits gobio (Pisces, Cottidae). Anim. Behau. 36, 1352- 1360. Bisazza. A,. Marconato. A,, and Marin. G . (1989a). Male competition and female choice in Padogobius murtensi (Pisces. Gobiidae). Aninf. Behuu. 38(3). 406-413. Bisaua, A,. Marconato. A,. and Marin. A. (1989b). Male mate preferences in the rnosquitofish Gamhrtsia holbrooki. Ethology 83, 335-343. Bowman, M. L. ( 1979).Life history of the black redhorse. Moxosromu dirquesnei (Lesueur). in Missouri. Truns. Am. Fish. Soc. 99, 546-559. Bradbury, W. J.. and Gibson, R. M. (1983). Leks and mate choice. In “Mate Choice” (P. Bateson, ed.). pp. 109-138. Cambridge Univ. Press, Cambridge. Brantley. R. K . . and Bass. A. H. (1991). Secondary sex characters in a vocalizing fish: Intra- and intersexual dimorphism and role of androgens. I n “Proc. Fourth Intl. Symp.: Reproduction and Physiology of Fish” (A. P. Scott. J . P. Sumpter, D. E. K i m . and M. S . Rolfe. eds.). pp. 197-199. FishSymp 91. Sheffield. Brawn. V. M. (1961). Reproductive behaviour of the cod (Gudrrs ~ u l l u r i uL.~ ). Behauiour 18, 177-198.

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

83

Breder, C. M., and Rosen, D. E. (1966). “Modes of Reproduction in Fishes.” Natural History Press, Garden City, NY. Brichard, P. (1979). Unusual behaviors in Lake Tanganyika cichlids. Buntbarsche Bull. 74, 10-12.

Burchard, J. E., Jr. (1965). Family structure in the dwarf cichlid Apistogramma trifasciatum Eigenmann and Kennedy. Z. Tierpsychol. 22(2), 150-162. Burchard, J. E., Jr. (1967). The family Cichlidae. In “Fishand Fisheries of Northern Nigeria” (W. Reed, ed.), pp. 128-144. Min. Agric. Northern Nigeria. Cameron, W. M. (1940). A preliminary investigation of the natural spawning, incubation and alevinage of the pink salmon (Oncorhynchus gorbuscha). Fish. Res. Board Can., Manuscr. Rep. 229, 1-29. Carlisle, T. R. (1985). Parental response to brood size in a cichlid fish. Anim. Behau. 33, 234-238.

Caro, T. M., and Bateson, P. (1986). Organization and ontogeny of alternative tactics. Anim. Behau. 34, 1483-1499. Carr, M. H. (1942). The breeding habits, embryology and larval development of the largemouth black bass in Florida. Proc. N . Engl. Zool. Club 20, 43-77. Carr, M. H. (1946). Notes on the breeding habits of the eastern stumpknocker, Lepomis punctatus punctatus (Cuvier). J . Florida Acad. Sci. 101-106. Chan, T.-Y. (1987). “The Role of Male Competition and Female Choice in the Mating Success of a Lek-Breeding Southern African Cichlid Fish Pseudocrenilabrus philander (Pisces: Cichlidae).” Ph.D. Thesis. Rhodes University, Grahamstown, South Africa. Chan, T.-Y., and Ribbink, A. J. R. (1990). Alternative reproductive behaviour in fishes with particular reference to Lepomis macrochira and Pseudocrenilabrus philander. Environ. Biol. Fishes 28, 249-256. Chebanov, N. A. (1980). Spawning behavior of the pink salmon Oncorhynchus gorbuscha. J . Ichthyol. 20(6), (A-73. Chebanov, N. A., Vamavskaya, N. V., and Vamavskiy, V. S. (1983). Effectiveness of spawning of male sockeye salmon, Oncorhynchus nerka (Samonidae), of differing hierarchical rank by means of genetic-biochemical markers. J. Ichthyol. 23, 51-55. Chesser, R. K., Smith, M. W., and Smith, M. H. (1984). Biochemical genetics of mosquitofish. 111. Incidence and significance of multiple insemination. Genetica (The Hague) 64,77-8 1.

Childers, W. F. (1967). Hybridization of four species of sunfishes (Centrarchidae). Bull.-Ill. Nut. Hist. Surv. 29, 159-214. Choat, J. H., and Robertson, D. R. (1975). Protogynous hermaphroditism in fishes of the family Scaridae. 1n“Intersexuality in the Animal Kingdom” (R. Reinboth, ed.), pp. 263-283. Springer-Verlag, Heidelberg. Cichocki, F. P. (1976). Cladistic history of cichlid fishes and reproductive strategies of the American genera Acarichthys, Biotoma and Geophagus. Ph.D. Thesis, University of Michigan, Ann Arbor. Clavijo, I. E. (1983). Pair spawning and formation of a lek-like mating system in the parrotfish Scarus vetula. Copeia, 1983, 253-256. Clutton-Brock, T. H. (1991). “The Evolution of Parental Care.” Princeton Univ. Press, Princeton, NJ. Clutton-Brock, T. H., and Vincent, A. C. J. (1991). Sexual selection and the potential reproductive rates of males and females. Nature (London) 351, 58-60. Colditz, G. (1986). Der “Kuckuck” im Aquarium: Ein Fiederbartwels als “Nestparasit”. Aquarien Magazin 20(9), 372-375.

84

MICHAEL TABORSKY

Cole, K. S. (1982). Male reproductive behaviour and spawning success in a temperate zone goby Coryphoprerus nicholsi. Can. J . Zool. 60, 2309-2316. Coleman, R. M., Gross, M. R., and Sargent. R. C. (1985). Parental investment decision rules: A test in bluegill sunfish. Behau. Ecol. Sociobiol. 18, 59-66. Colin. P. L.. and Bell, L. J. (1991). Aspects of the spawning of labrid and scarid fishes (Pisces: Labroidei) at Enewetak Atoll. Marshall Islands with notes on other families. Enuiron. Biol. Fish. 32, 229-260. Colin, P. L.. and Clavijo, I. E. (1978). Mass spawning of the spotted goatfish Pseudopeneus maculatus (Pisces: Mullidae). Bull. Mar. Sci. 28, 780-782. Colin. P. L.. and Clavijo, I. E. (1988). Spawning activity of fishes producing pelagic eggs on a shelf edge Coral reef, south western Puerto Rico. Bull. Mar. Sci. 43(2), 249-279. Collins, H . L.. and Braddock, J. C. (1%2). Notes on fostering experiments with the cichlid fishes, Tilapia sparrmani and Aequidens porralegrensis. A m . Zool. 2, 400. Colombe, J.. and Allgayer. R. (1985). Description de Variabilichromis, Neolamprologus, et Paleolamprologus genres nouveaux du lac Tanganyika, avec redescription des genres Lamprologus Schilthuis, 1891 et Lepidiolamprologus Pellegrin, 1904 (Pisces, Teleostei, Cichlidae). Rev. Fr. Cichlidophiles 49, 9-16 and 21-28. Constantz, G. D. (1975). Behavioral ecology of mating in the male Gila topminnow, Poeciliopsis occidentalis (Cyprinodontiformes; Poeciliidae). Ecology 56, 966-973. Constantz. G . D. (1979). Social dynamics and parental care in the tessellated darter (Pisces: Percidae). Proc. Acud. Nar. Sci. Philadelphia 131, 131-138. Constantz, G. D. (1984). Sperm competition in poeciliid fishes. I n “Sperm Competition and the Evolution of Animal Mating Systems“ (R. L. Smith, ed.), pp. 465-485. Academic Press. Orlando. FL. Constantz, G. D. (1985). Alloparental care in the tesselated darter, Erkeosroma olmstedi (Pisces: Percidae). Enuiron. B i d . Fishes 142-3). 175-183. Coyne, J . A., and Sohn. J . J. (1978). Interspecific brood care in fishes: Reciprocal altruism or mistaken identity? A m . Nar. 1l2, 447-450. Dalley, D. L., Andrews. C. W.. and Green. R. H. (1983). Precious male Atlantic salmon parr (Salmo salar) in insular Newfoundland. Can. J . Fish. Ayuat. Sci. 40, 647-652. Daniels. R. A. (1978). Nesting behaviour of Hurpagijer bispinis in Arthur Harbour, Antarctic Peninsula. J . Fish Biol. U ,465-474. Daniels. R. A. (1979). Nest guard replacement in the antarctic fish Harpagifer bispinis: Possible altruistic behavior. Science 205, 831-833. Dawkins, R. (1980). Good strategy or evolutionarily stable strategy? I n “Sociobiology: Beyond Nature Nurture?” (G. W. Barlow and J. Silverberg, eds.), pp. 331-367. Westview Press, Boulder, CO. De Boer, B. A. (1981). Influence of population density on the territorial, courting and spawning behavior of male Chromis cyonea (Pomacentridae). Behaviour 77, 99-120. De Jonge. J . , and Videler, J. J . (1989). Differences between the reproductive biologies of Tripferygion tripternorus and T. delaisi (Plsces. Perciformes, Tripterygiidae): the adaptive significance of an alternative mating strategy and a red instead of a yellow nuptial colour. Mar. B i d . 100, 431-437. DeMartini, E. E. (1985). Social behavior and coloration changes in painted greenling, Ox>?lebius pictus (Pisces. Hexagramrnidae). Copeia, 1985, 966-975. DeMartini. E. E., and Patten. B. G. (1979). Egg guarding and reproductive biology of the red Irish Lord. Hemilepidorus hemilepidorus (Tilesius). Syesis 12, 41-55. de Oliveira. R. F. D.. Alrnada, V. C.. and de Fatima, G. M. (1993). The reproductive

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

85

behavior of the longspine snipe fish, Macrorhamphosus scolopax (Syngnathiformes, Macrorhamphosidae). Env. Biol. Fishes 337-343. Dieke, H.-J. (1993). Kinderraub bei Apistogrammn borellii. DATZ 46(7), 471. Dipper, F. A. (1981). The strange sex lives of British wrasse. New Sci. 90, 444-445. Dipper, F. A., and Pullin, R. S . V. (1979). Gonochorism and sex-inversionin British Labridae (Pisces). J. Zool. 187, 97-112. Dominey, W. J. (1980). Female mimicry in male bluegill sunfish-A genetic polymorphism? Nature (London) 284, 546-548. Dominey, W. J. (1981). Maintenance of female mimicry as a reproductive strategy in bluegill sunfish (Lepomis macrochirus). Environ. Biol. Fishes 6, 59-64. Dominey, W. J. (1984). Alternative mating tactics and evolutionarily stable strategies. A m . ZOO^. 24,385-3%. Donaldson, T. J. (1987). Social organization and reproductive behavior of the hawkfish Cirrhitichthys fulco (Cirrhitidae). Bull. Mar. Sci. 41, 531-540. Dugatkin, L. A. (1992). Sexual selection and imitation: Females copy the mate choice of others. Am. Nut. 139, 1384-1389. Dunbar, R. I. M. (1982). Intraspecific variations in mating strategy. In “Perspectives in Ethology V ” (P. P. G. Bateson and P. H. Klopfer, eds.), pp. 385-431. Plenum, New York. Dupuis, H. M. C., and Keenleyside, M. H. A. (1988). Reproductive success of nesting male longear sunfish (Lepomis megalotis peltastes). 1. Factors influencing spawning success. Behav. Ecol. Sociobiol. 23(2), 109-116. Emlen, S. T., and Oring, L. W. (1977). Ecology, sexual selection, and the evolution of mating systems. Science 197, 215-223. Endler, J. A. (1980). Natural selection on color patterns in Poecilia reticulata. Evolution (Lawrence, Kans.) 34, 76-91. Endler, J. A. (1983). Natural and sexual selection on color patterns in poeciliid fishes. Environ. Biol. Fish. 9, 173-190. Endler, J. A. (1987). Predation, light intensity and courtship behaviour in Poecilia reticulata (Pisces: Poeciliidae). Anim. Behav. 35, 1376-1385. Fahy, W. E. (1954). The life history of the northern greenside darter, Etheostorna blennioides blennioides Rafinesque. J . Elisha Mitchell Sci. Soc. 70(2), 139-205. Farr, J. A. (198Qa).Social behavior patterns as determinants of reproductive success in the guppy, Poecilia reticulata Peters (Pisces, Poeciliidae): An experimental study of the effects of intermale competition, female choice, and sexual selection. Behaviour 74, 38-91. Farr, J. A. (1980b). The effects of sexual experience and female receptivity on courtship-Rape decisions in male guppies, Poecilia reticulatu (Pisces: Poeciliidae). Anim. Behav. 28, 1195-1201. Farr, J. A., Travis, J., and Trexler, J. C. (1986). Behavioural allometry and interdemic variation in sexual behavior of the sailfin molly, Poecilia latipinna (Pisces, Poeciliidae). Anim. Behav. 34,497-509. Fernald, R. D., and Hirata, N. R. (1977). Field study of Haplochromis burtoni: Quantitative behavioural observations. Anim. Behav. 25, 964-975. Fiedler, K. (1964). Verhaltensstudien an Lippfischen der Gattung Crenilabrus (Labridae, Perciformes). Z . Tierpsychol. 21, 521-591. Finck, E.-M. (1985). “Fortpflanzungsverhalten, Gonadengewicht und Sexualdimorphismus bei Tripterygion tripteronotus und Triptetygion delaisi.” Staatsexamensarbeit, AlbertLudwigs-Univ., Freiburg, Germany.

86

MICHAEL TABORSKY

Finley. L. ( 1984). Reproduction in Synodontis rniclripuncratrrs Boitlenger-An apparent “Cuckoo” relationship with maternal mouthbrooding cichlids. Freshnwter M a r . Aquarium 7f6). 22-25. 63. 66-74. Fischer. E . A. ( 1980).The relationship between mating system and simultaneous hermaphroditism in the coral-reef fish, Hj,poplectrirs nigricuns (Serranidae). Anim. Behau. 28, 620-633. Fischer. E. A. (1984). Egg trading in the chalk bass. Serranrrs tortugarum, a simultaneous hermaphrodite. Z. Tierpsgchol. 66, 143-151. Fischer. E. A. ( 1986). Mating systems of simultaneously hermaphroditic serrdnid fishes. In “lndo-Pacific Fish Biology: Proceedings of the Second International Conference on Indo-Pacific Fishes“ (T. Ugeno. R. Avai, T . Tanunchi. and K. Matsuura, eds.), pp. 776-784. Ichthyol. Soc. Jpn.. Tokyo. Fishelson. L. (1963). Observations on the littoral fishes of Israel. I. Behaviour of Blennius pauo Risso. Isr. J . Zool. 12, 67-80. Foote, C. J. (1990). An experimental comparison of male and female spawning territoriality in a pacific salmon. Behaviour 115, 283-314. Footc. C. J . . and Larkin, P. A. (1988). The role of male choice in the assortative mating of anadromous and non-anadromous sockeye salmon (Oncorhynchrrs nerka). Brhuviour 106.43-62. Fricke. H . W . (1979). Mating system. resource defence and sex change in the Anemonefish Amphiprion akullopisos. Z. Tierpsq.chol. 50, 313-326. Fricke. H. W. ( 1980). Mating systems. maternal and biparental care in triggerfish (Balistidae). Z . Tierpsychol. 53, 105-122. Frischknecht, M. ( 1993). The breeding coloration of male threespined sticklebacks (Gastercisleus acu/earrts)as an indicator of energy investment and vigour. Euol. E c d . (in press). Garratt. P. A. (1991). Spawning behaviour of Cheimerirrs nufar in captivity. Enuiron. B i d . Fish. 31, 34.5-353. Gadgil, M. (1972). Male dimorphism as a consequence of sexual selection. Am. Nat. 106, 574-580. Gebhards, S. V. (1960). Biological notes on precocious male chinook salmon pdrr in the salmon river drainage, Idaho. Prog. Fish-Cult. 22, 121-123. Gerald. J . W. ( 1970). Species isolating mechanisms in the genus Lepomis. Ph.D. Thesis, Univ. of Texas. Austin, Texas. Gladstone, W. (1987a). Role of territoriality in social and mating systems of Canthigaster ualentini (Pisces: Tetraodontidae): Evidence from field experiments. M a r . Biol. %(2), 185-191. Gladstone. W . ( 1987b). The courtship and spawning behaviors of Canthigaster ualenrini (Tetraodontidae).Environ. Biol. Fishes 20(4), 255-261. Glebe, 9 . D.. Saunders, R. L.. and Sreedharon. A. (1978). Genetic and environmental influence in expression of precocious sexual maturity of hatchery-reared Atlantic salmon (Salmo salur) parr. Can. J . Genet. Cytol. 20, 444. (;off. G. P. (1984). Brood care of longnose gar (Lepisosterrs osseus) by smallmouth bass Micropterus dolornieuif. Copeia 1 W , 149-152. Goldschmidt, T.. and Bakker. T. C. M. (1990). Determinants of reproductive success of male sticklebacks in the field and in the laboratory. N e f h . J . Zool. 40, 664-687. Cioldschmidt. T.. Foster. S. A., and Sevenster. P. (1992). Internest distance and sneaking in the three-spined stickleback. Anim. Behau. 44, 793-795. Cioldschmidt, T.. Bakker. T. C. M . . and Feuth-de Bruijn (1993). Selective copying in mate choice of female sticklebacks. Anim. Behciu. 45, 541-547.

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

87

Greenberg, B. (1963). Parental behavior and imprinting in cichlid fishes. Behauiour 21, 127-144. Gross, M. R. (1979). Cuckoldry in sunfishes (Lepomis: Centrarchidae). Can. J . Zool. 57, 1507- 1509. Gross, M. R. (1982). Sneakers, satellites and parentals-Polymorphic mating strategies in North-American sunfishes. Z. Tierpsychol. 60, 1-26. Gross, M. R. (1984). Sunfish, salmon, and the evolution of alternative reproductive strategies and tactics in fishes. In ”Fish Reproduction: Strategies and Tactics” (G. W. Potts and R. J. Wootton, eds.), pp. 55-75. Academic Press, London. Gross, M. R. (1985). Disruptive selection for alternative life histories in salmon. Nature (London) 313,47-48. Gross, M. R. (1991). Evolution of alternative reproductive strategies: Frequency-dependent sexual selection in male bluegill sunfish. Philos. Trans. R . Soc. London. Ser. B 332, 59-66. Gross, M. R., and Charnov, E. L. (1980). Alternative male life histories in bluegill sunfish. Proc. Nutl. Acad. Sci. U.S.A. 77, 6937-6940. Gross, M. R., and Dueck, L. (1989). Evolution of alternative reproductive strategies: How successful are cuckolders? Abstr. P14, Behav. Ecol. Conf., 3rd., Uppsala, Sweden. Hackney, P. A., Tatum, W. M., and Spencer, S. L. (1967). Life history study of the river redhorse, Moxostoma carinatum (Cope), in the Cahaba River, Alabama, with notes on the management of the species as a sport fish. Proc. 21st Annu. Conf. S.E. Assoc. Game Fish. Comrnun. 324-332. Hankinson, T. L. (1920). Report on the investigations of fish of the Galien River, Berrien County, Michigan. Occas. Pap. Mus. Zool. Univ. Mich. 89, 1014. Hanson, A. J., and Smith, H. D. (1967). Mate selection in a population of sockeye salmon (Oncorhynchus nerka) of mixed age-groups. J . Fish. Res. Board Can. 24, 1955-1977. Harcourt, A. H. (1987). Cooperation as a competitive strategy in primates and birds. In “Animal Societies” (Y. Ito, J. L. Brown, and J. Kikkawa, eds.), pp. 141-157. Jpn. Sci. SOC.Press, Tokyo. Harcourt, A. H. (1988). Helpful fish. BBC Wild/. 6. Hartman, G. F. (1969). Reproductive biology of the Gerrard stock rainbow trout. In “Symposium on Salmon and Trout in Streams” (T. G. Northcote, ed.), British Columbia University, Vancouver, R.C., Canada. Harvey, P. H., and Pagel, M. D. (1991). The Comparative Method in Evolutionary Biology. Oxford Series in Ecology and Evolution. Oxford University Press, Oxford. Hastings, P. A. (1988). Female choice and male reproductive success in the angel blenny. Coralliozetus angelica (Teleostei: Chaenopsidae). Anim. Behau. 36, 115-124. Heard, W. R. (1972). Spawning behavior of pink salmon on an artificial redd. Trans. A m . Fish. SOC. 101, 276-283. Heinrich, W. (1967). Untersuchungen zum Sexualverhalten in der Gattung Tilupiu (Cichlidae, Teleostei) und bei Artbastarden. Z. Tierpsychol. 24, 684-754. Heinrich, W., and Schroder, J. H. (1986). Tentantive findings on the phylogenetic relationship within the genus Xiphophorus with regard to the frequency distribution of sexual behavior patterns. Ber. nut.-med. Veiren Innsbruck 73, 187-197. Heinsohn, R. G. (1991). Kidnapping and reciprocity in cooperatively breeding white-winged choughs. Anim. Behau. 41, 1097-1100. Helas, T., Lejeune, P., Michel, C., and Voss, J. (1982a). A propos de quelques poissons de la Mediterrannee: Symphodus (Crenilubrus) roissali Risso, 1810. Rev. Fr. Aquariol. 9(1), 29-32.

88

MICHAEL TABORSKY

Helas, T., Michel. C.. Lejeune, P., and Voss, J . (1982b). A propos de quelques poissons de la Mediterrannee: Symphodus (Crenilabrus) tinca (Linne, 1758). Rev. Fr. Aquariol. %2). 37-40. Herr, E. (1985).Individual recognition of helpers by the breeders in the cichlid fish Lamprologrrs brichardi (Poll. 1974). Z . Tierpsychol. 68, 3 13-325. Heschl, A. (1989).Integration of “innate” and ”learned” components within the IRME for mussel recognition in the European bitterling Rhodeus amarus (Rloch). Ethology 81, 193-208. Hillden, N . - 0 . (1981 ). Territoriality and reproductive behaviour in the goldsinny, Ctenolabrus rupestris L. Behav. Proc. 6, 207-221. Hillden. N.-0. ( IY84a). “Behavioural Ecology of the Labrid Fishes (Teleostei: Labridae) a6 Tjarno on the Swedish West Coast.” Ph.D. Thesis. University of Stockholm, Stockholm, Sweden. Hillden. N.-0. (1984b). An ethological study of the goldsinny, Cienolahrus rupesiris (Pisces, Labridae) in Swedish waters. N e d . , I. A . F. E. 7(2). 15. Hughes. A. L . (1985). Male size, mating success and mating strategy in the mosquitofish Cambusia afinis (Poeciliidae). Behav. E d . Sociobiol. 17, 271-278. Hunter. J. R., and Hasler, A. D. (1965). Spawning association of the redfin shiner, Notropis umbratilis, and the green sunfish, Lepomis cyanellus. Copeia, 1965, 265-281. Hutchings, J. A,. and Myers, R. A. (1988). Mating success of alternative maturation phenotypes in male Atlantic salmon, Salmo salar. Oecologicc, 75(2), 169-174. Iwamoto. R. N.. Alexander, B. A , , and Hershberger. W. K. (1983). Genotypic and environmental effects of the incidence of sexual precocity in coho salmon. In “Salmonid Reproduction: An International Symposium,” p. 8. Univ. of Washington, Seattle, WA. Jamieson, I. G., and Colgan, P. W. (1989). Eggs in the nests of males and their effect on mate choice in the three-spined stickleback. Anim. Behau. 38(5), 859-86s. Janzow. F. T. (1982). Sexual interference by non-territorial honey gourami (Colisa c h u m ) males. Anim. Behav. 30, 633. Jenkins. R. E . (1970).Systematic studies of the catostomid fish tribe Moxostotnaiini. Ph.D. Thesis, Cornell University, Department of Zoology, Ithaca. NY. Jenkins. R. E.. and Jenkins, D. J . (1980). Reproductive behavior of the greater redhorse, Moxostoma ualencitwnesi, in the Thousand Islands region. Can. Field Nai. 94, 426-443. Jennings, M. J . , and Philipp, D. P. (1992a). Reproductive investment and somatic growth rates in longear sunfish. Enuiron. B i d . Fish.35, 257-271. Jennings. M. J., and Philipp. D. P. (1992b). Female choice and male competition in longear sunfish. Behav. Ecol. 3(1). 84-94. Jewell. E. D. (1968). SCUBA diving observations on lingcod spawning at a Seattle breakwater. Wash. State Dep. Fish., Fish-Res. Pap. 3(1). 27-36. Jones, G. P. (1981). Spawning site choice by female Pseudolabrus celidotus (Pisces, Labridae) and its influence on the mating system. Behav. Ecol. Sociobiol. 8, 129-142. Jones, J . W. (1959). “The Salmon.” Collins New Naturalist, London. Jones, J . W.. and Ball. J. N . (1954). The spawning behaviour of brown trout and salmon. Br. J . Anim. Behav. 2, 103-114. Jones, J. W., and King. G. M. (19SOa). Progeny of male salmon parr. A comparison with those from normal adults. Salmon Trout Mag. Us,24-26. Jones. J. W., and King, G. M. (1950b). Further experimental observation of the spawning hehaviour of Atlantic salmon.Proc. Zoo/. Soc. London UO,317-223. Jones, J. W.. and King. G. M. (1952a). Spawning of the male salmon parr (Salmo salar I h n . juv.). Nature (London) 169, 882.

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

89

Jones, J. W., and King, G. M. (1952b). The spawning of the male salmon parr (Salmo salar Linn. juv.). Proc. Zool. SOC.London 122, 615-619. Jonsson, B., and Hindar, K. (1982). Reproductive strategy of dwarf and normal arctic cham (Saluelinus alpinus) from Vangsvatnet Lake, Western Norway. Can. J. Fish. Aquat. S C ~39, . 1404-1413. Jordan, W. C., and Youngson, A. F. (1992). The use of genetic marking to assess the reproductive success of mature male Atlantic salmon parr (Salmo salar, L.) under natural spawning conditions. J . Fish. Biol. 41, 613-618. Kalas, K. (1976). Brutpflegehelfer und Polygamie beim afrikanischen Buntbarsch Lamprologus brichardi. Natunvissenschaften 63, 94. Katano, 0. (1983). Spawning of the dark chub Zacco temmincki (Temminck et Schlegel) (Pisces: Cyprinidae) with special reference to male social status. Physiol. Ecol. J p n . 20, 33-52. Katano, 0.(1990). Dynamic relationships between the dominance of male dark chub, Zacco temmincki, and their acquisition of females. Anim. Behau. 40, 1018-1034. Katano, 0. (1992). Spawning tactics of paired males of the dark chub, Zacco remmincki reflect potential fitness costs of satellites. Environ. Biol. Fishes 35, 343-350. Kazakov, R. V. (1981). Peculiarities of sperm production by anadromous and parr Atlantic salmon (Sulmo sular L.) and fish cultural characteristics of such sperm. J . Fish Biol. 18(1), 1-7. Keenleyside, M. H. A. (1972). Intraspecific intrusions into nests of spawning longear sunfish (Pisces: Centrarchidae). Copeia, 1972, 272-278. Keenleyside, M. H. A. (1979). I n “Diversity and Adaption in Fish Behavior,” p. 208. Springer-Verlag, Berlin. Keenleyside, M. H. A. (1991). Parental care. I n “Cichlid Fishes: Behaviour, Ecology and Evolution” (M. M. A. Keenleyside, ed.), pp. 191-208. Chapman Hall, London. Keenleyside, M. H. A., and Dupuis, H. M. C. (1988). Courtship and spawning competition in pink salmon (Oncorhynchus gorbuscha). Can. J . Zool. 66, 262-265. Kirkpatrick, M. (1982). Sexual selection and the evolution of female choice. Evolution (Lawrence, Kans.) 36, 1-12. Kirkpatrick, M. (1987). The evolutionary forces acting on female mating preferences in polygynous animals. I n “Sexual Selection: Testing the Alternatives” (J. W. Bradbury and M. B. Anderson, eds.), pp. 67-82. Wiley, Chichester. Knowlton, N. K., and Greenwell, S. R. (1984). Male sperm competition avoidance mechanisms: The influence of female interests. I n “Sperm Competition and the Evolution of Animal Mating Systems” (R. L. Smith, ed.), pp. 62-84. Academic Press, Orlando, FL. Kodric-Brown, A. (1977). Reproductive success and the evolution of breeding territories in pupfish. (Cyprinodon). Evolution (Lawrence, Kans.) 31, 750-766. Kodric-Brown, A. (1981). Variable breeding systems in pupfishes (genus Cyprinodon): Adaptions to changing environments. I n “Fishes in North American Deserts” (R. T. Naiman and D. L. Soltz, eds.), pp. 205-235. John Wiley and Sons, New York. Kodric-Brown, A. (1986). Satellites and sneakers: Opportunistic male breeding tactics in pupfish (Cyprinodon pecosensis). Behav. Ecol. Sociobiol. 19,425-432. Kodric-Brown., A. (1992). Male dominance can enhance mating success in guppies. Anim. Behau. 44, 165-167. Kondo, T. (1986). Feeding habits of Lamprologus savoryi (Teleostei: Cichlidae) with reference to its social behaviour. Physiol. Ecol. Jpn. 23(1), 1-15. Koprowski, J. L. (1993). Alternative reproductive tactics in eastern gray squirrels: “making the best of a bad job.” Behav. Ecol. 4, 165-171.

90

MICHAEL TABORSKY

Kramer, R. J.. and Smith. L . L.. Jr. (1960). Utilization of nests oflargemouth bass, Microptt rus salmoides. by golden shiners, Notemigonus crysoleucas. Copeiu. 1%0, 73-74. Kratt, L. F.. and Smith. R. J. F. (1980). An analysis of the spawning behavior of the arctic grayling Thymallits urcticus (Pallas) with observations on mating success. J . Fish. Biol. 17,661-666.

Kuwamura, T. 11986). Parental care and mating systems of cichlid fishes in Lake Tanganyika: A preliminary field survey. J . Ethol. 4(2). 129-146. Kuwamura, T. (1987). Male mating territory and sneaking in a maternal mouthbrooder, I’seudosimochromis curuifrons (Pisces: Cichlidae). J . Ethol. 5 , 203-206. Kwak. T. J., and Skelly. T. M. (1992). Spawning habitat, behavior and morphology as isolating mechanisms of the golden redhorse. Moxostomu erythrirrum and the black redhorse, M. duquesnei, two syntopic fishes. Enu. Biol. Fishes 34, 127-137. Ladich. F.. and Taborsky, M. (1991). Broodcare and egg cannibalism in Neolamprologiis brichordi (Poll 1986). Ann. Mus. R. Afr. Cent., Sci. Zoo/.,262, 73-74. Landc, R. (1987). Genetic correlations between the sexes in the evolution of sexual dimorphism and mating preferences. In “Sexual Selection: Testing the Alternatives” ( J . M. Bradbury and M. P. Andersson, eds.). pp. 83-94. Wiley, Chichester. Langlois, T. H. (1929). Breeding habits of the northern dace. Ecology 10, 161-163. Lejeune, P. (1982). A propos d e quelques poissons de la Mediterranee: Note concernant le comportement social et reproducteur de Coris jirlis L. Reu. Fr. Aquariol. 9(1), 27-28. Lejeune, P. (1985). Etude ecoethologique des comportements reproducteurs et sociaux des Labridae Mediterraneens des genres Symphodus Rafinesque. 1810 et Coris Lacepede, 1802. Cah. Ethol. Appl. 5 , 1-208. Lejeune, P. (1987). The effect of local stock density on social behaviour and sex change in the Mediterranean labrid Coris julis. Enuiron. Biol. Fishes 18, 135-141. Lewis, D. S. C. (1980). Mixed species broods in Lake Malawi cichlids-An alternative to the cuckoo theory. Copeia, 1980, 874-875. Li, S. K.. and Owings, D. H. (1978a). Sexual selection in the three-spined stickleback. I. Normative observations. Z. Tierpsychol. 46, 359-371. Li. S. K., and Owings, D. H . (3978b). Sexual selection in the three-spined stickleback. 11. Nest raiding during the courtship phase. Behauiour 64, 298-304. Liley. N. R. (1966). Ethological isolating mechanisms in four sympatric species of poeciliid fishes. Behauiour (Suppl.) 13, 1-197. Limberger, D. (1983). Pairs and harems in a cichlid fish, Lamprologus hrichardi. Z. Tierpsychol. 62, 115-144. Lobel, P. S. (1989). Spawning behavior of Chaetodon multicinctus (Chaetodontidae); Pairs and intruders. Enuiron. Biol. Fishes 25, 125-130. Lorenzen. E. (1989).Datz 42(1), 16. Lundqvist. H . , and Fridberg, G. (1982). Sexual maturation versus immaturity: Different tactics with adaptive values in Baltic salmon (Salmo salar L.) male smolts. Can. J. ZOO^. 60, 1822-1827. Luyten, P. H.. and Liley, N. R. (1985). Geographic variation in the sexual behaviour of the guppy, Poecilia reticdata (Peters). Behauiour 95, 164-179. MacWhirter, R. B. (1989). On the rarity of intraspecific brood parasitism. Condor 91, 485-492. Maekawa. K. (1983). Streaking behaviour of mature male parrs of the Miyabe charr, Suluelinus malma miyahei, during spawning. Jpn. J. Ichthyol. 30, 227-234. Maekawa. K.. and Hino. T. (1986). Spawning behaviour of Dolly Varden in southeastern Alaska, with special reference to the mature male parr. Jpn. J . Ichrhyol. 32, 454-458.

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

91

Maekawa, K., and Hino, T. (1990). Spawning tactics of female Miyabe charr (Saluelinus malma miyabei) against egg cannibalism. Can. J. Zool. 68, 889-894. Maekawa, K., and Onozato, H. (1986). Reproductive tactics and fertilization success of mature male Miyabe charr, Saluelinus malma miyabei. Enuiron. Biol. Fishes 15, 119- 129. Magnhagen, C. (1992). Alternative reproductive behaviour in the common goby, Pomatoschistus microps: An ontogenetic gradient? Anim. Behau. 44, 182-184. Magnhagen, C., and Kvarnemo, L. (1989). Big is better: the importance of size for reproductive success in male Pomatoschistus minutus (Pisces: Gobiidae). J . Fish Biol. 35, 755-763. Magurran, A. E. (1986). Individual differences in fish behaviour. In “The Behaviour of Teleost Fishes” (T. J. Pitcher, ed.), pp. 338-365. Croom Helm, London and Sydney. Marconato, A. (1982). Preliminary investigation on the reproductive behavior of Aphanius fasciatus Nardo (Cyprinodontidae). Newsl. I. A . F. E. 5(2), 24-26. Marconato, A., and Bisazza, A. (1986). Males whose nests contain eggs are preferred by female Cottus gobio L. (Pisces, Cottidae). Anim. Behau. 34(5), 1580-1582. Marconato, A., Bisazza, A,, and Marin, G. (1989). Correlates of male reproductive success in Padogobius martensi (Gobidae). J . Fish. Biol. 34, 889-899. Martin, E., and Taborsky, M. (1993). Alternative mating tactics of male cichilids: A comparison of reproductive effort and success in Peluicachrvmis pulcher. In preparation. Maynard Smith, J. (1977). Parental investment: a prospective analysis. Anim. Behau. 25, 1-9. Maynard Smith, J. (1982). “Evolution and the Theory of Games.” Cambridge Univ. Press, Cambridge. Mboko, K. (1989). Preliminary report on the social organization of a cichlid fish, Telmatochromis temporalis, in Lake Tanganyika. In “Ecological and Limnological Study on Lake Tanganyika and Its Adjacent Regions VI” (H. Kawanabe, ed.), pp. 36-37. Kyoto University, Kyoto, Japan. McCart, P. J. (1969). Digging behaviour of Oncorhynchus nerka spawning in streams at Babine Lake, British Columbia. In “Symposium on Salmon and Trout in Streams” (T. G. Northcote, ed.), pp. 39-51. University of British Columbia, Vancouver, BC, Canada. McCart, P. J. (1970). A polymorphic population of Oncorhynchus nerka at Babine Lake, B.C., involving anadromous (sockeye) and non-anadromous (kokanee) forms. Ph.D. Thesis, University of Vancouver, B.C. Canada. McKaye, K. R. (1977). Defense of a predator’s young by a herbivorous fish: An unusual strategy. Am. Nat. 111, 301-315. McKaye, K. R. (1979). Defense of a predator’s young revisited. Am. Nat. 114, 595-601. McKaye, K. R. (1981). Natural selection and the evolution of interspecific brood care in fishes. In “Natural Selection and Social Behaviour” (R. D. Alexander and D. W. Tinkle, eds.), pp. 173-183. Chiron Press, New York. McKaye, K. R. (1983). Ecology and breeding behavior of a cichlid fish, Cyrtocara eucinostomus, on a large lek in Lake Malawi, Africa. Enuiron. Biol. Fishes 8, 81-96. McKaye, K. R. (1984). Behavioural aspects of cichlid reproductive strategies: Patterns of territoriality and brood defense in Central American substratum spawners and African mouth brooders. In “Fish Reproduction: Strategies and Tactics” (R.J. Wootton and G. W. Potts, eds.), pp. 245-273. Academic Press, New York. McKaye, K. R. (1985). Cichlid-catfish mutualistic defense of young in Lake Malawi, Africa. Oecologia 66(N3), 358-363.

92

MICHAEL TABORSKY

McKaye, K. R . , and McKaye. N. M. (1977). Communal care and kidnapping of young by parental cichlids. Evolution (Lawrence, Kans.) 31, 674-681. McKaye. K. R . , and Oliver. M. K. (1980). Geometry of a selfish school: Defense of cichlid young by a bagrid catfish in Lake Malawi. Anim. Behau. 28, 1287. McKaye, K. R., Mughogho. D. E., and Lovullo, T. J. (1992). Formation of the selfish school. Env. Biol. Fishes 35, 213-218. Meyer. K. A. (1977). Reproductive behaviour and patterns of sexuality in the Japanese labrid fish Thalnssoma crrpido. Jpn. J . Ichrhyol. 24, 101-112. Michei. C., and Voss, J. (1982). Observation en baie de Calvi du comportement social chez Synphodus (Crenilabrus) cinereus (Bonnaterre 1788) (Plsces: Labridae). Cab. Erhol. A@. 2(1), 17-35.

Michel. C.. Lejeune. P., and Voss, J. (1987). Biologie et comportement des labrides Europeens. Rev. Fr. Aquariol. 14, 1-80. Milinski, M. (1987). Tit-for-tat in sticklebacks and the evolution of cooperation. Nature (London)325, 433-434. Milinski, M., Muger. D.. Knelling. D., and Kettler, R. (1990). Do sticklebacks cooperate repeatedly in reciprocal pairs? Behau. Ecol. Sociobiol. 27(1), 17-21. Miller. H. C . (1963). The behavior of the pumpkinseed sunfish Lepomis gibbosus (Linnaeus), with notes on the behavior of the other species of Lepomis and the pigmy sunfish Elassoma euegladei. Behauiour 22, 88-151. Miller. P. J. (1984). The tokology of gobiid fishes. I n “Fish Reproduction: Strategies and Tactics” iG. W. Potts and R. J. Wootton. eds.). pp. 119-154. Academic Press, New York. Mohr, H. P. ( 1986). Kleptogamie-Vergleich dreier mediterraner Arten der Tripterygiidae (Perciformes, Teleostei). Diplomarbeit. Albert-Ludwigs-Univ., Freiburg, Germany. M@ller,A. P. (1987). lntraspecific nest parasitism and anti-parasite behaviour in swallows, Hirundo rusrica. Anim. Behau. 40, 1035-1047. Moodie, G. E. E. (1972). Predation, natural selection and adaptation in an unusual stickleback. Heredity 28 ,155-167. Moms, D. (1952). Homosexuality in the ten-spined stickleback (Pygusreus pungitius L.). Brhaviour 4, 233-261. Moyer. J. T . (1979). Mating strategies and reproductive behavior of ostraciid fishes at Miyake-Jima, Japan. Jpn. J . Ichrhvol. 26, 148-160. Moyer, J. T. (1981). Interspecific spawning of the pygmy angelfishes Centropyge shepardi and Cenzrupyge bispinosus at Guam. Micronesica 17, 119-124. Moyer, J . T. (1Y84). Social organization and reproductive behavior of ostraciid fishes from Japan and the western Atlantic Ocean. J . Erhol. 2,85-98. Moyer. J. T., and Shepard, J. W. (1975). Notes on the spawning behavior of the wrasse Cirrhilabrus temminckii. Jpn. J . Ichrhyol. 22(1), 40-42. Moyer, J. T.. Thresher, R. E., and Colin, P. L . (1983). Courtship, spawning and inferred social organization of American angelfishes (genera Pornacanthus, Holacnnrhus and Cuntropyge. Pomacanthidae). Enuiron. Biol. Fishes 9, 25-39. Mrowka, W. (1987a). Brood adoption in a mouthbrooding cichlid fish: Experiments and a hypothesis. Anim. Behau. 35(3), 922-923. Mrowka, W. (1987b). Egg stealing in a mouthbrooding cichlid fish. Anim. Behau. 35(3), 923-925.

Myers. R. A. (1986). Game theory and the evolution of Atlantic salmon (Salrno s d a r ) age ai maturation. In “Salmonid Age at Maturity” (D. J. Meerburg, ed.), Can. Spec. Publ. f i ~ b Aqiiat. . Sci. 89, 53-61.

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

93

Myers, R. A., and Hutchings, J. A. (1987). Mating of anadromous Atlantic salmon, Salmo salar L., with mature male parr. J. Fish Biol. 31, 143-146. Myrberg, A. A., Jr. (1964). An analysis of the preferential care of eggs and young by adult cichlid fishes. Z. Tierpsychol. 21, 53-98. Neudecker, S., and Lobel, P. S. (1982). Mating systems of chaetodontid and pomacanthid fishes at St. Croix. Z. Tierpsychol. 59, 299-318. Naokes, D. L. G., and Barlow, G. W. (1973). Cross-fostering and parent-offspring responses in Cichlasoma citrinellum (Pisces, Cichlidae). Z . Tierpsychol. 33, 147-152. Noble, G. K., and Curtis, B. (1939). The social behaviour of the jewel fish, Hemichromis bimaculatus. Bull. A m . Mus.Nut. Hist. 76, 1-76. Noltie, D. B. (1989). Intrapopulation variation in the breeding of male pink salmon (Oncorhynchus gorbuscha) from a Lake Superior tributary. Can. J . Fish. Aquat. Sci. 47(1), 174-179. Olla, B. L., Samet, C., Bedja, A. J., and Studholme, A. L. (1977). Social behaviour as related to environmental factors in the tautog, Tautoga onitis. In “Behaviour of Marine Organisms” (Annu. Northeast. Reg. Meet. Anim. Behav. SOC.1977), MSRL Tech. Rep. 20, pp. 47-99. Orr, J. W., and Ramsey, J. S. (1990). Reproduction in the greenbreast darter Etheostoma jordani (Teleostei: Percidae). Copeia 1990, 100-107. Orton, J. M., Jones, J. W., and King, G. M. (1938). The male sexual stage in salmon parr (Salmo salar L. juv.). Proc. R. SOC.London, Ser. B 125, 103-114. Page, L. M., and Johnston, C. E. (1990). Spawning in the creek chubsucker, Erimyzon oblongus, with a review of spawning behavior in suckers (Catostomidae). Enuiron. Biol. Fishes 27(4), 265-272. Parker, G. A. (1974). Assessment strategy and the evolution of fighting behaviour. J. Theor. Biol. 47, 223-243. Parker, G. A. (1982). Phenotype-limited evolutionarily stable strategies and fitness in three types of intrasexual competition. I n “Current Problems in Sociobiology” (Kings College Sociobiology Group, eds.), pp. 173-202. Cambridge Univ. Press, Cambridge. Parker, G. A. (199Oa). Sperm competition games: Raffles and roles. Proc. R. SOC.London, Ser. B 242, 120-126. Parker, G. A. (199Ob). Sperm competition games. Sneaks and extra-pair copulations. Proc. R. SOC.London, Ser. B 242, 127-133. Parzefall, J. (1979). Zur Genetik und biologischen Bedeutung des Aggressionsverhaltens von Poecilia sphenops (Pisces, Poeciliidae). Z . Tierpsychol. 50, 399-422. Petersen, C. W. (1987). Reproductive behaviour and gender allocation in Serranusfasciatus, a hermaphroditic reef fish. Anim. Behau. 35, 1601-1614. Petersen, C. W. (1990). The relationship among population density, individual size, mating tactics and reproductive success in a hermaphroditic fish, Serranus fasciatus. Behauiour 113, 57-80. Petersen, C. W . , and Fischer, E. A. (1986). Mating system of the hermaphroditic coral-reef fish, Serranus baldwini. Behau. Ecol. Sociobiol. 19, 171-178. Petravicz, W. P. (1938). The breeding habits of the black-sided darter, Hadropterm maculatus Girard. Copeia, 1938,40-44. Pinxten, R., Eens, M., and Verheyen, R. F. (1991). Conspecific nest parasitism in the European starling. Ardea 79, 15-30. Pitcher, T. (1992). Who dares, wins: The function and evolution of predator inspection behaviour in shoaling fish. Neth. J. Zool. 42, 371-391. Pottle, R. A., and Green, J. M. (1979a). Field observations on the reproductive behaviour

94

MICHAEL TABORSKY

crf the cunner. Tucttogolabrus udspersus (Walbaum). in Newfoundland. Can. 1.Zoo/. 57, 247-256. Pottle, R. A,. and Green, J . M. (1979b). Territorial behaviour of the north temperate labrid Taiiiogolabrus udspersus. Can. J . Zoo/. 57, 2337-2347. Pottle. R. A , . Green, J . M., and Martel. G . (1981). Dualistic spawning behaviour of the cunner, Tuitrogolabrus udspersfis (Pisces: Labridae). in Bonne Bay, Newfoundland. Can. .I. Zoo/. 59, 1582-1585. Pressley. P. H. (1981). Parental effort and the evolution of nest-guarding tactics in the threespine stickleback. Gasterosieus aculeaius L . EU0/utiOn 35, 282-295. Randall. J . E.. and Randall, H . A. (1963). The spawning and early development of the Atlantic parrotfish Sporisonia ritbripinne, with notes on other scarid and labrid fishes. Zoologica ( N .Y . ) 48, 49-60. Raney. E. C.. and Lachner. E. A. (1946). Age, growth and habits of the hog sucker, Hvpenielium nigricans (Le Sueur). in New York. A m . Mid/. Nut. 36, 76-86. Raney. E. C.. Backus, R. H.. Crawford, W., and Robins, C. R. (1953). Reproductive Y.) 38.97-104. behdviour in Cyprinodon uariegatus Lacepi.de in Florida. Zodogicn (N. Reeves. C. D. (1907). The breeding habits of the rainbow darter (Erheostoma caeruleirm Storer), a study in sexual selection. B i d . ELI//.(Woods Hole, Mass.) 14, 35-59. Reighard, J. (1920). The breeding behaviors of the suckers and minnows. I . The suckers. H i d . Bull. (Woods Hole, Mass.) 38, 1-33. Reighard, J. ( 1943). The breeding habits of the river chub, Noconzis micropogon (Cope). Pup. Mich.Acad. Sci., Arts Leti. 28, 397-423. Reinboth. R. ( 1962). Morphologische und funktionelle Zweigeschlechtlichkeit bei marinen Teleostiern (Serranidae, Sparidae, Centracanthidae. Labridae). Z o o / . Jahrh., Aht. AUg. Zoo/. Physiol. Tiere 69, 405-480. Reinboth. R. (1973). Dualistic reproductive behavior in the protogynous wrasse Tholassoma hiJasciatum and some observations on its day-night changeover. Helgol. Wiss. Meeresunters. 24, 174-191. Reyer, H.-U., Dittami. J. P., and Hall, M. R. (1986). Avian helpers at the nest-Are they psychologically castrated? Eihology 71, 2 16-228. Reynolds. J. D.. Gross, M. R., and Coombs, M. J . (1993). Environmental conditions and male morphology determine alternative mating behaviour in Trinidadian Guppies. Anim. Aehau. 45, 145-152. Ribbink, A. J. (1971). The hehaviour of Hernihaplochromis philander, a South African cichlid fish. Zoo/. Afr. q 2 ) . 263-288. Ribbink. A. J . (1975). ”A Contribution to the Understanding of the Ethology of the Cichlids of Southern Africa.” Ph.D. Thesis. Rhodes University, Grahamstown, South Africa. Ribbink. A. J. (1977). Cuckoo among Lake Malawi cichlid fish. Narure (London) 267, 243-244. Ribbink. A. J . , Marsh. A. C., Marsh. B.. and Sharp, B. J . (1980). Parental behavior and mixed broods among cichlid fish of Lake Malawi. 5 . Afr. J . Zoo/. 15, 1-6. Ribbink. A. J.. Marsh, A. C., and Marsh, B. A. (1981). Nest-building and communal care of young by Tilapia rendalli Dumeril (Pisces. Cichlidae) in Lake Malawi. Enuiron. B i d . Fishes 6, 219-222. Rico, C., Kuhnlein. U., and FitzGerald, G. J. (1992). Male reproductive tactics in the threespine stickleback-an evaluation by DNA fingerprinting. Mol. Eco/. 1,79-87. Ridley, M., and Rechten, C. (1981). Female sticklebacks prefer to spawn with males whose nests contain eggs. Behauioirr 76, 152-161. Robertson, D. R. (1983). On the spawning behavior and spawning cycles of 8 surgeonfishes i Acanthuridae) from the Indo-Pacific. Enuiron. B i d . Fishes 9, 193-223.

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

95

Robertson, D. R., and Choat, J. H. (1974). Protogynous hermaphroditism and social systems in labrid fish. Proc. I n t . Coral Reef Symp., 2nd., 1973, Vol. 1, pp. 217-225. Robertson, D. R., and Hoffman, S. G. (1977). The roles offemale mate choice and predation in the mating systems of some tropical labroid fishes. Z . Tierpsychol. 45, 298-320. Robertson, D. R., and Warner, R. R. (1978). Sexual patterns in the labrid fishes of the western Caribbean. Part 11. The parrotfishes (Scaridae). Smithson. Contrib. Zool. 255, 1-26. Robertson, D. R., Polunin, N. V. C., and Leighton, K. (1979). The behavioral ecology of three Indian Ocean surgeonfishes (Acanthurus lineatus, A . leucosternon, Zebrasoma scopas): Their feeding strategies, and social and mating systems. Environ. Biol. Fishes 4, 125-170. Robertson, D. R., Reinboth, R., and Bruce, R. W. (1982). Gonochorism, protogyneous sexchange and spawning in three sparisomatinine parrotfishes from the western Indian Ocean. Bull. Mar. Sci. 32, 868-879. Rohwer, S. (1978). Parent cannibalism of offspring and egg raiding as a courtship strategy. Am. Nut. 112,429-440. Rohwer, S. (1986). Selection for adoption versus infantcide by replacement “mates” in birds. I n “Current Ornithology,” Vol. 3, (R. F. Johnston, ed.), pp. 353-395. Plenum, New York. Romer, U. (1993). Erste Beobachtungen an einer neu eingefuhrten Art der Gattung Nannacara Regan, 1905. In “Buntbarschjahrbuch 1994” (W. Staeck, ed.), Bede Berlag, Kollnburg, Germany (in press). Ross, M. R. (1977). Aggression as a social mechanism in the creek chub (Semotilus atromaculatus). Copeia 1977, 393-397. Ross, M. R. (1983). The frequency of nest construction and satellite male behaviour in the fallfish minnow. Enuiron. Biol. Fishes 9, 65-70. Ross, M. R., and Reed, R. J. (1978). The reproductive behavior of the fallfish Semotilus corporalis. Copeia 1978, 215-221. Rowland, W. J. (1979). Stealingfertilizations in the fourspine stickleback, Apeltes quadracus. Am. Nat. 114,602-604. Royce, W. F. (1951). Breeding habits of lake trout in New York. Fish. Bull. 59/52, 59-76. Rubenstein, D. I. (1980). On the evolution of alternative mating strategies. I n “Limits to Action: The Allocation of Individual Behavior” (J. E. R. Staddon, ed.), pp. 65-100. Academic Press, New York. Ryan, M. J., and Causey, B. (1989).Alternative mating behaviour in the swordtailsXiphophorus nigrensis and X . pygmaeus (Pisces: Poeciliidae). Behau. Ecol. Sociobiol. 24, 341-348. Ryan, M. J., Hews, I). K., and Wagner, W. E., Jr. (1990). Sexual selection alleles that determine body size in the swordtail Xiphophorus nigrensis. Behau. Ecol. Sociobiol. 26,231-237. Ryan, M. J., Pease, C. M., and Moms, M. R. (1992). A genetic polymorphism in the swordtail, Xiphophorus nigrensis: testing the prediction of equal fitnesses. Am. Nut. 139, 21-31. Santos, R. S. (1985). Parentais e satklites: Thcticas alternativas de acasalamento nos machos de Blennius sanguinolentus Pallas (Pisces: Blenniidae). Arquiptlago S.C.N. 6 , 119-146. Santos, R. S., (1986). Alternative mating strategies in males of Blennius sanguinolentus (Pisces: Blenniidae) in the Azores. Abstr. Unpublished manuscript. Santos, R. S., and Almada, V. C. (1988). Intraspecific variations in reproductive tactics in males of the rocky intertidal fish Blennius sanguinolentus in the Azores. In “Behavioral

96

MICHAEL TABORSKY

Adaptation to Intertidal Life” (G. Chelazzi and M. Vannini, eds.), pp. 421-447. Plenum, New York. Sargent. R. C. (1982). Temtory quality, male quality, courtship intrusions, and female nestchoice in the threespine stickleback Gasrerosteus aculeafus. Anim. Behau. 30,364-374. Sargent. R. C. (1988). Paternal care and egg survival both increase with clutch size in the fathead minnow, Pimephales promelas. Behau. Ecol. Sociobiol. 23, 33-37. Sargent. R. C. (1989). Allopaternal care in the fathead minnow, Pirnephales promelas: stepfathers discriminate against their adopted eggs. Behau. Ecol. Sociobiol. 25,379-386. Sargent, R. C . , and Gebler, J. B. 11980). Effects of nest site concealment on hatching success, and paternal behavior of the threespine stickleback, Gasterosteus nculeatus. Behav. Ecol. Sociobiol. I, 137-142. Sato, T. (1986).A brood parasitic catfish of mouthbrooding cichlid fishes in Lake Tanganyika. Nature (London) 323.58-59. Sato, T. (1987). A brood parasitic catfish of the mouthbrooding cichlid fishes in Lake Tanganyika. In “Ecological and Limnological Study on Lake Tanganyika and Its Adjacent Regions. IV.” (H. Kawanabe and M. Nagoshi. eds.). pp. 43-44. Kyoto University, Kyoto, Japan. Sato, T. (1988). Mating system and parental care of a shell-brooder, Lamprologus cullipterus. In “Ecological and Limnological Study on Lake Tanganyika and Its Adjacent Regions. V.” (H.Kawanabe and M. K. Kwetuenda. eds.). pp. 23-24. Kyoto University. Kyoto, Japan. Sato, T. (1991). Prominent polygyny and male alternative reproductive strategies in a shell brooding cichlid fish, Lamprologus callipterus. in Lake Tanganyika. In “Ecological and Limnological Study on Lake Tanganyika and Its Adjacent regions. VII. (H. Kawanabe, and M. Nagoshi, eds.), p. 112. Kyoto University, Kyoto Japan. Schaffer, W. M., and Elson. P. F. (1975). The adaptive significance of variation in life history among local populations of Atlantic salmon in North America. Ecology, 56, 577-590. Schiefer, K. (1971). Ecology of Atlantic salmon. with special reference to occurrence and abundance of grilse, in North Shore Gulf of St. Lawrence River. Ph.D. Thesis, University of Waterloo, Ontario, Canada. Schrader, E. (1993). Untersuchungen zum Brutparasitismus von Fiederbartwelsen. DATZ 46(6), 426-434; and 46(8), 492-495. Schroder, S. L. (1981). “The Role of Sexual Selection in Determining Overall Mating Patterns and Mate Choice in Chum Salmon.” University of Washington, Seattle. Schroder, S. L. (1982). The influence of intrasexual competition on the distribution of chum salmon in an experimental stream. In “Salmon and Trout Migratory Behavior Symposium” (E. L. Brannon and E. 0. Salo, eds.), pp. 275-285. School of Fisheries, University of Washington, Seattle. Schroder, S. L., and Duker, G. J . (1979). “Assessment of Production of Chum Salmon Fry from the Big Reef Creek Spawning Channel,” Annual Report, pp. 1-14. Fisheries Research Institute, Coltege of Fisheries, University of Washington. Seattle. Semler, D. E. (1971). Some aspects of adaptation in a polymorphism for breeding colours in the threespine stickleback (Gasterosteus aculeatus). J . Zool. 165, 291-302. Shaw. J . (1836). An account of some experiments and observation on the parr, and on the ova of the salmon, proving the parr to be the young of the salmon. Edinburgh New Philos. J . 21, 99-1 10. Shoemaker, H. H. (1947). Pickerel and pumpkinseed coaction over the sunfish nest. Copein 1947, 195-1%. Shustcr, S. M. (1989). Male alternative reproductive behaviors in a marine isopod crustacean (Paracerreis sculpta). Evolution (Lawrence, Knns.) 34, 1683-1698.

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

97

Shuster, S. M., and Wade, M. J. (1991). Equal mating success among male reproductive strategies in a marine isopod. Nature (London) 350, 606-611. Shute, P. W., Shute, J. R., and Lindquist, D. G. (1982). Age, growth and early life history of the Waccamaw darter, Etheostoma perlongum. Copeia 1982, 561-567. Sigujonsdottir, H., and Gunnarsson, K. (1989). Alternative mating tactics of arctic charr, Salvelinus alpinus, in Thingvallavatn, Iceland. Environ. Biol. Fishes 26(3), 159-176. Sikkel, P. C. (1988). Factors influencing spawning site choice by female garibaldi, Hypsypops rubicundus (Pisces: Pomacentridae). Copeia 1988, 710-718. Sikkel, P. C. (1989). Egg presence and developmental stage influence spawning-site choice by female garibaldi. Anim. Behav. 38,447-456. Sikkel, P. C. (1990). Social organization and spawning in the Atlantic sharpnose puffer Canthigaster rostrata (Tetraodontidae). Environ. Biol. Fishes 27(4), 243-254. Sjolander, S. (1972). Feldbeobachtungen an einigen westafrikanischen Cichliden. Aquarien Terrarien 19,42-45, 86-88, 116-118. Smith, 0. R. (1941). The spawning habits of cutthroat and eastern brook trouts. J. Wildl. Manage. 5(4), 461-471. Smith, P. J. (1986). Spawning behaviour of snapper, Chrysophrys auratus, in captivity. N . Z. J. Mar. Freshwater Res. 20(3), 513-551. Smith, R. L. (1984). “Sperm Competition and the Evolution of Animal Mating Systems.” Academic Press, Orlando, FL. Soljan, T. (1930a). Nestbau eines adriatischen Lippfisches (Crenilabrus ocellatus Forsk.). Z . Morphol. Oekol. Tiere 17, 145-153. Soljan, T. (1930b). Die Fortpflanzung und das Wachstum von Crenilabrus ocellatus Forsk., einem Lippfisch des Mittelmeeres. Z . Wiss. Zool. 137, 150-174. Soljan, T. (1931). Brutpflege durch Nestbau bei Crenilabrus quinquemaculatus Risso, einem adriatischen Lippfisch. Z. Morphol. Oekol. Tiere 20, 132-135. Soltz, D. (1974). Variation in life history and social organization of some populations of Nevada pupfish, C,yprinodon nevadensis. Ph.D. Thesis, University of California, Los Angeles. Staats, F. (1988). Beobachtungen zum Brutparasitismus von Synodontis multipunctatus. DATZ 41, 402-403. Steele, B. D. (1978). The reproductive strategy and energetics of rosefin shiners, Notropis ardens, in Barebone Creek, Kentucky. Ph.D. Thesis, University of Louisville, Louisville, KY. Steele, B. D., and Pearson, W. D. (1981). Commensal spawning behavior of the rosefin shiner, Notropis ardens, in nests of longear sunfish, Lepomis megalotis. In “Ecology and Ethology of Fishes” (D. L. G. Noakes and J. A. Ward, eds.), p. 138. Junk, The Hague, The Netherlands. Susuki, K., Tanaka, Y.,and Hioki, S. (1980). Spawning behavior, eggs and larvae of the butterflyfish, Chaetodon nippon, in an aquarium. Jpn. J. Ichthyol. 26, 334-341. Taborsky, M. (1984a). Broodcare helpers in the cichlid fish Lamprologus brichardi: Their costs and benefits. Anim. Behav. 32, 1236-1252. Taborsky, M. (1984b).Temtory holders, sneakers and satellites: Alternative male strategies in a Mediterranean wrasse. Newsl. I.A.F.E. 7(3), 32-33. Taborsky, M. (1985a). Breeder-helper conflict in a cichlid fish with broodcare helpers: An experimental analysis. Behaviour 95, 45-75. Taborsky, M. (198513). Male reproduction in Symphodus (Crenilabrus) ocellatus. Newsl. I.A.F.E. 8(2), 20-21. Taborsky, M. (198%). On optimal parental care. Z. Tierpsychol. 70, 331-336. Taborsky, M. (1987). Cooperative behaviour in fish: Coalitions, kin groups and reciprocity.

98

MICHAEL TABORSKY

In “Animal Societies: Theoriesand Facts’’ ( Y . Ito, J . L . Brown. and J. Kikkawa, eds.), pp. 229-237. J p n . Sci. Soc. Press. Tokyo. Taborsky, M.. and Limberger, D. (1980). Lamprologus hrichardi, a cichlid fish with facultative polygyny and helpers at brood care and territory maintenance. Neivsl. I . A . F. E . 3, 63-64. Taborsky, M.. and Limberger, D. (1981). Helpers in fish. Behau. Ecol. Sociobiol. 8, 113- 145, Taborsky. M.. and Taborsky, B. (1993). Kooperation und Helfen bei Fischen. Prax. NaturH.~SS.B i d . 42, 26-29. Taborsky, M.. Hert. E., von Siemens. M., and Stoerig, P. (1986). Social behaviour of Lamprologus species: Functions and mechanisms. Ann. M u s . R . Afr. Cent., Sci. Zool. 251-7-1 I . Taborsky. M . , Hudde. B . . and Wirtz. P. (1987). Reproductive behaviour and ecology of Symphodus (Crenilabrus) ocellafrts, a European wrasse with four types of male behaviour. Behauioitr 102, 82-1 18. ten Cate, C.. and Taborsky. M. (1992).Toraise or to abandon a reduced clutch: A theoretical approach illustrated using ringed turtle-doves (Sfrepfopeliarisoriu). Auk 109, 594-600. Thorpe, J. E.. and Morgan. R. I. G. (1978). Parental influence of growth rate, smoking rate and survival in hatchery reared juvenile Atlantic salmon, Salmo salar L. J . Fish Biol. 13, 549-556. Thorpe. J. E . . and Morgan. R. I. G. (1980). Growth rate and smoking rate of progeny of male Atlantic salmon parr. Salmo salur L. J . Fish Biol. 17, 451-459. Thorpe. J . E., Morgan. R. I . G.. Talbot. L.. and Miles, M. S. (1983). Inheritance of developmental rates in Atlantic salmon (Snlmo sulnrj. Aquaculfitre 33, 119-128. Thresher. R. E . (1979). Social behavior and ecology of two sympatric wrasses (Labridae: Halichoeres spp.) off the coast of Florida. Mar. B i d . 53, 161-172. Thresher, R. E. (1984). “Reproduction in Reef Fishes.” TFH Publications. Neptune City, NJ. Thresher. R. E. (1985). Brood-directed parental aggression and early brood loss in the coral reef fish, Amntliochrornis polyacanfhus (Pomacentridae).Anim. Behau. 33, 897-907. Thresher, R. E . , and Moyer. J. T. ( 1983). Male success. courtshipcomplexity and patterns of sexual selection in three congeneric species of sexually monochromatic and dichromatic damselfishes (Pisces: Pomacentridae). Anim. Behau. 31, 113-127. Travis. J., and Woodward. B. D. (1989). Social context and courtship flexibility in male sailfin mollies. Poecilia latipinna (Pisces: Poeciliidae). Anim. Behav. 38(6), 1001-101 I . Trivers, R. (198.5). “Social Evolution.” Benjamin/Cummings. Menlo Park, CA. Turner. G. F. (I986a). Teleost mating systems and strategies. I n “The Behaviour of Teleost Fishes” (T. J. Pitcher. ed.), pp. 253-274. Croom Helm, London and Sydney. Turner. G. F. (1986b). Territorial dynamics and cost of reproduction of the colonial nesting mouthbrooder Oreochroniis mossambicus (Peters). J . Fish B i d . 29, 573-587. Unger. L. M.. and Sargent. R. C. (1988). Allopaternal care in the fathead minnow, Pimephales promelas: Females prefer males with eggs. Behau. E d . Sociobiol. 23, 27-32. van den Assem. J. (1967). Territory in the three-spined stickleback Gusferosteus acitleatus L.. Belzuuior. Suppl. 16, 1-164 van den Berghe, E. 1’. (1988). Piracy: A new alternative male reproductive tactic. Nature (London) 334, 697-698. van den Berghe. E. P. (19921. Parental care and the cost of reproduction in a Mediterranean fish. Behau. Ecol. Sociohiol. 30, 373-378. van den Berghe, E. P., Wernerus, F., and Warner, R. R. (1989). Female choice and the mating cost of peripheral males. Anim. Behau. 38, 875-884.

PARASITIC AND COOPERATIVE BEHAVIOR IN FISH REPRODUCTION

99

von Siemens, M. (1990). Broodcare or egg cannibalism by parents and helpers in Neolamprologus brichardi (Poll 1986) (Pisces: Cichlidae): A study on behavioural mechanisms. Ethology 84, 60-80. Voss, J. (1980). “Color Patterns of African Cichlids.” T.F.H. Publications, Neptune, NJ. Wallin, J. E. (1989). Bluehead chub (Nocomis leptocephalus) nests used by yellowfin shiners (Notropis lutipinnis). Copeia 1989, 1077-1080. Wallin, J. E. (1992). The symbiotic nest association of yellowfin shiners, Notropis lutipinnis and bluehead chubs, Nocomis leptocephalus. Enuiron. Biol. Fishes 33, 287-292. Waltz, E. C. (1982). Alternative mating tactics and the law of diminishing returns-The satellite threshold model. Behau. Ecol. Sociobiol. 10, 75-83. Waltz, E. C., and Wolf, L. L. (1984). By Jove-Why do alternative mating tactics assume so many different forms? Am. 2001.24, 333-343. Ward, J. A., and Barlow, G. W. (1967). The maturation and regulation of glancing off the parents by young orange chromides (Etroplus macu/atus:Pisces, Cichlidae), Behauiour 29, 1-56. Ward, J. A., and Wyman, R. A. (1975). The cichlids of the Resplendent Isle, Oceans 8, 42-47. Ward, J. A., and Wyman, R. A. (1977). Ethology and ecology of cichlid fishes of the genus Etroplus in Sri Lanka: Preliminary findings. Enuiron. Biol. Fishes 2, 137-145. Warner, R. R. (1982). Mating systems, sex change and sexual demography in the rainbow wrasse, Thalassoma lucasanum. Copeia 1982, 653-661. Warner, R. R. (1990). Male versus female influences on mating site determination in a coral reef fish. Anim. Behau. 39, 540-548. Warner, R. R., and Downs, I. F. (1977). Comparative life histories: Growth versus reproduction in normal males and sex-changing hermaphrodites in the striped parrotfish, Scarus croicensis. R o c . Int. Coral Reef Symp., 3rd, 1977. Vol. 1 , pp. 275-282. Warner, R. R . , and Harlan, R. K. (1982). Sperm competition and sperm storage as determinants of sexual size dimorphism in the dwarf surfperch, Micrometrus minimus. Evolution (Lawrence, Kans.) 36, 44-55. Warner, R. R., and Hoffman, S. G. (1980a). Local population size as a determinant of mating system and sexual composition in two tropical marine fishes (Thalassoma s p p ) . Evolution (Lawrence, Kans.) 34, 508-518. Warner, R. R., and Hoffman, S. G. (1980b). Population density and the economics of territoral defense in a coral reef fish. Ecology, 61,772-780. Warner, R. R., and Le,jeune, P. (1985). Sex change limited by paternal care: A test using four Mediterranean labrid fishes, genus Symphodus. Mar. Biol. 87,89-99. Warner, R. R., and Robertson, D. R. (1978). Sexual patterns in the labroid fishes of the western Caribbean. I. The wrasses (Labridae). Smithson. Contrib. 2001.254, 1-27. Warner, R. R., Robertson, D. R., and Leigh, E. G. J. (1975). Sexchange and sexual selection. Science 190,633-638. Wedekind, C. (1992). Detailed information about parasites revealed by sexual ornamentation. Proc. R. Soc. London 247, 169-174. Weigmann, C., and Lamprecht, J. (1991). Intraspecific nest parasitism in bar-headed geese, Anser indicus. Anim. Behau. 41,677-688. Wernerus, F. M. (1989). Etude des mecanismes sous-tendant les systemes d’appariement de quatre esptces de poissons labrides mediterranbens des genres Symphodus Rafinesque, 1810 et Thalassoma Linnt, 1758. Cah. Ethol. Appl. 9, 117-320. Wernerus, F. M., and Tessari, V. (1991). The influence of population density on the mating system of Thalassoma pavo, a protogyneous Mediterranean labrid fish. P . S . 2. N . I . : Mar. Ecol. l2(4), 361-368.

100

MICHAEL TABORSKY

Wemerus, F. M., Michel, C., and Voss, J. (1987). Introduction a \’etude de la selection sexuelle chez Syrnphodus ocellarus (Forskal 1755) et S. rnelanocercus (Risso 1810). poissons labrides mediterrankens. Cah. Erhol. Appl. 7(2), 19-38. Wicket, W. P. (1959). Observations on adult pink salmon behaviour. Fish. Res. BoardCan., Prog. Rep. Pac. B i d . Stn. 113, 6-7. Winn, H . E. (1958a). Observations on the reproductive habits of darters (Pisces: Percidae). Am. Midi. Nut. 59, 190-212. Winn, H. E. (1958b). Comparative reproductive behavior and ecology of forteen species of darters (Pisces: Percidae). Ecol. Monogr. 28, 155-191. Wirtz, P. (1978). The behaviour of the Mediterranean Tripterygion species (Pisces, Blennioidei). Z. Tierpsychol. 48, 142-174. Wisenden, B. D., and Keenleyside, M. H . A. (1992). Intraspecific brood adoption in convict cichlids: A mutual benefit. Behau. Ecol. Sociobiol. 31, 263-269. Witte, K. E.. and Schmidt, J. (1992). Eetra brownorurn. a new species of anabantoids Creleostei: Belontiidae) from northwestern Borneo, with a key to the genus. Ichrhyol. Explor. Freshwaters 2(4), 305-330. Wittenberger, J . F. (1979). The evolution of mating systems in birds and mammals. In “Handbook of Neurobiology” (P. Marler and J. Vandenbergh, eds.), Vol. 3, pp. 271-349. Plenum. New York. Woodhead. A. D., and Armstrong, N. (1985). Aspects of the mating behaviour of male mollies (Poecilia s p p ) . J. Fish Biol. 27, 593-601. Wootton, R. J. (1971). A note on nest-raiding behavior of male sticklebacks. Can. J. 2001. 49, %0-%2. Wootton. R. 3 . (1984). “A Functional Biology of Sticklebacks.” Croom Helm, London. Yamagishi, S. (1988). Polyandry and helper in a cichlid fish Julidochrornis rnarlieri. In “Ecological and Limnological Study on Lake Tanganyika and Its Adjacent Regions. V . ” (H. Kawanabe and M. K. Kwetuenda. eds.), pp. 21-22. Kyoto University, Kyoto, Japan. Yanagisawa, Y . (1985a). Parental strategy of the cichlid fish Perissodus rnicrolepis, with particular reference to intra-specific brood “farming out.” Environ. Fishes l2,241-249. Yanagisawa, Y. (1985b). Mating system of a cichlid fish, Xenorilupiu sp. In “Ecological and Limnological Study of Lake Tanganyika and Its Adjacent Regions. 111” (H. Kawanabe. ed.), pp. 22-23. Kyoto University, Kyoto, Japan. Yanagisawa, Y. (1986). Parental care in a monogamous mouthbreeding cichlid Xenotilapia fiauipinnis in Lake Tanganyika. Jpn. J. lchthycil. 33, 249-261. Yanagisawa, Y. (1987). Social organization of a polygyneous cichlid Lamprologus furcifer in Lake Tanganyika. Jpn. J. Ichrhvol. 34, 82-90. Yanagisawa, Y.. and Nshombo. M. (1983). Reproduction and parental care of the scaleeating cichlid fish Perissodus rnicrolepis in Lake Tanganyika. Physiol. Ecol. Jpn. 20, 23-3 I . Yanagisawa, Y .. and Ochi, H. (1986). Step-fathering in the anemone fish Arnphiprion clurkii: A Removal study. Anim. Eehau. 34(60). 1769-1780. Zimmerer. E. J . (1982). Size related courtship strategies in the pygmy swordtail, Xiphophorus niprensis. A m . Zoo/. 22(4), 910. Zimmerer. E. J.. and Kallmann, K. D. (1989). The genetic basis for alternative reproductive tactics in the pygmy swordtail, Xiphophorus nigrensis. Evolution (Lawrence, K a n s . ) 43, 1298-1307.

ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 23

Behavioral Ecology and Levels of Selection: Dissolving the Group Selection Controversy LEEALANDUGATKIN CENTER FOR EVOLUTIONARY ECOLOGY T . H . MORGAN SCHOOL OF BIOLOGICAL SCIENCE UNIVERSITY OF KENTUCKY LEXINGTON, KENTUCKY 40506

HUDSONKERNREEVE MUSEUM OF COMPARATIVE ZOOLOGY HARVARD UNIVERSITY CAMBRIDGE, MASSACHUSETTS 02138

We have good evidence that there are these two types of social or subsocial interactions among animals; the self centered, egoistic drives which lead to personal achievement and self-preservation and the group-centered, more-or-less altruistic drives that lead to the preservation of the groups or some members of it perhaps at the sacrifice of many others. The existence of egoistic forces in animal life has long been recognized. It is not so well known that the idea of the group-centered forces in animal life also has a respectable history. (Allee, 1943, p. 519) [I# should be possible to show that every adaptation is calculated to maximize the reproductive success of the individual, relative to other individuals, regardless of what effect this maximization has on the population. (Williams, 1966, p. 160) Populations may last a long while, but they are constantly blending with other populations and so losing their identity. They are also subject to evolutionary changes from within. A population is not a discrete enough entity to be a unit of natural selection, not stable and unitary enough to be “selected” in preference to another population. (Dawkins, 1982, p. 100) The fact that the population genetics of group selection, sex ratio theory, inclusive fitness, reciprocity and game theory, are all so similar to one another represents a unification of previously discrepant theories. In a slightly different intellectual climate, this would be regarded as a welcome event. So many people think of group selection as a bogey man, however, that they are reluctant to accept any connection with their own favored ideas, and all sorts of efforts go toward finding differences. (Wilson, 1983, p. 180) 101

Copyright 0 1994 by Academic Press, Inc. All rights of reproduction in any form reserved.

102

LEE ALAN DUGATKIN AND HUDSON KERN REEVE

I . INTRODUCTION How are behavioral and evolutionary ecologists to decipher the preceding series of seemingly contradictory quotes from some of the most respected names in the field? The levels-of-selection question-whether natural selection operates at the gene, individual, or group level-. is one that has daunted evolutionists since the time of Darwin. However, simply because a question proves difficult to answer, and continually divides a group of scholars into different ”camps.” does not necessarily make it an important question. In this chapter we address the question of what, if anything, behavioral ecologists stand to gain by incorporating a “hierarchical” approach (i.e.. considering selection at levels above that of the gene or individual) into their research program. Will such an approach generate new testable hypotheses regarding the evolution of social behavior? Will it allow empiricists to interpret prior studies in a new light? Or is the whole levels-of-selection debate “merely” an issue for theoreticians and philosophers‘? Before such questions can be addressed in any detail, a history of the levels-of-selection debate is desirable. We begin with a digression. Major issues in a discipline are often debated via a series of original papers, critiques of such papers, counter-critiques, and so on. We feel that although this procedure has some merit. there are alternative methods for addressing important issues, particularly those that seem to polarize their respective disciplines. One such method is to have proponents of different viewpoints collaborate on a review/perspective paper on the controversial issue at hand. That is what we are attempting here. One of us (HKR) was trained in the “individual/gene” camp of natural selection, while the other (LAD) was trained within the new group-selection school-yet we seemed to have converged on similar views with respect to many of the important issues in behavioral and evolutionary ecology. Such a collaboration is not easy as it requires participants to agree on how best to present the argument. If this were a simple task, the collaboration would not be needed in the first place. Yet if those who argue that levels of selection is a focus of their research program cannot come together and present the issues in a coherent manner, how can we expect others ever to do so? 11.

HISTORICAL PERSPECTIVE

Behavorial ecologists are primarily interested in adaptation and the process of natural selection in relation to behavior, including social behavior (Brown, 1975; Krebs and Davies, 1991; Alcock, 1989). When ad-

BEHAVIORAL ECOLOGY AND LEVELS OF SELECTION

103

dressing the question of the level(s) at which natural selection can operate, one might instinctively look to Darwin’s work for an answer. Although often thought of as a champion of the individual selection paradigm (Williams, 1966; Dawkins, 1976), Darwin’s work (1859, 1871) shows that, at times, he adopted a more hierarchical view in which selection might act at the level of the individual, the kin group, or the population. Though he clearly stressed selection at the individual level, Darwin recognized that some traits defied understanding at that level. For example, Darwin was fully aware of the altruistic (often self-sacrificial)behavior that characterizes much of the social insect world and felt that such behavior was a special difficulty which at first appeared to me insuperable, and actually fatal to the whole theory (Darwin, 1859, p. 228).

Darwin (1859) suspected that this type of social behavior could be explained by natural selection at the level of the family group, yet over a century elapsed before a convincing theory of kin-selected altruistic behavior emerged (Hamilton, 1964). In addition to individual and kin selection, Darwin (1871), at times also adopted a surprisingly modern group-selection perspective with respect to understanding the evolution of moral virture in primitive societies (Richards, 1987). Darwin argued that bravery may have been selected against within tribes (as brave individuals are more likely to be killed in battle), but that tribes with brave individuals were more likely to emerge victorious in any given battle (Darwin, 1871; Richards, 1987). After Darwin, and until the 1960s, evolutionary biologists and ecologists did not spend much effort arguing about the level at which natural selection acts. A peaceful coexistence seemed to exist between group- and individualselectionists, partly because the levels-of-selection question was simply not recognized. However, when it was recognized, each side seemed to present evidence in support of their argument without denying the importance of other levels (e.g., see Allee, 1943). As such, the rampant group selection of Kropotkin (1908), Allee (1943), and Emerson (1960) stood side by side with studies on selection at the level of the individual. (One exception was in the realm of human behavior, where biologists, psychologists, and sociologists argued vociferously on whether humans were inherently cooperative as a result of group selection-like forces or aggressive as a result of selection perpetuating selfish, violent behavior (see Mitman, 1988, for a review). Unfortunately, aside from Wright’s (1945) brief “island model” of group selection, virtually no theoretical work addressed the levels-of-selection question until the 1960s and 1970s. Peaceful coexistence disappeared in the 1960s with the works of Wynne-

104

LEE ALAN DUGATKIN AND HUDSON KERN REEVE

Edwards (1%2), Brown (1966), Hamilton (1964, 1975), Maynard Smith (19641, and Williams (1966, 1971). In the longest, most vehement, but least rigorous argument to date, Wynne-Edwards (1962) presented his case for group selection as the major force controlling population size. His most celebrated example of such population regulation was the daily chorusing behavior of numerous species of birds. He labeled such displays “epideictic” and felt they were a mechanism of censusing population density to avoid overpopulation, overutilization of resources, and eventual population level extinction. In his most detailed example, Wynne-Edwards (1962) also argued that territoriality was a mechanism preventing grouplevel extinction [see Weins (1966)and Brown (1969a,b)for early alternative views on the evolution of territoriality]. Unfortunately, Wynne-Edwards provided no theoretical underpinnings to his ideas [but see Pollock (1989) for what such a model might look like] and yet felt free to make broad sweeping, unsubstantiated remarks on the efficacy of group selection [see Wynne-Edwards (1986) for a toned-down version of these arguments]. By 1967, group selection had been dealt an apparently devastating set of blows by the works of Hamilton (1964) and Williams (1966). Williams (1966) eloquently argued that virtually all cases of purported group selection can be understood at the level of the individual. Invoking Occam’s Razor, he proposed that we need not invoke group selection as an explanation in such cases. With minor qualifications . . . , it can be said that there is no escape from the conclusion that natural selection. as portrayed in elementary texts and in most of the technical contribution of population geneticists, can only produce adaptations for the genetic survival of individuals. (Williams, 1%. pp. 7-8)

Given the impact of Adaptation und Natural Selection, it is interesting to note, that Wynne-Edwards (1962), Williams (1966) relies upon verbal arguments and a review of the evidence available at the time, rather than mathematical models, to support his argument (see Sober, 1984, for more on this). Williams’ case against group selection rests on two premises. (1) Although selection above the level of the individual is theoretically possible, in practice such selection will be weak because of the relative speed of within- versus between-population selection and (2) virtually no evidence exists that cannot be understood using the individual selectionist paradigm. Hamilton’s (1944) inclusive fitness models, however, provided a detailed mathematical argument supporting the claim that most altruistic behavior can be understood by supplementing “individual fitness” with inclusive fitness. Inclusive fitness takes into account the effect of an allele not only on its bearer, but on the relatives of its bearer as well (see Section

BEHAVIORAL ECOLOGY AND LEVELS OF SELECTION

105

II1,D for more details). Hamilton’s theory was seen as strong support for the individual selection school as, inclusive fitness was interpreted as being an expansion, rather than a replacement, of the classic view of individual fitness. Maynard Smith (1964) further separated inclusive fitness from group selection by coining the phrase “kin selection,” which he contrasted with his “haystack” model of group selection [see Dawkins (1976, 1979), Vehrencamp (1979), and Maynard Smith (1980, 1983) for a more detailed statement of this position]. The effect of Williams’ prose and Hamilton’s mathematical models on the status of group selection cannot be overstated-it is a blow that group selectionists are still recovering from today. As D. S. Wilson noted, “For the next decade, group selection rivaled Lamarckianism as the most thoroughly repudiated idea in evolutionary biology” (Wilson, 1983, p. 159).

Surprisingly, at the same time that Dawkins (1976) was developing his selfish gene theory as an extension of Williams (1966) ideas, a “new group selection” school rose from the ashes in the mid-1970s. While Dawkins (1976) was presenting his reductionist view that genes, but not groups (or for that matter individuals), qualified as “replicators” (i.e., enduring units of self-interest), the new group selection school was being spearheaded by D. S. Wilson’s trait-group models (1975, 1976, 1977) and the empirical and theoretical work of Wade (1977, 1978, 1979). Two critical features separate Wilson and Wade’s view of group selection from that of their predecessors, namely Wynne-Edwards, Allee, etc. First and foremost, they provided detailed genetic models that partitioned variance into within- and between-group components. Such models allow detailed predictions of the circumstances favoring the evolution of individual versus group beneficial traits [also see Price (1972), Eshel (1972), Boorman and Levitt (1973a,b), Levin and Kilmer (1974), Gadgil (1975), Cohen and Eshel ( 1976), and Matessi and Jayakar (1976) for other contemporary models of group selection and altruism]. Second, the definition of group was no longer confined to a reproductively isolated deme. Wilson (1975) introduced the term “trait-group” and defined it as a population “within which every individual feels the effect of every other individual” (Wilson, 1980, p. 22). As such, groups need not be spatially or temporally isolated and group selection need not be restricted to wholesale extinction of entire groups. Importantly, trait-groups are seen as embedded within a larger interbreeding population, and thus models of trait-group evolution are sometimes called intrademic selection models to distinguish them from the interpopulational or interdemic models such as those implied in the arguments of Wynne-Edwards (1962). The difference between “old”

106

LEE ALAN DUGATKIN AND HUDSON KERN REEVE

"Old" Group Selection

Trait-Group Selection

SSASS

c Population 3 goes extinct.

*

c

Area recobnized by dispersers from population 2

*

1

2

AAAAS

3

Reproduction in trait groups

c ASAAS AS

AAAAA SAAAS

A = .Altruist S = Selfish

c

Solid circle = Population Broken circle = Trait group

FIG. I . In the old group-selection model presented here. populations are reproductively isolated from one another. The probability that a group goes extinct is proportional to the frequency of altruists within it. Here, population 3 is assumed to go extinct (as it contains the greatest frequency of selfish types). Population 3's "patch" is then recolonized by individuals from population 2 (the population with the greatest frequency of altruists). Two factors make altruism unlikely to evolve under this scenario. First. because groups are

BEHAVIORAL ECOLOGY AND LEVELS OF SELECTION

107

group selection and trait-group selection is shown diagrammatically in Fig. 1). Much of the debate on levels of selection in the 1980s arose from apparent collisions between the ideas of Hamilton, Dawkins, Maynard Smith, and Williams and those of the new (intrademic) group-selection school. In the following we examine more closely the logical relationships among these ideas and assess whether there is in fact any foundation for the most recent controversy over levels of selection. 111. SOMEISSUESSURROUNDING THE CONTROVERSY

A. WHAT IS STIRRING ALL THE CONTROVERSY? THEEQUIVALENCE OF INDIVIDUAL AND INTRADEMIC GROUPSELECTION Much needless confusion has resulted from discussions treating individual and intrademic group selection as opposing evolutionary mechanisms. It seems likely that resistance to the new group selection models draws power from the earlier, triumphant, detonation of simplistic and improbable interdemic group selection models, even though the two kinds of group selection are quite different. (Of course, such reflexive resistance might have been avoided had the term “group” not been included as a descriptor of the new intrademic selection models.) In any event, the false opposition between individual versus the new group selection has created the illusion that the current levels-of-selection debate is empirical in nature. As a result, two evolutionary camps have arisen, one insisting that most evolution arises from selfish reproductive competition among individuals within a breeding population (from here on, we refer to this view as “broadsense individual selection”), the other asserting that selection results from a balance of between-group and within-group fitness variation (we refer

reproductively isolated, over time any group containing even one S type rapidly moves toward fixation of S . Second, groups with many altruists are unable to export their productivity (Wilson et al., 1992), since extinction of moderately sized groups is assumed to be rare. In the trait-group model, trait-groups are embedded within a population. Reproduction occurs within these groups. Again, selfish types increase in frequency, but trait-groups with many altruists produce more offspring than those with few altruists. After reproduction, trait-groups dissolve and a mixing phase occurs. As the mixing phase may occur numerous times during the life span of an individual, altruistic groups are provided numerous opportunities to export their productivity. After the mixing phase, trait-groups are reformed. Altruism evolves under this scenario when the increased productivity of groups with high frequencies of altruists outweighs the within-group advantage to selfish types.

I08

LEE ALAN DUGATKIN AND HUDSON KERN REEVE

to this view as “trait-group selection” or “new group selection”). The conceptual terms employed by the two different camps tend to be different, and even when similar the shared terms refer to different phenomena (Wilson and Dugatkin, 1993). The differences in meanings of the shared terms (such as “altruism” and “individual selection”) unfortunately have served as barriers to recognition of the fundamental identity of the approaches of the two camps. A number of theoretical investigations (Alexander and Borgia, 1978; Uyenoyama and Feldman, 1980; Wilson, 1980; Colwell, 1981; Crow and Aoki, 1982; Michod, 1982, Wade, 1985; Maynard Smith, 1987a; Queller, 1992ah) have shown that the mathematics of the gene-, individual-, kin-, and new group-selection approaches are equivalent. We will show (as have others in different ways; e.g., Grafen, 1984; Ratnieks and Reeve, 1991; Queller, 1992a,b) that this must be the case and that individual and trait-group selection are not alternative evolutionary mechanisms; rather, they are alternative pictures of the same underlying mechanism. As Grafen points out: Once the basis of the new group selection is understood . . , most kin selectionists should realize they have been new group selectionists all their lives (Grafen, 1984, pp. 83-84).

Wilson and Sober (1989) enthusiastically reiterate this point, which might suggest that theorists in the two camps are not far from merging ideas. However, most empirical and theoretical behavioral ecologists in the two camps appear to remain xenophobic. Logical equivalence of individual-selectionists and trait-groupselectionist pictures of selection does not mean that the two pictures are heuristically equivalent. Indeed, as we will attempt to show, the value of the distinction is that by emphasizing different aspects of selection, the two pictures are each especially well suited to modeling or summarizing different kinds of evolutionary scenarios. However, the demonstration of the logical translatability of one picture into the other does mean that evolutionary biologists are no longer justified in making certain strong claims, namely, (i) individual-selectionistscan no longer dismiss the theoretical results of the new group selection models as irrelevant on the premise that group selection is less theoretically robust and consequently less important than individual selection, and (ii) new group-selectionists can no longer maintain that they have discovered a truly new mechanism of trait evolution that serves as an empirical alternative to the broadsense “individual selection” conceived of by individual selectionists. In particular, claims of the sort “my system provides an example of intra-

BEHAVIORAL ECOLOGY AND LEVELS OF SELECTION

109

demic group selection, not (broad-sense)individual selection!” should be viewed as neither right nor wrong on the facts of the case, but as empirically empty. The crux of the problem is that individual-selectionists and the new group-selectionists differ only in that the latter dissect selective mechanisms in ways that the former do not. To illustrate this, imagine a diploid population with an allele A underlying some trait of interest. Suppose we draw imaginary boundaries around subsets of individuals belonging to this population, thus creating multiple, nonoverlapping groups of arbitrary size (ranging from one individual to the entire population). Let pi be the frequency of the allele in the ith groups and pi. be the frequency of this allele in the ith group after selection. Obviously, natural selection favors the trait when the overall mean frequency after selection exceeds the overall mean frequency before selection, that is, when (ZpJVc)/ZNi‘> (ZpiNj)/ZNl,

(1)

where Nir and Ni are, respectively, the number of individuals in the ith group after and before selection. If we multiply both sides by the identity 2/2 = 1, we can describe (1) in terms of the number of alleles as 2(Cp,Ni,)/2ZNi’> 2(ZpiNi)/2ZNi,

(2)

Inequality (2)simply and straightforwardly expresses the condition “total number of A alleles after selection, divided by the total number of alleles in the gene pool after selection, must exceed the total number of A alleles before selection, divided by the total number of alleles in the gene pool before selection.” There is nothing mysterious or controversial about the latter condition, and it makes no references to individuals, groups, or allele frequencieswithin those groups, and yet, by virtue of its identity with condition (I), it simultaneously encompasses both broad-sense individual selection and any form of trait-group selection that one may care to envision. In fact, this condition also encompasses “within-individual’’ or “gene-level’’ selection arising from intragenomic conflicts, such as conflicts between maternally and paternally inherited genes within an individual (Haig and Westoby , 1989). If broad-sense individual selection, genic selection, and trait-group selection all can be represented by a single condition based only on allelefrequencies, then they cannotfundamentally differ from one another. The feeling that broad-sense individual selection and trait-group selection are somehow different appears to have been fostered by the different meanings of “individual selection” adopted by the two evolutionary camps (also see Grafen, 1984). For broad-sense individual-selectionists, “individual selection” refers to evolution resulting from the higher popula-

110

LEE ALAN DUGATKIN AND HUDSON KERN REEVE

tion wide average fitness of individuals bearing some trait; for the new group-selectionists, “individual selection” more narrowly refers to selection only within groups (Wilson, 1983; Wilson and Sober, 1989)-the component of selection left over is then designated “between-group’’ selection. Thus, the new group-selectionists have simply partitioned the individual-selectionist’s concept of individual selection into two components, only one of which is called individual selection. This has unfortunately created the appearance that trait-group selection and broad-sense individual selection are opposing mechanisms rather than one piece of the mechanism and the whole mechanism, respectively. The resulting confusion has led both camps to eye each other with distrust.

B. THEEVILSOF NEWGROUPSELECTION FROM G E N E ’ S EYEPERSPECTIVE

THE

INDIVIDUAL OR

Some of the controversy surrounding the levels-of-selection debate stems from the belief that adopting one particular level of analysis may slow down or even halt our flow of progress in understanding the process of natural selection. This view is held by both camps. The fear of broadsense individual-selectionists seems to be that adopting a group-selection approach to the study of social behavior is, at best, a superfluous parlorroom game and, at worst, erases much of the progress made in evolutionary ecology since the publication of Williams’ (1966) classic book. This appears to be the most commonly held view on the subject within behavioral ecology (as a scan of virtually any text in the field will indicate). Consider the following two quotes: The reason for the vehemence with which Williams, Ghiselin and Lack and other opponents of group selection have argued their case is, 1 think, their conviction that group selection assumptions, often tacit or unconscious. have been responsible for the failure to tackle important problems. (Maynard Smith. 1976, p. 277) As for group selection itself, my prejudice is that it has soaked up more theoretical ingenuity than its biological interest warrants . . . I hope 1 may be forgiven for wondering whether part of group selection’s enduring romantic appeal stems from the authoritative hammering the theory has received ever since Wynne-Edwards (1962) did us the valuable service of bringing it out into the open. (Dawkins, 1982, p. 115).

But the feelings surrounding this issue are more volatile than this. For example, when discussing a paper on levels of selection by Sober (1987), Maynard Smith argues: It is therefore perfectly justified to study eyes (or, for that matter, ribosomes, or

BEHAVIORAL ECOLOGY AND LEVELS OF SELECTION

111

foraging behaviors) on the assumption that these organs adapt organisms for survival and reproduction. But it would not be justified to study fighting behavior of spiders on the assumption that this behavior evolved because it insures survival of the species, or to study the behavior of earthworms on the assumption that it evolved because it improves the efficiency of the ecosystem. This point may seem so obvious as not to need stressing. I can only say that was not obvious to everyone twenty years ago. If Sober’s way of describing the world is taken seriously, it will again cease to be obvious and someone (not me, next time) will have to do the job over again. (Maynard Smith, 1987b, pp. 147-148)

It appears from this quote (and others like it) that many of those who adopt a broad-sense individualist view of the universe apparently believe that group-selectionists uncritically assume that selection at higher levels is responsible for the evolution of virtually all interesting social behavior. This view is not totally unwarranted since many of the writings of groupselectionists like Wynne-Edwards (1962, 1986),Allee (195 l), and Lovelock (1979) have fueled such fears. It should be noted, however, that to remedy the excesses of the past, some advocates of the new group selection have been quite careful to avoid promoting the view that nature is replete with group-level adaptations and superorganisms. Social interactions may sometimes evolve as group-level adaptations, increasing the fitness of some groups relative to others, but the sweeping interpretation of most social interactions in this fashion is doomed to failure. Grandiose superorganism theories can be most effectively refuted by insisting on evolutionary mechanisms. Flat assertions that superorganisms lie outside of Darwinian theory or require such implausible conditions that they never exist in nature are quite unnecessary. (Wilson and Sober, 1989, p. 352)

The new group selection approach cannot be viewed as an idle parlor game, and it cannot force us to accept the ubiquity of group-level adaptations, because trait-group selection is just a different version of the same selective process that underlies individual or gene selection (see previous sections). This new version simply encourages a more flexible partitioning of natural selection and so can yield important, novel insights into its operation. In short, the fears of individual- or gene-selectionists are groundless.

C. THE EVILSOF BROAD-SENSE INDIVIDUAL OR GENICSELECTION IN THE EYESOF NEW GROUP-SELECTIONISTS New group-selectionists fear that a strict individual or gene’s eye view of natural selection, by compressing multiple, hierarchically arranged levels of selection into a single level, will cause us to miss many interesting

1 I:!

LEE ALAN DUGATKIN A N D HUDSON KERN REEVE

behavioral phenomena. The root of this discomfort is that selfish-gene (or individual selection) theory simply averages across all levels above the gene (or individual) to calculate the average effect of competing alleles (or competing individual traits). Such an approach answers “why does a trait evolve?” by “it has the largest average effect” (Wilson and Sober, 1989; see Fisher, 1941 for more on average effects). This is of course true, but this approach myopically focuses only on allele frequencies (or on overall individual reproductive success) and thus potentially fails to detect interesting effects of selection above the level of the gene (or individual), if they exist (Sober, 1984; Wilson and Sober, 1989). Dawkins himself recognizes this: 1 said that I preferred to think of the gene as the fundamental unit of natural selection and therefore the fundamental unit of self-interest. What I have done is to define the gene in such a way that I cannot help being right! (Dawkins, 1989, p. 33: author’s italics)

Dawkins incorporated hierarchy into his approach by introducing the concept of a “vehicle of selection,” defined as Any unit, discrete enough to seem worth naming, which houses a collection of replicators and which works for the preservation of and propagation of those replicators. . . . A vehicle’s success is measured by its capacity to propagate the replicators that ride inside it. (Dawkins. 1982, p. 114)

Dawkins argues that the controversy about group selection and individual selection is a “controversy about rival claims of suggested kinds of vehicles” (Dawkins, 1982, p. 82). To many new group-selectionists, the phrase “kinds of vehicles” suggests material for natural selection to operate on. Yet to their chagrin, Dawkins refuses to grant any real power to “betweenvehicle” selection above the level of the individual. The rivalry between individual organism and groups of organisms for the vehicle role, being a real rivalry, can be settled. As it happens the outcome, in my view, is a decisive victory for the individual organism. The group is too wishy-washy an entity. (Dawkins, 1989. pp. 254-255)

This is a critical point for new group-selectionists, because their attempts to apply adaptationist analyses to groups require that groups can be treated as vehicles and that between-vehicle selection produces adaptation at the group level. We emphasize that the latter notion of a group-level adaptation

BEHAVIORAL ECOLOGY AND LEVELS OF SELECTION

113

does not constitute a retreat to the old group-selectionism; it follows from our earlier arguments that such adaptations will always also be interpretable (although perhaps not as picturesquely) as broad-sense individual-level adaptations. However, a new group selection interpretation may be of greater heuristic value in some cases. By erecting the concept of a “vehicle,” Dawkins in effect began to build a bridge between his gene-selectionism and the new groupselectionism. When the bridge was half-built, however, he effectively abandoned it, apparently sensing that a completed bridge would lead him to a land inhabited by the dreaded group selection dragons. As argued in the previous section, such a fear was unnecessary, as is the resultant fear by group-selectionists that gene-selectionism is inherently too barren to illuminate many interesting evolutionary phenomena. The concept of a nested hierarchy of vehicles, each vehicle engaged in competition with other vehicles at the same level, has a close affinity with the levels-ofselection approach of trait-group-selectionists.To put this another way, Dawkins’ concept of the ‘‘extendedphenotype” itself extends far enough to connect the two supposedly divergent evolutionary camps, if only the members of both camps would fully grasp its implications. The concepts of the gene as a replicator and the individual or trait-group as a vehicle could be accepted by both camps without either camp having to make any theoretical concessions! OF FITNESS IN D. THEDIFFERENT FLAVORS

A SOCIAL

ENVIRONMENT

As stated in Section III,A, numerous papers have shown that the mathematics used in different levels-of-selectionmodels are equivalent. It is not our intention to review these papers, but one question-how inclusive fitness and trait-group fitness are related-arises repeatedly (Maynard Smith, 1976; Dawkins, 1979; Wilson, 1980; Grafen, 1984; Wade, 1985; Queller, 1992a,b). We believe that much of the misunderstanding surrounding this question results from confusion over how these different versions of fitness are calculated. Behavioral ecologists need to pay particular attention to how kinship fits into the level-of-selection debate, since much work in this field centers on social behavior and relatedness-in fact, Hamilton’s original paper (1964) is often cited as marking the birth of the field of behavioral ecology and sociobiology. Fitness in a selectively potent social environment has usually been framed in terms compatible with broad-sense individual selection, in one of two ways (Maynard Smith, 1980, 1983). As Maynard Smith notes in a discussion on models of evolution:

114

LEE ALAN DUGATKIN AND HUDSON KERN REEVE

‘‘If the interacting individuals are relatives. there is again a choice between the exact ‘neighbour-modulated fitness’ approach and the more intuitive ‘inclusive fitness’ method“ (Maynard Smith. 1983, p. 315).

The most common way to model interactions among relatives is to view individuals expressing a trait as entities that are selected to maximize the summed effects of the trait on the reproduction of their relatives and themselves, each effect being weighted by a coefficient of genetic relatedness (Hamilton, 1964). Although Maynard Smith (1980) refers to this as the “inclusive fitness” method, we refer to this more narrowly as the “kin selection” approach, because Hamilton ( 1964) intended inclusive fitneTs to be a broader term than kin selection (see the following). The kin selection approach must be amended when costs and benefits are nonadditive or selection is strong (Grafen, 1984; Queller 1984, 1985, 1992b). A second, and equally valid, way to characterize inclusive fitness in individual-selectionist terms is simply to average personal fitness over all individuals possessing the trait’s genotype, not just those expressing the trait; selection acts to maximize this averaged, classical personal fitness. This is the “neighborhood-modulated fitness approach” (Hamilton, 1964; Maynard Smith. 1980,1983; Grafen, 1984).This trait-group selection approach partitions selection, as does the kin selection (or, more generally, the inclusive fitness) approach, but the partitioning is performed with reference to groups of individuals, not with reference to an “individual‘seye view” of its relatives. For example, Johnson and Brown (1980) use this trait-group selection approach when examining aid-giving behavior in grey-crowned babblers, Pornastostomus temporulis, in which “social units are generally families. but not invariably so” (Johnson and Brown, 1980, p. 95). In trait-group selection, the partitioning is in terms of fitness within versus between (kin) trait-groups, which loosely parallels the personal and kin components, respectively, of inclusive fitness (when kin interact). The logical relationships among neighbor-rnodulated fitness, inclusive fitness (when kin are or are not involved), and within and between trait-group fitnesses are pictured in Fig. 2. Trait-group-selectionistshave argued that kin groups are just one type of trait-group and that kin selection theory is that subset of trait-group selection theory dealing with kin groups. Hamilton himself makes statements akin to this view. [KJinship should be considered just one way to getting positive regression of genotype in the recipient. . . . Thus the inclusive fitness concept is more general than “kin selection.” (Hamilton, 1975. pp. 140-141) []It obviously makes no difference if altruists settle with altruists because they are

BEHAVLORAL ECOLOGY AND LEVELS OF SELECTION

115

b

0 Neighbor-Modulated Fitness

Inclusive Fitness

“Broad-sense“ individual selection approach

Within (w)and between (b) Trait-Group Fitness

Trait-group selection approach

FIG. 2. The three flavors of fitness in a social environment: neighbor-modulatedfitness, inclusive fitness, and within- and between-trait-group fitness. Neighbor-modulated and inclusive fitness approaches are two versions of the broad-sense individual-selection approach. Neighbor-modulated fitness: Arrows represent fitness effects from the social environment on a focal individual (solid circle) bearing the genotype for the behavior. Fitness effects result from the behavior of conspecifics (straight arrows from open circles) such as kin and from the individual’s own behavior, if expressed (curved arrow). These fitness effects are averaged over all individuals bearing the behavioral genotype, regardless of whether that genotype is expressed. Inclusive fitness: Fitness effects are partitioned into effects of an individual who expresses the behavior (hatched circle) on other possessors of the behavioral genotype, such as kin (thick arrows), and into effects on the focal individual itself (thin, curved arrow). When the interactants are kin, the former effects are part of the kin component of inclusive fitness, the latter effects are part of the personal component of inclusive fitness, and kin selection is said to operate. Within- and between-trait-group fitness: Fitness effects are partitioned into those on an individual (solid circle) within a group, relative to others within the same group (w) and into those on the group considered as a whole (b). Note that if one draws a cricle around the inclusive fitness diagram and draws an arrow corresponding to b on the trait-group diagram, the inclusive fitness case = trait-groups of kin. related (perhaps never having parted from them) or because they recognize fellow altruists as such or settle together because of some pleiotropic effect of the gene on habitat preference. (Hamilton, 1975, p. 141)

Kin-selectionistsrespond by claiming that the only plausible trait-groups are kin groups, at least with regard to the evolution of altruism in the strongest sense. The argument is often stated as follows: Kin groups segregate altruists from nonaltruists “automatically ”-that is, individuals are born into groups that are already segregated by type; learning of kincorrelated cues during this period may enable relatives to segregate even after a period of separation. In the case of non-kin group selection, however, altruists and nonaltruists must be segregated into groups by some

116

LEE ALAN DUtiATKlN AND HUDSON KERN REEVE

other mechanism. In particular, for “strong” altruism (altruism that ha a negative effect on the absolute reproduction of the donor; Wilson, 1980 to evolve under this scenario, an altruist allele must have pleiotropit effects, that is, the allele must code for altruism and some other character istic that segregates altruists. However, if the allele has pleiotropic effects individuals who show the correlated behavior, but do not display altruism. will invade the population and altruism will spiral down to extinction (Dawkins, 1982). Does the evolution of strong altruism therefore require the existence of kin-groups? The large between-group variance needed to select for strong altruism in the absence of (the effect of) kinship per se can be created in a number of ways that do not involve pleiotropic effects. For example, consider altruistic and selfish behaviors conforming to the payoffs displayed in Fig. 3. Altruists pay a cost for their actions, but pairs of altruists outproduce all other pairs. As such, Fig. 3 qualifies as a Prisoner’s Dilemma. Now, consider a tit-for-tat (TFT) rule that instructs a player to be altruistic the first time it meets someone and to copy their behavior thereafter. If the number of expected encounters is above some threshold level, conditional altruism evolves. Thus, even when groups (in this case pairs) are formed randomly, conditional altruism is maintained because TFT segregates at the behavioral level, matching moves of altruism with moves of altruism and moves of selfishness with moves of selfishness [it should be noted, however, that for TFT to invade a population it must surpass (a generally very low) threshold frequency]. Note that an individual-selectionist cannot object that “The spread of the TFT strategy is a straightforward case of individual-level,not trait-group, selection!” Our point is that there exists a completely equivalent explanation for the maintenance of TFT in terms of trait-group selection; it is mere historical accident, not logical necessity, that the broad-sense, individualselectionist description was used initially. Thus it is simply false that trait-group selection cannot plausibly lead to strong altruism without kin selection, unless we reject the plausibility of the TFT theory for the evolution of cooperation! In sum, the inclusive fitness approach of broad-sense individualselectionism partitions selection in a way that is somewhat similar to that of trait-group selection models (although the terminology is quite different), and the less-used, neighbor-modulated fitness approach of individual-selectionism does not partition selection at all, as in most discussions of individual selection when kin are not involved. Thus, the theory of fitness in social environments comes in three flavors: inclusive fitness, neighbor-modulated fitness (both individual-selectionist), and within- and between-group fitness (trait-group-selectionist) approaches. All three are fundamentally equivalent.

BEHAVIORAL ECOLOGY AND LEVELS OF SELECTION

117

Player 2 Altruistic

Selfish

Altruistic

Player 1

Selfish

FIG. 3. A matrix for the Prisoner’s Dilemma. Arbitrary payoffs for player 1 are shown. The matrix qualifies as a Prisoner’s Dilemma if b > c > 0.

Readers therefore should be wary of claims that any one of the versions of fitness must be retranslated into one of the other versions. For example, it is not true that inclusive fitness arguments are inherently preferable to trait-group selection arguments (contra West Eberhard, 1981). Conversely, it is misleading to say that “maximizing inclusive fitness requires between-group selection” (Wilson and Sober, 1989; our emphasis). It is perfectly legitimateto frame one’s explanation completely within the terms of one version of fitness. The translatability of one version into another does not mean that it is necessary to perform the translation, although it might be heuristically useful in some cases, as we shall now argue. The equivalence of the different versions of inclusive fitness is sometimes obscured by debates that seemingly pit one version of fitness against another, but that in reality concern the relative importance of the components of fitness within one version. For example, suppose a trait-groupselectionist stresses that differential kin group productivity is the driving force for the evolution of altruistic behavior, whereas a kin-selectionist points to the coefficient of relatedness (r) as the critical feature driving altruism. Is this adebate about the utility of trait-group versus kin selection explanations? No. This controversy could be framed either within the context of trait-group selection explanations or within the context of kin selection explanations. In the context of trait-group selection, the debate is over whether altruism spreads primarily because of the differential productivity among or the genetic homogeneity within groups (both of which figure into the strength of between-group selection); in the context

I18

LEE ALAN DUGATKlN A N D HUDSON KERN REEVE

of kin selection, the debate is about whether altruism spreads because of the altruist’s relatedness to the beneficiaries or because of the reproductive benefits accruing to these relatives as a result of the altruistic acts. Only by keeping in mind the translation procedures that allow clear passage among the different inclusive fitness versions can such potential confusions be avoided.

E. THEROLE OF GAMETHEORY As game theory is perhaps the most popular modeling tool employed by behavioral ecologists, the relationship between such models of social behavior and individual and trait-group models of behavior also merits discussion. Though game theory models are typically represented as individualistic in nature (Maynard Smith, 1982), others have argued that twoperson games are equivalent to two-person trait-groups (Wilson, 1983; Maynard Smith, 1983; Wilson and Sober, 1989; See Dugatkin, 1990 for a discussion of this question in relation N-person games). For example, as shown in Section III,D, the Prisoner’s Dilemma can be viewed as a traitgroup model in which altruism is selected against within mixed groups {containingan altruist and a cheater: b > (b-c)/2),but favored by betweengroup selection (as b-c > 0). One behavior that helps illustrate the relationship between game theory, trait-group selection, and broad-sense individual selection is predator inspection in fishes. In many species of schooling fish, one to a few individuals (“inspectors”) approach a potentially dangerous predator to obtain information about the threat it poses. Since inspectors pass on at least some of the information they obtain to noninspectors (i.e., all school members obtain this benefit; Magurran and Higgrnan, 1988), but only inspectors assume the considerable risk associated with approaching the predator (Dugatkin, 1992), inspection behavior likely qualifies as an N person Prisoner’s Dilemma game. A broad-sense individual approach to this N-person game would examine the fitness of inspectors (averaged over all groups that contain inspectors) versus the fitness of noninspectors (again,averaged over all groups that contain noninspectors). Under certain cost-benefit schemes, inspection could evolve and would be viewed as an individually advantageous trait. Under the trait-group selection interpretation of this N-person game, it is explicitly recognized that within any group containing both inspectors and noninspectors, the latter have an advantage (since they do not pay the costs but obtain the benefits of inspection). It is only the fact that groups with many inspectors are better able to avoid predators than groups with no inspectors (i.e., betweengroup selection) that allows inspection to evolve. Thus, when between-

BEHAVIORAL ECOLOGY AND LEVELS OF SELECTION

119

group selection for inspection outweighs within-group selection against it, inspection evolves, otherwise it does not (see Section IV,B,I for more levels-of-selection and predator-inspection behavior).

EXPLANATIONS FOR THE EVOLUTION OF Iv. ALTERNATIVE BEHAVIOR: ANALOGIES AND EXAMPLES A. THE ROADMAP ANALOGY Next we outline some situations in which each view, new group and broad-sense individual selection, enhances understanding or facilitates modeling of the selective basis of a trait. Since the two approaches to phenotype evolution yield different (but not contradictory!) information about the same process, we argue that both should be included in the conceptual toolbox of evolutionary and behavioral biologists. The two pictures of selection can perhaps usefully be compared to two road-maps of differing detail: the broad-sense, classical individual selection view is like a highly schematized roadmap that omits the small towns and streets that connect segements of the main highway, whereas the new group selection view is more like a detailed roadmap that includes such features. These maps provide different amounts of information, even though both can be used to move from the point of origin to the same destination. Before describing the utility of each picture of selection, however, it is useful to give specific examples of how explanations under one picture translate into explanations under the other. We provide a number of such examples (Table I) in the hopes of weakening learned resistances to one or the other set of explanations, especially for those who have been trained (like each of us) to accept and promulgate one view. Admittedly, the reader might well have to exert considerable self-discipline in the attempt to accept the paired explanations as fundamentally identical. The key is to extinguish conditioned negative responses to the terms used by the opposing camp; once this barrier is hurdled, insights into the translation procedures are likely to be gained. The translation procedures are relatively straightforward. In passing from an individual-selectionist to a trait-group-selectionist view, we (1) draw imaginary boundaries around collections of interacting individuals, thus forming groups (our boundaries may or may not correspond to real geographic boundaries), (2) describe how the behavior of interest affects the reproductive output of an individual within the group relative to other members of the group, and ( 3 ) describe how the behavior affects the reproductive output of the group as a whole. Thus, in the vampire bat

120

LEE ALAN DUGATKIN A N D HUDSON KERN REEVE

TABLE I MAPPINGSOF BROAD-SENSE INDIVIDUAL-SELECTIONIST EXPLANATIONS ONTO TRAIT-GROUP-SELECTIONIST EXPLANATIONS: SOMEEXAMPLESO Broad-sense individual-selectionist view 1. Honeybee workers increase their

inclusive fitness by altruistically aiding their mother queen.

2. Foraging-specialist ant queens in groups of unrelated queens maximize their personal fitness by foraging (otherwise no workers are produced and the colony is susceptible to brood-raiding). 3. A bird feeding in a flock searches for a rich patch of food because this increases its offspring production (even though other birds parasitize this information on the location of the rich food patch). 4. Vampire bats exhibit reciprocal blood-feeding because of the longrun gains in personal fitness (cheaters are punishable). 5 . The tit-for-tat strategy in pairwise territorial disputes is evolutionarily stable against invasion by cheating strategies, if individuals have a high probability of interaction. 6. Colonial birds prey on neighboring pairs’ brood because this (a) increases their own offspring production or (b) reduces competition for scarce resources.

Trait-group-selectionist view 1. Altruistic helping by honeybee

workers is disfavored by withincolony selection but is more strongly favored by between-colony selection (the latter being potent because of genetic similarity within colonies). 2. Foraging is favored by strong between-colony selection (even though disfavored by within-colony selection). 3. Searching for food is favored by between-flock selection even though information parasitism results in negative within-flock selection. 4. Reciprocal blood-feeding is favored

by between-pair selection; checks against cheating prevent significant negative within-pair selection. 5 . The tit-for-tat strategy (when predominant) is favored over cheating when positive between-pair selection outweighs negative withinpair future selection for cheaters (Fig. 1). 6. Colonial bird predation on neighboring broods is favored by within-colony selection even though it is slightly disfavored by betweencolony selection.

References: ( 1 ) Ratnieks and Reeve (1991);(2) Rissing et (4) Wilkinson (1984); ( 5 ) Pollock (1988). Dugatkin (1990).

a / . (1989); ( 3 ) Barnard (1984);

BEHAVIORAL ECOLOGY AND LEVELS OF SELECTION

121

example (example 4 in Table I), individuals are grouped into reciprocally feeding pairs, and blood-feeding is seen as it behavior that reduces withinpair relative output (though only slightly, given checks against cheating) but increases between-pair reproductive output. The reverse translation is also simple: Take an individual’s fitness within the group (e.g., its proportionate offspring representation), multiply this relative fitness by the group’s overall absolute output, and sum across groups. Thus, in the vampire bat example, we would focus on the effect of blood-feeding (versus not blood-feeding) on a vampire bat’s overall expected personal reproduction.

B. USESOF THE INDIVIDUAL AND NEWGROUPSELECTION “PICTURES” 1 . Assessing the Utility of a Picture The two pictures of selection differ in their heuristic value. There are at least three ways by which their relative utility can be assessed: (1) economy of explanation, (2) modeling simplicity, and (3) hypothesisgenerating potential. By “economy of explanation” we mean the amount of verbal information needed to describe a selective mechanism adequately. Typically, the two pictures of selection differ in their economy of explanation, depending on the selective mechanism being considered. Obviously, more economical descriptions will tend to be more desirable as both memory and pedagogical aids. We might also prefer more economical explanations as a way of minimizing the number of postulated theoretical entities, in keeping with the philosophical principle of parsimony (Occam’s razor). By “modeling simplicity” we refer to the degree of difficulty in constructing mathematical models of the given selective mechanism. The two pictures of selection can lead to modeling approaches that differ dramatically in their complexity; we argue that on some occasions the individual selection picture suggests the more efficient modeling strategies; on other occasions, the new group selection picture is much more useful. By “hypothesis-generating potential,” we refer to the ability of a picture to suggest new phenomena worthy of scientific investigation. The two pictures of selection often differ in their research fertility, or at least in the kinds of hypotheses they generate. We will present a few examples that illustrate the differing hypothesis-generating abilities of the individual and new group selection views. a. Economy ofExplanation. On the surface, it might seem as if broadsense, classical, individual-selectionism would provide the most economical descriptions of evolutionary mechanisms, since this approach bores through all intervening levels of selection to reach the “bottom-line’’

122

LEE ALAN DUGATKIN AND HUDSON KERN REEVE

measure of overall classical fitness. Sometimes this is true. For example, suppose we wish to explain why unrelated males in a group compete for control of territories at sites characterized by high rates of encounter with females. The explanation: “Males compete for territory ownership (versus doing something else) because this maximizes their mating success, hence their expected reproductive success.” What more needs to be added to this explanation? There seems to be little point to an explanation of the form: “Males seek to acquire territories because positive within-malegroup selection overrides between-male-group selection (the latter being nonexistent if all females at a site are successfully inseminated). The last part of the preceding explanation adds nothing, since it does not refer to any causally important selective factors. This point can be made more mathematically rigorous by using condition (1) for the spread of an allele. Suppose that. for a large population, we seek the conditions under which a strategy is evolutionarily stable, that is, resistant to invasion by rare mutant alleles prescribing alternative behaviors (Maynard Smith, 1982). If we let allele A in condition (1) be that prescribing the strategy of interest. the condition for this strategy being evolutionarily stable reduces to

because, when the mutant is rare, I; N,. is approximately equal to C N;. In esence, condition (3) says that evolutionarily stable strategies (ESSs) will be those that tend to maximize the product p N (at least when those strategies are common), where p is the within-group frequency of the allele underlying the strategy and N is the reproductive output of the group (representing between-group selection; see Wilson, 1990, for a similar argument). It is important to note that in addition to total number of group members remaining after selection, the reproductive output of a group might include new groups founded by “propagules” sent out by the group. Now both p and N may be affected by the behavior of interest, so we would say that selection will tend to maximize the product p ( x ) N ( x ) ,where p(x) and N(x) are p and N, respectively, as functions of the level of the behavior, x. Now. returning to our point, when the group output N is unaffected by the behavior x (as in the preceding male competition example), the evolutionarily stable value of x will be that which maximizes p f x ) N = p ( x ) times a constant. The constant pIays no mathematical role in the maximization since the constant “drops out” when the derivative of the product is set equal to zero. Since N therefore is superfluous, there is no need to mention any between-group selection. The only critical term is p(x13 which is directly proportional to an individual male’s expected

BEHAVIORAL ECOLOGY AND LEVELS OF SELECTION

123

mating success; this provides the theoretical footing for our assertion that an explanation couched in terms of male mating success maximization contains all the information one needs to understand the underlying selective mechanism. In sum, when there is no or negligible between-group selection (in mathematical terms, when the N(x) component of the product p(x)N(x) does not depend strongly on the behavior, x, and thus can be treated as a constant), broad-sense, individual-selectionist explanations will be the more economical. Now suppose a behavior strongly affects both within-group allele frequency and group output, that is, selection acts to maximizep(x)N(x), with both components depending on the level of the behavior. For example, suppose, as in example 2 of Table I, that one ant queen in a foundress association of several unrelated queens takes on the risky task of foraging (as in Rissing et al., 1989). How do we explain why the single queen forages, given that it incurs a risk not accepted by the other queens? Suppose it turns out that foraging increases the rate of worker production (of all queens, not only the forager) and that more workers are better able to defend the colony, thereby increasing colony survivorship. In this case, foraging decreases the within-colony frequency of the foraging allele (because of predation risks while foraging) but increases the overall colony output (by increasing colony survivorship). In other words, foraging reduces p but increases N. Is it sufficient to explain foraging (versus not foraging) simply as a maximization of overall personal fitness? As we have seen from the formal equivalence of individual and new group-selection models, it would be absolutely correct to say this. However, this answer does not adequately explain why the foraging is favored because it glosses over important structural complexity in the way selection has acted. The trait-groupselection picture reveals this structural complexity in a systematic, hierarchical way. Thus, the trait-group-selection explanation for the foraging, that is, that it is favored because between-colony selection for an adequate worker defense force has overwhelmed within-colony selection against risk-taking foragers, is useful in that it places markers at the important foci of selection. Of course, one could describe these opposing selective forces without any reference to groups: “foraging by the foraging-specialist is favored over her not foraging because losses in queen expected survival due to foraging are, on average, compensated by increased expected survival of her reproductive brood due to enhancement of the worker defense force.” The latter explanation is perfectly adequate in that it identifies the same two relevant foci of selection as does the trait-group-selection explanation, but from a different angle: “negative within-colony selection” corre-

124

LEE A L A N DUGATKIN AND HUDSON KERN REEVE

sponds to “reduced individual survivorship of the forager” and “positive between-colony selection” corresponds to “increased survivorship of offspring due to enhanced worker defense force.” The key to the individuallevel explanation is that the two components of fitness are each averaged over and compared between individuals that choose to forage and those that choose not to forage, given that no other queen forages. In other words, this individual-level explanation can adequately reveal the structural complexity in the selective mechanism without any direct reference to groups or to the reproduction of the other nestmate queens. (This ant example has been touted by some as the quintessential example of traitgroup selection, as if this were an empirical discovery; of course, it could equally well be seen as an example of broad-sense individual selection.) Thus our point is not that one must employ the trait-group-selectionpicture (since the structure of selection can be adequately represented in several ways I, but rather that the latter picture provides a ready-made, systematic procedure for unveiling structural complexity in selective mechanisms. It is possible that in certain ecoiogical scenarios, within-group and between-group selection may act in the same direction. For example, suppose that in a mutualisitc insect society all the co-nesting females share equally in the reproduction and that total group reproduction is enhanced by group defense of the nest against predators. Suppose further that females failing to participate in group defense are killed by nestmates. In this case, both within- and between-group selection favor group defense. A hierarchical approach to this scenario, once again, allows one to separate causal factors. Finally, it should be mentioned that there exists a class of evolutionary explanations requiring a hybridization of the individual- and group-level pictures, that is, explanations of within-individual selection or intragenomic conflict (involvingphenomena such as meiotic drive or conflicts arising from genomic imprinting; see Haig and Westoby, 1989). In such cases an individual in a sense becomes a group, with between-group selection corresponding to effects of genetic elements on overall, averaged individual fitness and within-group selection corresponding to effects of genetic elements on their representation within an individual’s reproductive output. (Of course, it is only because genes within individuals cooperate so often that individuals and, at a higher level, groups of individuals can usefully be treated as vehicles at all.) Here again, adequately structured evolutionary explanations must refer to at least two components of selection, and, once again, a levels-of-selection approach allows achievement of this goal. b. Modeling Simplicity. It is obvious that the construction of mathematical models of the evolution of behavior (even social behavior) can

BEHAVIORAL ECOLOGY AND LEVELS OF SELECTION

125

proceed in a straightforward way without any reference to groups; there exist many such examples of both optimization (e.g., Stephens and Krebs, 1986; Mange1 and Clark, 1988) and game-theoretic models (Maynard Smith, 1982). What may not be so obvious is that some selective mechanisms are much easier to model with the trait-group selection picture than with the broad-sense individual selection picture. We illustrate two cases. Example 1: It is known that multiple queens often peacefully coexist in social insect colonies, dividing up the reproduction (Reeve and Ratnieks, 1993). Why don’t the queens attempt to kill each other to obtain complete reproductive control? Suppose that there are two queens per colony and that they are haplodiploid full siblings. Also suppose that fighting reduces total colony output by a factor Ilk and that a fighter wins a fight with a nonfighter with probability 1/2 + a, where a measures a “first-blow advantage” (Reeve and Ratnieks, 1993). How do we go about modeling the conditions under which a rare fighting strategy would be prevented from spreading? One option might be to use the usual additive, kin selection version of inclusive fitness (additive in the sense that the fitness effects of an action on self and on relatives are added together after the latter effect has been weighted by the genetic relatedness). However, this will not work in this case because the relevant fitness effects are nonadditive because of lethal fighting. Either terms will need to be added to the standard inclusive fitness expression to account for this nonadditivity (Queller, 1985) or, as shown by Queller (1992b), partial regressions of a queen’s fitness on both the frequency of the fighter allele in itself and on the frequency of the fighter allele in its partner will need to be calculated (the partial regression coefficients yield fitness terms that can be additively combined). At least for this example, a trait-group-selection picture suggests an easier modeling strategy: First list the different colony types and their relative frequencies At, which will be functions of the genetic similarity of the queens. For example, when the fighter allele is rare (frequency = p), colonies with one fighter and one nonfighter can be produced only by matings in which a heterozygous female mates with a male containing a nonfighter allele. This mating frequency occurs with approximate frequency 2p, and half of the colonies spun off from this mating will consist of one fighter and one nonfighter, so the overall frequency of colonies with a fighter and a nonfighter will be 2p x 1/2 = p. Next, calculate the frequency of the fighting allele within each type of colony after fights (i.e., pr) and calculate the colony productivity due to fighting in each type of colony (i.e., Nr = N if fights occur and kN > N if fights do not occur), then multiply these values together for each colony type, and sum the resulting products, weighting each colony type by its relative frequency.

126

LEE ALAN DUGATKIN AND HUDSON KERN REEVE

In the trait-group selection picture the allele fails to spread when Zcf;p,N,J Zf,.N,,< p . In this case, the solution is that fighting does not invade when a < 2(k - 1) (Reeve and Ratnieks, 1993). Note that the correct way to proceed is less obvious if one models this situation using an “individual’s eye” approach: “Now the focal queen’s fitness if it fights is 1/2 + a . . . but this is only if it faces a nonfighting queen, and the probability that it faces a nonfighting queen depends on the genetic relatedness. . . . Example 2: Suppose that we wish to explain why sterile social insect workers in a colony with a single, multiply mated queen fail to favor raising their full sisters over their half sisters. Let us assume that such intracolonial nepotism entails a cost in overall colony productivity, because nepotistic workers are less efficient at working for their colony than are nonnepotistic workers (e.g., see Ratnieks and Reeve, 1991). The neighbor-modulated fitness version of the individual selection approach provides little intuition for modeling this situation, because reproducing individuals bearing the nepotistic genotype (but necessarily not expressing it) will receive complex reproductive contributions from individuals expressing the genotype (i .e., nepotistic, nonreproducing workers). At first glance, the kin selection version of individual selection might appear to provide a simple modeling strategy, since the insect colony is just a group of relatives. However, these hopes again are dashed once it is realized that complex interactions among nepotistic and nonnepotistic workers within the colony invalidate use of the usual additive version of inclusive fitness. (Interactions among nepotists and nonnepotists are complicated because nepotists may produce changes in the within-colony frequency of the nepotism allele that are nonlinearly related to the frequency of nepotists within the colony.) The amended inclusive fitness for a nepotistic worker must again either (1) include additional terms (Queller, 1985) or (2) be framed in terms of partial regressions of individual fitness on individual and group gene frequencies (Queller, 1992b). The trait-group-selection approach suggests a much more straightforward modeling strategy for the reasons given in the first example. It turns out that, indeed, the latter picture yields the most tractable analysis of the conditions inhibiting the spread of intracolonial nepotism (Ratnieks and Reeve, 1991). In general, the trait-group-selection picture will be most useful (perhaps necessary) when complex interactions between many strategists and nonstrategists within a group nonlinearly affect both the intragroup frequency of the alleles underlying the strategy and the net group reproductive output. c. Hypothesis-Generating Potential. The individual- and trait-groupselectionist pictures are each useful in stimulating new hypotheses and 3,

BEHAVIORAL ECOLOGY AND LEVELS OF SELECTION

127

guiding future research. As implied in the foregoing, the broad-sense individual selection picture keeps our eyes trained on potentially the most economical explanations for behavioral phenomena. For example, this picture would lead us first to consider the possibility that a sufficient account of a behavior could be couched solely in terms of long-term personal reproductive success (or inclusive fitness), even if the behavior is imbedded in a complicated web of social interactions encompassing a large, cohesive group of conspecifics. As hinted in the preceding male competition example, the existence of physically delimited social groups does not guarantee that the trait-group selection picture will be the more useful. Thus, the value of the individual selection picture lies in its ability to generate hypotheses stripped of unnecessary references to betweengroup selection. In short, the individual selection picture encourages parsimonious explanations. A second advantage of the individual selection picture lies in its ability to prevent lapses into old-style (“good-of-the-species”) group selection arguments-the trait-group picture does not support such arguments but nevertheless is more likely to let such arguments “slip through” if the boundaries of the trait-group are uncritically allowed to expand to the limits of the entire breeding population. However, the individual selection picture may often fail to generate hypotheses that adequately account for the behavioral phenomenon of interest. For example, this picture alone is bound to be unhelpful in cases of intragenomic conflict (which may have behavioral consequences). In such cases we can employ the trait-group selection picture (i.e., a levelsof-selection approach). In general, a partitioning of fitness into withinversus between-group components may yield important insights into the structure of the selective mechanism (as when we tried to account for the absence of intracolonial nepotism in a genetically diverse insect society). One of these components may turn out to be superfluous, but the point remains that the richness of trait-group-selection explanations lies in their ability to isolate potentially important selective submechanisms at different hierarchical levels. Let us now compare the hypothesis-generating potentials of the broadsense individual and trait-group selection approaches in the particular case of predator inspection behavior, which was described in Section II1,E. Consider two ecological scenarios in which inspectors and noninspectors exist within groups at nonzero equilibria1 frequencies. In the first scenario, inspectors and noninspectors coexist in small isolated pools that fill during rainfalls, but always contain enough water to maintain fish predators and prey. Each pool contains one group of prey fish. In the

128

LEE A L A N DUGATKIN AND HUDSON KERN REEVE

second, imagine larger streams with many schools of prey and their predators. Applying the broad-sense individual-selection approach to the case of isolated pools, we would hypothesize that the net benefit to inspectors must equal the net benefit to noninspectors in each and every group, as coexistence within pools would not otherwise be possible. If we then examined the risk of predation (assumed to be the primary selective force in both scenarios) to inspectors and noninspectors we would predict no difference in mortality rate within pools. Although no data are available on inspection in such isolated ponds, we would guess that this broadsense individual-selection hypothesis would be quite accurate in these cases. In our larger streams, many groups of prey fish coexist, each remaining together for some period of time. Offspring produed by “parent schools” mix freely and form new schools (either randomly or via some rule for assorting). Since inspectors and noninspectors coexist, a broad-sense individual selection approach would, as described earlier, tend to generate the hypothesis that fitness of both types would be equal within groups and that the summed fitnesses of group members would not differ across groups. In contrast, the trait-group approach would not necessarily begin by assuming that inspectors and noninspectors have equal fitness within groups-although that possibility would not be ruled out a priori. A traitgroup hypothesis would postulate that inspectors paid some cost not paid by noninspectors, but that schools of inspectors outproduced schools of noninspectors. Thus in a predation experiment on prey mortality due to inspection, a trait-group hypothesis would predict that within any school containing both types, inspectors would have relatively low survival rates, but that schools with many inspectors would have higher overall survival rates than schools with few inspectors. Data available on mortality rates due to inspection in the guppy (Poecilia reticulata) support this hypothesis (Dugatkin, 1992). Ira terms of hypothesis-generatingpotential, then, the broad-sense individual-selection hypothesis seems to lead us on the right road in the case of predator inspection in isolated pools, whereas trait-group hypotheses appear to be better guides in the case of inspection in larger streams. Of course, as always, one can translate across approaches. For example, in the isolated pool case, trait-group-selectionists can argue that their approach produces the same hypothesis as the broad-sense individual model, once it is realized that this case is an example of pure within-group selection. Likewise, broad-sense individual-selectionists, just as in the case of the foraging queens, could provide a lengthy, perhaps convoluted, argu-

BEHAVIORAL ECOLOGY AND LEVELS OF SELECTION

129

ment that would make predictions similar to those generated by the traitgroup perspective for the case of inspection in streams. The point, however, is that the individual selection approach is ideally structured to generate potentially insightful hypotheses in the case of isolated pools, while the same is true of trait-group models and inspection in streams. Some appear to believe that trait-group-selection explanations predispose us to focus on between-group selection and overlook the potential for within-group conflicts. For example, West Eberhard (1981) expresses concerns that colony-level (group selection) models will tend to gloss over potentially important conflicts within insect societies. We claim that just the opposite is true, because the trait-group selection picture, when correctly applied, forces us to consider simultaneously both within- and between-group selective forces; this picture does not automatically entail the predominance of between- over within-group selection. Thus, the traitgroup selection picture does not compel us to accept a sociobiological universe filled with cooperation as opposed to selfish manipulation. Any assertion that between-group selection is the predominant mode of selection in nature is a different and, indeed, questionable empirical claim. V. CONCLUSION

It is not our intention to convert individual-selectionists to the hierarchical approach or vice versa. We chose the word “convert” in the previous sentence with some care. It appears that players in both camps of this debate approach the questions addressed here with an almost religious fervor. Perhaps this should not be surprising-after all, the question of at what level(s) natural selection can operate is a critical one in both evolutionary biology and evolutionary/behavioralecology. We have argued that to understand the controversy surrounding this issue, one must first understand the history of this tenacious debate. Only with this historical perspective can one understand the terminology used by participants and the questions that are truly at issue. The question is not whether one side is “correct” but rather whether behavioral ecologists ever stand to gain anything by employing a levels-of-selection approach. We believe the answer is a resounding yes. When studying certain types of social behavior the hierarchical approach may generate more insight and testable hypotheses than the individuallgene perspective, while under other scenarios, the reverse will be true. In no case will the two perspectives be mutually exlusive.

I30

LEE ALAN DUGATKlN AND HUDSON KERN REEVE

VI.

SUMMARY

The debate between individual- and group-selectionists has raged in evolutionary and behavioral biology for close to thirty years. Here, we attempt to present the debate in a manner that allows behavioral ecologists to judge whether incorporating the issue of levels of selection into their repertoire is beneficial. After placing the debate in both historical and theoretical perspective, we attempt to show that although the gene/individual and trait-group selection approaches are mathematically equivalent, each has its own heuristic value relative to (1) economy of explanation, (2) modeling simplicity, and (3) hypothesis-generating potential. Certain ecological scenarios are best approached from the individual selection road, whereas others are best traversed from a trait-groupselection path. Behavioral ecologists can only profit by having the “nuts and bolts” of each approach in their conceptual toolbox. Acknowledgments H.K.R. was supported by a Junior Fellowship from the Harvard University Society of Fellows. L.A.D. was supported by an NSERC International Postdoctoral Fellowship and a National Science FoundationlEPSCoR Grant to the Center for Evolutionary Ecology at The University of Kentucky. We thank Jerram L. Brown. Alan Grafen. David Haig, Manfred Milinski, David Pfennig, Charles Snowdon. Elliott Sober, Paul Sherman, Janet ShellmanReeve. David Stern. David Westneat, and David Sloan Wilson for critical comments on the manuscript. We also thank Manfred Milinski for facilitating the writing of this paper.

References Alcock. J. (1989). “Animal Behavior.” Sinauer Associates. Sunderland. MA. Alexander, R..and Borgia, G. (1978). Group sclection. altruism and the levels of organization of life. Anntt. Rev. Ecol. S y s t . 9, 449-47s. Allee, W. C. (1943). Where angels fear to tread: A contribution from general sociology to human ethics. Science 97,517-525. Allee, W. C. (1951). .‘Cooperation among Animals.’’ Henry Schuman. New York. Barnard. C. J., ed. (1984). “Producers and Scroungers.” Croom Helm, London. Boorman. S., and Levitt. P. R. (1973a). A frequency-dependent natural selection model for the evolution of social cooperation networks. Proc. Nu!/.Acud. Sci. U.S.A. 70,187-189. Boorman, S . A , , and Levitt, P. R. (1973b). Group selection on the boundary of a stable population. Tlieor. Popul. Biol. 4, 85-128. Brown. J. L. (1966). Types of group selection. Nafirre (London) 211, 870. Brown, J . L. (1969a). The buffer effect and productivity in tit populations. Am. Nut. 103, 347- 354. Brown, J . L. (1969b). Territorial behavior and population regulation in birds. Wilson Bull. 81, 293-329. Brown. J. L,. (1975). ”The Evolution of Behavior.” Norton. New York.

BEHAVIORAL ECOLOGY AND LEVELS OF SELECTION

131

Cohen, D., and Eshel, 1. (1976). On the founder effect and the evolution of altruistic traits. Theor. Popul. Biot. 10, 276-302. Colwell, R. K. (1981). Group selection is implicated in the evolution of female-biased sex ratios. Nature (London) 290, 401-404. Crow, J., and Aoki, K. (1982). Group selection for apolygenic behavioral trait: A differential proliferation model. Proc. Natl. Acad. Sci. U.S.A. 79, 2628-2631. Darwin, C. (1859). “On the Origin of Species.” Murray, London. Darwin, C. (1871). “The Descent of Man and Selection in Relation to Sex.” Murray, London. Dawkins, R. (1976).“The Selfish Gene,” 1st ed. Oxford Univ. Press, Oxford. Dawkins, R. (1979). Twelve misunderstandings of kin selection. Z. Tierpsychol. 51,184-200. Dawkins, R. (1982). “The Extended Phenotype.” Oxford Univ. Press, Oxford. Dawkins, R. (1989). “The Selfish Gene,” 2nd ed. Oxford Univ. Press, Oxford. Dugatkin, L. A. (1990). N-person games and the evolution of cooperation: A model based predator inspection behavior in fish. J. Theor. Biol. 142, 123-135. Dugatkin, L. A. (1992). Tendency to inspect predators predicts mortality risk in the guppy, Poecilia reticulafa. Behau. Ecol. 3, 124-128. Emerson, A. E. (1960). The evolution of adaptation in population systems. In “Evolution after Darwin’’ (S. Tax, ed.), pp. 307-348. Chicago Univ. Press, Chicago. Eshel, I. (1972). On the neighbor effect and the evolution of altruistic traits. Theor. Popul. Biol. 3, 258-277. Fisher, R. A. (1941). Average excess and average effect of a gene substitution. Ann. Eugen. (London) 11,53-63. Gadgil, M. (1975). Evolution of social behavior through interpopulational selection. Proc. Nafl. Acad. Sci. U.S.A. 72, 1199-1201. Grafen, A. (1984). Natural selection, kin selection and group selection. In “Behavioural Ecology: An Evolutionary Approach” (J. Krebs and N. Davies, eds.), pp. 62-84. Blackwell, London. Haig, D., and Westoby, M. (1989). Parent-specific gene expression and the triploid endosperm. Am. Nat. 134, 147-15.5. Hamilton, W. D. (1964). The genetical evolution of social behaviour. I and 11. J. Theor. Biol. 7, 1-52. Hamilton, W. D. (1975). Innate social attitudes in man: An approach from evolutionary genetics. In “Biosocial Anthropology” (R. Fox, ed.), pp. 133-155. Wiley, New York. Johnson, M. S., and Brown, J. L. (1980). Genetic variation among trait groups and apparent absence of close inbreeding in grey-crowned babblers. Behau. Ecol. Sociobiol. I , 93-98. Krebs, J . , and Davies, N., eds. (1991). “Behavioral Ecology: An Evolutionary Approach,” 3rd ed. Blackwell, London. Kropotkin, P. (1908). “Mutual Aid.” Heinemann, London. Levin, B. R., and Kilmer, W. L. (1974). Interdemic selection and the evolution of altruism: A computer simulation. Evolution (Lawrence, Kans.) 28,527-545. Lovelock, J. E. (1979). “Gaia: A New Look at Life on Earth.” Oxford Univ. Press, Oxford. Magurran, A. E., and Higgam, A. (1988). Information transfer across fish shoals under predator threat. Ethology 78, 153-158. Mangel, M., and Clark, C. (1988). “Dynamic Modeling in Behavioral Ecology.” Princeton Univ. Press, Princeton, NJ. Matessi, C., and Jayakar, S. D. (1976). Conditions for the evolution of altruism under Darwinian selection. Theor. Popul. Biol. 9, 360-387. Maynard Smith, J. (1964). Group selection and kin selection. Nature (London) 201, 1145-1146. Maynard Smith, J. (1976). Group selection. Q. Rev. Biol. 51, 277-283.

132

LEE ALAN DUGATKIN AND HUDSON KERN REEVE

Maynard Smith. J. (1980). Models of the evolution of altruism. Theor. Popul. B i d . 18, I5 1-159. Maynard Smith, J. (1982). “Evolution and the Theory of Games.” Cambridge Univ. Press, Cambridge. Maynard Smith, J. (1983). Models of evolution. Proc. R. Soc. London, Ser. B 219,315-325. Maynard Smith. J. (1987a). How to model evolution. In “The Latest on the Best: Essays on Evolution and Optimality” ( J . Dupre, ed.), pp. 119-131. MIT Press, Boston. Maynard Smith. J. (1987b). Reply to Sober. In “The Latest on the Best: Essays on Evolution and Optimality” ( J . Dupre, ed.), pp. 147-150. MIT Press, Boston. Michod, R. (1982). The theory of kin selection. Annu. Rev. Ecol. Syst. 13, 23-55. Mitman. G. (1988). From the population to society: The cooperative metaphors of W. C. Allee and A. E. Emerson. J . Hist. Biol. 21, 173-194. Pollock, G. (1988). Population, spite and the iterated Prisoner’s Dilemma. Am. J . Phys. Anthropol. 77,209-221. Pollock, G. (1989). Suspending disbelief-Of Wynne-Edwards and his critics. J . Euol. Biol. 2, 205-22 1. Price, G . R. (1972). Extension of covariance selection mathematics. Ann. Hum. Genet. 35, 485-490. Queller, D. C. i1984). Kin selection and frequency dependence: A game theoretic approach. Biol. J . Linn. Soc. 23, 133-413. Queller, D. C. ( 1985). Kinship, reciprocity and synergism in the evolution of social behavior. Nature (London) 318, 366-367. Queller, D. C. (1992a). Quantitative genetics, inclusive fitness and group selection. Am. Nar. l39, 540-558. Queller. D. C. i1992b). A general model for kin selection. Evolution (Lawrence, K a n s . ) 46, 376-380. Ratnieks. F. L.. and Reeve, H. K. (1991). The evolution of queen-rearing nepotism in social hymenopteran: Effects of discrimination costs in swarming species. J . Evol. B i d . 4, 93-1 15. Reeve, H. K.. and Ratnieks, F. L. W. (1993). Queen-queen conflicts in polygynous societies: Mutual tolerance and reproductive skew. In “Queen Number and Sociality in Insects” (L. Keller. ed.). Oxford Univ. Press. Oxford (in press). Richards, R. J. (1987). “Darwin and the Emergence of Evolutionary Theories of Mind and Behavior.” University of Chicago, Chicago. Rissing. S., Pollock, G.. Higgins, M., Hagen, R., and Smith, D. (1989). Foraging specialization without relatedness or dominance among co-founding ant queens. Nature (London) 338, 420-422. Sober. E. (1984). “The Nature of Selection: Evolutionary Theory in Philosophical Focus.” Bradford/ M IT, Cambridge. MA. Sober. E. (1987). Comments on Maynard Smith‘s “How to model evolution.” In “The Latest on the Best: Essays on Evolution and Optimality” (J. Dupre, ed.), pp. 133-146. MIT Press, Boston. Stephens, D.. and Krebs. J. (1986). “Foraging Theory,” Princeton Univ. Press. Princeton, NJ. Uyenoyama, M. K., and Feldman. M. W. (1980). Theories of kin and group selection: A population genetics prespective. Theor. Popul. Biol. 17, 380-414. Vehrencamp, S . (1979). The roles of individual. kin, and group selection in the evolution of sociality. In ”Handbook of Behavioral Neurobiology: Social Behavior and Communication” (P. Marler and J. Vandenbergh, eds.), pp. 351-394. Plenum, New York.

BEHAVIORAL ECOLOGY AND LEVELS OF SELECTION

133

Wade, M. J. (1977). An experimental study of group selection. Evolution (Lawrence, Kans.) 31, 134-153. Wade, M. J. (1978). A critical review of the models of group selection. Q. Rev. B i d . 53, 101-114. Wade. M. J. (1979). The primary characteristics of Tribolium populations group selected for increased and decreased population size. Evolution (Lawrence, Kans.) 33,749-764. Wade, M. J. (1985). Soft selection, hard selection, kin selection, and group selection. Am. Natl. U5,61-73. Weins, J. (1966). On group selection and Wynne-Edward’s hypothesis. Am. Sci. 54,273-287. West Eberhard, M. J. (1981). Intragroup selection and the evolution of insect societies. In “Natural Selection and Social Behavior” (R. D. Alexander and D. W. Tinkle, eds.), pp. 3-17. Chiron Press, New York. Wilkinson, G. (1984). Reciprocal food sharing in vampire bats. Nature (London) 308, 181-184. Williams, G. C. (1966). “Adaptation and Natural Selection.” Princeton Univ. Press, Princeton, NJ. Williams, G. C. (1971). “Group Selection.” Aldine, Chicago. Wilson, D. S. (1975). A general theory of group selection. Proc. Natl. Acad. Sci. U.S.A. 72, 143-146. Wilson, D. S . (1976). Evolution on the level of communities. Science 192, 1358-1360. Wilson, D. S. (1977). Structured demes and the evolution of group-advantageoustraits. Am. Nat. 111, 157-185. Wilson, D. S . (1980). “The Natural Selection of Populations and Communities.” Benjamin/ Cummings, Menlo Park, CA. Wilson, D. S. (1983). The group selection controversy: History and current status. Annu. Rev. Ecol. Syst. 14, 159-187. Wilson, D. S . (1990). Weak altruism, strong group selection. Oikos 59, 135-140. Wilson, D. S., and Dugatkin, L. A. (1992). Altruism. In “Key Words in Evolutionary Biology” (E. F. Keller and E. A. Lloyd, eds.), pp. 29-35. Harvard Univ. Press, Cambridge, MA. Wilson, D. S., and Sober, E. (1989). Reviving the superorganism. J. Theor. B i d . 136, 337-356. Wilson, D. S., Pollock, G. B., and Dugatkin, L. A. (1992). Can altruism evolve in truly viscous populations? Evol Ecol. 6, 331-341. Wright, S . (1945). Tempo and mode in evolution: A critical review. Ecology 26, 415-419. Wynne-Edwards, V. C. (1962). “Animal Dispersion in Relation to Social Behavior.” Oliver & Boyd, Edinburgh. Wynne-Edwards, V. C. (1986). “Evolution through Group Selection.” Blackwell, Oxford.

This Page Intentionally Left Blank

ADVANCES IN THE STUDY OF BEHAVIOR. VOL. 23

Genetic Correlations and the Control of Behavior, Exemplified by Aggressiveness in Sticklebacks THEOC. M. BAKKER ABTEILUNG VERHALTENSOKOLOGIE ZOOLOGISCHES INSTITUT UNIVERSITAT BERN CH-3032HINTERKAPPELEN, SWITZERLAND

I. GENETICCORRELATIONS AS A TOOLI N STUDYING THE CONTROL OF BEHAVIOR Many behavioral traits show continuous variation in a population. Such quantitative traits are under polygenic control, that is, are influenced by many genes of which each has a small effect on the phenotype. The continuous variation of quantitative traits is the result of their multifactorial inheritance on which is superimposed additional variation caused by environmental influences. These two causes of variation make the translation of genotype into phenotype rather indirect and leave much room for secondary influences on the expression of quantitative traits. The many genes involved in the long and indirect path from genotype to phenotype do not operate in isolation and each of them is amenable in varying degree to environmental influences. For a particular behavioral phenotype each step (gene) in this path may be influenced by genes that control other phenotypes. The interaction between genes that control different phenotypes can be very direct; the most extreme case being a gene that affects more than one phenotype, and is thus directly involved in the expression of different phenotypes. Geneticists call this pleiotropic gene action. Genes that are involved in the control of different phenotypes can also influence one another in more indirect ways, for instance, through nonrandom association of alleles at different loci, which is denoted by the technical term linkage disequilibrium. This disequilibrium between genes that are involved in the control of different phenotypes may be the direct consequence of physical linkage, thus decreasing the chance that recombination breaks down the gene associations, or may be upheld by selective forces that favor particular combinations of genes. Pleiotropy 135

Copyright 0 1994 by Academic Press. Inc. All rights of reproduction in any form reserved.

136

T H E 0 C . M. BAKKER

and linkage disequilibrium are influences of genes on different phenotypes that may be transmitted to the next generation. The interdependence of traits can be studied using the methods of quantitative genetics and becomes especially clear when one tries to change the expression of a particular trait by directional selection. As a rule, artificial selection changes not only the trait chosen as the criterion of selection. An array of other traits will also be affected by the selection regime. Which traits will show correlated responses and to what extent cannot be deduced alone from their phenotypic correlations with the selected trait. Although phenotypic correlations may be indicative of the strengths of correlated responses (e.g., Cheverud, 1988; Falconer, 1989), this is not necessarily so and may be misleading (Willis et al., 1991 ;Spitze et al., 1991). Among other factors, the correlated responses depend on the extent to which the variation of the trait directly subjected to artificial selection and of associated traits is influenced by common (pleiotropic) genes, that is. on the degree of genetic correlations between traits. Linkage disequilibrium is often thought to be unimportant for maintaining genetic correlations in approximate equilibrium (Turelli, 1985; Hastings, 1989; Burger, 1989). The chief cause for genetic correlation is therefore the manifold or pleiotropic action of genes (e.g., Bulmer, 1974). The degree of genetic correlation relates, though with some reservation especially when genetic correlations are low, to the proportion of genes that two traits have in common (Carey. 1988). Negative genetic correlations indicate a trade-off between traits. An example is given in Fig. 1. Consider the trade-off in energy used for reprodution (R) and for survival ( S ) . The fraction of energy allocated to reproduction is assumed to be genetic and also determines (by what is left) the fraction allocated to survival. For any fixed level of the total amount of energy, the genetic correlation between the traits R and S is negative (Bell and Koufopanou, 1986; van Noordwijk and de Jong, 1986; Houle, 1991: Stearns et al., 1991). This genetic correlation does not mean, however, that R and S are controlled by the same genes. The interdependence of traits. which is especially evident with quantitative traits, automatically means that natural and sexual selection do not act on single traits. Recent theoretical models and considerations of multivariate evolution have made it clear that knowledge of genetic correlations among traits is essential to understanding both the potential and constraints for phenotypic evolution (e.g., Lande, 1979, 1982; Cheverud, 1982, 1984; Maynard Smith ef al., 1985; Clark, 1987). The major detrimental effect of genetic correlation is to delay adaptation (e.g., Lande, 1982; Via, 1984; Via and Lande, 1985; Arnold, 1987).

GENETIC CORRELATIONS AND THE CONTROL OF BEHAVIOR

137

SURVIVAL

REPRODUCTION

\/ tA

S = (1 - B)A

R = BA

ENEFIGY

FIG. 1 . The allocation of energy between reproduction (R) and survivial (S). A stands for the total amount of energy required (A = R + S), and B is the fraction of energy allocated to R. The trade-off in energy used for reproduction and survival results in a negative genetic correlation between the traits reproduction and survival. Adapted from Stearns et al. (1991).

An important application of the quantification of genetic correlations is the identification of developmentally or functionally integrated traits (Stearns et al., 1991). There is a growing body of data concerning genetic correlations within suites of traits at various levels or combinations of levels: biochemical traits (Clark, 1990), life-history traits (e.g., Service and Rose, 1985; Dingle et al., 1988; Scheiner et al., 1989; Snyder, 1991), morphological traits (Atchley et al., 1982; Lavin and McPhail, 1987) and their development (e.g., Cheverud, 1982; Atchley, 1984), behavioral traits (Arnold, 1981; Via, 1984; Gromko and Newport, 1988; Stevens, 19891, integration of behavioral and morphological traits (Brodie, 1989, 1992), integration of behavioral and physiological traits (e.g., Garland, 1988), integration of behavioral and neuroanatomical traits (e.g., Crusio et af., 1989), integration of behavioral, physiological, and morphological traits (Fairbairn and Roff, 1990), integration of behavioral, life-history, and morphological traits (Palmer and Dingle, 1989), integration of behavioral, life-history, and physiological traits (e.g., Lynch and Roberts, 1984; Sulzback and Lynch, 1984), and the integration of behavioral and life-history traits (Gromko et al., 1991). These studies have greatly increased our understanding of multivariate evolution. The application of genetic correlations is largely unexplored in causal studies of behavior (but see, e.g., Lynch and Roberts, 1984; Sulzbach and Lynch, 1984; Garland, 1988; Crusio et al., 1989; Fairbairn and Roff, 1990), in which genetic correlations

138

T H E 0 C. M. BAKKER

may be used as tools in studying the physiological and neurobiological control of behavior. In the following section I present the results of a behavior-geneticstudy in sticklebacks that was set up to quantify the genetic variance-covariance structure of different forms of aggressiveness. The results suggest the integration of a complex suite of behavioral, life-history ,and endocrinologicai traits. The genetic variance-covariance structure of this complex of traits augments the insight into a motivational system. Additionally, evolutionary implications with respect to stickleback aggression can be deduced. Quantitative genetic methods are potentially powerful in studying the interplay of causal, functional, and evolutionary aspects of behavior. 11.

W H Y S T U D Y STICKLEBACK

AGGRESSION'?

Among the stickleback species, male three-spined sticklebacks (Gasterosteus nculeatus) have the highest levels of aggression, the most pronounced breeding coloration, and the best-developed morphological defense mechanism against vertebrate predators (Bell and Foster, 1993). It has been suggested that the relative freedom from predators has facilitated the change from breeding in areas of dense vegetation to the open (Morris, 1958: Wilz. 1971: Wootton. 1976, 1984). where competition for females would be more intense. This habitat shift may have permitted the evolution of male traits that enhance competitive abilities (e.g., high aggression levels: see the following) and attractiveness toward females (e.g., red breeding coloration: Milinski and Bakker, 1990). Furthermore, this fish species is remarkably variable for a wide array of features, including the aforementioned traits, and is actually a large complex of differentiated allopatric populations and biological species (Bell, 1984; Bell and Foster, 1993). Variation among three-spined stickleback populations is (like the exaggeration of the aforementioned male traits) often interpreted in terms of adaptation (Bell and Foster, 1993), but the genetics of most traits has not been studied. In studying the evolution of stickleback aggression, the assessment of heritable variation in male territorial aggression would be a necessary first step, but would be of limited value in understanding its evolution because natural and sexual selection do not act on single traits (e.g., Lande, 1988). Our understanding of the evolution of territorial aggression would gain substantiallyby knowing the important genetic relationships between territorial aggressiveness and other traits. Through reproductive physiology, territorial aggression has obvious links with other aspects of reproductive

GENETIC CORRELATIONS AND THE CONTROL OF BEHAVIOR

139

biology such as male courtship and coloration (e.g., Munro and Pitcher, 1983; Villars, 1983). Laboratory and field research on stickleback aggression and the situations in which it may occur has been strongly biased toward territorial aggression of reproductively active males for reasons of both conspicuousness and interest. Its occurrence among juveniles or subadults and among adult females is less well known. These other forms of aggression can be very pronounced (Bakker, 1986, 1993a; Bakker and Feuth de Bruiljn, 1988) and cannot be neglected when studying the evolution of aggression In this species. Thus, sticklebacks of both sexes show aggressive behavior in a variety of contexts. Consequently, aggression is subject to diverse selective forces. This diversity can be expected to be reflected in the underlying causal mechanisms and genetic bases of different forms of aggression, making stickleback aggressiveness a suitable example with which to study multivariate evolution. I have used multiple artificial selection experiments to evaluate the extent of common genetic control of different forms of aggression and to examine the underlying hormonal influences on aggressive behavior.

111. LIFE-CYCLEAND AGGRESSIVE BEHAVIOR OF STICKLEBACKS The three-spined stickleback, Gasterosteus aculeatus L., is a small fish (5-10 cm) that inhabits waters of the Northern Hemisphere and breeds in fresh or brackish water. In spring, male sticklebacks typically develop conspicuous nuptial coloration consisting of an orange-red throat and forebelly and blue-green eyes. Males interact aggressively while estabishing territories in shallow water, subsequently building a tunnel-shaped nest of plant materials that they glue together with a kidney secretion. The territory and nest are vigorously defended against intruders (rival males, large juveniles, females, other fish species). Also during courtship males may show aggressive behavior against the female. Males spawn with multiple females (up to 20: Kynard, 1978; or even more: T. C. M. Bakker, unpublished data), after which they care for the eggs and young, aggressively defending them against predators, which include cannibalistic conspecifics in many cases. Males may complete several breeding cycles during the breeding season and have thus a higher potential reproductive rate than females, although these are capable of spawning several times in a single season. According to expectation (Clutton-Brock and Vincent, 1991 ; Clutton-Brock and Parker, 1992), males compete aggressively for the females. In contrast, females are relatively rarely observed to compete aggressively for males, except courting females late in the breeding season

140

THE0 C. M. BAKKER

(Bakker, 1993a). Aggression is not restricted to reproductively active fish; it also occurs among subadults, nonbreeding adults, and juveniles (Bakker, 1993a).Consequently, aggression appears to occur throughout the stickleback life cycle and in nearly every social context. The diversity of stickleback aggression may also be reflected in the control of different forms of aggression. Estimating the genetic correlations among the different forms of aggression would be a powerful method of tackling this issue of common causality.

Iv.

CHOICE OF THE

BREEDING DESIGN

Because in sticklebacks, as is the case in most other organisms, inbred or otherwise genetically well-defined strains are not available, the refined behavior-genetic analyses that are feasible with, for instance, fruitflies or house mice cannot be done with this species. Information on the genetic architecture of behavioral traits in sticklebacks is therefore necessarily less detailed. However, many behavior-genetic studies on fruitflies and house mice had been started from a purely genetic interest often at the expense of their value for ethological, behavioral ecological, and evolutionary issues. Behavior-geneticstudies on less standard organisms necessarily start from natural variation in behavior and are often driven by ethological, behavioral ecological, or evolutionary questions. This enhances their chances of making significant contributions in these fields. When inbred or otherwise genetically well-defined strains are not available, there are two options left for quantitative genetic studies (Falconer, 1989). One is based on the resemblance between individuals of different degrees of relationship. This is the only possible approach when the possibility of (selective) breeding is restricted or precluded, but the relationship of individuals is known. The other option is artificial directional selection, which involves starting with a heterogeneous base population and in each successive generation choosing individuals at one extreme of the distribution of phenotypic values as parents for the next generation (for more sophisticated selection designs, see, e.g., Falconer, 1989). Directional selection effects the concentration of increasing alleles for the behavioral trail in question in the line selected in the upward directio? and of decreasing alleles in the one selected in the downward direction. Artificial selection provides the most unambiguous evidence for the contribution of additive genetic variation (variation of individual genetic differences that will be passed on to the offspring) to the phenotypic variation. Although selection experiments are not designed to unravel the genetic architecture in great detail, the resulting selection lines are suitable material for further genetic. ethological, ecological, or physiological studies. Additional infor-

GENETIC CORRELATIONS AND THE CONTROL OF BEHAVIOR

141

mation can be obtained when one trait is selected in opposite directions in independent selection lines: limitations and asymmetries in two-way responses contain information about the action of natural selection (e.g., Broadhurst and Jinks, 1974; Falconer, 1989; Frankham, 1990). The experimental methods that I used were partly dictated by the aims and the experimental animal and these methods of quantitative genetics will be treated in some detail in the following sections. In short, they consisted of a series of double, two-way selection experiments each based on the levels of particular forms of aggression. Some explanation of the terms “two-way,” “double,” and “series” is in place here in order to make their meanings clear. These terms will occur regularly later in the paper. By “two-way” I mean that, startingfrom the same base population, two independent selection lines were founded, one line for enhanced levels and one for reduced levels of a particular form of aggression. The term “double” describes a selection experiment involving two different forms of aggressiveness; one line (or pair of lines in the case of two-way selection) was selected for one form of aggressiveness and screened for another form of aggressiveness not directly selected for, while the other line was selected the other way round, thus selected for the other form of aggressiveness and the correlated response measured for the form of aggressiveness that served as the criterion of selection in the other line. The term “series” denotes that several double, two-way selection experiments were performed each with a different combination of criteria of selection. The experiments started from a natural stickleback population. Sticklebacks are suitable study objects because they possess a suite of attributes that makes quantitativegenetic studies with this species feasible; they can be kept in relatively large numbers in the laboratory under seminatural conditions, they have a great reproductive capacity, their generation time can be reduced to about 6 months in the laboratory, they can reproduce in the laboratory year-round by appropriate manipulation of day length and temperature, their behavior can be quantified reliably, and their ethology, ecology, morphology, and endocrinology are well studied (Wootton, 1976, 1984; Bell and Foster, 1993). V. PROSAND CONSOF ESTIMATING GENETICCORRELATIONS FROM SELECTION DESIGNS

This section is an account of methods for and pitfalls in estimating genetic correlation and is at times rather technical. Readers who are not interested in these methodological problems and details can skip the technical parts of this section without losing the thread and the essence of the chapter.

I42

T H E 0 C. M. BAKKER

Genetic correlations can be estimated in three ways (Falconer, 1989). The first method is based on the resemblance of related individuals and is thus analogous to the estimation of heritability that is defined as the proportion of variation in a phenotypic character in a population that is due to individual genetic differences that will be inherited by the offspring (Ridley, 1993). If a set of data permits the estimation of heritability, genetic correlations may be estimated from the same data set when two or more traits are measured on each individual. Instead of computing the components of variance of one trait from an analysis of variance, the components of covariance of two traits are computed from an analysis of covariance. The offspring-parent relationship can also be used for estimating genetic correlations. This is done by computing a so-called cross-covariance obtained from the product of the value of trait x in parents and the value of trait y in offspring, or from the reciprocal situation (trait y in parents multiplied by trait x in offspring). Usually either the geometric or arithmetic mean of the two cross-covariances is considered, but a separate analysis of the two estimates of the genetic correlation based on the two crosscovariances is useful for checking possible errors of the estimates arising from similarities between relatives caused by common environments (van Noordwijk, 1984). For example, in a natural population of great tits, the genetic correlation between the size of the eggs laid by the mother and body size of her daughters is higher than when we calculate the same genetic correlation the other way round, that is, with the size of the eggs laid by the daughters and the body size of the mother. Apparently, nongenetic influences bias the former genetic correlation (van Noordwijk, 19841. In the second method, genetic correlations are estimated from the responses to selection in a manner analogous to the estimation of realized heritability. If selection is exerted on trait x and the correlated response of trait y is measured, then the heritabilities of both traits must be known in order to compute the genetic correlation. Even if the heritability of trait y is not known. the correlated response of y gives valuable information about the maximum value of the genetic correlation and whether the genetic correlation between x and y is positive, negative, or. with some reservation, absent. Genetic correlations can also be computed from selection experiments without knowledge of the heritabilities of the traits. In that case a socalled double selection experiment (Falconer, 1989) has to be carried out, that is, line X is selected for trait x and screened for trait y , while line Y is selected for trait y and screened for trait x. So both lines are screened for both traits, but each is selected for a different trait (Fig. 2). Then both the direct and the correlated responses of each trait can be measured and

GENETIC CORRELATIONS AND THE CONTROL OF BEHAVIOR

LINE SELECTED FOR X

143

LINE SELECTED FOR Y

A

B

C >

B X

8 u1

8

FIG. 2. Possible outcomes of a double, one-way selection experiment when there exists (A) a strong negative genetic correlation (ra = - 1) between the traits X and Y, (B) a strong positive genetic correlation (ra = + l), and (C) when a genetic correlation between X and Y is absent (ra = 0). The responses of trait X and the correlated responses (in bold) of trait Y to selection for enhanced levels of X are shown on the left side. On the right side are the responses of trait Y and the correlated responses (in bold) of trait X to selection for enhanced levels of Y.

a joint estimate of the genetic correlation obtained from the equation (Falconer, 1989) r2A

=

(CR,/R,) (CRJR,) ,

where rA is the genetic correlation between traits x and y, and R symbolizes the response and CR the correlated response with subscripts x and y

I44

THE0 C. M. BAKKER

according to the trait referred to. When the genetic correlation between x and y is + 1 (Fig. 2B), then direct selection for x (in the line selected for x) and indirect selection for x (in the line selected for y) are equally effective in changing x. The same is true for trait y. With a genetic correlation of - 1 (Fig. 2A) the magnitude of the changes in x (or y) produced by direct and indirect selection for x (or y) are the same but in opposite directions. Without genetic correlation between x and y (Fig. 2C), selection for x has no effect on y, and vice versa. Hill (1971) provides a formula for the standard error of genetic correlation that is estimated this way. Since in the formula the standard errors of the two heritabilities appear in the numerator, an experiment designed to minimize the sampling variance of heritability will also have the optimal design for the estimation of a genetic correlation. There exists some scepticism around the estimation of genetic correlations for the following reasons. Estimates of genetic correlations are usually subject to large sampling errors and are therefore seldom very precise. Quantitative genetic studies based on the resemblance of related individuals require fairly large sample sizes relative to phenotypic studies (e.g., Klein et al., 1973; Gianola, 1979; Cheverud, 1988; Falconer, 1989). This is a consequence of a two-level sampling problem. Both the number of offspring per family and the number of families are important in determining the sampling variation of genetic estimates (e.g., Falconer, 1989). The method of two-way selection is somewhat less demanding as to the sample sizes necessary for reliable genetic estimates, because it accumulates genetic differences over a number of generations in either direction. Heritabilities that are estimated from two-way selection studies have smaller sampling variances than estimates from parent-offspring, full-sib, or half-sib analyses with the same total number of individuals recorded (Hill, 1971). The estimation of the sampling variance of heritability from selection studies is, however, not straightforward. In selection studies, realized heritability is usually estimated from the regression of cumulative response on cumulative selection differential (e.g., Falconer, 1989; for alternative estimators, see Hill, 1972a,b). The sampling variance of the regression coefficient, which had frequently been used as the sampling variance of the realized heritability, seriously underestimates the sampling variance of the realized heritability because of autocorrelation and may be one-tenth or less of the correct value (Dickerson, 1969; Hill, 1971, 1972a,b). Because in selection studies the number of selected parents is small, the variance of the population mean increases each generation as a result of genetic drift, and the generation means become correlated. In standard regression analysis the observations are assumed to have equal variance and be uncorrelated. After a few generations of selection most

GENETIC CORRELATIONS AND THE CONTROL OF BEHAVIOR

145

variance is contributed by drift. Hill (1972a,b) gives the correct expressions (though approximations) for calculating the sampling variance of heritability with different selection designs (one-way selection with or without a control line, two-way selection, selection in one or both sexes). The best estimate of the sampling variance of heritability is the direct estimate obtained from the variance between replicate selection lines. One or a few replicates will not be sufficient for this purpose. Two-way selection is the most efficient (ie., produces the smallest sampling variance of heritability) selection design (Hill, 1972b). The estimation of genetic correlations from the correlated responses of traits not directly selected for may pose interpretative problems (Henderson, 1989). In two-way selection experiments, modest but significant line differences in traits not directly selected for may be due to genetic drift and thus totally irrelevant to the originally selected trait. This problem is especially relevant when new traits are being investigated in already existing selection lines (most of which show considerable inbreeding) for which replicate lines are not available. Henderson (1989) provided a helpful decision diagram for evaluating genetic correlations between selected traits and correlated traits. The ideal experimental design would consist of replicate high- and low-selected lines. Consistency of direct and correlated responses between replicates would rule out the possibility that the responses were produced by genetic drift. An unselected control line may serve as an unselected replicate line (see the following)but is less powerful than the use of replicate high- and low-selected lines unless the unselected lines score well below or above high and low groups (Henderson, 1989). The size of the drift effect will be considerably smaller than that produced by common genetic influences when even a modest (0.25 < r, < 0.40) genetic correlation exists between two traits x and y (Henderson, 1989). When replicate lines are not available, effect size can be used as a guide to interpreting whether significant high- and low-selection line differences in y are likely to be due to drift or pleiotropy (Henderson, 1989). An experimental design consisting of double selection experiments reduces the aforementioned difficulties. In double selection experiments, the responses and correlated responses serve as mutual controls for drift effects and compensate for not having replicate selection lines, especially when a series of double, two-way selections are run (Bakker, 1986). The reason why such a design controls for genetic drift is the same as for replicate selection lines: consistency of direct and correlated responses in both designs reduce the probability that the observed changes are produced by chance effects. Let us consider a double, two-way selection experiment involving the traits x and y. In one pair of lines trait x serves as the criterion of selection (one line selected for high x and one for low

I46

T H E 0 C . M. BAKKER

x) and the correlated responses of y are measured in both lines. The other pair of selection lines is selected for high y and low y , respectively, and screened for changes in x. If genetic drift always occurs but in an arbitrary direction, then, after one generation of selection, there would be a 50% chance in each of the four selection lines of x and y changing in the same direction, that is of high x going with high y, or low x with low y. Since we have four independent selection lines, there would be a (0.514x 100% = 6.25% chance of a “perfect” association in all four lines by chance alone (i.e., always the appropriate association of high and low x’s and y’s). Probabilities are of course much less when observations are made over more than one generation. Probabilities are further reduced when we consider a series of double, two-way selection experiments, that is. they involve more lines selected for different traits. Including a third pair of selection lines for. say. trait z, and screened for the other two traits x and y (the lines selected for x and y should then also be screened for z), would make genetic drift a very unlikely cause for consistent direct and correlated responses. This is in essence the design that I applied in studying common genetic control of different forms of aggression in sticklebacks. The potential problems and difficulties in estimating genetic variances and covariances may be a considerable obstacle to initiating quantitative genetic studies of behavior with organisms that are not commonly used in genetics. This constraint becomes especially clear in studies on sexual selection through female choice. There exist a plethora of population genetic models for the evolution of male sexual ornaments through female choice (e.g., Maynard Smith, 1991). In all of these models, assortative mating will generate a positive genetic correlation through linkage disequilibrium between male ornaments and female preference for them as long as there is genetic variation for these traits. Although ten years ago Arnold (1983) advocated the measurement of this genetic correlation and gave a guideline of how to measure it, it was only recently estimated at the withinpopulation level (Bakker, 1993b). In cases where genetic correlations are caused by linkage disequilibrium, artificial selection may be a less appropriate method to estimate genetic covariance, unless special care is taken to maintain the genetic correlation. The estimation of genetic correlations between the sexes using artificial selection experiments has recently been debated with respect to male and female mating speeds in Drosophila melanogaster (Arnold and Halliday. 1992; Gromko, 1992; Butlin, 1993: Stamencovic-Radak ef d., 1993). At the species level, a positive genetic correlation between senders and receivers of sexual (in particular, acoustic and chemical) signals is of interest because it produces prezygotic reproductive isolation and thus

GENETIC CORRELATIONS AND THE CONTROL OF BEHAVIOR

147

promotes speciation. Hybrid females between closely related taxa show in many cases a mating preference for secondary sexual traits of hybrid males (e.g., Butlin and Ritchie, 1989; Boake, 1991; Ritchie, 1992). Neuroethologists have raised the possibility that this genetic correlation could result from pleiotropic effects of genes influencing the neural networks that control sending and receiving. The concept of common genetic or physiological control of these male and female behaviors has been termed genetic coupling. The evidence for genetic coupling is weak (Butlin and Ritchie, 1989), and backcrosses and recombination will be necessary to test linkage of genes.

AND THE CAUSATION OF AGGRESSIVE VI. GENETICCORRELATIONS BEHAVIOR: DOUBLESELECTION EXPERIMENTS

A. METHODS

I . Experimental Design To start the selection experiments with a genetically heterogeneous population, the base population was derived from laboratory-bred progeny of a large sample of parents (25 mating pairs) collected from a freshwater stickleback population (near Vaassen, Netherlands). I started independent selection lines for each of three forms of aggression each of which were selected for enhanced and reduced levels of aggression. In each generation about three parental pairs with extreme levels of aggression were selected to propagate the selection lines. In each line and in every generation about 15 progeny of either sex were tested for their aggression levels. Five generations (the base population and four selected generations) were involved. There are several possibilities for a control line in the analysis of direct and correlated responses to selection (e.g., Falconer, 1989; Gromko et al., 1991); (a) paired high and low lines may be used as controls for one another; (b) an unselected line may be maintained at the same effective population size as the selection lines (an inbred control); and (c) an unselected line may be maintained at a larger effective population size than the selection lines (an outbred control). I used a combination of options (a) and (c), because I was interested in the direct and correlated responses to selection for enhanced and reduced levels of aggression. Since the paired high and low lines were propagated from about the same number of parents, they provided a control for inbreeding depression. Consistent differences between paired high and low lines are most likely to be due

I48

THE0 C. M. BAKKER

to selection, regardless of whether selection has produced inbreeding depression or not. Responses to two-way selection for fitness traits are expected to be asymmetrical, with greater responses in the direction of lower fitness (e.g., Falconer, 1989; Frankham, 1990). Therefore, in addition to the two-way (also called bidirectional or divergent) selection lines, a separate control line was maintained that was propagated by about 10 different parental pairs. The use of such an outbred control line allows the analysis of asymmetry. The control line was maintained throughout the generations of selection and used partly to detect environmental deviations and partly for assessing responses and correlated responses in the separate high and low lines. For these purposes it was not necessary (though it would have been better) to screen the control animals in every generation; the control line was measured at generations 0 and 2, and average control line levels were calculated (Bakker, 1986). A further extension of the experimental design involved the screening of the fish in every generation and in each line for all the investigated forms of aggression. The study thus consisted of a series of double selection experiments permitting the estimation of genetic correlations among different forms of aggression. For practical reasons I had to refrain from the use of replicates, but the design of double selection experiments compensates for not having such replicate selection lines (see foregoing). This is especially true when the same traits are screened in several different two-way selection lines. Although inbreeding was avoided as much as possible, in later generations some inbreeding could not be avoided (generation 1 , no inbreeding; range of coefficients of inbreeding among lines in generation 3,0.12-0.25; Bakker, 1986). Further details on experimental design can be found in Bakker (1986).

2 . Behavioral Tests and Selection Lines To standardize rearing conditions, that is, to exclude paternal effects, clutches of eggs were transferred to an artificial hatching system shortly after fertilization.Juveniles used to establish each generation were isolated well before the onset ofjuvenile aggression in small (10 liter) tanks. Several forms of aggressiveness were quantified in standardized aggression tests (van Iersel, 1958). In these tests, the fish were offered an appropriate opponent in a glass tube inside their tanks or in a polyacrylic plastic chamber hung on the front of their tanks. When the fish reached the opponent, the duration of the aggressive acts of biting and bumping at the opponent was scored during 5 min. During the juvenile stage the aggressiveness of juvenile males and females toward a juvenile opponent was screened (juvenileaggressiveness).

GENETIC CORRELATIONS AND THE CONTROL OF BEHAVIOR

149

Adult females were presented with a subadult (female aggressiveness). The aggressiveness of reproductive, territorial males was measured toward a rival male (territorialaggressiveness)and toward a ripe female (courtship aggressiveness). Additionally, reproductive males were tested for their dominance ability. If two reproductive, isolated males are simultaneously introduced into a tank unfamiliar to both and just large enough for the settlement of one territory, then one of the males usually dominates the other after a short and intense fight (Bakker and Sevenster, 1983). The dominant male begins nest building, while the inferior male remains quiet at the water surface or hidden between plants where he is attacked by the dominant male if he moves. Dominance ability was measured by making all pairwise comparisons of relative dominance among a group of about 15 individually isolated males. The males can then be arranged in a linear order of dominance based on the probability of winning the dominance contests (Bakker and Sevenster, 1983; Bakker, 1985, 1986). Before starting the selection experiment, I measured the consistency of aggressive behavior in each of the test situations by calculating repeatabilities (Falconer, 1989). The levels of aggression in each situation were significantlyrepeatable (Bakker, 1986). Because I used the mean of several aggression tests per individual as the aggression score in the selection experiments, the repeatability of these scores would have been even higher (Falconer, 1989). Independent selection lines, one each for enhanced and reduced levels of aggression, were established for each of three forms of aggression. Juvenile aggressiveness in juveniles of both sexes served as the criterion of selection in founding lines with high (JH) and low (JL) levels ofjuvenile aggression. Similarly, territorial aggressiveness of adult males and female aggressiveness of adult females were used as the criteria of selection in the high (TH) and low (TL) territorial aggression lines. In establishing the high (DH) and low (DL) dominance lines, the male’s dominance ability was used as the criterion of selection. In addition to these six selection lines, an unselected control (C) line was maintained by breeding randomly selected adults. In the juvenile and territorial aggression lines both sexes were selected. Though facilitating quick responses to selection, this procedure complicated the analyses of the selection lines (details in Bakker, 1986).

B. DIRECT AND CORRELATED RESPONSESTO SELECTION I . Direct Responses Selection for reduced juvenile aggressiveness produced significant divergence from the control line in both sexes after one generation of selection. The differences increased in the ensuing two generations (Figs. 3A

150

THE0 C . M. BAKKER

FIG. 3. Responses to three generations of selection (solid circles) for high and low levels of juvenile ( J H and JL lines) and territorial (TH and T L lines) aggression. Aggressiveness in an unselected control iine ( C ) is indicated in generations 0 and 2 (open circles). Juvenile aggressiveness was measured and selected for in juvenile (A) males and (B) females. Territorial aggressiveness was measured and selected for in reproductive (C) males and (D) females. Aggressiveness is expressed as the mean percentage of biting and bumping time against an opponent during weekly 5-min standardized aggression tests. Error bars represent one standard error of the generation mean. Adapted from Bakker (1993a) by permission of Oxford University Press.

and 3B; Bakker, 1986). Selection for enhanced juvenile aggressiveness was less successful, producing significant divergence from the control line only in the second generation. In the third, levels of aggression were similar to those in the control line in the second generation. Similarly, selection for reduced territorial aggression produced significant divergence from the control line in reproductive males, but selection for enhanced aggression did not (Fig. 3C; Bakker, 1986). In the females, however, selection in both low and high territorial aggression lines produced signifi-

GENETIC CORRELATIONS AND THE CONTROL OF BEHAVIOR

151

cant differences from the control line by the third generation of selection (Fig. 3D; Bakker, 1986). So significant responses were obtained after a few generations of selection for reduced levels of juvenile aggression in both sexes, reduced levels of male territorial aggression, and reduced and enhanced levels of female aggression. The levels of aggression in the control line were in general rather constant, indicating limited environmental changes during the period of selection (Fig. 3). Territorial aggressiveness of control line males was probably underestimated in the base population (Fig. 3C) because of some slight methodological differences between aggression tests in generation 0 and later generations (Bakker, 1986). Because the standard errors of realized heritabilities in Bakker (1986) were estimated from the regression analyses, I reanalyzed the data to estimate correct standard errors according to expressions given in Hill (1972a,b). Furthermore, I used a square-root transformation of the aggression scores (increased by 0.5) of adult females to meet the normality assumptions of the analysis. Realized heritabilities were estimated from the regression of cumulative response on cumulative weighted selection differential (Table I); I have not included estimates for the high juvenile aggression line because of the apparently nonlinear response in both juvenile males (Fig. 3A) and females (Fig. 3B). The combined two-way responses yielded estimates that ranged from 0.23 to 0.37. These values agree with h2 estimates for aggressiveness in other species (reviewed in Bakker, 1986) and lie around the mean value for behavioral traits in general (Mousseau and Roff, 1987). Because of asymmetry in two-way response the estimates from the single selection lines had a much wider range (0-0.64). Analysis of dominance abilities was less straightforward because dominance had to be measured in contests between two individual males. The outcome of any contest therefore depends on the phenotypes of both males. In each generation, the joint response to two-way selection for dominance was determined from interline dominance tests with males randomly chosen from both lines. Selection for low and high dominance ability produced significant divergence between the two lines by the third generation (Fig. 4), at which time males from the high dominance line dominated males from the low dominance line in 19 out of 24 dominance tests (x’ = 8.17, P < 0.01; Bakker, 1986). In the second selected generation, males from the high dominance line won 5 out of 10 contests against control males, while males from the low dominance line won only 3 out of 10 contests. This nonsignificant trend was confirmed by the results of dominance tests between both dominance lines and the other selection lines in the third generation (see the following) suggesting that the divergence in the high and low dominance lines was due to a decrease in the dominance ability of males from the low dominance line rather than an

152

T H E 0 C. M . BAKKER

TABLE I REALIZEDHERITABILITIES (h2) FOR DIFFERENT

FORMS OF AGGRESSIVENESS~ Type of selection JL males JH + JL males JL females JH + JL females JL males +females TH males TL males TH + TL males TH femalesh TL femalesh TH + TL femalesb DH + DL males

hZ 2 SEc

*

0.51 0.28 0.37 2 0.19 0.64 2 0.28 0.25 2 0.14 0.57 2 0.20 -0.01 k 0.11 0.58 f. 0.18 0.23 2 0.1 I 0.34 f. 0.12 0.27 t 0.15 0.31 2 0.12 0.34'

Fd

Pd

43.17 46.76 73.51 1.73 80.76 0.01 592.83 21.07 164.96 4403.30 422.98

c0.012 CO.011 <0.007 >O. 15 <0.007 >0.46 <0.001 <0.023 <0.003 <0.001

<0.002

" Estimations from the regression of the selection response on the cumulative selection differential of the various selection lines. SE, standard error; F, variance ratio: P , one-tailed probability; JH, high juvenile aggression line; JL, low juvenile aggression line; TH, high territorial aggression line; TL, low territorial aggression line; DH. high dominance line; DL, low dominance line. Aggression scores (increased by 0.5). square-root-transformed. ' SE calculated according to Hill (1972a,b). F and P from regression analysis. Approximation based on ranks (see Bakker. 1986). t~

increase in that of males from the high dominance line. The ranking scale of dominance ability does not allow the formal estimation of heritability; an approximation based on ranks (see Bakker, 1986) was calculated and this fell within the range of heritabilities of other forms of aggressiveness (Table I). These results demonstrate that there is heritable variation for each of different forms of aggressiveness. The apparent lack of response of males to selection for enhanced territorial aggressiveness (Fig. 3C) and dominance ability (Fig. 4) can probably best be explained as a consequence of long-term selection for high levels of territorial aggression and dominance ability in the natural population (Bakker, 1986). An obvious relationship between aggressiveness and fitness operates via territory size of reproductive stickleback males. In homogeneous habitats, males with large territories initiated more attacks toward rivals (van den Assem, 1967; Black, 1971) and experienced superior reproductive success through enhanced courtship success and enhanced parental success (less often victim of

GENETIC CORRELATIONS AND THE CONTROL OF BEHAVIOR

153

FIG. 4. The response to three generations of selection for high (DH line) and low (DL line) dominance ability. Dominance ability in each generation was measured as the proportion of cases in which reproductive males from the high dominance line won dominance contests against males from the low dominance line in a small tank unfamiliar to both males (including the first unselected generation). Males from the high and low dominance lines were also scored for dominance abilities against the control line (C) in the second selected generation. Adapted from Bakker (1993a) by permission of Oxford University Press.

sneakers and less disturbed paternal care) both under seminatural conditions in the laboratory (van den Assem, 1967) and under field conditions (Goldschmidt and Bakker, 1990).In contrast, the apparent lack of response to selection for increased levels of juvenile aggression in both males and females (Figs. 3A and 3B) may result from reduced embryonic viability and correlated increases in female aggressiveness during courtship that reduce the probability of successful spawning. Both were observed in the high juvenile aggression line (Bakker, 1986). Moreover, juvenile aggressiveness, male territorial aggressiveness, and male dominance ability were each significantly greater in laboratory-bred offspring from wild-caught parents originating from the population used in the selection study than they were in offspring from an allopatric Dutch population (Bakker, 1993a). This result may suggest that natural and sexual selection have favored high aggression levels in the population used in the selection experiments.

154

THE0 C. M. BAKKER

2. Correlated Responses The fish in each selection line and in the control line were screened for all the investigated forms of aggression. The experimental design consisted in fact of a series of double selection experiments permitting estimation of genetic correlations (Falconer, 1989). Thus, each female in each line was assayed for both juvenile and female aggressiveness. Males were assayed for juvenile aggressiveness, territorial aggressiveness, dominance ability. and courtship aggressiveness. From these data, the mean score for each form of aggression was calculated for each generation in each line. A single mean for each form of aggression was calculated for each sex in the control line. The selection lines were then compared pairwise for two forms of aggression; one form of aggression had been directly selected for in one line, and the other form had been used as a criterion of selection in the second line. Thus, for example, as a direct response to selection, mean male temtorial aggression was determined for each of three generations of the low territorial aggression line. Three mean scores were also calculated for the correlated response ofjuvenile aggressiveness in the same line so that the correlation between the two forms of aggression could be examined. The same comparison was made for the reverse case, that is, between the direct response of juvenile aggressiveness in males of the low juvenile aggression line and the correlated response of territorial aggressiveness in the same line. We then have the same situation as was visualized in Fig. 2. When we combine the left and right graphs of Fig. 2 and plot trait x on say the x axis and trait y on the y axis, we can calculate two regression lines, one for each selection line. For selection line X the slope equals (correlated response of y)/(response of x), whereas for line Y this is (response of y)/(correlated response of x). From the theory of quantitative genetics (e.g., Falconer, 1989) it can be deduced that the more the slopes of the two regression lines differ from each other the smaller is the genetic correlation between x and y; the regression coefficients are equal and positive when r, = + 1, equal and negative when r A = - 1, and maximally different when rA = 0. Plotting the relationships between the direct and correlated responses of two forms of aggression in a double selection experiment gives us a first impression of the genetic relationships between different forms of aggression. The advantage of this method is that all lines can be included, thus also the ones that did not show significant heritability (namely, the high juvenile aggression line and high territorial aggression line). The graphs give us an impression of how closely the direct and correlated responses were matched among the generations of selection, and of the direction and extent of the genetic correlations. The estimation of genetic

GENETIC CORRELATIONS AND THE CONTROL OF BEHAVIOR

155

correlation only makes sense with selection experiments that demonstrate significant heritability. I therefore used only the low selection lines in estimating genetic correlations. The genetic relationship between juvenile aggressiveness in females and adult female aggressiveness was assessed using the means of the control line, high and low juvenile aggression lines, and high and low territorial aggression lines. The regression of juvenile aggression on female aggression was similar among the generation means of the juvenile aggression lines and those of the territorial aggression lines (Fig. 5A), despite the nonlinear response in females of the high juvenile aggression line (Fig. 3B). This indicates that the direct and correlated responses of both juvenile and female aggression were about equally strong, suggesting that the same loci affect both forms of aggression. The genetic correlation between female aggressiveness and juvenile aggressiveness was calculated using the most reliable data: because of the nonlinearity of response and correlated response in the high juvenile aggression line, only the data from the low juvenile aggression line and low territorial aggression line were used. The genetic correlation was estimated according to Falconer (1989) from the mean responses and correlated responses per generation (calculated from the regression of mean aggression on generations of selection; Nagai et al., 1978) and amounted to 1.05 (Table 11). The standard error of 0.12 is probably underestimated because the genetic correlation is close to one (Hill, 1971). In contrast, the correlated responses of both juvenile and temtorial aggression in males were less strong than were the direct responses of these two forms of aggression (Fig. 5B), suggesting that in males juvenile aggressiveness is only partly governed by the same genetic factors as territorial aggressiveness. The genetic correlation between juvenile aggression and territorial aggression in males was calculated using the mean responses and correlated responses per generation in the low juvenile aggression line and low territorial aggression line, and was 0.50 (Table 11). The significant positive correlation between territorial aggression (direct responses) and courtship aggression (correlated responses) among generation means of the high and low territorial aggression lines (Fig. 5C) suggests a positive genetic correlation between these two forms of aggression and may point to common genetic influences between these two forms of aggression (see also Sevenster, 1961). Selection for dominance ability produced little correlated change in other forms of aggressiveness; fish (males and females) from the high and low dominance lines did not differ significantly in any of the other forms of aggressiveness (males: juvenile, territorial, and courtship aggressiveness; females: juvenile and female aggressiveness) (Bakker, 1986).This suggests that male dominance ability is affected by genetic factors different from

156

THE0 C. M. BAKKER

8 0

1 0 20 30 4 0 5 0 60 JUVENILE AGGRESSIVENESS

I .

0 20 30 4 0 5 0 JUVENILE AGGRESSIVENESS 10

0

C

0 1 0 2 0 30 4 0 5 0 6 0 TERRiTORlAL AGGRESSIVENESS

FIG. 5. Phenotypic correlations between (A) juvenile and female aggressiveness, (B) juvenile and male territorial aggressiveness, and ( C )territorial and courtship aggressiveness. The data points (circles) on each graph represent the means of the aggression scores of all individuals in a single generation in a selected line. A single mean was calculated for the control line (open squares) using scores of individuals in the base population and in the second generation of the control line. (A) Relationship between aggression of juvenile and adult females in three generations of lines selected for high and low juvenile aggression (0pencircles;y = -0.151 + 0 . 8 6 0 = ~ ~0.93.F ~ = 65.90,d.f. = 1,5,P <0.001,2-tailed) and for high and low territorial aggression (solid circles; y = 2.293 + 0.880x,r2 = 0.54, F = 5.80, d.f. = 1.5, P = 0.06). (B) Relationship between aggression ofjuvenile males and territorial aggression of reproductive males in three generations of lines selected for high and low juvenile aggression (open circles; Y = 30.333 + 0.246x,r2 = 0.21, F = 1.32, d.f. = 13, P = 0.30) and for high and low territorial aggression (solid circles: y = 10.146 + 1.597x,r’ = 0.39, F = 3.17,d.f. = 1,5,P = 0.14). (C) Relationshipbetween territorial and courtship aggression of reproductive males in three generations of lines selected for high and low territorial aggression (solid circles; y = 7.253 + 0.259x,rZ = 0.70, F = 11.57.d.f. = 1.5. P = 0.02). Adapted from Bakker (1993a) by permission of Oxford University Press

157

GENETIC CORRELATIONS AND THE CONTROL OF BEHAVIOR

TABLE I1 REALIZEDGENETICCORRELATIONS (rA)BETWEEN DIFFERENTFORMS OF AGGRESSIVENESS~ ~

Direct response

Correlated response

Juvenile aggression

- 5.27

- 5.73

Female aggressionb

- 0.36

- 0.36

Juvenile aggression

-4.90

-2.90

Territorial aggression

-9.81

- 4.07

Form of aggression

r, r SE'

Females 1.05

* 0.12

Males 0.50 & 0.26 Realized genetic correlations were estimated according to Falconer (1989) from the mean direct and correlated responses per generation of males and females in the low juvenile and low temtorial aggression lines. Aggression scores (increased by O S ) , square-root-transformed. SE was calculated according to Hill (1971).

those affecting juvenile, territorial, or courtship aggressiveness. The apparent absence of genetic correlation between juvenile aggressiveness and dominance ability was further substantiated by the outcomes of interline dominance tests in the third generation of selection. Males from the high and low juvenile aggression lines had dominance abilities similar to males from the high dominance line (Bakker, 1986). Because a genetic correlation between juvenile aggressiveness and dominance ability was not detectable in males of four selection lines (high and low juvenile aggression lines, and high and low dominance lines), and because selection for dominance in males did not lead to significant changes in aggression scores of females of the high and low dominance line, it is unlikely that the absence of genetic correlation between the two forms of aggression was caused by genetic drift. Two-way selection for territorial aggressiveness resulted, however, in parallel changes in dominance ability; males from the high territorial aggression line had dominance abilities that were similar to (or even higher than) the dominance abilities of males of the high dominance line, whereas the dominance abilities of males from the low territorial aggression line were similar to those of males from the low dominance line (Bakker, 1986). This seems in conflict with the apparent absence of genetic correlation between territorial aggressiveness and dominance ability when one

158

T H E 0 C. M. BAKKER

considers the levels of territorial aggression in the high and low dominance lines (see the foregoing). In this population, dominance ability correlated significantly positively with the degree of red breeding coloration but was not significantly correlated with territorial aggressiveness (Bakker and Sevenster, 1983: Bakker, 1986, 1993a). Selection for dominance ability may thus have directly acted on genetic determinants of brightness of red breeding coloration that do not affect territorial aggression levels. Also, in the interline comparisons of dominance ability, differences between lines in the degree of red coloration explained the greater part of the variation in dominance ability (? = 0.92 and 0.76 in tests against males from the high and low dominance lines, respectively; Bakker, 1986,1993a). The conflicting correlated responses in the territorial aggression lines and the dominance lines are counterintuitive and may suggest a different genetic causation of color changes in these lines (see the following), leading to the apparent absence of genetic correlation between territorial aggressiveness and dominance ability in the dominance lines, but a positive one in the territorial aggression lines via the association of both traits with the degree of red breeding coloration. When interpreting the presence or absence of correlated responses one has to be aware of the pitfall of genetic drift, which may generate significant correlated responses in the absence of genetic correlations or prevent correlated responses from becoming significant in the presence of genetic correlations. In standard quantitative genetic analysis the effect of genetic drift is estimated from the consistency of correlated responses in replicate selection lines (Falconer, 1989). My choice of multiple double, bidirectional selection yielded more information than could have been obtained with the same number of tested individuals in a replicate selection design and a t the same time allowed for the control of drift effects. Drift effects are, however, less obvious in this design because they become visible through the inconsistency of responses and correlated responses in the double, two-way selection lines. Thus more reasoning is required in weighing selection effects against drift effects than in a design where drift effects are made directly visible in replicate lines. The up and down selection lines showed in general consistent correlated responses. Moreover, by the use of double two-way selection, direct and correlated responses could be compared in two sets of selection lines. The general consistency of the results suggests a limited influence of genetic drift. There is one weakness in the experimental design, namely, the application of simultaneous selection for male territorial aggressiveness and female aggressiveness in the territorial aggression lines. In view of the strong positive genetic correlation between female aggressiveness and juvenile aggressiveness (see the foregoing), some correlated responses of juvenile

GENETIC CORRELATIONS AND THE CONTROL OF BEHAVIOR

159

aggressiveness in males of the territorial aggression lines could have been expected even if juvenile and territorial aggressiveness in males were genetically uncorrelated. Does this mean that the calculated positive genetic Correlation between juvenile and territorial aggressiveness in males (Table 11)is nonexistent? No, because the other changes of aggressiveness in the juvenile and territorial aggression lines seem to support a positive genetic correlation between juvenile and territorial aggressiveness in males. If juvenile and territorial aggressiveness were genetically uncorrelated, then the responses of female aggressiveness in the territorial aggression lines (only one sex would have been selected in the absence of genetic correlation between juvenile and temtorial aggressiveness in males) would be half the correlated responses of female aggressiveness in the juvenile aggression lines (both sexes selected for the same trait) (but see Fig. 5A). Second, in the absence of genetic correlation between juvenile and territorial aggressiveness in males one would not expect a correlated response of territorial aggressiveness in the juvenile aggression lines (but, although not decisive, see Fig. 5B). The presence of some positive genetic correlation between juvenile and territorial aggressiveness in males therefore seems plausible, although influences of genetic drift are difficult to judge in this case. The genetic correlations among different forms of aggression were comparable in sign and magnitude with the corresponding phenotypic correlations in the base population (Bakker, 1985, 1986) as often is the case (e.g., Cheverud, 1988; Falconer, 1989; but see Willis et al., 1991). The genetic analysis suggests that stickleback aggression is characterized by a complex genetic correlation structure among different forms of aggression that may constrain the evolutionary response of specific forms of aggression. The demands made upon the levels of different forms of aggression may be quite different. The evolutionary trajectory that each form of aggressiveness follows might not only be dictated by the selection regimes acting upon the different forms of aggressiveness, but rather be a compromise imposed by the genetic relationships within this behavioral complex. Analyses of hormonal influences on stickleback aggression (see the following) suggest that the genetic correlation structure of aggression is part of a larger complex involving reproductive behaviors and several life-history characters. C. HORMONAL INFLUENCES Before treating hormonal influences on stickleback aggression, two points should be made clear. First, I have not directly measured hormone levels, so hormonal involvement in the control of aggressiveness was

160

T H E 0 C. M. BAKKER

deduced from a number of correlated changes in the various selection lines combined with data from the literature and should be considered as circumstantial evidence. Second, in the interpretation of causal mechanisms, some effects may have been due to genetic drift. Because I did not directly select for the life-history and morphological traits involved, the relationships of these traits with aggressive behavior could only be deduced from their correlated responses to selection for aggressiveness. The consistent correlated responses of the up and down selection lines as compared with control line levels, and the plausible biological interpretation that emerges, give faith in a correct interpretation that may guide more detailed studies. There existed differences in a number of life-history and morphological traits among the various lines selected for different forms of aggression. These differences suggested the involvement of two classes of hormones in the control of stickleback aggression (Bakker, 1985,1986, 1993a).Juvenile aggressiveness was significantly negatively correlated with the age at sexual maturity among generation means of the low and high juvenile aggression Lines (Bakker, 1986, 1993a). Thus, selection for juvenile aggressiveness was accompanied by a change in the age at sexual maturity, such that after three generations of selection both sexes of fish from the high juvenile aggression line matured on average about 2 weeks earlier than fish from the low juvenile aggression line (Bakker, 1986, 1993a).Additionally, after three generations of selection, the onset of juvenile aggression was on average about a week later in fish from the low juvenile aggression line as assessed in standardized groups of juveniles (Bakker. 1986). Finally, the incidence of female ripeness was significantly lower in fish from the low juvenile aggression line than it was in fish from the high juvenile aggression line (Bakker, 1986, 1993a).These results suggest that selection for juvenile aggressiveness has acted on the (effective) level of gonadotropic hormones because teleost gonadotropins induce spermatogenesis, spermiation, and testicular steroidogenesis in males and vitellogenesis, ovarian estrogen secretion, oocyte maturation, and ovulation in females (Idler and Ng, 1983; Ng and Idler, 1983). Gonadotropins are pituitary hormones, whose secretion is triggered by light (e.g., Slijkhuis, 1978; Borg et af., 1987).Secretion of gonadotropins in turn stimulates the production of gonadal hormones (androgens). Under winter conditions (short photoperiod, low temperature) androgens have a positive-feedback effect on gonadotropin synthesis (Borg et al., 1986). During the breeding season (long photoperiod, relatively high temperatures) androgens inhibit gonadotropic cells (Borg et al., 1985). Attainment of sexual maturity was delayed in the low juvenile aggression line, suggesting that selection on this aspect of life history could produce

GENETIC CORRELATIONS AND THE CONTROL OF BEHAVIOR

161

a correlated response in juvenile aggressiveness (and correlated forms of aggression). Similarly, selection on body size could indirectly influence juvenile aggressiveness because growth slows after attainment of sexual maturity (e.g., Wootton, 1976, 1984), and later-maturing individuals tend to be larger. Additionally, females in the high juvenile aggression line appeared to mature clutches more rapidly than did females in the low juvenile aggression line. This could indicate a positive genetic correlation between aggression and clutch maturation rate, or could simply result from smaller females producing smaller clutches at a higher frequency than larger females (but see Wootton, 1973). Data from the high and low juvenile aggression lines, and control line in the second selected generation where differences among lines in the level of juvenile aggression were greatest (Fig. 3), offered some support for the former explanation (Bakker, 1993a). Selection on territorial aggressiveness has apparently affected the level of androgen production rather than the level of gonadotropin production. Males from the high territorial aggression line had significantly enlarged kidneys (corrected for differences in body size) relative to control line males, a condition indicative of elevated levels of androgen production (Wai and Hoar, 1963; Mourier, 1972; de Ruiter and Mein, 1982), whereas males from the low territorial aggression line had significantly smaller kidneys than those of the controls (Bakker, 1986, 1993a).Parallel changes in the degree of red breeding coloration in males from the high and low territorial aggression lines (Bakker, 1986, 1993a) further support this hypothesis. Although genetic correlations between aggressiveness and most aspects of reproductive behavior have yet to be investigated, their common hormonal control and signaling system (red nuptial coloration of the male; Bakker, 1993a) make it likely that some aspects of aggressive and reproductive behavior will prove to have evolved in concert. The probable hormonal cause of such character correlations are the androgens, which have been implicated as determinants of aggressiveness, nest building, nest-directed activities, courtship behavior, and the secondary sexual characteristics of the male (see references in Bakker, 1993a).Thus, androgen levels, levels of some forms of aggression, nuptial coloration, and reproductive behavior tend to covary over the life cycle of the male. It follows from this that differences in androgen levels among individuals or populations can lead to positive correlations between these characters (Rowland, 1984; Giles and Huntingford, 1985; McLennan and McPhail, 1989). The interrelationships among the pituitary-gonadal axis, the aggressive system, the sexual system, and the breeding coloration of the male three-

162

T H E 0 C. M. BAKKER

spined stickleback in the sexual (empty nest) phase are summarized in Fig. 6. The proximate model in Fig. 6 is based on experimental data from the stickleback literature combined with the findings of this study. The model shows only those relationships that have been investigated in sticklebacks and is a simplification compared with possible relationships known from studies in other teleost fishes. For example. the influence of hormones on behavior is not one- but two-sided, that is, there is a feedback from behavior to hormone levels (e.g., Munro and Pitcher, 1983: Villars, 1983). Long daylength triggers the production of gonadotropins in the pituitary, which in turn stimulates the production of androgens in the testes. During the breeding season there is a negative feedback of androgens on gonadotropin synthesis. Both hormones of the pituitary-gonadal axis are assumed to have a positive effect on the aggressive system, which is activated by stimuli that threaten a male’s resources, in the case considered in Fig. 6 a male’s territory and nest. Every conspecific intruder in a male’s territory

ENDOCRINE SYSTEM

BEHAVIORAL SYSTEMS

SECONDARY SEXUAL CHARACTERISTICS

STIMULI

FIG. 6. Proximate model of the interrelationships among the pituitary-gonadal axis. the aggressive system. the sexual system. and red breeding coloration of the male three-spined stickleback in the sexual (empty nest) phase. Plus sign indicates stimulatory and minus sign inhibitory effects. The aggressive system is usually stimulated much more strongly by rival males than it is by ripe females. whereas for the sexual system the reverse holds (e.g., Sevenster. ]%I). The intensification of coloration is greater after activation of the sexual system than it is after activation of the aggressive system (e.g.. Bakker, 1993a). For further explanations see text.

GENETIC CORRELATIONS AND THE CONTROL OF BEHAVIOR

163

is a potential threat and will be attacked, most vigorously in the case of rival males. Territorial aggression is much less but not absent toward ripe females (compare levels of territorial aggression and courtship aggression in Fig. 5C).A conspecific intruder may on the other hand increase a male’s reproductive success and activate his sexual system. This increase in sexual activity is most evident when the appropriate stimuli are offered, that is, a female with a swollen abdomen that assumes a head-up courtship posture when attacked by the territory owner, but against male intruders some sexual activity can also be shown. Androgens but not gonadotropins are assumed to control the sexual system. Within an individual male there exists a mutually inhibitory relationship between the tendencies to show aggressive and sexual behavior preventing the simultaneous occurrence of enhanced levels of territorial aggression and enhanced levels of sexual activity. Finally, the male’s red breeding coloration, an androgendependent secondary sexual trait, is intensified after activation of the aggressive and the sexual behavioral systems. The greater intensification after activation of the sexual system points to the role of coloration in female choice (Bakker, 1993a). Although some forms of aggression are affected by androgen levels, gonadotropins may affect levels of most forms of aggression and exert a primary effect on some (see earlier, Fig. 6). For this reason, not all forms of aggression are positively correlated with reproductive behavior. For example, in lines selected for low levels of juvenile aggression and low levels of territorial aggression, males of the third and fourth selected generations displayed similarly low levels of aggressive activity during courtship. However, males from the low juvenile aggression line and the low territorial aggression line differed significantly in direct (courtship intensity expressed as the number if zigzags) and indirect (particular nestdirected activities) measures of sexual tendency; compared with males of the control line or the corresponding high line males, males from the low juvenile aggression line tended to display an enhanced sexual activity, whereas the sexual activity of males from the low territorial aggression line tended to be reduced (Bakker and Sevenster, 1989). These findings and the extensive literature on the hormonal control of male aggressiveness outside and during the reproductive period (reviewed in Bakker, 1993a) support the interpretation that levels of juvenile aggression are affected primarily by gonadotropin levels, whereas territorial aggression is also affected by androgen levels (Fig. 6). Only when territorial aggression was selected against was there a parallel correlated response in sexual activity, a character known to be affected by androgen levels (see the foregoing). Because there exists a mutually inhibitory relationship between the tendency to behave aggressively toward a stimulus

164

THE0 C. M. BAKKER

and the tendency to behave sexually over a short time period and within an individual male stickleback (see references in Bakker and Sevenster, 1989: Fig. 6), a one-sided reduction of the aggression level as in males from the low juvenile aggression line causes an increase in sexual activity. Thus, although males from both selection lines were equally aggressive during courtship, the different selection regimes had opposite effects on sexual activity.

VII.

CONCLUDING REMARKS

This study showed significant additive genetic variation in each of the different forms of stickleback aggressiveness. So natural and sexual selection can potentially change the levels of stickleback aggression in different social contexts. although there appeared little or no genetic variation for enhanced levels of most forms of aggression in this population. Furthermore, stickleback aggressiveness was characterized by a complex system in which different forms of aggressiveness were genetically correlated to varying degrees with each other. Natural or sexual selection pressures acting on particular forms of aggressiveness may thus have consequences for other forms of aggressiveness. For instance, selection for reduced levels of juvenile aggression will automatically result in reduced aggression levels of adult females, will also affect but to a lesser extent the level of male territorial aggression, but will have no influence on the males’ dominance abilities. Because aggressiveness seemed to be genetically integrated into a complex character suite involving sexual behavior, secondary sexual traits, and several life-history characters, the same selection for reduced levels of juvenile aggression may have, at first sight unexpected, consequences for characters other than aggressive behavior such as the age at first reproduction. Quantitative genetics thus provides us with tools that enable the study of the evolution and control of behavior. Of central importance is the estimation of genetic correlations. Knowledge of the degree and direction of genetic correlations is essential if we are to understand multivariate evolution. Knowledge of the physiological basis of genetic correlations is essential to understand the causal relationships between traits. In my study of stickleback aggressiveness the genetic correlations in the complex suite of aggression, sexual behavior, and life-history traits were partly based on the multiple influences of hormones of the pituitary-gonadal axis. A comparison of two stickleback populations that differ in their lifehistory mode (migratory versus nonmigratory) may illustrate the genetic

GENETIC CORRELATIONS AND THE CONTROL OF BEHAVIOR

165

integration of diverse traits and its evolutionary implications. Anadromous and freshwater populations should exhibit differences in life-history and behavioral characters because of the higher costs of a migratory life-style (e.g., Steams, 1976; Roff, 1988), which leads to a so-called migratory life-history syndrome, in which the relevant life-history characters show positive genetic correlations (e.g., Dingle, 1988). Juveniles of anadromous populations often show reduced aggression levels (sticklebacks: Honma and Tamura, 1984; Bakker & Feuth-de Bruijn, 1988, unpublished data; Bakker, 1993a;salmonids: e.g., Keenleyside, 1979). Low levels ofjuvenile aggression may be favored because juveniles in anadromous stickleback populations migrate to the sea in large shoals (e.g., Daniel, 1985). Lines selected for enhanced and reduced levels of juvenile aggression suggested that juvenile aggressiveness was genetically correlated with the migratorylife history syndrome via gonadotropins (see foregoing). So, low levels of juvenile aggression in anadromous sticklebacks may also have been evolved as a correlated response to selection acting on life-history traits of the migratory-life history syndrome. I compared aggression levels in laboratory-bred offspring from a Dutch freshwater and anadromous population (Bakker and Feuth-de Bruijn, 1988, unpublished data; Bakker, 1993a).The laboratory-bred freshwater fish were significantlymore aggressive and territorial during the juvenile stage than were the laboratorybred anadromous fish. Sexual maturity of fish (males and females) of the anadromous population was significantly delayed, and freshwater males had slightly but significantly higher levels of territorial aggression but significantly lower levels of courtship activity. The differences in levels of territorial aggression did not appear to be attributable to differences in androgen levels as mean kidney size (corrected for body size) in the anadromous population did not differ significantlyfrom that in the freshwater population. The marked differences in juvenile aggressiveness make it likely that differences in the level of (or sensitivity to) gonadotropins were responsible for the difference in territorial aggressiveness between the populations (see Fig. 6). The higher sexual activity of the anadromous males agrees with this interpretation (see Fig.6). These results suggest that (direct or indirect) selection for reduced or enhanced aggression in juvenile sticklebacks, or a more general difference in life-history pattern, has implications for the aggression and the sexual activity of mature fish.

VIII. SUMMARY Quantitative genetic methods allow us to investigate the causative associations between behavioral traits. Genetic correlations between traits

166

T H E 0 C. M. BAKKER

are useful but not often applied tools in studying the physiological and neurobiological control of behavior. In the methodological section, methods to estimate genetic correlations and pitfalls in their estimation are discussed. A breeding design consisting of double, bidirectional selection lines is advocated as being particularly suitable to causal and evolutionary studies of suites of behaviors. I used this design in studying the genetic and physiological control of functionally similar behaviors in sticklebacks, namely, aggressive behavior (biting and bumping) shown in various social contexts. The study suggests that stickleback aggression can be characterized by a complex genetic correlation structure among different forms of aggression that may constrain the evolutionary response of specific forms of aggression. Additionally. aggressiveness seemed to be part of a complex character suite in which different forms of aggressiveness, sexual behavior, the intensity of red breeding coloration and several life-history characters are genetically correlated to varying degrees. It is argued that the genetic correlations in this complex are partly based on the multiple influences of hormones of the pituitary-gonadal axis. Increased knowledge of the causation of stickleback aggression may lead to new insights and predictions with regard to multivariate evolution in this species. Acknowledgments

I thank Enja Feuth-de Bruin for kindly permitting me the use of some of our unpublished data, and Chris Boake, Susan Foster, Charles Goodnight, Manfred Milinski. Tim Mousseau, and Mike Ritchie for helpful comments on earlier versions of the manuscript. Manuscript preparation was funded by the Swiss National Science Foundation.

References

Arnold. S . J. (1981). Behavioral variation in natural populations. I . Phenotypic, genetic and environmental correlations between chemoreceptive responses to prey in the garter snake. Tharnnuphis elegans. Euolitrion (Laitv-ence, Kans.) 35, 489-509. Arnold. S. J. (1983). Sexual selection: The interface of theory and empiricism. In “Mate Choice” (P. Bateson. ed.). pp. 67-107. Cambridge Univ. Press, Cambridge. Arnold. S. J . (1987). Genetic correlation and the evolution of physiology. In “New Directions in Ecological Physiology” (M. E. Feder. A. F. Bennett, W . W. Burggren, and R. B . Huey. eds.). pp. 189-215. Cambridge Univ. Press. Cambridge. Arnold. S . J.. and Halliday. T. (1992). Multiple mating by females: The r‘esign and interpretation of selection experiments. Anim. Behou. 43, 178-179. Atchley. W. R. ( 1984). Ontogeny, timing of development. and genetic variance-covariance structure. A m . N n r . 123, 519-540. Atchley. W. R.. Rutledge. J. J., and Cowley. D. E. (1982). Multivariate statistical analysis of direct and correlated response to selection in the rat. Euolitrion (Lawrence, Kans.) 36. 677-698.

GENETIC CORRELATIONS AND THE CONTROL OF BEHAVIOR

167

Bakker, T. C. M. (1985). Two-way selection for aggression in juvenile, female and male sticklebacks (Gasterosteus aculeatus L.), with some notes on hormonal factors. Behuuiour 93, 69-8 1 . Bakker, T. C. M. (1986). Aggressiveness in sticklebacks (Gasterosteus aculeatus L . ): A behaviour-genetic study. Behaviour 98, 1-144. Bakker, T. C. M. (1993a). Evolution of aggressive behavior in the threespine stickleback. In “The Evolutionary Biology of the Threespine Stickleback” (M. A. Bell and S. A. Foster, eds.), pp. 345-380. Oxford Univ. Press, Oxford. Bakker, T. C. M. (1993b). Positive genetic correlation between female preference and preferred male ornament in sticklebacks. Nature (London) 363,255-257. Bakker, T. C. M., and Feuth-de Bruijn, E. (1988). Juvenile territoriality in stickleback Casterosteus aculeatus L . Anim. Behau. 36, 1556-1558. Bakker, T. C. M., and Sevenster, P. (1983). Determinants of dominance in male sticklebacks (Gasterosteus aculeatus L.). Behauiour 86, 55-71. Bakker, T. C. M., and Sevenster, P. (1989). Changes in the sexual tendency accompanying selection for aggressiveness in the three-spined stickleback, Gasterosteus aculeatus L. J . Fish Biol. 34, 233-243. Bell, G . , and Koufopanou, V. (1986). The costs of reproduction. Oxford Suru. Euol. Biol. 3,83-131. Bell, M. A. (1984). Evolutionary phenetics and genetics: The threespine stickleback, Gasterosteus aculeatus and related species. In “Evolutionary Genetics of Fishes” (B. J. Turner, ed.), pp. 431-528. Plenum, New York. Bell, M. A,, and Foster, S. A,, eds. (1993). “The Evolutionary Biology of the Threespine Stickleback.” Oxford Univ. Press, Oxford. Black, R. (1971). Hatching success in the three-spined stickleback (Gasterosteus aculeatus) in relation to changes in behaviour during the parental phase. Anim. Behuu. 19,532-541. Boake, C. R. B. (1991). Coevolution of senders and receivers of sexual signals: Genetic coupling and genetic correlations. Trends Ecol. Euol. 6, 225-227. Borg, B., Reschke, M., Peute, J . , and van den Hurk, R. (1985). Effects of castration and androgen-treatment on the pituitary and testes of the three-spined stickleback, Gasterosteus aculeatus L., in the breeding season. Acta Zool. (Stockholm) 65,47-54. Borg, B., Paulson, G., and Peute, J. (1986). Stimulatory effects of methyltestosterone on pituitary gonadotropic cells and testes Leydig cells of the three-spined stickleback, Gasterosteus aculeatus L., in winter. Gen. Comp. Endocrinol. 62, 54-61. Borg, B., Peute, J . , Reschke, M., and van den Hurk, R. (1987). Effects of photoperiod and temperature on testes, renal epithelium, and pituitary gonadotropic cells of the threespine stickleback, Gasterosteus aculearus L. Can. 1.Zool. 65, 14-19. Broadhurst, P. L., and Jinks, J. L. (1974). What genetical architecture can tell us about the natural selection of behavioural traits. I n “The Genetics of Behaviour” ( J . H. F. van Abeelen, ed.), pp. 43-63. North-Holland Pub., Amsterdam. Brodie, E. D., I11 (1989).Genetic correlations between morphology and antipredator behaviour in natural poulations of the garter snake Thamnophis ordinoides. Nuture (London) 342,542-543. Brodie, E. D., I11 (1992). Correlational selection for color pattern and antipredator behavior in the garter snake Thamnophis ordinoides. Evolution (Lawrence, Kans.)46,1284-1298. Bulmer, M. G. (1974). Linkage disequilibrium and genetic variability. Genet. Res. 23, 281-289. Burger, R. (1989). Linkage and the maintenance of heritable variation by mutation-selection balance. Genetics 121, 175-184.

168

T H E 0 C. M. BAKKER

Butlin, R. K. (1993). A comment on the evidence for a genetic correlation between the srxes in Drosophila melanogaster. Anim. Behav. 45, 403-404. Butlin, R. K., and Ritchie, M. G. (1989). Genetic coupling in mate recognition systems: What is the evidence? Biol. J . Linn. SOC.37, 237-246. Carey, G. (1988). Inference about genetic correlations. Behav. Gener. 18, 329-338. Cheverud, J. M. ( 1982). Phenotypic. genetic. and environmental morphological integration in the cranium. Evolution (Lawrence, Kans.) 36,499-516. Cheverud. J. M. (1984). Quantitative genetics and developmental constraints on evolution by selection. J. Theor. Biol. 110, 155-171. Cheverud. J. M . (1988). A comparison of genetic and phenotypic correlations. Euolution (Lawrence, Kans.) 42, 958-968. Clark. A. G. (1987). Genetic correlations: The quantitative genetics of evolutionary constraints. In “Genetic Constraint on Adaptive Evolution” ( V . Loeschcke, ed.), pp. 25-45. Springer-Verlag. Berlin. Clark, A. G . (1990). Genetic components of variation in energy storage in Drusophila melanogaster. Evolution (Lawrence, Kans.) 44, 637-650. Clutton-Brock, T. H., and Parker. G. A. (1992). Potential reproductive rates and the operation of sexual selection. Q. Reu. B i d . 67, 437-456. Clutton-Brock, T. H., and Vincent, A. C. J. (1991). Sexual selection and the potential reproductive rates of males and females. Nature (London) 351, 58-60. Crusio, W. E., Schwegler, H., Brust. I.. and van Abeelen. J. H. F. (1989). Genetic selection for novelty-induced rearing behavior in mice produces changes in hippocampal mossy fiber distributions. J. Neurogener. 5, 87-93. Daniel, W. (1985). Fragen zum Wanderverhalten des dreistachligen Stichlings (Gusrerusteus uculearus L.). Faun.-Oekol. Mitt. 5, 419-429. de Ruiter. A. J . H.. and Mein, C. G. (1982). Testosterone-dependent transformation of nephronic tubule cells into serous and mucous gland cells in stickleback kidneys in vivu and in uirro. Gen. Cornp. Endocrinol. 47, 70-83. Dickerson. G. E. (1969). Techniques for research in quantitative animal genetics. In “Techniques and Procedures in Animal Production Research.” pp. 36-79. Am. SOC. Anim. Sci., New York. Dingle. H. (19881. Quantitative genetics of life history evolution in a migrant insect. I n “Population Genetics and Evolution” ( G . de Jong, ed.). pp. 83-93. Springer-Verlag, Berlin. Dingle. H.. Evans, K. E.. and Palmer, J. 0. (1988). Responses to selection among lifehistory traits in a nonmigratory population of milkweed bugs (Oncopelfusfasciarus). Evolution (Lawrence, Kans.) 42, 79-92. Fairbairn, D. J . , and Roff, D. A. (1990). Genetic correlations among traits determining migratory tendency in the sand cricket, Gpllus firmus. Evolurion (Lawrence, Kans.) 44, 1787- 1795. Falconer, D. S. (1989). “Introduction to Quantitative Genetics,” 3rd ed. Longman, Harlow. Frankham, R. (1990). Are responses to artificial selection for reproductive fitness characters consistently asymmetrical? Ganef. Res. 56, 35-42. Garland, T., Jr. (1988).Genetic basis ofactivity metabolism. I. Inheritance of speed, stamina, and antipredator displays in the garter snake Thamnophis sirralis. Evolurion (Lawrence, Kans.) 42, 335-350. Gianola. D. ( 1979). Estimation of genetic covariance from joint offspring-parent and sib-sib statistics. Generics 93, 1039-1049. Giles, N., and Huntingford, F. A. (1985). Variability in breeding biology of three-spined

GENETIC CORRELATIONS AND THE CONTROL OF BEHAVIOR

169

Sticklebacks(Gasterosteus aculeatus): Problems with measuring population differences in aggression. Behciuiour 93, 57-68. Goldschmidt, T., and Bakker, T. C. M. (1990). Determinants of reproductive success of male sticklebacks in the field and in the laboratory. Neth. J. Zool. 40, 664-687. Gromko, M. H. (1992). Genetic correlation of male and female mating frequency: Evidence from Drosophila rnelanogaster. Anim. Behau. 43, 176-177. Gromko, M. H., and Newport, M. E. A. (1988). Genetic basis for remating in Drosophila melanogaster. 111. Correlated responses to selection for female remating speed. Behau. Genet. 18, 633-643. Gromko, M. H., Briot, A., Jensen, S. C., and Fukui, H. H. (1991). Selection on copulation duration in Drosophila rnelanogaster: Predictability of direct response versus unpredictability of correlated response. Evolution (Lawrence, Kans.) 45, 69-81. Hastings, A. (1989). Linkage disequilibrium and genetic variances under mutation-selection balance. Genetics 121, 857-860. Henderson, N. D. (1989). Interpreting studies that compare high- and low-selected lines on new characters. Brhau. Genet. 19, 473-502. Hill, W. G. (1971). Design and efficiency of selection experiments for estimating genetic parameters. Biornetrics 27, 293-31 1. Hill, W. G. (1972a). Estimation of realised heritabilities from selection experiments. I. Divergent selection. Biometrics 28, 747-765. Hill, W. G. (1972b). Estimation of realised heritabilities from selection experiments. 11. Selection in one direction. Biornetrics 28, 767-780. Honma, Y. and Tamura, E. (1984). Anatomical and behavioral differences among threespine sticklebacks: The marine form, the landlocked form and their hybrids. Acta Zool. (Stockholm)65979-87. Houle, D. (1991). Genetic covariance of fitness correlates: What genetic correlations are made of and why it matters. Evolution (Lawrence, Kans.) 45, 630-648. Idler, D. R., and Ng, T. B. (1983). Teleost gonadotropins; Isolation, biochemistry, and function. In “Fish Physiology” (W. S. Hoar, D. J. Randall, and E. M. Donaldson, eds.), Vol. 9, pp. 187-221. Academic Press, New York. Keenleyside, M. H. A. (1979). “Diversity and Adaptation in Fish Behaviour.” SpringerVerlag, Berlin. Klein,T. W.,DeFries, J. C., and Finkbeiner, C. T(1973). Heritabilityandgeneticcorrelation: Standard errors of estimates and sample size. Behau. Genet. 3, 355-364. Kynard, B. E. (1978). Breeding behavior of a Lacustrine population of threespine sticklebacks (Gasterosteus aculeatus L.). Behauiour 67, 178-207. Lande, R. (1979). Quantitative genetic analysis of multivariate evolution applied to brain:body size allometry. Euolution (Lawrence, Kans.) 33, 402-416. Lande, R.(1982). A quantitative genetic theory of life history evolution. Ecology 63,607-615. Lande, R. (1988). Quantitative genetics and evolutionary theory. In “Proceedings of the Second International Conference on Quantitative Genetics” (B. S. Weir, E. J. Eisen, M. M. Goodman, and G. Namkoong, eds.), pp. 71-84. Sinauer Associates, Sunderland, MA. Lavin, P. A., and McPhail, J. D. (1987). Morphological divergence and the organization of trophic characters among lacustrine populations of the threespine stickleback (Gasterosteus aculeatus). Can. J . Fish. Aquat. Sci. 44, 1820-1829. Lynch, C. B., and Roberts, R. C. (1984). Aspects of temperature regulation in mice selected for large and small size. Genet. Res. 43, 299-306. Maynard Smith, J. (1991). Theories of sexual selection. Trends Ecol. Euof. 6, 146-151.

170

T H E 0 C. hl. BAKKER

Maynard Smith. J.. Burian. R.. Kauffman. S . . Alberch. P.. Campbell. J.. Goodwin. B., Lande. R.. Raup. D.. and Wolpert, L. (1985).Developmental constraints and evolution. Q.Rev. Biol. 60. 265-287. McLennan. D. A . . and McPhail. J . D. (1989).Experimental investigationsofthe evolutionary significance of sexually dimorphic nuptial colouration in Casrerosreiis uculea/us (L.): The relationship between male colour and male behaviour. Can. J . Z o o / . 67,1778-1782. Milinski. M.. and Bakker. T. C . M. (1990). Female sticklebacks use male coloration in mate choice and hence avoid parasitized males. Nurirre (London) 344, 330-333. Morris. D. ( 1958). The reproductive behaviour of the ten-spined stickleback (Pvgosterts pi/ngiri/t.s L..). Behaciour, Suppl. 6, 1-154. Mourier. J.-P. (1972). Etude de la cytodifferenciation du rein de I'epinoche femelle apres tiaitement par la methyltestosterone. Z. Zellforsch. Mikrosk. Anar. 123, 96-11 1. Mousseau. T. A,. and Roff, D. A . (1987). Natural selection and the heritability of fitness components. Heredify 59, 181-197. Munro. A . D.. and Pitcher. T. J . (19831. Hormones and agonistic behaviour in teleosts. In "Control Processes in Fish Physiology" ( J . C. Rankin. T. J. Pitcher. and R. T. Duggan. eds.). pp. 155-175. Croom Helm. London. Nagai, J.. Eisen. E. J., Emsley. J . A. B., and McAlister. A . J . ( 1978). Selection for nursing ability and adult weight in mice. Genetics 88, 761-780. Ng. 1 . B.. and Idler. D. R . (1983). Yolk formation and differentiation in teleost fishes. In "Fish Physiology" ( W . S. Hoar. D. J . Randall. and E. M. Donaldson, eds.), Vol. 9. pp. 373-404. Academic Press. New York. Palmer. J . 0..and Dingle. H . (1989). Responses to selection on flight behavior in a migratory population of milkweed bug (Oncopelfus.fiisciatirs). Euoliition (Lawwnce. Kans.) 43, 1805- 1808, Ridley. M. i1993). "Evolution." Blackwell. Oxford. Ritchie. M. G. ( 1992). Behavioral coupling in tettigoniid hybrids (Orthoptera).Behau. Gene!. 22, 369-379. Roff. D. A . (1988). The evolution of migration and some life history parameters in marine fishes. Enuiron. B i d . Fis/ws 22, 133-146. Rowland. W. J. i 1984).The relationships among nuptial coloration. aggression. and courtship of male three-spined sticklebacks. Gcrsferosretrs aculetirtrs. Crin J . Zoo/. 62, 999- 1004. Scheiner. S. M.. Caplan, R. L.. and Lyman. R. F. (1989). A search for trade-offs among life history traits in Urosop/ii~anielanognsrer. E u d . Ecd. 3, 5 1-63, Service. P. M.. and Rose. M. R. (1985). Genetic covariation among life-history components: The effect of novel environments. Euolrttion (Lnn,rrnce.Kuiis.) 39, 943-945. Sevenster. P. ( I 9 6 1 ). A causal analysis of a displacement activity (fanning in Gasrerosteris cicitleiitits L.). Bc/iucI'ortr. Suppl. 9, 1-170. Slijkhuis. H . (1978). Ultrastructural evidence for two types of gonadotropic cells in the pituitary gland of the male three-spined stickleback. Go.\teros/err.s acit1eatu.s. Gen. C'OnfP. End#CritfO/. 36,639-641. Snyder. R. J . (1991).Quantitative genetic analysis of life histories in two freshwater populations of the threespine stickleback. Copcin. pp. 526-529. Spitre. K.. Burnson. J.. and Lynch. M. (1991). The covariance structure of life-history characters in Dnphnici pule.v. Euolirfiori (Lm.renc,e.Katrs.) 45, 1081-1090. Starnencovic-Radak. M.. Partridge, L.. and Andjelcovic. M. ( 1993). Genetic correlation between the sexes in Drosophila me/onog!nsrer: A reply to Butlin. Anini. Behau. 45, 405. Steams. S. L.. (1976). Life-history tactics: A review of the ideas. Q. Reu. B i d . 51, 3-47.

GENETIC CORRELATIONS AND THE CONTROL OF BEHAVIOR

171

Steams, S., de Jong, G., and Newman, B. (1991). The effects of phenotypic plasticity on genetic correlations. Trends Ecol. Euol. 6, 122-126. Stevens, L. (1989). The genetics and evolution of cannibalism in flour beetles (genus Tribofiurn).Evolution (Lawrence, Kans.) 43, 169-179. Sulzbach, D. S., and Lynch, C. B. (1984). Quantitative genetic analysis of temperature regulation in Mus musculus. 111. Diallel analysis of correlations between traits. Evolution (Lawrence, Kans.) 38,541-552. Turelli, M. (1985). Effects of pleiotropy on predictions concerning mutation-selection balance for polygenic traits. Genetics 111, 165-195. van den Assem, J. (1967). Territory in the three-spined stickleback, Gasterosteus aculeatus L. An experimental study in intra-specific competition. Behauiour, Suppl. 16, 1-164. van Iersel, J. J. A. (1958). Some aspects of territorial behaviour of the male three-spined stickleback. Arch. Neerl. Zool. 13, 383-400. van Noordwijk, A. J. (1984). Quantitative genetics in natural populations of birds illustrated with examples from the great tit, Parus major. I n “Population Biology and Evolution” (K. Wohrman and V. Loeschcke, eds.), pp. 67-79. Springer-Verlag, Heidelberg. van Noordwijk, A. J., and de Jong, G. (1986). Acquisition and allocation of resources: Their influence on variation in life history tactics. Am. Nut. 128, 137-142. Via, S. (1984). The quantitative genetics of polyphagy in an insect herbivore. 11. Genetic correlations in larval performance within and among host plants. Evolution (Lawrence, Kans.) 38, 896-905. Via, S., and Lande, R. (1985). Genotype-environment interaction and the evolution of phenotypic plasticity. Evolution (Lawrence, Kans.) 39, 505-522. Villars, T. A. (1983). Hormones and aggressive behavior in teleost fishes. I n “Hormones and Aggressive Behaviour” (B. B. Svare, ed.), pp. 407-433. Plenum, New York. Wai, E. H., and Hoar, W. S. (1963). The secondary sex characters and reproductive behavior of gonadectomized sticklebacks treated with methyl testosterone. Can. J . Zool. 41, 611-628. Willis, J. H., Coyne, J. A., and Kirkpatrick, M. (1991). Can one predict the evolution of quantitative characters without genetics? Evolution (Lawrence, Kans.) 45, 441-444. Wilz, K. J. (1971). Comparative aspects of courtship behavior in the ten-spined stickleback, Pygosteus pungitius (L.). Z . Tierpsychol. 29, 1-10. Wootton, R. J. (1973). The effect of size of food ration on egg production in the female three-spined stickleback, Gasterosteus aculeatus L. J . Fish Biol. 5 , 89-96. Wootton, R. J. (1976). “The Biology of the Sticklebacks.” Academic Press, London. Wootton, R. J. (1984). “A Functional Biology of Sticklebacks.” Croom Helm, London.

This Page Intentionally Left Blank

ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 23

Territorial Behavior: Testing the Assumptions JUDY STAMPS SECTION OF EVOLUTION A N D ECOLOGY UNIVERSITY OF CALIFORNIA AT DAVIS DAVIS, CALIFORNIA 95616

I. INTRODUCTION We know a great deal about territorial behavior, but perhaps not as much as we think we do. The main theme of this chapter is that current ideas about the behavior of territorial animals are based on a series of assumptions, and that in some cases these assumptions have not been adequately tested. By virtue of repetition, untested assumptions have a tendency to solidify into “quasi-facts” and assume a stature that their original authors never intended (Yasukawa and Searcy, 1985; Davies, 1989). For instance, in 1967 van den Assem suggested that if territorial animals settle simultaneously, they might achieve higher densities than if they settle sequentially. In subsequent years, this paper was frequently cited as evidence that settlement patterns affect both population density and territory size, ideas that eventually were incorporated into a variety of theoretical models and field studies (review in Stamps, 1992). In point of fact, van den Assem’s own study of sticklebacks failed to reveal any effect of settlement patterns on the density of territory owners, and the handful of empirical studies on this question have produced equivocal and contradictory results (Stamps, 1992). This is but one of many cases in which assumptions about territorial animals have been widely accepted without benefit of critical reexamination or empirical validation. In this type of review, there is a danger of losing the reader in either a sea of generalizations or a forest of specifics. On one hand, assumptions about territorial animals are interesting and important only if they pertain to a wide range of species, so that some degree of generality is clearly desirable. On the other hand, assumptions about territorial behavior are embedded within a complex logical framework, in which ideas about I73

Copyright 0 1994 by Academic Press, inc. All rights of reproduction in any form reserved.

174

JUDY STAMPS

territory quality. habitat selection, assessment, the significance of communication signals and the function of animal contests are intimately intertwined with one another. These complicated relationships can be most easily grasped if put into concrete terms. using information from a single, well-studied territorial species. Hence, this chapter considers issues that are relevant to many territorial species, and then illustrates these using information from juvenile Anolis aeneus lizards, a species with which I have worked for many years. The body of the chapter is organized into three sections. Section I1 begins with an overview of hypotheses and assumptions about the function of territories, the adaptive significance of habitat selection, and the role of assessment in choosing habitats and territories. This is followed by a discussion of how these concepts have been addressed and tested in a series of field and laboratory studies ofjuvenile A . iieneiis lizards. Section III focuses on territorial behavior, that is, the behavior patterns exhibited by territorial animals as they settle in territorial neighborhoods and acquire and maintain territories. Examples of territorial behavior include chases and fights, advertisement signals, or territorial patrols. As before, the first portion of this section considers some widely held assumptions about territorial behavior, while the second discusses tests of these assumptions usingjuvenile lizards. Finally, Section IV considers some exciting new lines of research that are emerging as field workers and theoreticians begin to reexamine oft-cited assumptions about territorial behavior and habitat selection.

I i . TERRITORY FUNCTION, HABITATSELECTION, A N D ASSESSMENT A.

ASSUMPTIONS

I . LkjinifionY Although there are several different ways to define territoriality (Noble, 1939: Emlen. 1952; Pitelka, 1959), Noble's definition of a territory as "any defended area" fits the needs of the current chapter. As a further refinement, I will concentrate on territories that are held for an appreciable period of time, as opposed, for example, to the defense of localized but ephemeral resource5 such as basking sites, areas around receptive females, or transitory food supplies. "Habitat selection" is defined as the process by which animals actively choose habitats in which they will conduct particular activities. Habitat selection can occur on various spatial and temporal scales, ranging from the choice of microhabitat in which an animal forages from one moment

TERRITORIAL BEHAVIOR

I75

to the next, to the range of communities used by the members of a species over evolutionary periods of time (Wiens et al., 1986; Addicott et al., 1987; Morris, 1987; Orians, 1991). This paper considers an intermediate temporal and spatial scale, encompassing the period during which animals acquire and maintain territories, and the neighborhood in which those territories will be located. At the ultimate level, one can ask why animals might choose to live in a particular area, and why individuals should attempt to exclude particular categories of conspecifics or heterospecifics from some portion of that area. In an ideal world, one might answer these questions by comparing the lifetime reproductive success of individuals who adopted one option (i.e., the defense of a territory of a particular type and size) with the success of otherwise identical individuals in the same habitat who followed a different spacing option. However, students of territorial species are still far from this goal. A few field workers have estimated the lifetime reproductive success for animals with different spacing behavior (e.g., Smith and Arcese, 1989; Koenig, 1990), but the subjects of these unmanipulated field studies were probably not identical with respect to morphology, physiology, prior experiences, abilities, aptitudes, and other factors that might have influenced their reproductive success and spacing behavior. Hence, observed correlations between territorial behavior and lifetime reproductive success could have been due to the underlying effects of individual differences in other traits on both of these variables. It is possible to design experimental field studies that control for individual differences in size, experience, and other factors, but this approach is rarely used to study the relationship between spacing behavior and lifetime reproductive success. Instead, students of territorial species usually measure variables that they assume contribute to an individual’s lifetime reproductive success; examples include growth rates, foraging rates, survival, or the number of offspring produced each season. They then assume that selection would tend to favor the behavioral phenotypes or strategies that would maximize these components of fitness (e.g., see Maynard Smith, 1978; Stephens and Krebs, 1986; Mitchell and Valone, 1990; Hoffman, 1991;Rosenzweig, 1991). Most current theory in habitat selection and territorial behavior follows this convention (see the following) and assumes that behavioral “phenotypes” such as habitat preference, the presence or absence of territory defense, or territory size have been shaped by natural selection so as to increase or maximize individual or inclusive fitness (Davies and Houston, 1984; Pulliam and Danielson, 1991). In addition, selection is presumed to favor the spacing behavior that yields the highest “payoff,” where payoff refers to the change in fitness expected as aresult of adopting a particular behavior (Maynard Smith, 1982).

I76

JUDY STAMPS

2 . Territory Function Speculation about the function of defended areas is an honored tradition

in the territorial literature. In his review of territorial behavior in birds, Howard (1920) described two primary functions: to facilitate pair formation and ensure adequate supply of food for the young. Suggestions about the adaptive significance of territories proliferated in subsequent years (Burt, 1943; Carpenter, 1958), and in his classic review, Hinde (1956) identified ten different hypotheses about the adaptive significance of territory defense in birds. Hinde concluded this review with two main points: (1) territory function varies enormously among avian species and (2) detailed knowledge about life history is required to understand the function of territory in any particular species. Although these conclusions are probably valid, they understandably discouraged theoreticians and field workers from trying to formulate global theories about the functions of animal territories. A swing back toward generality was made in the 1960s, in papers that applied an economic approach to the problem of territory function (Brown, 1964; Brown and Orians, 1970). The economic approach assumes that one can measure the costs and benefits of territory defense. In this original set of papers, the benefits of territory defense were expressed in terms of access to limited resources or mates, whereas costs were expressed in terms of the behavior patterns used to exclude other animals from the territory (e.g., aggressive behavior, advertisement, patrols, etc. ; Brown, 1964; Brown and Orians, 1970; Table I). Most of the examples in these papers mentioned food, leaving readers with the impression that food was an important, if not the most important, resource with respect to the evolution of territorial behavior (Brown, 1964; Brown and Orians, 1970). The economic approach to the study of territory function revolutionized the field. It provided an explanation for the variation in territorial behavior within and among species, identified categories of benefits and costs that might apply to a variety of territorial animals, and led to testable predictions about the adaptive significance of territorial defense. In subsequent years, this approach was refined and formalized (e.g., Schoener, 1983, 1987; Davies and Houston, 1984; Carpenter, 1987; Hixon, 1987), and much of our current thinking about territory function is based on this framework. In contrast to the earlier comparative approach to the study of territory function, the economic approach provided models that could be applied to a wide variety of species and situations. However, the shift from a comparative approach to an economic approach also led to a reduction in the number and type of environmental and social factors thought to affect the benefits or costs of temtorial behavior. Of the ten hypotheses

TERRITORIAL BEHAVIOR

177

TABLE I TRADITIONAL ASSUMPTIONS ABOUT TERRITORIAL ANIMALS Temtory Function 1 . Benefits equal access to resources (food)." 2. Costs equal defense costs." Habitat Selection 1. Habitat quality is determined by resources (food)." 2. Conspecifics are competitors.' 3. Fitness is inversely related to population density. Assessment 1 . Settlers can accurately assess the environmental factors that determine habitat and temtory quality.' The Function of Territorial Behavior 1. Territorial behavior is aversive. 2. Newcomers avoid established residents and their advertisement signals.' 3. Animals who win contests win space.' ~~~~

a

Assumptions tested using juvenile lizards.

that Hinde listed for territory function, some sound quite familiar to modern ears (e.g., the defense of food, nest sites, or mates). However, other hypotheses on his list have faded from view. For example, Tinbergen (1952) suggested that the overdispersion of nests in species with cryptic eggs, young, or incubating parents might reduce the effects of predation, whereas Tavistock (1931) noted that the spacing produced by temtorial behavior might reduce epidemic diseases. In the years following Hinde's review, relatively few workers have considered the effect of territory size or interneighbor distances on the risk of predation to territory owners or their young (Krebs, 1970, 1971; Taylor, 1988; Arcese et aE., 1992). This lack of interest is surprising, given a burgeoning literature indicating that predators increase the time spent searching in a patch after encountering prey in that area (Stephens and Krebs, 1986), as well as field studies showing that avian nest predators concentrate their search in habitats in which they have previously discovered nests with eggs (e.g., Martin, 1988). If the rate of offspring production is determined by predation rather than starvation, it would seem logical to ask how territory sizes affect rates of predation on eggs and young, rather than focusing on relationships between food levels and territory size. However, to date only a handful of studies have taken this approach (e.g., Arcese et ui., 1992).

178

JUDY STAMPS

If predators have been relegated to the sidelines in the recent territorial literature. parasites have virtually disappeared from the playing field. One possible reason is that the effects of disease or macroparasites on spacing originally were couched in the vaguely group selectionist terms of the early part of this century. However. it is a simple matter to rephrase these hypotheses in terms more acceptable to a modern audience. If territory owners or their offspring suffer costs as a result of infection during the territorial season, and if contact with conspecifics or their body products increases the probability that animals will become infected, then parasites could easily favor the establishment and maintenance of a large exclusive area around territory owners and their young. This is especially true if intruders into established territories can be repelled by advertisement behaviors that do not require physical contact (e.g., song, scent marks. or visual displays. see Section 111). Recently there has been a resurgence of interest in micro- and macroparasites among biologists interested in population dynamics (Anderson and May. 1981; Begon et ul., 1992), behavior (Freeland, 1976: Hart. 1990; Mgller, 19901, and the evolution of physiological defense systems (Williams and Nesse, 1991 ). The time is ripe for a reexamination of the role that parasites might play in the evolution and maintenance of spacing patterns in territorial animals. The ascendency of the economic approach also affected assumptions about the effects of conspecifics on one another. In the first part of this century. students of territoriality were willing to entertain the possibility that conspecifics might have positive as well as negative effects on one another in territorial species. Although there was general agreement that territorial behavior tended to keep conspecifics out of the territory proper, many field workers had the impression that territorial animals were attracted to one another, forming integrated social groups within territorial neighborhoods (review in Stamps, 1988a). They even suggested ways that territorial animals might benefit from the presence of neighboring conspecifics. for example, that males in groups might attract more females than isolated males (May, 1949; Lack, 1948: Durango, 1950), that social stimulation from neighbors might synchronize reproduction within colonial species (Darling, 1952: Fisher, 1954). or that groups of adjacent territory owners might cooperate to expel intruders from their neighborhood (Healy. 1967). With the advent of the economic approach, the emphasis shifted to the competitive effects of conspecifics on one another. Neighbors, floaters, intruders, etc.. all entered the equation in the same place, as factors contributing to the cost of territorial defense. Practical concerns may have also contributed to the heavy emphasis on competitive behavior in territorial animals. Aggressive behavior is more conspicuous and easily

TERRITORIAL BEHAVIOR

179

measured than the more subtle benefits that neighboring territory owners might provide to one another. It is easier to count an animal’s fights with its neighbor than to measure the effect of neighbors on an owner’s ability to detect and expel intruders from its own territory (Hart, 1987; Eason and Stamps, 1993:). At the risk of oversimplification, the economic approach to territory function can be summarized as follows. In principle, the costs and benefits of a particular spacing option may be measured in terms of their effect on an individual’s lifetime reproductive success. The net payoff for a territory of a particular type, size, etc., is measured by the benefit-cost of defending that territory, and this payoff is then compared to the payoffs for other possible types of spacing behavior. It is assumed that a territory of type X will be defended if the payoff is greater than the payoffs for any other spacing behavior that might be adopted by that same individual. Of course, the alternative spacing behaviors vary, depending on the species and the situation. If the question is whether or not a particular individual should defend a territory, alternative spacing behaviors might include inhabiting an undefended home range, floating in a territorial neighborhood, or joining a flock. Alternatively, one might ask whether territory X would provide a higher payoff than a territory in the same place but of a different size or shape, or than a territory with comparable dimensions in another location. Finally, for species in which individual home ranges overlap to a greater or lesser extent, one might ask whether territory X provides a higher payoff than a comparable territory with greater overlap with neighbors, or with different numbers of individuals sharing the territory. Traditionally, the benefits of territory defense were assumed to involve access to limited resources (especially food), and costs were measured in terms of the behavior patterns used to defend the territory (Table I). Though this formulation has proven quite useful, it should not be uncritically applied to every species in which animals defend territories containing food, nest sites, or other potential resources. For instance, spacing itself could be a benefit of territorial behavior, if the risk of predation or parasitism were inversely related to territory size (see the foregoing). Conversely, aggression, advertisement, and other conspicuous “defensive” behavior patterns are not the only possible costs of territorial behavior. Other potential costs include the time and energy required to sample different areas prior to acquiring a territory (see also Section II,A,4). The economic approach does not specify the nature of the costs and benefits of temtorial behavior, but rather provides a framework within which many possible costs and benefits can be identified and examined. In the last few years, workers have taken advantage of this approach to propose and study

180

JUDY STAMPS

a variety of “nontraditional” costs and benefits of territorial behavior (see Section IV). It is also worth noting that the costs and benefits of territorial behavior need not occur at the same time. In seasonally territorial species, the costs of territory establishment are likely to be concentrated during the settlement period, whereas the benefits of territory ownership may not begin to accrue until later in the season (MacLean and Seastedt, 1979; Orians and Wittenberger, 1991). Aggressive or defensive behavior that makes no sense when measured in terms of short-term costs and benefits may be adaptive if net payoffs are measured over the period of territory tenure, or over the animal’s lifetime (Stamps and Tollestrup, 1984; Houston et al., 1985; Townshend, 1985). Hence, longitudinal or long-term studies of territorial species are usually preferable to those that attempt to measure costs and benefits on a short-term basis. 3. Habitat Selection in Territorial Species The subject of habitat selection is intimately tied to questions about temtory function. Territories are embedded within larger habitats, and prospective temtory owners must choose a habitat before they can establish a territory within that habitat (Orians and Wittenberger, 1991). Hence, assumptions about territory quality and the benefits of temtory defense are closely related to assumptions about the ultimate factors that are responsible for habitat selection. a . Brief Overview of Habitat Selection and Quality. From an early date, it was obvious that a variety of environmental factors might affect the survival, growth, or reproduction of the individuals within a species (Shelford, 1913). In the ecological literature, this approach eventually led to the idea of a “multidimensional niche,” the suite of environmental factors that determine (1) whether or not individuals within a species can survive within a particular habitat and (2) the relative fitness of individuals living in microhabitats within the range of acceptable habitats (Hutchinson, 1957; see Pianka, 1988, for an introduction to the basic concepts). This multivariate approach to habitat quality has also been applied to territorial species (Lack, 1937; Klopfer and Hailman, 1965; Hilden, 1965; Cody, 1985; Klopfer and Ganzhorn, 1985). However, as was the case with the topic of territory function, studies or reviews that stressed the compiexity and diversity of the factors affecting habitat quality tended to discourage attempts to develop general, testable models of habitat selection for temtorial species. At roughly the same time as the economic models of territorial defense appeared on the scene, a related conceptual framework was developed for habitat selection (Brown, 1%9; Fretwell and Lucas, 1970; Fretwell,

TERRITORIAL BEHAVIOR

181

1972). These models assumed that one could determine the “intrinsic quality” of different habitats for the members of a given species, where intrinsic quality was determined by the density or amount of limiting resources, as well as any other environmental factors that might affect growth, survival, or reproduction. In principle at least, the intrinsic quality of a given habitat could be described in terms of the expected fitness of an individual living in that habitat in the absence of conspecifics. Conspecifics entered into these habitat selection models solely in their capacity as competitors. Brown’s model (1969) and the temtorial models of Fretwell and Lucas (1970) and Fretwell (1972) assumed that within any given habitat, individual fitness monotonically declined as the density of conspecifics increased. In turn, this assumption was based on several other implicit assumptions, for example, that individual fitness is primarily determined by access to nondivisible resources (especially food), and that within any given habitat, the amount of resources is limited with respect to the number of individuals requiring those resources. As a result, it was assumed that conspecifics have only negative effects on one another’s survival, growth, or reproduction. Actually, Fretwell and Lucas (1970) did briefly consider another model of habitat selection in which individuals might benefit from the presence of conspecifics in the same habitat. This “Allee-type ideal free distribution model” assumed that the fitness of individuals first increases and then later declines as a function of the density of conspecifics living in a given patch. As the name implies, it was based on extensive work by Allee showing that the individuals of many species have beneficial effects on one another, especially when living at low to intermediate densities (Allee, 1931; Allee el al., 1949). Although Fretwell and Lucas seemed to be intrigued by the Allee model, they were unable to incorporate it into their general habitat selection theory. Two problems were cited: (1) the curves did not yield unique solutions to analytical equations predicting the density of animals in different habitats and (2) given the other assumptions of the models, slight differences in population density would tend to produce large changes in the distribution of animals in different habitats. Hence, it appears that Allee-type models of habitat selection were abandoned because they were intractable, not because they were biologically unrealistic. The Fretwell and Lucas “ideal free” and “ideal despotic” models have provided the basis for much of the subsequent work on habitat selection in the ecological literature. Even today, many theoretical models and experiments are based on this framework. The assumption that individual fitness always declines monotonically as a function of conspecific density has become firmly entrenched in this literature (Maynard Smith, 1974;

I82

JUDY STAMPS

Parker and Sutherland, 1986; Rosenzweig, 1985, 1991). Moreover. the tendency to focus on food to the exclusion of other potentially important environmental factors affecting habitat quality has, if anything. become stronger in the years since these models were first introduced. For instance, Rosenzweig ( 1985) commented that “habitat selection is really a branch of optimal foraging theory,” and then went on to suggest that the premier factor determining habitat suitability is the amount of food within a given area. Recently, habitat selection theory has followed the lead of the optimal foraging literature in adding predators to the list of environmental factors that might affect patch choice. However, even these models still focus on only a handful of biotic factors (food, predators, intra- and interspecific competitors) when studying habitat quality and habitat selection, and they continue to assume that fitness monotonically declines as a function of population density (Morris, 1987, 1989; Abramsky et uf., 1990: Milinski and Parker, 1991; Sutherland and Parker, 1992). In summary. traditional habitat selection theory often assumes that habitat quality is determined by the availability of limited resources (especially food) and that conspecifics primarily act as competitors for those resources (Table I). Given these assumptions, it follows that conspecifics would usually have adverse effects on one another, so that the fitness of individuals would be inversely related to the number of competitors in the same patch. h. Relationships bet#Feen Habitat Qiiality, Terrirory Quality, and the Flrncrion uf Territorial Defense. A multidimensional approach to habitat quality can be quite useful for sorting out the complicated relationships between habitat quality, habitat selection, and territorial behavior. In principle, all the environmental factors affecting the fitness of a particular individual can be incorporated into a multidimensional description of habitat quality for that individual. Some environmental factors may set both upper and lower limits to habitat quality. For instance, one can identify upper and lower lethal thermal limits for many animals; within these limits, components of fitness (growth, survival, reproduction) often vary as a function of temperature (Ursin, 1979; Huey, 1982. 1991). Other environmental factors set either upper or lower limits (but not both) to habitat quality. Food is often assumed to have a sigmoid, saturating relationship with fitness: beiow a certain food density, starvation is certain; at somewhat higher densities, food and fitness may be positively related; and above some density, further increases in food levels would have no further effect on individual growth, survival, or reproduction. Environmental factors can interact in complex ways with respect to habitat quality and habitat selection. Examples appear with spiders, in which interactions between temperature and food availability affect habitat

TERRITORIAL BEHAVIOR

183

quality (Riechert and Tracy, 1975), or with fish, in which depth preferences vary as a function of the availability of cover (Sale, 1968). In addition, the environmental factors affecting habitat quality may be nested in importance, as is reflected in the familiar ecological concept of “limiting factors.” Some factors may place strict and absolute limits on the habitats in which survival or reproduction is possible, whereas others affect the growth, survival, or reproduction of individuals who live within those habitats. So far I have focused on the environmental factors that might determine “intrinsic habitat quality,” the fitness expected in a particular microhabitat in the absence of conspecifics. A basic assumption of both territory and habitat selection theory is that conspecifics do make a difference, and that the fitness of a given individual living in a patch of microhabitat varies as a function of the number of conspecifics that share a patch. The potential effects of conspecifics on habitat quality are most easily envisioned by considering the environmental factors that affect fitness in a given species. For each factor, one can ask whether and how fitness would be affected by the number or density of animals sharing the same area. Even the most simple-minded application of this procedure suggests that some environmental factors might have critical, density-independent effects on habitat quality. For instance, temperature might restrict the microhabitats that are suitable for stream-dwelling fish, but all else being equal, one would not expect a fish’s thermal regime to be much affected by the number of other fish living in the same stream. From the perspective of territorial defense, the most relevant environmental factors are those for which individual fitness would be adversely affected if that animal shared a particular area with conspecifics. Even at an intuitive level, it is clear that food might be a good candidate for this type of environmental factor. As noted earlier, it is often assumed that territorial animals live in habitats in which growth, survival, or reproductive rates are positively related to food availability. Since food eaten by one animal is unavailable to conspecifics, the addition of conspecifics to a patch of habitat would tend to decrease the food available to other animals living in that patch. Given this set of assumptions, it is easy to see why animals might benefit by excluding conspecifics from a foraging area. However, food is not the only environmental factor that might have density-dependent effects on individual fitness. Some of these environmental factors are similar to food, in that they affect fitness but cannot be shared among different individuals. Examples of potentially limited resources include structural features of the habitat, such as nest sites large enough for a single brood, thermal refuges or basking sites too small to

184

JUDY STAMPS

accommodate more than one individual. or predator refuges that are most effective if used by a single animal. If these types of sites are in short supply relative to the number of potential users, and if the availability or quality of these sites affects components of fitness, then individuals might benefit from the defense of these sites. The potential importance of a nontrophic resource on habitat and territory quality is nicely illustrated by a recent experimental study of redcockaded woodpeckers (Walters et al., 1992). This species excavates nest cavities in living pine trees, a process that requires 10 months to several years. Walters et al. were able to induce birds to establish territories in previously vacant areas by the simple expedient of drilling new nest cavities to the birds’ specifications in appropriate trees. This is only one of a long series of studies suggesting that avian habitat selection may be influenced by the availability of suitable nest sites (Hilden, 1965; Klopfer and Ganzhorn, 1985). In some situations, members of the opposite sex might also be envisioned as “resources.” For instance, if male reproductive success is limited by access to females, and if females are highly sedentary and are spaced uniformly across the landscape, then males with large exclusive territories may enjoy higher reproductive success than those with smaller territories (Stamps, 1983~;Davies and Lundberg, 1984; Hixon, 1987; review in Davies, 1991). However, females differ from food, nest sites, and other resources in that they are themselves active participants in habitat, territory, and mate choice. Males may defend territories around females, but even so, those females may choose to mate with neighboring territory owners (Gibbs et al., 1990; Gowaty and Bridges, 1991). Viewing females as a resource is even more problematical in species in which males defend small mating territories and are visited by females (e.g., Gibson, 1992). In this situation, the presumptive “resource” clearly has a major say in whether or not it will be accessible to any particular territory owner! Females differ from more typical sorts of resources in another important respect. If females take an active role when choosing habitats, territories, and males, then male mating success need not decline as a function of the density of males in a patch. For instance, if females prefer to search for mates in habitats that contain many territory owners, then male density and mating success might be related via an Allee-type curve, with highest male success at intermediate densities and territory sizes (see Section II,A,3a). I will return to this idea later on, when discussing recent developments in territorial behavior (Section IV). For now, the point to remember is that interactions between the sexes are considerably more complex than interactions between a territory owner and the types of environmental factors that fit the usual definition of a resource.

TERRITORIAL BEHAVIOR

185

Some environmental factors may favor territorial defense without fitting anyone’s definition of a “limited resource.” Examples include any factor that affects components of fitness and for which individuals would benefit by keeping intruders and neighbors at some distance from their center of activity. As was discussed earlier, infectious microparasites or certain types of predators might encourage territory owners to maintain exclusive areas around themselves and their offspring. Cannibalism in seabirds provides another example of this phenomenon. Hemng gulls (Larus argentatus) nest in colonies, in which the most important source of mortality on eggs and young is predation by conspecific intruders and neighbors (Burger, 1984). In this situation, it is not surprising that the presence of close neighbors increases the probability of cannibalism on eggs and young. Predation on young by conspecifics is thought to be a major factor contributing to the defense of a territory and optimal territory sizes in this species (Burger, 1984; see also FitzGerald et al., 1992, for a comparable example in sticklebacks). Thus far, I have focused on environmental factors that might lead to negative relationships between density and individual fitness. This emphasis is understandable, since the most obvious manifestation of territorial behavior is the maintenance of a more or less exclusive area by the owner. However, there are other environmental or social factors that might contribute to positive relationships between density and fitness. Over the years, field workers have suggested a variety of reasons why territory owners might benefit from the presence of neighbors on adjacent territories (Stamps, 1988a; see also Section II,A,2). Although the presence of territorial defense clearly implies that conspecifics compete with one another at some level, final territory sizes and spacing patterns may result from a combination of positive and negative effects of conspecifics on one another’s fitness. For instance, male insects may form territorial aggregations in order to produce the signals likely to attract females, but then compete with one another for the females’ attention after they amve (Bailey, 1991). Even cannibalistic herring gulls cooperate with one another in some situations, for example, when mobbing predators or when chasing intruders out of one another’s territories (Burger, 1984). In recent years, there has been a resurgence of interest in situations that might encourage territorial animals to settle near one another, leading to higher local densities and smaller territory sizes than one would expect if all interactions among settlers were competitive. These recent developments are considered in more detail in Section IV. If nothing else, this section highlights the potential dangers of oversimplification when studying habitat and territory quality in territorial animals. Since many different environmental and social factors are likely to contrib-

186

JUDY S-TAMPS

Ute to habitat and territory quality, experimental manipulations of one factor at a time may not tell us much about the relative importance of that factor under natural conditions. For instance, many field workers have studied the effects of food on habitat selection and territorial behavior by adding food supplements to an area that already contains territory residents (Boutin. 1990). However. even if territory sizes or population densities change in response to this experimental manipulation. one could not conclude that food was a major determinant of habitat or territory quality in that species. If territorial animals are already living in an area. this fact alone implies that the most critical determinants of habitat or territory quality are satisfied in that area. The most important determinants of habitat selection, intrinsic habitat quality, and intrinsic territory quality are those environmental factors that lead to the rejection or acceptance of an area by every potential settler. not those factors that produce minor variations in density, temlory size, or spacing patterns within habitats that every settler views as acceptable. To show that food is the critical determinant of habitat or territory quality, it would be necessary to show that animals are willing to settle and establish territories in previously vacant areas in direct consequence to the addition of food in those areas. Of course. the same criterion applies to any other factor that is thought to be a critical determinant of habitat or territory quality. The best way to determine the environmental factors critical for habitat and territory selection is to experimentally change a previously unacceptable area to one that is avidly preferred by territory owners (e.g., Pleszczynska, 1978; Lindstrom, 1992; Walters et al., 1992). Sometimes, the results of this type of experiment are rather surprising. For instance. nectivorous birds are considered to be classic exemplars of animals with feeding territories. However, recent studies show that perches, not food supplies, limit territory establishment in several species. Areas with abundant food but no perches are used but not defended (Goldberg and Ewald. 1991; Armstrong, 1992). The problem of attracting settlers to previously unused areas is of more than academic interest. since a goal of many conservation studies is to add animals to empty habitats, in the hopes that they will settle and establish viable populations there. Unless field workers have a reasonable idea about the environmental and social factors that affect habitat and territory selection under natural conditions, many of these attempts will be doomed to failure. 4 . rl S S e S S l l l en 1

An issue that links proximate and ultimate questions in territorial behavior and habitat selection is assessment. If areas vary with respect to the

TERRITORIAL BEHAVIOR

187

environmental and social factors likely to affect individual fitness, then animals may assess these environmental factors when they choose habitats and territories. Assessment provides a bridge between ultimate and proximate issues, because animals must rely on proximate behavioral processes to estimate the environmental and social factors that ultimately determine habitat quality, territory quality, and individual fitness. The assumption that animals assess important environmental and social factors when selecting habitats, territories, and spacing patterns underlies many theoretical and empirical studies of territorial behavior (Table I). Much of the habitat selection literature explicitly assumes that potential settlers are able to accurately estimate intrinsic habitat quality and competitor density, and then use this information to choose the habitat that will yield the highest payoff for that individual (Morris, 1989; Abramsky et al., 1990; Rosenzweig, 1991). Similarly, economic models of territorial behavior assume that animals accurately estimate resources and intruder pressure, and then adjust their territory sizes and spacing patterns accordingly (Davies and Houston, 1984; Hixon, 1987; Schoener, 1987). Assumptions about assessment have also been incorporated into experimental field studies of territorial species, in which manipulations of food levels are used to investigate questions about the ultimate significance of spacing patterns or habitat quality. Implicit in this approach is the assumption that if food levels are ultimately important for territorial behavior or habitat selection, then at the proximate level animals will accurately assess food levels and then alter their behavior in an appropriate manner when choosing habitats, establishing territories, or changing the sizes of established territories (e.g., Boutin, 1990). Recall that the payoffs for both territorial behavior and habitat choice are supposed to be measured over the lifetime of the animal in question. At the very least, they are supposed to be measured over the period when the subjects use the habitat or territory. However, many species choose habitats and territories within a fairly brief period of time, and then use those habitats and territories over a much longer time in the future. Similarly, territory sizes, boundaries, overlaps with neighbors, etc., may be established during the settlement period, and then maintained with few if any changes in subsequent days, months, or years. If the members of a species choose a habitat or establish a territory at one time, and then inhabit those areas for a relatively long period in the future, direct assessment of important ecological and social factors becomes problematical. Direct assessment of important environmental or social factors is possible only if two conditions are met: (1) these factors are apparent when the animal first settles and establishes its territory and (2) these factors do not change over the period of habitat or territory tenure (Orians and Wittenberger, 1991). However, in many species the

188

JUDY STAMPS

social and ecological factors affecting habitat and territory quality are likely to change after individuals settle and acquire territories (Svardson, 1949; HildCn, 1965; MacLean and Seastedt, 1979; Myers et al., 1979; Slagsvold, 1986;Orians and Wittenberger, 1991).These and other workers are willing to grant that food, predators, intruder pressure, and other factors might affect habitat quality, territory quality, or the benefits of territory defense. However, they also suggest that territory owners would be hard-pressed to directly assess all the relevant factors affecting territory and habitat quality prior to settling and choosing particular territories. If prospective settlers are unable to directly assess all the environmental and social factors that might affect habitat or territory quality, they might use indirect cues to assess those factors (Hilden, 1965;Pulliam and Parker, 1979; Smith and Shugart, 1987). “Indirect cues” are stimuli that can be monitored during the period of habitat selection and territory acquisition and that are related to or correlated with habitat quality, territory quality, or the benefits and costs of defending particular territories. Examples of potentially useful indirect cues to habitat and territory quality include structural features of the habitat (Smith and Shugart, 1987),the presence of conspecific residents in the same patch of habitat (Kiester, 1979; Stamps, 1988a; Raimondi, 1988; Shields e t a / . , 1988), or the presence of a previous owner on a particular territory (Stamps, 1987a). The notion of indirect assessment forces us to come face to face with the important distinction between proximate and ultimate factors in habitat selection and territorial behavior. If animals use one set of cues (e.g., vegetation height, foliage density) to assess a different set of ecological factors (e.g., food levels, predation risk for eggs or fledglings, or the probability of surviving winter storms), then the proximate reasons why animals prefer particular habitats or territories may be quite different from the ultimate reasons why they prefer those habitats and territories. This important distinction between proximate and ultimate factors was acknowledged and discussed by many of the early students of habitat selection in territorial species (e.g., review in Hilden, 1965). However, it is sometimes overlooked in current studies of habitat and territory selection. The distinction between direct and indirect assessment is especially relevant when studying species living in disturbed habitats. An indirect cue is only useful if there is a correlation between that cue and other environmental or social factors that affect components of fitness. In disturbed habitats these correlations may no longer exist, as a result of recent changes in food levels, predators, competitors, or other important selective pressures. If the mechanisms that are used to assess habitat and territory quality are inherited via genetic or other pathways, animals may

TERRITORIAL BEHAVIOR

189

continue to rely on their original set of cues when assessing habitats and territories, even if these processes do not produce optimal habitat or territory selection under current conditions. The potential importance of indirect assessment in disturbed habitats can be illustrated using a favorite subject for territorial studies, the great tit (Parus major). In the primordial forests that once covered much of northern Europe, the population densities and territorial behavior of great tits were probably governed by predators (Wesolowski e f al., 1987). The subsequent fragmentation of forests, extermination of many important predators, and provision of predator-safe nest boxes have apparently led to elevated population densities within the patches of woodland where these birds are typically studied, relative to the densities experienced by great tits over most of their evolutionary history. Hence, it is not surprising that recent studies of this species suggest that food and competitors are important determinants of reproductive success (Krebs and Perrins, 1978; Klomp, 1980; McCleery and Perrins, 1985). However, even if food and competitors are currently the most important factors affecting offspring production, the great tit’s habitat and territory selection behavior need not have adjusted to this new situation. If great tits originally chose habitats and territories based on indirect cues that were correlated with the risk of predation for themselves and their offspring, there is no guarantee that the birds would now base these decisions on assessments of food supplies and competitor density. In this and other species in which selective pressures have dramatically changed over recent evolutionary time, reliance on indirect cues for habitat and territory might lead to “suboptimal” behavior, for example, animals may choose particular habitats, territories, or spacing patterns even though other better options are available to them (Table 11).

B. TESTING ASSUMPTIONS ABOUT TERRITORY FUNCTION, HABITAT SELECTION, AND ASSESSMENT USINGJUVENILELIZARDS By now, it should be apparent that territory function, habitat selection, and assessment are related in complicated ways. This section traces the relationships among these variables in one territorial animal, juvenile A. aeneus lizards. Habitat selection, territory selection, and the function of territorial behavior in these lizards have been studied for the last twenty years in the laboratory and the field. Although some of these results are incomplete or preliminary, they show how many of the questions introduced in the previous section can be addressed and studied in freeliving subjects under natural conditions.

190

JUDY STAMPS

TABLE I1 ALTERNATIVE IDEAS

ABOUT

TERRITORIAL ANIMALS

Territory Function I . Benefits of neighbors in territorial species. 2. Expanding the list of potential costs and benefits. 3. Costs and benefits over the long term. Habitat Selection 1. Multivariate. nested approaches to describing habitat quality 2. Distinguishing between habitat and territory quality. 3 . Allee-type relationships between fitness and density. Assessment I . Indirect assessment and conspecific cuing. 2 . Assessment processes and “suboptimal” behavior. The Function of Territorial Behavior I . Cooperative aspects of territorial behavior. 2. Attraction of newcomers to temtory owners and their advertisement signals. 3 . Persistence and space acquisition.

I . Habitat Selection and Territory Function in Jrivenile Anolis aeneirs u. Nutirrul History. The species Anolis aeneiis is one member of a large genus of iguanid lizards, many of which are endemic to particular islands or island banks in the West Indies. This species is native to the island of Grenada, a small island at the southern end of the Lesser Antilles. Adult female A . aeneiis live in territories in woodland habitats and lay one egg at a time at about 2-week intervals within their woodland territories (Stamps. 1976). After about a month, a small (0.2 g, 20-mm snout-vent length) hatchling emerges from its egg. After hatching, the young lizards emigrate to clearings. where they attempt to acquire territories. A juvenile who successfully acquires a territory defends it for 2 to 6 months, the length of time it takes to grow to a snout-vent length of 30 mm (Stamps and Tollestrup, 1984). Virtually all territory owners survive until they reach 30 mm, whereas non-territorial animals typically emigrate (or die) prior to reaching this size. When territory owners have grown to 30 mm in size, they leave their territories and migrate back into the woodland habitat, where they acquire new, subadult territories (Stamps, 1983b). The territories held by hatchlings have nothing to do with mating and breeding; adult territories are in different habitats and are spatially and temporally separate from the territories held by juveniles. The rate at which juveniles arrive at clearings depends on a number of factors. including the time of year and the density of reproducing females

TERRITORIAL BEHAVIOR

191

in the woodland surrounding the clearing (Stamps, 1991). Frequently more juveniles arrive at clearings than are successful in getting territories there. Some of these juveniles become floaters, others become subordinates within the territories of other individuals, and still others emigrate from the clearing and attempt to settle elsewhere (Stamps, 1983a, 1991). Experimental removals of territory owners show rapid replacement within hours or days by floaters, subordinates, neighbors, or newcomers to the clearing (Stamps, 1983a). These results imply that under natural conditions there is often a shortage of territories relative to the number ofjuveniles hatching in surrounding woodland habitats. b. Habitat Selection and Habitat Quality for Juvenile A . aeneus. One distinct advantage of using a small animal to study habitat selection and territorial behavior is that one can experimentally manipulate structural and other habitat features in the field, and then ask whether prospective settlers accept these artificial microhabitats as home ranges or territories. Over the years, these manipulative studies have provided a “window” on habitat selection that is not available in field studies that rely entirely on correlations bet ween particular environmental features and population density. In addition to identifying the environmental factors with the greatest impact on juvenile space use, density, and territorial behavior, these studies provided “blueprints” for artificial microhabitats that are highly attractive to the hatchlings of this species. Studies of habitat selection usually begin with observations of the habitats and microhabitats in which a species naturally exists, and this study was no exception. One of the most striking aspects of juvenile A . aeneus behavior is their strong preference for clearings (Stamps, 1978, 1983b). Even though clearings make up less than 5% of the available habitat in undisturbed areas, hatchlings travel considerable distances to reach them and are unwilling to establish permanent home ranges, let alone defend these home ranges, unless they live in a clearing. Artificial clearings work just as well as natural ones, and within hours to days of cutting a clearing in a woodland habitat, hatchlings arrive and begin to establish territories there. Within clearings, juveniles require perches (Stamps, 1978). Arboreal perches are essential for all members of this species, but whereas adults use tree trunks and bushes, hatchlings require twigs or small herbs with woody stems. As is the case with adults, juveniles are particular about the height, diameter, and textures of their perches, and they prefer sets of twigs and plants that are connected with one another to form a series of arboreal “highways.” The density and configuration of perches are strongly related to the density and spacing patterns of the juveniles within a microhabitat (Ono, 1981; Eason and Stamps, 1992). Another important

I92

JUDY STAMPS

environmental factor is insolation. Juveniles almost never bask but are subject to overheating, and they can only use areas that are shaded from the direct sun during the hottest hours of the day (Stamps, 1983b). These are only a few of the environmental factors that contribute to the spatial distributions and habitat preferences of juvenile A . aeneus. Some habitat requirements place strict limits on space use (e.g., suitable microhabitat must be in a clearing, must contain suitable perches, and must be shaded during the middle of the day), whereas other factors affect the strength of the juveniles’ preference for particular microhabitats (e.g., perch configurations, the amount of foliage or leaf litter, or the type of habitat surrounding a clearing). As a result of ail of these factors, only a very small proportion of the habitat as a whole is suitable for juveniles, and preferred microhabitats are rarer still. Hence, the shortage of available space for juveniles is not simply a function of competition for a single important “limiting resource,” but rather the result of a complex, nested set of habitat requirements, all of which affect the microhabitats that are acceptable to juveniles. As a result of all of these requirements, the final amount of preferred microhabitat in clearings is usually small in comparison to the number ofjuveniles arriving at those clearings. Hence, juvenile A. aeneus is one of many species in which competition for space is a direct outcome of habitat selection behavior. The preceding studies provided insights into the proximate factors affecting habitat selection in this species. The next step was to consider the ultimate reasons why this particular set of environmental factors might affect intrinsic habitat or territory quality. This type of question can be approached using a combination of observational and experimental techniques. One of the first questions is why juvenile A. aeneus show such a strong preference for clearings, when all the other members of this species are found in the surrounding woodland habitat. After an initial series of studies documented the types and sizes of the food items eaten and preferred by juveniles (Stamps et d.,1981). surveys of prey items suitable forjuveniles were collected in woodland and adjacent clearings. These studies indicated that prey of the sizes and taxa eaten by juveniles were equally abundant in both habitats (Stamps, 1983b). Similarly, there was no difference in the availability of suitable perches in the two types of habitat. If anything, the higher temperatures in clearings would decrease intrinsic habitat quality, since most juvenile thermoregulatory behavior is directed toward avoiding overeating. However, there was one major difference between woodland and clearings, in that the density of sit-and-wait predators was much higher in the woodland habitats. Juveniles are eaten by adult male and female Anolis richardi, as well as by adult males of their own species. With the exception of large adult

TERRITORIAL BEHAVIOR

193

male A . richardi, these predators are unable to eat juveniles larger than 30 mm in size (Stamps, 1983b). Hence, one possible benefit of the preference for clearings is that juveniles living there are protected from an important set of predators. However, it should be emphasized that the juvenile preference for complex perches in brightly lit areas does not require any exposure to predators if it is to be expressed. Juveniles raised from eggs hatched in the laboratory exhibit this same habitat preference and begin to climb the walls of the experimental rooms when they reach 30 mm in size, even though bright lights, perches, food, and water are still located on the floor! Also, it is possible that other environmental factors, such as vulnerability to macro- or microparasites, might account for the strong preference of juveniles for clearings. The effect of parasites on habitat selection is a topic that has been little studied (but see Valladares and Lawton, 1991), but it certainly deserves more attention in this and other species. At a finer scale, the same approach can be used to suggest reasons why particular structural features of a habitat might affect juvenile space use and habitat preferences. Perches provide a useful illustration of this point. Juveniles use their perches in a variety of ways: as highways to move around their home ranges, as refuges from rain and wind, as escape routes from avian or mammalian predators, as vantage points for spotting prey items or conspecifics, and as retreats from nocturnal predatory snakes. Laboratory and field experiments show that juveniles prefer particular types of complex perch configurations in the absence of predators, and that their preferences for these types of homesites do not vary as a function of the spatial or temporal distribution of food in the habitat (Stamps, 1983a; Eason and Stamps, 1992). However, the preference for a topographically complex microhabitat is accentuated in the presence of predatory birds or lizards, suggesting that one of the functions of complex perch configurationsis protection against potential predators (Stamps, 1983a). c . Territory Function, Food, and Growth Rates. Assume for the moment that most of the important environmental factors that affect the space use and habitat preferences of juvenile Anolis aeneus have been identified. This information can then be used to identify potential functions of territorial defense in this species. As was discussed earlier, one useful approach is to consider whether the effects of particular environmental factors on individual fitness are likely to be density dependent. Several of the factors with important effects on juvenile space use and habitat preferences are probably density independent with respect to their effects on juvenile growth or survival. One example is temperature. If an area exceeds 31°C for more than a few minutes, any and all of the juveniles in that area emigrate. Similarly, drought and rainstorms affect the availabil-

i 94

JUDY STAMPS

ity of drinking water, and hence juvenile survival and growth rates, independently of the density of juveniles that live in a patch of habitat. The same is true of the perches used to shelter from inclement weather or for traveling around a home range; the use of these perches by one individual by no means precludes their use by other juveniles. However, the same cannot he said of a food supply. Juveniles are sit-and-wait predators who prefer flying insects such as homopterans, dipterans, and hymenopterans, and the presence of other nearby animals can interfere with an individual’s abi1it.y to stalk and capture these wary prey (Stamps et nl., 1981; Stamps, 1984a). Other types of prey (e.g.. arthropods that live in the leaf litter) may not be renewable over the short term, so that sharing an area with conspecifics is likely to reduce the amount of these prey available to any given juvenile. Of course, food is not the only environmental factor with potential interactions with juvenile density. If juveniles packed into an area, they might attract more predators and suffer higher per capita rates of predation than if they preserved reasonable distances between one another. However. juveniles alter their display rates when they spot a predator (Stamps, 1983a), so it is also possible that juveniles living at higher than normal densities might suffer lower per capita predation rates, as a result of monitoring the behavior of their neighbors. The interaction between predators and density in territorial species is not easy to study, and it is particularly difficult in this species, since residents have high survival rates and predation is rarely observed. The same problem arises when considering density-dependent interactions between parasites and juvenile fitness. It is possible that juveniles might suffer higher costs from micro- or macroparasites if they did not preserve relatively exclusive areas around themselves, but currently there is no information on this point. A priori. it seemed that access to a food supply might be one possible advantage of territory defense in this species. As was noted earlier, this hypothesis only makes sense if some important component of juvenile fitness is food-limited under natural conditions. The two most obvious components of fitness for these animals are growth and survival, and since juveniles in clearings have high rates of survival, most of the following studies used growth as an index of fitness. Juvenile growth rates in the field are often food-limited, and under natural conditions many juveniles grow more sIowly than individuals who receive food ad libitirrn in the laboratory (Stamps and Tanaka, 1981h). Access to drinking water also limits juvenile growth, so that juvenile growth rates are curtailed during the dry season even if prey is abundant (Stamps and Tanaka, 1981b). The potential effects of drinking water on grovYth were controlled in subsequent field experiments by sprinkling

TERRITORIAL BEHAVIOR

195

microhabitats with water on any day in which rain or dew was not available. Further studies showed that temporal fluctuations in the prey taxa and sizes preferred by juveniles occur simultaneously across wide areas, in response to daily, monthly, and seasonal variation in rainfall patterns. There is a curvilinear, saturating relationship between prey abundance and juvenile growth rates, whether prey abundance is measured by weekly sweep samples across a range of habitats (Stamps and Tanaka, 1981b) or by sticky traps sampled daily within individual clearings (Stamps and Eason, 1989). Within clearings, juvenile growth rates were highly correlated with temporal variation in food levels, as measured by the proportion of days within each juvenile’s growth interval in which food levels were high enough to sustain maximum growth rates (Stamps and Eason, 1989). These results indicated that temporal variation in food levels not only had a dramatic effect on juvenile growth rates, but that this variation would have to be taken into account in any further field studies that used growth rates as an index of juvenile fitness. d . Contender Pressure, Spacing Patterns, and Growth Rates. One of the basic assumptions of the economic models of territorial behavior is that conspecific competitors affect spacing patterns (Table I). For species in which territories are established during an initial settlement period and then maintained for the rest of the territorial “season,” this implies that the number of competitors (contenders) during the settlement period should influence territory sizes, territory overlap, and the final density of settlers that live in a given patch of habitat. Juvenile A . aeneus were ideal for testing this sort of hypothesis, because it was possible to control for habitat quality by designing artificial patches of preferred habitat and to control for differences among the prospective settlers by collecting size-matched individuals in an area far removed from a patch, and then releasing them into it (Stamps, 1990). Since the patches of habitat were designed on the basis of previous studies ofjuvenile habitat preferences, naive juveniles readily settled in these microhabitats, and it was not necessary to enclose the patches with walls or other barriers to prevent emigration. Juveniles were free to leave the patch and return at will, just as free-living settlers do when choosing habitats and territories. As a result, the final densities and spacing patterns obtained in each patch were determined by the subjects, not by the experimenter. In contrast, when experiments are conducted on real islands or within enclosures (e.g., “hard-edged” habitats, sensu Stamps et al., 1987),individual movements, spacing patterns, and final densities may not reflect those that occur when animals naturally settle in “soft-edged” patches of habitat (Stamps and Buechner, 1985; Stamps and Krishnan, 1990; Stamps, 1992).

196

JUDY STAMPS

The results of this experiment provided strong support for the assumption that competitors affect spacing patterns (Stamps, 1990). For any given patch, subtrials with many initial contenders ended up with smaller territory sizes, greater territory overlap, and more residents than subtrials with fewer initial contenders. Temporal fluctuations in food levels were monitored throughout each trial, but food levels had no effect on final spacing patterns or the density of settlers. An example for a 1.8-m2patch that was used in a series of three subtrials is illustrated in Fig. 1. In the first and third subtrials (BI, B3), 6 juveniles were released in the patch, whereas in the second subtrial (B2) 12 prospective settlers were added to the patch. When few contenders were added to the patch, fewer animals settled, territory sizes were larger, and overlap among neighbors was smaller than when many contenders were added to the patch (Fig. 1). Given that contender pressure affects spacing patterns in this species, the next question is whether spacing patterns affect components of fitness.

1 meter a

SUBTRIAL B1

l

a

SUBTRIAL 8 2

SUBTRIAL 8 3

FIG. 1. The effects of contender pressure on territory size and overlap. Three subtrials, in which 6, 12, and 6 contenders were added to a patch of juvenile microhabitat, were conducted in the West Indies. The polygons reflect the final temtory boundaries for the animals that eventually settled in the patch in each subtrial; the dotted line in subtrial B2 represents the home range of a floater. Reproduced from Stamps (1990); 0 1990 by the University of Chicago. All rights reserved.

TERRITORIAL BEHAVIOR

197

The answer to this question is a qualified “yes” (Stamps and Eason, 1989). Although the story is complicated, the basic outlines are fairly straightforward. For juveniles living in patches of artificial habitat in either the field or the laboratory, the best spacing option seems to be an exclusive territory with an area of about 0.6 m2. Overlap between adjacent territories or home ranges results in lower growth rates for the members of all status groups (territory owners, subordinates, floaters; Stamps 1984a,b; Stamps and Eason, 1989). Similarly, growth rates are reduced if home ranges or temtories are either much larger or much smaller than 0.6 m2 (Stamps and Eason, 1989). Social status influences growth rates indirectly by affecting the chances that an individual is able to maintain an exclusive territory of the optimal size. Low-status animals (subordinates and floaters) are more likely than higher-status animals (territory owners) to end up with substantial home range overlap, or home ranges much smaller or larger than those that produce relatively high growth rates. As a result, territory owners have higher average growth rates than floaters or subordinates living in the same habitat at the same time (Stamps, 1984a,b; Stamps and Eason, 1989). Why is an exclusive territory of about 0.6 m2the best option forjuveniles living in this type of microhabitat? Indirect evidence suggests that this spacing pattern allows juveniles ample time and space for effective foraging, in areas isolated from direct or indirect interference from conspecifics (Stamps and Eason, 1989). In contrast, juveniles who overlap extensively with other animals travel long distances chasing one another, spend time in social interactions that might have been spent foraging, and may lose prey items to animals with whom they share an area. Similarly, animals with large territories travel long distances while patrolling or defending their territory, and may suffer reduced growth rates as a result of elevated travel costs or less time available for foraging. Hence, we have returned by a circuitous route to our original question about the function of territory defense. So far, the available evidence for these animals supports the traditional economic view of foraging territories, in which an important benefit is access to a food supply, whereas costs are affected by competition with conspecifics. Juveniles with exclusive territories of average size were able to forage without interference from conspecifics and grow at relatively high rates. In contrast, individuals who settled with many competitors (high contender pressure) ended up with smaller territories, extensive overlap, and relatively low growth rates. e . Conclusions. These studies of habitat selection and spacing patterns in juvenile A. aeneus highlight the important distinction between environmental factors that affect habitat quality and those that affect the benefits or costs of territory defense. Juvenile A. aeneus exhibit very strong prefer-

198

JUDY STAMPS

ences for clearings and for particular types of perches within those clearings, and the ultimate reasons for these preferences probably have more to do with avoiding predators and overheating than with locating microhabitats with an adequate food supply. In fact, prey of the types and sizes preferred by juveniles are abundant in a variety of areas that juveniles assiduously avoid (woodland habitats, areas in clearings that lack perches, and areas in clearings that receive full midday sun). Food is important only because juveniles insist upon living in clearings and in particular types of microhabitats within those clearings. As a result of these habitat preferences, more juveniles arrive at suitable microhabitats than can be accommodated in the available space. That is, after all of their other requirements have been satisfied, there is finally the potential forjuveniles to interfere with one another when foraging on the leaves and leaf litter around their perches. This species provides a graphic illustration of how one set of environmental factors can place restrictions on areas suitable for long-term use, while another set of factors determines whether the animals living within those areas might benefit from territorial defense. 2. Assessment

In this section. we turn from the social and environmental factors that might affect habitat and territory quality to the question of how animals might assess these factors prior to settling in a habitat and acquiring a territory. Given that juvenile lizards might benefit from assessing certain factors at the time of settlement, how might they be able to do this? a. Direct Assessment. Juvenile A . aeneus may be able to assess some aspects of habitat and territory quality quickly and easily. With respect to their attraction to clearings, the proximate cue for juveniles is probably fairly simple: move toward brighter areas. This is suggested by laboratory experiments. in which juveniles hatched in captivity move to brightly lit areas, even if temperatures, food, perches, etc., are held constant ( J . A. Stamps, unpublished data). Similarly, perches, foliage, and other structural features of the habitat may be assessed directly via visual and tactile inspection. Frequently, new settlers move actively around the perches in a habitat for several hours before they begin to give territorial displays and to defend that area (Stamps, 1987a). Other types of important environmental factors are probably more difficult to assess accurately and quickly. The patterns of sun and shade over a given patch of habitat change over the course of a day, and they cannot be evaluated during overcast conditions. Direct assessment of the thermal regime on a given homesite requires 6-8 hours on a sunny day, but if juveniles settle during overcast conditions, it may take them several days to discover that their newly acquired territories exceed lethal temperatures when exposed to the midday sun.

TERRITORIAL BEHAVlOR

199

Direct assessment of the food supply would be very difficult forjuveniles under natural conditions. Since prey levels fluctuate in response to rainfall (see earlier), there is no guarantee that encounter rates on the first day or two in a habitat would be correlated with the mean or variance in encounter rates during the following 2 to 6 months, the period when a juvenile can expect to use its territory. In addition, microhabitats that meet all the other juvenile habitat requirements usually have sufficient food to support reasonable growth rates. Hence, it is not apparent that juveniles should invest much time in sampling the food supply when they first arrive in a habitat or a territory. In fact, there is no indication that juveniles sample food when they choose habitats or territories. Upon entering a vacant potential territory, juveniles move actively around the perches, inspect potential hiding places under leaves, and otherwise behave as if they were “exploring” the novel microhabitat. However, many juveniles begin to give territorial advertisement displays and chase conspecifics before they have captured or attempted to capture a single prey item (Stamps, 1987a,b; unpublished data). Many other aspects of habitat selection and spacing behavior are unrelated to food levels when juveniles first arrive at a patch; these include (1) the probability of settling in a patch (Stamps, 1991), (2) the final density of settlers in a patch (Stamps, 1990), and (3) territory sizes and territory overlap (Stamps and Eason, 1989). In addition, food levels tend to be higher and juvenile growth rates less food-limited in the wet season than in the dry season, yet there are no seasonal differences in juveniles’ territory sizes or overlap (Stamps, 1978; Stamps and Tanaka, 1981b; Stamps, 1984b). At this point, there is no indication that juveniles assess food directly when choosing habitats or territories. Although juveniles appear to ignore food supplies when they choose habitats and territories, they are not totally oblivious to spatial variation in prey abundance. Once in a while, as when ants or termites are moving their eggs and Larvae to another nest, juveniles are presented with a transitory, localized, abundant food source. In this situation, they move their center of activity to include this food source. One can mimic this situation by adding artificial point sources of food to the habitat, and juveniles respond appropriately by temporarily shifting their home ranges to take advantage of the windfall (Stamps and Tanaka, 1981a). On a slightly larger scale, if one artificially adds extra food to a patch of habitat within a clearing, animals from neighboring areas move into the area, leading to local decreases in territory size and increases in territory overlap (Stamps and Tanaka, 1981a; see also Stamps, 1990). These experiments show that juveniles have the capacity to respond to pronounced spatial variation in food levels, even though this type of variation is not particularly relevant to the situation when hatchlings normally settle in clearings or acquire

'00

JUDY STAMPS

territories. They also indicate that responses to artificial food manipulations on a small spatial scale may not necessarily provide a good model of the situation when animals naturally choose habitats or territories. Since juveniles are capable of adjusting their spacing behavior in response to artificial food supplements, this leaves us with the question of why they do not appear to sample food supplies prior to settling and acquiring feeding territories. One possibility is that the food that juveniles require does not vary appreciably on a spatial basis within the microhabitats that fulfill all of their other requirements. Another possibility is that these animals are able to use indirect cues such as the abundance of leaves or the density of leaf litter to reliably predict future food levels in their temtories. In addition, the arthropods preferred by these animals are probably difficult to sample during the settlement period and are apt to fluctuate widely on a temporal basis after the settlement period. Other insectivores may experience the same problems when selecting habitats and territories. Many insectivorous birds do not even respond to artificial manipulations of prey abundance by changing their territory sizes or reproductive output (Franzblau and Collins, 1980;Arvidsson and Klaesson, 1986; Rodenhouse and Holmes, 1992). In contrast, direct assessment of food levels may be more likely to occur in species whose food is easily assessed at the time of settlement and remains predictable during the period following settlement. Examples include seed-eating birds with winter feeding territories, in which the seed crop for the rest of the year is already present at the time when territories are established; in this situation artificial seed provisioning during the appropriate season may have substantial effects on territory sizes and local population density (e.g., Enoksson, 1990). b. Indirect Assessment and Conspecific Citing. As should be obvious from the previous section, juveniles may have difficulty directly assessing many of the environmental factors that affect habitat and territory quality. However, observations of these and other Anolis lizards suggested a way that naive newcomers might indirectly assess many of the environmental features that affect habitat and territory quality: conspecific cuing (Kiester, 1979; Stamps, 1987a). The concept of conspecific cuing rests upon two assumptions: ( I ) conspecifics of the same size, age, etc., have similar requirements for habitats or territories, and (2) temporal variation in habitat quality occurs on a scale that is long relative to the period of territory tenure. With respect to habitat and territory selection, conspecific cuing may occur on at least three different spatial scales: (1) choice of a patch of habitat. (2) choice of a territory location within a patch, and (3) choice of a particular territory. Conspecific cuing may occur when animals are choosing patches of

TERRITORIAL BEHAVIOR

20 1

habitat in which to settle. In the case of juvenile A . aeneus, important environmental factors such as the presence of resident sit-and-wait predators vary from one clearing to another, so at the very least the presence of established territory owners within a clearing would indicate that particular clearing was safe for occupancy. At this level, conspecific cuing predicts that the presence of established residents within a patch of habitat would increase the probability that subsequent new arrivals would settle in the same patch. This prediction was tested by establishing preferred juvenile habitat within clearings and then monitoring hatchlings from surrounding woodlands as they entered and settled in those clearings (Stamps, 1991). The conspecific cuing hypothesis predicted that hatchlings would be less likely to settle if they were among the first juveniles to arrive in a clearing and more likely to settle if they arrived later, after a number of residents had already established territories in the clearing. As expected, the probability of settlement increased as the clearings filled, until the last new arrivals settled and the probability of settlement abruptly declined to zero (Stamps, 1991). Conspecific cuing can also occur at a smaller spatial scale, as animals choose territory locations within a patch of territorial habitat. If habitat quality varies on a scale that extends beyond the boundaries of a given territory, the conspecific cuing hypothesis predicts that settlers would prefer territories next to established conspecifics to equivalent, isolated territories located elsewhere in the same patch of habitat. For instance, if a juvenile A . aenrws has a territory that is shaded from the midday sun, an adjacent cluster of plants and perches would also probably be suitable from a thermal point of view. The same might not be true for another set of perches located elsewhere within the same clearing. Hence, conspecific cuing predicts that hatchlings entering a clearing would be more likely to settle next to established territory owners rather than on equivalent territories located elsewhere in the same clearing. This prediction was tested by establishing two-sided, clear-walled, plastic enclosures within the center of clearings (Stamps, 1988a). Each side of the enclosure contained habitat suitable for juveniles, and homesites suitable for a single territory were established around the outside of each enclosure. After preparations were complete, a coin toss determined which side of the enclosure received juvenile lizards as residents (Fig. 2). These animals established territories inside the “experimental” side of the enclosure, and their behavior could be readily observed by outside juveniles through the clear plastic walls. The conspecific cuing hypothesis predicted that new arrivals to the clearing would prefer to settle on the territories surrounding the side of the enclosure containing the residents.

202

JUDY STAMPS

FIG. 2. Experimental design for field studies of conspecific attraction. E = the side of the ciear-walled enclosure that contained resident juveniles: C = the side that was vacant. Reproduced from Stamps ( 1988a): 0 1988 by the University of Chicago. All rights reserved.

When the juveniles within the enclosure were of the same size as the hatchling settlers, the results supported the prediction (Stamps, 1988a). In all eight trials, the first settler arrived at the experimental side of the enclosure. Over the course of each trial, the homesites around the experimental side filled up first, and juveniles who arrived on the experimental side moved less prior to settling than did juveniles who first arrived on the control side. These results are illustrated for one of these trials (Fig. 3). On the other hand, when the juveniles inside the enclosure were a few millimeters larger than the new settlers, newcomers did not prefer to settle near those residents (Stamps. 1988b). This makes sense in the light of studies of the social behavior of this species, since under natural conditions hatchlings who settle among larger residents are likely to end up as subordinates or floaters (Stamps, 1978). and since subordinates and floaters have lower growth rates than territory owners (Stamps, 1984a.b; Stamps and Eason, 1989). Of course, conspecific cuing of habitat or territory quality is only one of a number of possible reasons why territorial animals might prefer to settle next to conspecifics. Conspecific attraction would also be expected if settlers benefited by living next to neighbors during the period of territory tenure (see review in Stamps, 1988a), or if settlers decreased their future defense costs by settling next to conspecifics (Stamps and Krishnan, 1990).

203

TERRITORIAL BEHAVIOR

DAY

I

/I

6 0

.--VOMO

8 r

o+o

P

?B

FIG.3. An example of settlement patterns and conspecific attraction in lizards. Symbols for the enclosure are as in the legend to Fig. 2; events are summarized for each day until all the homesites were filled. Each symbol represents a different animal; the locations of each symbol represent the homesites the animals used during the morning and afternoon activity periods each day. Arrows indicate the direction of movement when animals moved from one homesite to another over the course of a day. Reproduced from Stamps (1988a); 0 1988 by the University of Chicago. All rights reserved.

In the case of juvenile A . aeneus, it is not yet known whether juveniles grow or survive at higher rates if they live within territorial aggregations than if they live in equivalent, isolated territories, or whether juveniles who leave spaces between their territories lose more space to newcomers than those who initially defend larger territories with contiguous boundaries. Hence, even though the results of these experiments support the hypothesis that juveniles are attracted to conspecifics while settling, hypotheses other than conspecific cuing might account for the tendency of juveniles to choose territories next to previous settlers.

204

JUDY STAMPS

Finally, conspecific cuing may be used when animals choose particular territories. If the quality of a given temtory does not change appreciably on a temporal basis, then the best indication that a territory is of high quality may be the fact that it has already supported a satisfied resident. If territories vary in quality within a patch, the conspecific cuing hypothesis predicts that animals who are already living in a patch would prefer territories that had previously supported a resident to equivalent territories with no history of previous use. This prediction was tested by establishing matched pairs of territories in clearings surrounded by preferred juvenile habitat; each of these territories was surrounded by a clear-walled enclosure (Stamps, 1987a). A coin toss determined which of the two territories received a juvenile territory owner. Animals living in the clearing for the next week could see and interact socially with the territory owner inside the enclosure and observe both the occupied and vacant territory through the clear plastic walls. Then both enclosures were simultaneously removed, along with the owner of the occupied territory, so that the juveniles living in the clearing simultaneously had access to two newly vacant territories, one of which had previously supported a territory owner. As predicted, juveniles who had lived in the clearing and been exposed to the territory owner in the enclosure preferred the previously used to the unused territory (Stamps, 1987a). In contrast, hatchlings who arrived in the clearing after the enclosures and the territory owner had been removed did not show any preference for the previously used territory (Stamps 1987a.b). In summary, there are a number of indications in this species that prospective settlers monitor the presence, location, and behavior of previous settlers when choosing habitats and territories. Naive newcomers prefer to settle in habitats that already contain residents, in territories next to established residents, and in territories previously used by conspecifics. These results support the idea that juveniles use conspecifics as cues when choosing habitats and territories. However, we should keep in mind that conspecific cuing is only one of several hypotheses that predict that settlers should prefer territories next to established territory owners. 111. THEFUNCTION OF TERRITORIAL BEHAVIOR A.

ASSUMPTIONS

I.

iritrodrrctiori

Assumptions about the function of territorial behavior patterns are closely tied to assumptions about the function of territories. By now, it

TERRITORIAL BEHAVIOR

205

should be apparent that much of the literature on territorial behavior and habitat selection has emphasized the competitive, aversive effects of conspecifics on one another (Table I). Under the circumstances, it is not surprising that the behavior exhibited by territory owners is often assumed to have an aversive, aggressive, and competitive function. This idea comes across very strongly in the terms that are used to describe the behavior of territorial animals. The behavior patterns characteristic of territory owners are called fights, combat, challenges, contests, attacks, chases, patrols, territorial defense, or territorial advertisement. Many of the words used to describe the behavior that animals use when acquiring or maintaining territories come loaded with implied function: such behavior must be combative, aggressive, aversive. Indeed, these days, the “default option” with respect to the function of territorial behavior seems to be the following: the function of territorial behavior is to repel or expel competing conspecifics from the area containing the resources (Falls, 1978; Waser and Wiley, 1979; Morton, 1982; Searcy and Anderson, 1986; see Table I). The assumption that territorial behavior is aversive leads to a number of more specific ideas about the ways that territory owners ought to behave when settling into habitats and acquiring territories. The first and most obvious of these is that prospective territory owners should avoid settling in areas that already contain many territory residents. This assumption follows directly from habitat selection theory. If one assumes that individual fitness always declines as the density of conspecifics increases, then ergo, animals should prefer empty habitats to equivalent habitats that already contain territory residents. This assumption can also be derived from the standard economic approach to territorial behavior. If the only effect of conspecificsin territorial species is to increase defense costs, then animals would be better off settling in areas containingfew conspecifics, so as to reduce the costs of interacting with those individuals. The assumption that settlers avoid high-density areas can be traced back to some of the classic papers on territorial behavior. For example, Kluyver and Tinbergen (1953) interpreted patterns of habitat use in great tits (Parus major) by assuming that settlers had “an aversion from densely populated localities.” This assumption was eventually incorporated into Fretwell and Lucas’s models, and hence propelled into the modern literature on habitat selection. Kluyver and Tinbergen (1953) were also among the first to suggest that newcomers might use the territorial behavior of residents to assess population density, enabling newcomers to avoid densely settled areas without entering them and interacting with the residents. This intuitively satisfying assumption was incorporated into subsequent experimental

206

JUDY STAMPS

studies (e.g., Krebs, 1977)and leads to the related assumption that territorial advertisement signals (e.g., bird songs, mammalian scent marks, lizard head bobs, etc.) discourage intruders from entering territories or habitats. In the event that prospective settlers did venture into established neighborhoods, it was assumed that the aggressive behavior of the territory owners would discourage those newcomers from settling and establishing territories of their own. In other words, attacks by territory owners on newcomers were assumed to discourage newcomers from settling in the areas in which they were attacked (van den Assem, 1967; Brown and Orians, 1970; Patterson, 1980; Kaufmann, 1983). Similarly, animals were assumed to gain space in territorial neighborhoods by winning contests. This assumption eventually found its way into game theoretical models of contests involving territorial animals (Maynard Smith and Parker, 1976; Enquist and Leimar, 1983; Grafen, 1987; Ydenberg er al., 1988). Currently, the assumption that animals win space by winning fights is widespread in the theoretical literature, and many field workers have assumed that animals who win contests in an area end up in possession of that area (Ydenberg et al., 1988; Rohwer and Roskaft, 1989; Marden and Waage, 1990). In all fairness, 1 should point out that the territorial literature also contains alternative ideas about the significance of agonistic interactions in territorial species. For example, even within the game theory literature, there is not unanimous agreement that animals who win fights win space. A prime example is the chapter on territorial behavior in Maynard Smith (1982).Maynard Smith emphasized that since space is a divisible resource, one might expect social interactions during territory establishment to resemble bargaining and negotiation, rather than contests in which the “winner takes all.” He also noted that it was not possible to devise a good model of the process of territorial establishment given the almost total lack of empirical information on how this process works. That observation is as true today as it was then. 2 . Conspecifics: Aversion or Attraction The observant reader may recall some results from juvenile A . ueneus that bear on the effect of conspecifics on the probability of settlement. The experiments already discussed in the context of conspecific cuing show that newcomers in this species are attracted rather than repelled by the presence and advertisement behavior of conspecific residents. Indirect evidence from other territorial species also suggests that naive newcomers may be attracted to established territory owners (Hoeck, 1982; Shields et ctl.. 1988; reviews in Stamps, 1988a; Smith and Peacock, 1990).

TERRITORIAL BEHAVIOR

207

Other studies of A . aeneus have confirmed the assumption that when newcomers are attacked at high rates by previous residents, they are less likely to settle in an area than if they are attacked at lower rates (Stamps, 1992). However, this same experiment also indicated that the effects of social interactions are not entirely aversive, and that if newcomers are attacked by previous settlers at low to moderate rates, they are more likely to settle than if they enter an equivalent patch that contains no previous settlers. Studies of birds suggest that the effects of territorial advertisement signals on conspecifics depend a great deal on the context in which those signals are produced. If tape recordings of song are played within established territories, they often reduce the tendency of other individuals to enter those territories (review in Kroodsma and Byers, 1991). In this situation, neighbors, floaters, and other residents in the neighborhood have had an opportunity to interact with the resident, and they may have learned to associate his songs with the outcome of those social interactions. In any event, songs that emanate from within an established avian territory are apt to discourage intrusions into that territory by conspecifics. In contrast, if songs are played in an empty habitat, they may attract rather than repel conspecifics. Over the years, many field biologists have noted that bird songs appear to attract prospective territory owners (Lack, 1948; Durango, 1950; Svardson, 1949; Emlen, 1952). Recently, this idea has been explicitly tested in the field using tape recordings, and in at least four cases so far, newcomers seem to be attracted rather than repelled to loudspeakers playing the songs or calls of conspecific territory owners (Alatalo et al., 1982; Yasukawa, 1990; Goldberg and Ewald, 1991; Mountjoy and Lemon, 1991). Hence, when avian songs are played in a vacant habitat, they seem to encourage investigation, if not settlement, by newcomers to that habitat. The scent-marking literature for mammals has proceeded along similar lines. Although some mammals do avoid areas marked by a familiar, dominant individual, newcomers to an area are not necessarily repelled by the scent marks of previous settlers in that habitat (Gosling, 1982; Gosling and McKay, 1990). Currently, this literature is in a revisionist phase, as workers reexamine old assumptions about the function of the scent marks produced by territorial and other animals. Although exact parallels to the experiments with bird songs are lacking, recent studies of wild Mus musculus indicate that naive mice are not repelled by the odor of established male residents (Hurst, 1989, 1990), and that when prospective territory owners enter unsaturated habitats, they may be attracted (not

208

JUDY STAMPS

repelled) by the scent marks of established territory owners (J. L. Hurst, personal communication). These results imply that if one were to add the scent of a territory owner to an unoccupied habitat, this odor might be more likely to elicit approach and investigation than avoidance from prospective territory owners. Although definitive empirical studies are still quite rare, the available evidence suggests that newcomers are not always repelled by the presence and behavior of established territory owners. One possible explanation for this pattern is that some of our underlying assumptions about territory function, habitat selection, and assessment are incorrect. For instance, if fitness is related to density via an Allee-type curve, then it would not be surprising if newcomers in relatively empty habitats were attracted to established conspecifics. Similarly, if animals use conspecific cuing to assess habitat quality, then one would expect naive newcomers to be attracted. not repelled, by established conspecifics. However, there is a deeper problem here, one that cannot be addressed simply by reformulating assumptions about habitat quality, territory function, or assessment. Many current models of territorial behavior and habitat selection ignore the proximate behavior processes that territorial animals must use to establish spatial and social relationships with one another. That is. they have lost sight of the fact that territorial behavior is social behavior (Baerends and Baerends-van Roon, 1950). In most species, territory locations and boundaries are not defined by lines drawn upon the landscape, ready to be discovered and then defended by territory owners (but see Reid and Weatherhead, 1988). Instead, territories are the spatial outcome of a series of social interactions involving individuals who will eventually live together in the same patch of habitat. These social interactions determine which individuals will become the final owners of contested bits of real estate, the location of territory boundaries, and the relative status of the neighbors, floaters, subordinates, and other individuals that will inhabit that neighborhood during the period of territory tenure. If the social and spatial relationships in territorial species are determined by social interactions, then it is difficult to imagine how any newcomer could settle and acquire a territory without first interacting with other animals in that area. At the very least, a newcomer needs to determine which areas are already claimed by previous residents. Unless boundaries are strictly determined by environmental features, one of the simplest ways for a newcomer to discover which areas are already claimed is to approach established territory owners, who will then obligingly chase the intruder to their boundaries! In addition, a series of social interactions is often required to establish territory boundaries with neighbors, and the rate of these interactions declines dramatically once neighbors have estab-

TERRITORIAL BEHAVIOR

209

lished stable social and spatial relationships with one another (Stefanski, 1967; Nolan, 1978; Copenhaver and Ewald, 1980; Krebs, 1982; Hurst, 1990). If newcomers must interact with other settlers in order to establish social relationships and territory boundaries, then one would expect prospective settlers to initiate rather than avoid interactions with previous residents in the habitat. Conversely, if prospective settlers are attracted to territory owners and their advertisement signals, it may make more sense to view this behavior as an attempt to initiate a social relationship or elicit information from the owner, rather than as an attempt to steal resources or take over the territory. It is apparent that assumptions about the function of territorial behavior require more information than is currently available about the process of territorial establishment. Until we know how prospective territory owners acquire space, it will be difficult to interpret the responses of newcomers to previous residents, or the “fights,” “chases,” “contests,” etc., that occur as animals settle in territorial neighborhoods. 3 . Space Acquisition and the Outcome of Contests A common assumption in the territorial literature is that territorial males acquire space as a result of winning contests (see the foregoing). Most of the empirical evidence cited in support of this assumption does not come from studies of social interactions as animals first acquire territories, because such studies are still quite rare (but see van Buskirk, 1986; Arcese, 1987; Stutchbury, 1991). Instead, field and laboratory studies usually focus on social interactions between one animal who has inhabited a territory for some period of time (the “owner”) and a second individual who is unfamiliar with that territory (the “intruder”). In social interactions involving owners and intruders, three things usually happen: (1) the owner wins the encounter, (2) the owner remains in possession of the territory, and (3) the intruder leaves (at least temporarily) (e.g., Phillips, 1971; Boer and Heuts, 1973; Riechert, 1982). The implicit assumption here is that there is a causal relationship between these events. That is, the owner retains possession because it won the fight. Conversely, it is assumed that if the intruder ever won a contest, it would take full and permanent possession of its newly acquired territory. Even on the face of it, there are several problems with this approach to social interaction between owners and intruders. First, it ignores field studies showing that the same individual may repeatedly intrude into a territory, even after repeatedly losing interactions with the owner of that territory. An early account of the importance of persistence in space acquisition can be found in Huxley’s classic elastic disc paper (1934), in

210

JUDY STAMPS

which he described how coots took space away from previous residents by engaging in a long series of social interactions with those residents. Many other studies have also suggested that newcomers insert themselves into established neighborhoods by persisting in spite of repeated attacks by established residents (Nice. 1937; Patterson. 1980; LaPrade and Graves. 1982; see also the following). These and other accounts show that intruders do not always leave after being attacked by residents and imply that persistent individuals may be able to wrest space from previous owners even if they do not win contests (Francis, 1984; Barlow ef af., 1986; Walton and Nolan, 1986: Jakobsson, 1988: Johnson and Kermott, 1990). An experimental study by Beletsky and Orians (1987) also tends to refute the assumption that winning contests plays a critical role in space acquisition. This study followed a protocol first used by Krebs (19821, in which an owner is removed from its territory, a replacement animal acquires the vacant territory, and then the owner is released. Usually in this situation both the original and the replacement owner fight vigorously over the territory. while experimenters monitor the outcome of these contests. Typically workers assume that the individual who won the initial contest or series of contests would end up in possession of the territory. However, Beletsky and Orians’s experiment with red-winged blackbirds was unusual in that they watched the territory in the days and weeks following the experiment, so that they knew which of the two animals was eventually successful in the competition over space. In terms of the outcomes of the fights between the original and replacement owner, the results were definitive. The original owners did very poorly in the first contest with the replacement males: 85% of them lost the first interaction. Seventy five percent of the original owners were still losers after the first series of contests, and most of them disappeared from the marsh at the end of the day. However, 2 weeks later, almost all (86%) of the original owners had regained their original territories, and another 4% had regained a portion of their original territory from the replacement male. Beletsky and Orians commented on the persistence of the original owners. but they did not observe the social interactions that preceded the success of the original territory owners. This experiment was recently repeated with the same result: original owners had dismal success in initial contests with replacement males, but most of them eventually regained their territories (Shutler and Weatherhead, 1992). Unfortunately, the social interactions preceding the change in ownership were not observed in this case either. Although there is still very little information about the relationship between social interactions and space acquisition under natural conditions,

TERRITORIAL BEHAVIOR

21 1

there are a few tantalizing hints that winning fights may not be a critical component of this process. Arcese (1987) observed contests between floaters and owners in song sparrows and noted that “floaters returned from chases repeatedly to sing or perch in the owner’s territory. Contests ended when the floater no longer returned to the territory after a chase or when the owner no longer chased the floater.” A similar pattern was observed by Stutchbury (1991) in a study of territory acquisition in purple martins. Young birds acquired territories by repeatedly intruding into the territory of an older resident, until the older male eventually stopped attacking and ceded the newcomer a portion of its territory. There was no indication that successful newcomers ever won any of their fights with the previous resident. These and other descriptive studies of territory acquisition under natural conditions cast doubt on the assumption that success in aggressive contests is essential for space acquisition in territorial species. B. TESTINGASSUMPTIONS ABOUT THE FUNCTION OF TERRITORIAL BEHAVIOR USINGJUVENILE LIZARDS The process of territory acquisition has been studied in juvenile A . aeneus using methods that have been previously described. Patches of artificial, preferred habitat were established in the field, and then juveniles captured elsewhere were added to those patches. The position of each individual was recorded on a scale map every 20 min during the periods when the animals were active, and the location, type, and outcome of all social interactions were recorded. Then these data were analyzed using programs developed by my collaborator on this project, V. Krishnan. Among other things, this study provides the data necessary to study how social interactions affect space acquisition in this species. Here I summarize the most salient points from this study, details of which have been submitted for publication. 1 . Relationships between Social Behavior and Space Acquisition In theory, contests have two distinct functions in territorial animals: they can be used to acquire novel space or they can be used to defend familiar space. Hence, to determine if animals acquire space by winning encounters, one must consider each opponent’s prior experience with the area in question. In the current study, this was done by locating the coordinates of the first social encounter per dyad on a map of the study area, and then using the computer to “draw a circle” with a radius of 0.2 m around the location of this interaction. The familiarity of each member of the dyad with the

213,

JUDY STAMPS

vicinity of the first encounter was estimated by counting the number of 20-min location points that each animal spent within the circle prior to that interaction. For our purposes, two types of dyads were most relevant. First, we considered those dyads in which the two opponents first met in an area unfamiliar to both of them (no location points within the circle for either member of the dyad). If animals win space by winning fights, then when two animals meet in an area new to both of them, the individual who wins the contest should remain in that area, while the loser should abandon it. A second set of analyses focused on situations analogous to the resident-intruder interaction in established territorial neighborhoods. That is, we considered dyads in which one member was already familiar with the vicinity of the interaction (one or more location points within the circle), whereas its opponent was not (no location points in the circle). Since the acquisition and defense hypotheses both predict that animals familiar with an area should win interactions with individuals new to that area, in this analysis we only considered those dyads in which the “intruder” won the first interaction with the “resident.” If animals acquire space by winning fights, then if intruders win their first contest with a resident, the intruder should acquire the area in question while the resident should abandon it. We used two methods to estimate the space use of the winner and the loser after their first interaction. In the simplest analysis, we focused on those dyads that had only one social interaction with one another, and then determined the number of location points that each individual spent within a radius of 0.2 m of the interaction on the day following the interaction. The results of this analysis did not support the assumption that animals acquire space in which they won interactions. In fact, the typical response in this situation was for both the winner and the loser to abandon the area in which they had their first encounter. This was true when both individuals were new to the area of the interaction, and when the “intruder” won its first and only contest with the “resident.” Overall, both opponents abandoned the vicinity of their first interaction in 10 cases, winners remained and losers left in 4 cases, and in one dyad the loser remained in possession while the winner abandoned the area. Even though winners were a bit more likely to remain in possession than losers, it is hard to reconcile these results with the hypothesis that the function of contests is to acquire space! It might be argued that animals that had only a single interaction were not seriously competing for space, so a second set of analyses included dyads that engaged in a series of chases and fights with one another. This analysis considered only those dyads in which the individual who won

213

TERRITORIAL BEHAVIOR

the first interaction won every subsequent interaction with that opponent. Space acquisition was measured over a longer term than in the previous analysis: in this case, we noted whether or not the location of the first interaction was inside or outside each lizard’s final territory boundaries at the end of the trial. If winning a first encounter in an unfamiliar area plays a major role in space acquisition, then the location of the first contest should be located within the final territory boundaries of the winner, and outside the territory boundaries of the loser. The results of this analysis were similar to the previous one; in most cases, neither the winner nor the loser included the site of their first encounter within their final territory boundary. This was true when both animals were unfamiliar with the location of their first encounter (Table 111). Even though winners were more likely to remain in possession than losers, the difference was not significant because in most cases both animals retreated from the area in which they first encountered one another. Similarly, when “intruders” won their first contest with a “resident,” the most common result was for both animals to move away from the vicinity of that interaction, and in this case there was not even a hint of a difference in the tendency of winners and losers to use the area in the future (Table IV). These results imply that first contests between settlers may be largely irrelevant to space acquisition in this species. Instead, first encounters

TABLE 111 CONTESTOUTCOME A N D SUBSEQUENT SPACE USE IN JUVENILE LIZARDS: AREAS UNFAMILIAR TO BOTHOPPONENTSO

Loser remained Loser left

Winner remained

Winner left

1

1 15

7

The location, winner, and loser of the first interaction per dyad were recorded in the field, and the previous use of the space near this first encounter was determined for both opponents. In this analysis, neither opponent was familiar with the area around the first interaction. “Remained” indicates that the individual included the location of the first interaction within its final territory; “left” indicates that this location was not included in the final territory.

214

JUDY STAMPS

TABLE 1V CONTEST OUTCOME A N D SUBSEQUENT SPACEUSE I N JUVENILE LIZARDS: INTERACTIONS WONBY “INTRUDERS”“

Loser remained Loser left

Winner remained

Winner left

1

3 20

2

a In these dyads. one member was familiar with the vicinity of the first encounter, whereas the other (the “intruder”) was not. The table summarizes the space use following first interactions that were won by the “intruder.” See also Table 111.

are important not for space acquisition per se, but because they are used to establish dominance relationships among the settlers in the habitat. The outcome of the first interaction per dyad was an excellent predictor of the future dominance relationship between those two individuals. In fact, of the 169 dyads who had a first encounter that ended with a clear winner and loser, there were only 13 dyads in which the dominance relationship reversed later during the settlement period. Similarly, 73% of the 30 dyads that began with a fight ending in a draw maintained their codominant status for the remainder of the settlement period. These results suggest another way that contests might affect space acquisition. Perhaps these lizards use first encounters to establish dominance relationships, and then compete for space within the context of these dominance relationships. In particular, dominants might aggressively force subordinates out of space desired by both settlers. This hypothesis was tested by focusing on dyads that had multiple interactions that were all won by the same individual, and in which space transferred from one animal to the other during the settlement period. Space transfers were quantified by mapping each individual’s space use before they had their first encounter, and the noting those cases in which space originally used by one juvenile ended up within the final territory boundaries of the other individual. If animals win space by winning contests. then we would predict that dominants would take space away from subordinates. Actually. subordinates were surprisingly successful in taking space away from dominants. In 19 dyads, the dominant member of the pair ended up owning space that used to belong to the subordinate, but in 12 other dyads, the subordinate took space away from the dominant, despite

TERRITORIAL BEHAVIOR

215

the fact that the subordinate lost every single interaction! Again, the outcome of contests had little, if any, relevance to success in space acquisition. In these animals, persistence was more important in space acquisition than was the outcome of social interactions. In order to pry space away from subordinates, dominants had to repeatedly and vigorously attack them. Conversely, subordinates gained space by repeatedly returning to an area in the face of repeated attacks, until eventually the dominant stopped attacking and ceded some space to the subordinate. Regardless of the status of the individuals involved, taking space away from another individual was expensive, as measured by the number of interactions, days spent fighting and chasing, etc. This may explain why the most common result after lizards first encountered one another was for both individuals to move away from the site of that interaction (see Tables I11 and IV). It also helps explain why most of the settlers in this system did not attempt to take space away from other animals, but instead located and then settled in areas not yet claimed by other settlers. All of these results raise the question of why these animals fight one another during the settlement period, if the outcome of these interactions is largely irrelevant to space acquisition. The key to this question is that contests (especially first encounters) are used to establish dominance relationships among the settlers in a habitat. In turn, the dominant member of the dyad has substantial control over the time, place, and intensity of future interactions and, even if dominants end up ceding space to subordinates, the former may be able to retain more valuable areas for themselves. More important, previous studies have shown that dominance status does influence juvenile growth rates in the postsettlement period, probably because high-status individuals are more likely to achieve exclusive territories of the optimal size than are lower-status individuals (see Section II,B,l,d). Hence, winning a first encounter is important because it establishes a social relationship that may eventually affect a juvenile’s status in the neighborhood, territory size, and extent of overlap with neighboring individuals. However, winning contests is not the only factor that affects success in settlement and territory acquisition, and many juveniles who lose first encounters with other settlers eventually end up with perfectly respectable territories. In any case, the situation in this species is far more complex than envisioned in models that assume that animals acquire space as a direct consequence of winning contests.

2 . Reexamining Assumptions about Contests in Territorial Species It may be time to reexamine current assumptions about the functional significance of the “contests” that occur in territorial species. In particu-

216

JUDY STAMPS

Iar, it is worth asking whether the basic assumptions of current game theory models apply to territorial species. Not only is space a divisible resource (cf. Maynard Smith, 1982). but animals that live in a neighborhood are likely to repeatedly encounter the same “opponents” over an extended period of time. As has been pointed out in a number of different contexts, games between individuals who repeatedly encounter one another are apt to be different than those in which each interaction involves a different pair of individuals (van Rhijn and Vodegel, 1980; Axelrod and Hamilton, 1981: Trivers, 1985; Noe, 1990). If animals have contiguous territories, then it is certainly likely that neighbors will encounter one another repeatedly during the period of territory tenure. In addition, in many species owners and particular “intruders” may also interact repeatedly with one another (see Section 111,,4,3). For the sake of simplicity, theoretical or empirical studies often assume that intruders visit each territory only once, or that the outcome of each interaction involving an owner-intruder dyad is independent of any other interaction involving those two individuals (e.g., Rosenberg and Enquist. 19911. However. if floaters, subordinates, flocks, etc., repeatedly encounter the same territory owner, and if interactions involving the same two individuals are not independent of one another, then many of our current models and experimental designs may need to be revised. If individuals do have the opportunity to interact with one another on a repeated basis, and if space can be divided among two competing individuals, it is not clear why winning any particular contest should play a definitive role in space acquisition. Instead, it may be more useful to think of social interactions as a way to establish social relationships among the settlers in a habitat, to interrogate other settlers about the quality of their territories or their residency status, or to negotiate the division of space desired by more than one individual. Relaxing old assumptions can lead to a different series of questions about territorial behavior than have been asked in the past. For example, it is often assumed that escalated fights involving physical contact are more costly than chases or fights involving displays, but this assumption is predicated on the assumption that pairs of contestants have only one encounter with one another. As was noted earlier, juvenile A. aeneu.7 frequently engage in a series of interactions during the settlement period. Analyses of these interactions show that the number of interactions per dyad is strongly related to the type of the first interaction. If the first contest is an escalated fight, then this is usually the only interaction for that dyad. Typically, the two individuals establish a territory boundary near the location of the escalated fight, and thereafter limit their interactions to the exchange of territorial advertisement displays across this

TERRITORIAL BEHAVIOR

217

boundary. In contrast, if two animals begin with a chase, they usually engage in a series of chases or fights, and in some cases one member of the dyad takes space away from the other. In this species, beginning with an escalated fight rather than a chase might be a less expensive option in the long run, since the former quickly and decisively establishes a mutually respectful social and spatial relationship and allows both neighbors to begin accruing the benefits of territory ownership. Conversely, a chase may be less costly in the short run, but it is usually only the first salvo in a long series of interactions, which may result in the loss of space to the other individual. IV. FUTURE DIRECTIONS This is a very interesting and active time in the study of habitat selection and territorial behavior, as workers reexamine and test assumptions that have been taken for granted for the past twenty years. Ecologists are reevaluating the assumption that food is the most important determinant of habitat or territory quality (Huey, 1991; Schluter and Repasky, 1991). Behavioral ecologists are reemphasizing the distinction between the proximate mechanisms that govern territory and habitat selection and the ultimate factors that affect habitat and territory quality (Ruby, 1986; Deslippe and M’Closkey, 1991; Armstrong, 1992). The renewed interest in proximate mechanisms has encouraged studies of the cues that naive settlers use when choosing habitats and territories (Robinson, 1985; Erckmann et al., 1990; Orians and Wittenberger, 1991; Dickman, 1992; Walters et al., 1992). Conspecific cuing has attracted particular attention because neither the animals in question nor the human investigator must identify and sample all the environmental factors that might affect habitat quality in that particular species. Conspecific cuing implies that naive newcomers would be more attracted to areas containing residents than to comparable empty areas. This mechanism has obvious implications for ecologists working on the dynamics of metapopulations (Ray et al., 1991), as well as conservation biologists interested in attracting settlers to reserves or other vacant areas (Smith and Peacock, 1990; Alper, 1991; Verner, 1992). Another recent line of work in territorial behavior has focused on the benefits that territorial animals might gain as a result of sharing a neighborhood with conspecifics. While granting that territory owners do compete with one another, this approach considers the ways that owners might benefit from the presence of adjacent conspecifics. At the simplest level, these studies focus on reasons why territorial animals might incur shortterm defense costs to increase net payoffs over the period of territory

218

JUDY STAMPS

tenure. For example, one reason for settling next to conspecifics within relatively homogeneous habitats is to prevent later arrivals from inserting themselves between previous territory owners (Getty, 1981; Stamps and Krishnan, 1990). Interestingly. theoretical models show that settling next to neighbors leads to larger final territory sizes even if newcomers are unable to take any space away from previous settlers, and that both early and late arrivals into territorial neighborhoods can benefit by settling next to established territory owners (Stamps and Krishnan, 1990). Other studies suggest that animals with neighbors may experience reduced intrusion or predation rates, because ( I ) owners in territory aggregations can expel intruders more easily than the owners of isolated territories (Hart. 1987), (2) neighbors may act as “early warning systems” that permit adjacent owners to detect sneaky intruders into their own territories (Eason and Stamps, 19931, (3) territory owners expel intruders from their neighbor‘s territories (Colgan et ul., 1979: Patterson, 1980; Wiley and Wiley, 1980; Vines. 1981; Burger, 1984: Freeman, 1987; Getty, 1987), or (4) neighboring territory owners alert their neighbors to the presence of predators in the neighborhood (Smith, 1986: Beletsky, 1991). In species in which mating occurs on territories, there is renewed interest in the ways that male territorial behavior might affect male reproductive success. This new literature acknowledges that females are not inert objects in this process, but that their behavior may have a major impact on the spacing patterns of the opposite sex. One line of research considers the effect of male territory size on male mating success, in view of recent studies showing that females do not necessarily mate with the male whose territory they occupy. Studies of supposedly monogamous birds show that extrapair copulation and paternity are common, that neighboring territory hoiders are often responsible. and that females may venture out of their spouse’s territory while seeking extrapair copulations (Smith, 1988; Gibbs c>t (11.. 1990; Gowaty and Bridges, 1991). All else being equal, in this situation males defending large territories might father more of their own offspring than males with smaller territories (MBller, 1987, 1992b;but see Dunn, 1992). This prediction was recently confirmed in an experimental study of bluebirds, in which the distance between adjacent nestboxes was manipulated and the effects on extrapair paternity assessed (Gowaty and Bridges, 1991). As expected, closely spaced nests had a higher frequency of nondescendant offspring than widely spaced nests. Interestingly. although extrapair copulations accounted for most of this effect, closely spaced nests also suffered from intraspecific egg parasitism. These results suggest another potential benefit of spacing in territorial birds: the avoidance of “egg dumping” by conspecific females.

TERRITORIAL BEHAVIOR

219

The reverse of spacing is aggregation. Hence, we can turn the question around and ask whether there are situations in which animals with mating territories might benefit by settling at relatively close distances from one another. One possibility mentioned earlier is that males that settle in groups might attract more females than males who settle farther from one another. This idea has a long and distinguished history in the literature (May, 1949; Lack, 1948; Durango, 1950), and it has been incorporated into both theoretical and empirical studies of animals in which males defend small mating territories on leks (Bradbury, 1981; Bradbury and Gibson, 1993). Several studies indicate that females prefer larger clusters of male territories to smaller ones, and that this preference is strong enough to lead to higher per capita average encounter rates for the males on larger leks (Campbell, 1990;Alatalo et al., 1992; Lank and Smith, 1992; but see Jennings and Phillip, 1992). The idea that females might prefer groups of territorial males to isolated males has been extended to frogs and insects in which males advertise their presence using auditory signals (e.g., Given, 1988; Bailey, 1991; Tejedo, 1993). In some orthopterans, males in aggregations experience higher mating success than isolated males, and females prefer the calls produced by groups of males to those of solitary males (Bailey, 1991). An interesting feature of these studies is that females sometimes rely on auditory cues that unambiguously indicate the presence of more than one calling male. For instance, if females simply responded to sound intensity, a solitary, nearby male would be as attractive as a group of more distant males. However, the females of some orthopterans prefer the calls of an unstructured male chorus, or even random noise, to the calls of a single male, even when these sounds are played at the same intensity (Bailey, 1991). These results imply that these females are not simply interested in locating one potential mate of the appropriate species, but that they may prefer to choose a mate from within a cluster of likely prospects. For good practical reasons, most of the studies discussed so far have focused on species in which males defend relatively small mating and breeding territories. When males defend larger territories on which mating occurs, it is usually assumed that females choose mates independently of the number or distribution of other males in the same habitat. However, females in this situation might also be interested in having a choice of males within a particular habitat. The question of whether females in species with ‘‘large’’ territories prefer male aggregations is an open one, and it deserves further study. Of course, females could also exert other types of selective pressures

220

JUDY STAMPS

on male territorial behavior. In some species, fighting with neighbors may be incompatible with successful courtship (Rowland, 1988) or parental care (Wingfield et al., 1990). From the female perspective, a group of males who had worked out mutually satisfactory social and stable relationships with one another might be preferable to an equivalent group of males who were still fighting with one another. If that were true, then females should begin “mate hunting” in territorial neighborhoods in which the males indicate by their behavior that they have already settled their disputes and are ready to devote themselves to courtship or the responsibilities of parenthood. Cues that a neighborhood is settled include coordinated countersinging among males or the absence of aggressive signals indicative of temtorial contests (Kramer and Lemon, 1983). 1 am unaware of any empirical data on the question of whether females prefer “settled territorial neighborhoods” to equivalent areas in which the males are still fighting with one another. However, there is indirect evidence suggesting that clusters of red-winged blackbirds who are familiar with one another have higher reproductive success than unfamiliar groups, perhaps because the familiar clusters attract females earlier or attract higher-quality females than d o groups with unfamiliar neighbors (Beletsky and Orians, 1989). Males familiar with one another may engage in less fighting while females are choosing mates, since they have already established their social and spatial relationships in previous years (see also Godard, 1991). Of course, if females d o exert pressure on males to settle their disputes quickly and permanently, and prefer groups producing coordinated advertisement signals to those whose members are still fighting with one another, then our perspective on the functional significance of territorial behavior would need to be revised. In that situation, established owners and newcomers would both benefit by establishing mutually satisfactory social and spatial relationships with one another well in advance of the time when females begin to choose mates. Indeed, some recent studies of birds suggest that young males may settle, learn the dialects of a neighborhood, and establish territories in the fall of their first year, months before the onset of the breeding season (DeWolfe et al., 1989; M . Beecher, personal communication). Another new area of research focuses on variation among territory owners and asks whether territory owners might benefit from the presence of particular types of neighbors. As was noted earlier, juvenile A. aeneus do pay attention to the sizes of the previous residents in a patch and are reluctant to settle near juveniles larger than themselves. Settlers in other species may also attend to the relative competitive ability of established residents when choosing neighborhoods or neighbors. For instance, young

TERRITORIAL BEHAVIOR

22 1

black-headed grosbeaks live in areas segregated from those used by older individuals, and this segregation is maintained even if older birds are removed from their territories (Hill, 1988). Recent studies of birds in which females mate with the males on adjacent territories suggest another reason why established residents might respond differentially to various categories of potential neighbors. If a male is relatively attractive (e.g., older, more colorful, etc.), then he might increase his reproductive success by encouraging less attractive males to settle nearby, because of the opportunity of mating with the spouses of those unattractive neighbors (see also Meller, 1992a). This scenario has been offered as one reason why male purple martins might eventually give in to persistent younger male intruders and allow them to take space that used to belong to the older bird (Morton et al., 1990). All else being equal, this idea implies that attractive males should be more aggressive toward potential settlers on neighboring territories if they were attractive than if they were unattractive. However, before embarking upon studies of the responses of attractive males to different categories of potential neighbors, it might be advisable to consider the effects of male groups on female mate choice. For instance, if females prefer to search for mates within clusters of attractive males, then attractive males might do better by settling next to one another than by surrounding themselves with unattractive neighbors. Hence, as in the previous example, we need to know something about female behavior before we can say much about the spacing behavior of the males of the species. Many assumptions about territorial behavior and habitat selection are currently being reevaluated and tested, and new ideas about the functional significance of habitat choice and territorial behavior are appearing on a regular basis. Since we know so little about the behavioral processes that occur during settlement in free-living territorial animals, useful information can be obtained without a major outlay of time and expense. Indeed, several of the assumptions outlined in this paper can be tested via careful observation of individuals as they settle into habitats, choose territories, and establish social and spatial relationships with one another. When combined with experimental or statistical controls over important confounding variables, studies of how animals actually behave when choosing habitats and acquiring territories can go a long way toward confirming or rejecting some widely held assumptions about territorial animals. The combination of careful description and quantification of natural behavior with the techniques of experimental behavioral ecology will help put the study of habitat selection and territorial behavior on a firm empirical and theoretical foundation.

222

JUDY STAMPS

V . SUMMARY Over the years, students of territorial behavior have relied on a number of assumptions about the ultimate and proximate reasons why animals choose to live in particular habitats, why they defend particular types of territories within those habitats, and why they exhibit particular behavior patterns during the period of territory acquisition and maintenance. AIthough some of these assumptions have been tested, this is by no means true of all of them. After an initial period in which field workers described and categorized habitat selection and territorial behavior in a wide variety of species, simplified theoretical models were developed to study these phenomena. The original economic models of territoriality and ideal despotic models of habitat selection assumed that the ultimate benefits of habitat or territory choice can be measured in terms of access to limited resources (especially food), that conspecifics are competitors, and that individual fitness monotonically declines as a function of the number of individuals sharing a habitat (Table I ) . However, a series of field studies of juvenile lizards ( A . aeneus), as well as data from other territorial species, suggest that these assumptions bear closer examination. Habitat quality may be determined by a number of different environmental factors, some of which may be nested in importance, and habitat and territory quality need not be determined by the same suite of environmental factors. There are also a number of reasons why territorial animals might benefit from the presence of neighbors, in which case fitness would not decline monotonically as a function of density in territorial species (Table 11). A topic that links proximate and ultimate questions about habitat selection and territory function is assessment. It is often assumed that animals can accurately assess the relative quality of different habitats and territories and hence choose optimal habitats and spacing patterns (Table I). However, direct assessment of the environmental factors determining habitat and territory quality is difficult for many animals, includingjuvenile lizards. In this situation, prospective settlers may rely on indirect cues to habitat or territory quality, such as structural features of the habitat, or the presence or behavior of conspecifics in the area (conspecific cuing). If animals live in habitats in which selective pressures are changing (e.g., as a function of human disturbance), then indirect assessment rules could easily lead to "suboptimal" choices of habitats or spacing patterns (Table 11). As a result of &heheavy emphasis on competition in studies of territory function and habitat selection, many people have assumed that territorial behavior is aversive, for example. that contests, advertisement signals,

TERRITORIAL BEHAVIOR

223

and so forth function to keep conspecifics out of a habitat or a territory. More specific assumptions about territorial behavior are that potential territory owners avoid settling near conspecifics, that territorial advertisement signals deter newcomers from settling, and that animals acquire space by winning territorial disputes (Table I). In the rare instances when these assumptions have been tested in free-living species, they have not been validated. Recent studies of juvenile lizards and other species suggest that we may need to rethink our assumptions about the ultimate significance of territorial behavior patterns, and about the proximate mechanisms that are responsible for settlement, territory acquisition, and the maintenance of social and spatial relationships in territorial species (Table 11).

Acknowledgments Many of the ideas in this paper were developed in the course of conversations with colleagues; I am especially grateful to M. Beecher, R. Gibson, P. Gowaty, B. Hart, and V. Krishnan for ideas and discussion. Previous drafts of this paper were much improved by suggestions from D. Lott, M. Mangel, M. Milinski, K. Muller, M. Reid, C. Snowdon, P. Switzer, and P. Willmer. The experimental studies on juvenile Anolis aeneus were supported by grants from the National Science Foundation. I am also grateful to the University of Chicago Press for permission to reproduce Figs. I , 2, and 3.

References Abramsky, Z., Rosenzweig, M. L., Pinishow, B., Brown, J. S., Kotler, B., and Mitchell, W. A. (1990). Habitat selection: An experimental field test with two gerbil species. Ecology 71, 2358-2369. Addicott, J. F., Aho, J. M., Antolin, M. F., Padilla, D. K., Richardson, J. S., and Soluk, D. A. (1987). Ecological neighborhoods: Scaling environmental patterns. Oikos 49, 340-346. Alatalo, R. V., Lundberg, A., and Bjorklund, M. (1982). Can the song of male birds attract other males? An experiment with the pied flycatcher, Ficedula hypoleuca. Bird Behau. 4, 42-45. Alatalo, R. V., Hoglund, J., Lundberg, A,, and Sutherland, W. J. (1992). Evolution of black grouse leks: Female preferences benefit males in larger leks. Behuv. Ecol. 3, 53-59. Allee, W. C. (1931). “Animal Aggregations: A Study in General Sociology.” Univ. of Chicago Press, Chicago. Allee, W. C., Emerson, A. E., Park, O., Park, T., and Schmidt, K. P. (1949). “Principles of Animal Ecology.” Saunders, Philadelphia. Alper, J. (1991). To everything (tern, tern, tern) there is a season. Science 253, 740-741. Anderson, R. M., and May, R. M. (1981). The population dynamics of microparasites and their invertebrate hosts. Philos. Trans. R. SOC.London, Ser. B 291, 451-524. Arcese, P. (1987). Age, intrusion pressure and defence against floaters by territorial male song sparrows. Anim. Behau. 35, 773-784.

224

JUDY STAMPS

Arcese, P.. Smith, J. N. M.. Hochachka, W. M.. Rogers. C. M., and Ludwig, D. (1992). Stability. regulation, and the determination of abundance in an insular song sparrow population. Ecology 73, 805-822. Armstrong. D. P. (1992). Correlation between nectar supply and aggression in territorial honeyeaters: Causation or coincidence? Behau. E d . Sociohiol. 30, 95- 102. Arvidsson. B.. and Klaesson. P. (1986). Territory size in a willow warbler Phv//oscopits trochilus population in mountain birch forest in Swedish Lapland. Ornis Scand. 17, 24-30. Axelrod. R.. and Hamilton. W. D. (1981). The evolution of cooperation. Science, N. Y. 211, 1390-13%. Baerends, G. P.. and Baerends-van Room J. (1950). An introduction to the study of the ethology of cichlid fishes. Behauiour. Suppl. 1, 1-242. Bailey. W. J. (1991). Mate-finding: Selection on sensory cues. IN”Reproductive Behaviour in Insects” (W. J. Bailey and J. Ridsill-Smith. eds.). pp. 43-73. Chapman & Hall, London. Barlow, G. W., Rogers, W.. and Fraley. W. (1986).Do Midas chichlids win through prowess or daring-It depends. Behau. Ecol. Sociohiol. 19, 1-18. Begon, M.. Bowers. R. G.. Kodianakis. N., and Hodgekinson. D. E. (1992). Disease and community structure: The importance of host self regulation in a host-host-pathogen model. Am. Nar. 139, 1131-1150. Beletsky. L. D. (1991). Alert calls of male red-winged blackbirds-Call rate and function. Catl. 3. Z d . 69, 21 16-2120. Beletsky, L. D.. and Orians. G. H. (1987). Territoriality among red-winged blackbirds. 11. Removal experiments and site dominance. Behau. Ecol. Sociohiol. 20, 339-349. Beletsky. L. D.. and Orians, G. H . (1989). Familiar neighbors enhance breeding success in birds. Proc. Natl. Acad. Sci. U.S.A. 86, 7933-7936. Boer. J. N . . and Heuts. B. A. (1973). Prior exposure to visual cues affecting dominance in the jewel fish. Hemichromis bimacitlatits Gill 1862 (Pisces: Cichlidae). Behauioitr 44, 299-32 1. Boutin, S. ( 1990). Food supplementation experiments with terrestrial vertebrates: Patterns, problems, and the future. Can. J. Zoo/. 68, 203-220. Bradbury. J . W. (1981). The evolution of leks. I n “Natural Selection and Social Behavior: Recent Research and New Theory” (R. D. Alexander and D. W. Tinkle, eds.). pp. 138-169. Chiron Press. New York. Bradbury. J. W.. and Gibson, R. M. (1983). Leks and mate choice. I n “Mate Choice” (P. Bateson. ed.), pp. 109-138. Cambridge Univ. Press, Cambridge. Brown, J. L. (1964). The evolution of diversity in avian territorial systems. Wilson Bull. 76, 160- 169. Brown. J. L. (1969). The buffer effect and productivity in tit populations. A m . Nut. 103, 347-354. Brown. J . L., and Orians, G. H. (1970). Spacing patterns in mobile animals. Annrt. K e u . E c d . Svst. 1, 239-269. Burger. J . (1984). Pattern. mechanism and adaptive significance of territoriality in herring gulls (Larus orgentarus). Ornifhol. Monogr. 34, 1-92. Burt. W. H. (1943). Territoriality and home range concepts as applied to mammals. J. Mammal. 24, 346-352. Campbell, D. J. (1990). Resolution of spatial complexity in a field sample of singing crickets Teleogellus commodirs (Walker) (Gryllidae): A nearest neighbor analysis. Anim. Behau. 39, 1051-1057. Carpenter, C. R. (1958). Territoriality: A review of concepts and problems. I n “Behavior

TERRITORIAL BEHAVIOR

225

and Evolution” (A. Roe and G. G. Simpson, eds.), pp. 224-250. Yale Univ. Press, New Haven, CT. Carpenter, L. (1987). Food abundance and territoriality: To defend or not to defend? Am. ZOO/.27, 387-399. . Cody, M. L. (1985). An introduction to habitat selection in birds. In “Habitat Selection in Birds” (M. L. Cody, ed.), pp. 4-53. Academic Press, Orlando, FL. Colgan, P. W., Nowell, W. A., Gross, M. R., and Grant, J. N. A. (1979). Aggressive habituation and rim circling in the social organization of bluegill sunfish (Lepomis macrochirus). Environ. Biol. Fishes 4, 29-36. Copenhaver, C., and Ewald, P. W. (1980). Cost of territory establishment in hummingbirds. Oecologia 46, 155-160. Darling, F. F. (1952). Social behavior and survival. Auk 69, 183-191. Davies, N. B. (1989). Sexual conflict and the polygamy threshold. Anim. Behav. 38,226-234. Davies, N. B. (1991). Mating systems. In “Behavioral Ecology: An Evolutionary Approach” ( J . R. Krebs and N. B. Davies, eds.), 3rd ed., pp. 263-294. Sinauer, Sunderland, MA. Davies, N. B., and Houston, A. I. (1984). Territory economics. In “Behavioural Ecology: An Evolutionary Approach” ( J . R. Krebs and N. B. Davies, eds.), 2nd ed., pp. 148-169. Sinauer, Sunderland, MA. Davies, N. B., and Lundberg, A. (1984). Food distribution and a variable mating system in the dunnock, Prunella modularis. J . Anim. Ecol. 53, 895-912. Deslippe, R. J., and M’Closkey, R. T. (1991). An experimental test of mate defense in a iguanid lizard (Sceloporus graciosus). Ecology 72, 1218-1224. DeWolfe, B. B., Baptista, L. F., and Petrinovich, L. (1989). Song development and territory establishment in Nuttall’s white-crowned sparrows. Condor 91, 397-407. Dickman, C. R. (1992). Predation and habitat shift in the house mouse, Mus domesticus. Ecology 73, 313-322. Dunn, P. 0. (1992). Do male birds adjust territory size to the risk of cuckoldry? Anim. Behav. 43,857-859. Durango, S. (1950). F&gels&ngen.In “Svenska djur Fgglarna” (S. Durango and G. Svkdson, eds.), pp. 9-38. Udevalla, Stockholm. Eason, P. K., and Stamps, J. A. (1992). The effect of visibility on territory size and shape. Behau. Ecol. 3, 166-172. Eason, P. K., and Stamps, J. A. (1993). Territory neighbors as an early warning system. Anim. Behau. (in press). Emlen, J. T. (1952). Social behavior in nesting cliff swallows. Condor 54, 177-199. Enoksson, B. (1990). Autumn territories and population regulation in the nuthatch Sitta europaea: An experimental study. J. Anim. Ecol. 59, 1047-1062. Enquist, M., and Leimar, 0. (1983). Evolution of fighting behaviour: Decision rules and Theor. Biol. 102, 387-410. assessment of relative strength. .I. Erckmann, W. J., Beletsky, L. D., Orians, G. H., Johnsen, T., Sharbaugh, S . , and D’Antonio, C. (1990). Old nests as cues for nest-site selection: An experimental test with red-winged blackbirds. Condor 92, 113-117. Falls, J. B. (1978). Bird song and territorial behavior. In “Aggression, Dominance and Individual Spacing” (L. Kramer, P. Pliner, and T. Alloway, eds.), pp. 61-89. Plenum, New York. Fisher, J. (1954). Evolution and bird sociality. I n “Evolution as a Process” ( J . Huxley, A. C. Hardy, and E. B. Ford, eds.), pp. 71-83. Allen & Unwin, London. FitzGerald, G. J., Whoriskey, F. G., Morrissette, J., and Harding, M. (1992). Habitat scale, female cannibalism, and male reproductive success in three-spined sticklebacks (Gasterosteus aculeatus). Behav. Ecol. 3, 141-147.

226

JUDY STAMPS

R. C. (1984). The effects of bidirectional selection for social dominance on agonistic behavior and sex ratios in the paradise fish (Mucropodrts operc,it/uris).Behaviortr 90, 25-45. Franzblau. M . A . . and Collins, J. P. (1980). Test of a hypothesis of territory regulation in a n insectivorous bird by experimentally increasing prey abundance. Oeco/ogiu 46, IM- 170. Freeland, W. J. (1976). Pathogens and the evolution of primate sociality. Biotropica 8, 12-24. Freeman. S . (1987). Male red-winged blackbirds (Agekritrs phoenicerrs) assess the RHP of neighbors by watching contests. Behuu. Ecol. Sociohiol. 20, 43-52. Fretwell. S. D. (1972). "Populations in a Seasonal Environment." Princeton Univ. Press. Princeton. NJ. Fretwell. S. D.. and Lucas. H. L.. Jr. (1970). On territorial behavior and other factors influencing habitat distribution in birds. I . Theoretical development. Acra Biotheor. 19, 16-44. Get&. T. (lY8l).Competitive collusion: The preemption of competition during sequential establishment of territories. A m . Nut. 118, 426-431. Getty. T. (1987). Dear enemies and the prisoners' dilemma: Why should territorial neighbors form defensive coalitions'? A m . Zoo/. 27, 327-336. Gibbs. H. L.. Weatherhead. P. J., Boag. P. T.. White. B. N . . Tabak. L. M . , and Hoysak, 13. J . ( 1990). Realized reproductive success of polygynous red-winged blackbirds revealed by DNA markers. Science 250, 1394-1397. Gibson. R. M. (1992). Lek formation in sage grouse: The effect of female choice on male territory settlement. Anim. Behuu. 43, 443-450. Given. M. F. (1988). Territoriality and aggressive interactions of male carpenter frogs, Runa virguripes. Copeiu. pp. 41 1-421. Godard. R . ( 1991). Long-term memory of individual neighbors in a migratory songbird. Nuture (London)350, 228-229. Goldberg, T. L.. and Ewald. P. W. (1991). Territorial song in the Anna's hummingbird, C'dypte m n u : Costs of attraction and benefits of defense. At7im. Behuu. 42, 221-226. Gosling. L. M. (1982). A reassessment of the function of scent-marking in territories. %. 7ierpsy-hol. 60,87-1 18. Gosling, L. M.. and McKay. H. V . (1990). Competitor assessment by scent matching: An experimental test. Behuu. E c d . Sociobiol. 26, 41 5-420. Gowaty, P. A,. and Bridges. W. C. (1991). Behavioral. demographic. and environmental correlation of extrapair fertilization in eastern bluebirds. Siuliu siulis. Behuu. Ecol. 2, 339-350. Grafen. A. (1987). The logic of divisively asymmetric contests: Respect for ownership and the desperado effect. Anim. Behau. 35, 462-467. Hart. B. L. (1990). Behavioral adaptations to pathogens and parasites: Five strategies. Neirrosci. Biobchuu. Rev. 14, 273-294. Hart. D. D. (1987). Feeding territoriality in aquatic insects: Cost-benefit models and experimental tests. Am. Z o o / . 27, 371-386. Healv. M. C. (1967).Aggression and self-regulation of population size in deermice. E w l o g y 41). 377-392. Hilden. 0 . (1965). Habitat selection in birds. Ann. Zoo/. Fenn. 2, 53-75. Hill. G. E . (1988). The function of delayed plumage maturation in male black-headed grosbeaks. Auk 105, 1-10. Hindr. R. A. (1956). The biological significance of the territories of birds. lbis 98, 340-369. Hixon, M . A. (1987). Territory area as a determinant of mating systems. A m . Zoo/. 27, 129-247.

TERRITORIAL BEHAVIOR

221

Hoeck, H. N. (1982). Population dynamics, dispersal and genetic isolation in two species of hyrax (Heferohyraxbrucei and Procauiajohnstoni) on habitat islands in the Serengeti. Z. Tierpsychol. 59, 177-210. Hoffman, A. A. (1991). Heritable variation for territorial success in field-collected Drosophila melanogaster. A m . Nut. 138, 668-679. Houston, A. I., McCleery, R. H., and Davies, N. B. (1985). Territory size, prey renewal and feeding rates: Interpretation of observations on the pied wagtail (Motacilla alba) by simulation. J . Anim. Ecol. 54, 227-240. Howard, E. (1920). “Territory in Bird Life.” Dutton, New York. Huey, R. B. (1982). Temperature, physiology and the ecology of reptiles. I n “Biology of the Reptilia” (C. Cians and F. H. Pough, eds.), Vol. 12, pp. 25-91. Academic Press, London. Huey, R. B. (1991). Physiological consequences ofhabitat selection. Am. Nut. 137, s95-sl IS. Hurst, J. L. (1989). The complex network of olfactory communication in populations of wild house mice, Mus domesticus Rutty: Urine marking and investigation within family groups. Anim. Behav. 37,705-725. Hurst, J. L. (1990). Urine marking in populations of wild house mice Mus domesticus Rutty. I. Communication between males. Anim. Behau. 40, 209-222. Hutchinson, G. E. (1957). Concluding remarks. Cold Spring Harbor Symp. Quant. B i d . 22, 415-427. Huxley, J. S. (1934). A natural experiment on the territorial instinct. Br. Birds 27,270-277. Jakobsson, S. (1988). Temtorial fidelity of willow warbler (Phylloscopus trochilus) males and success in competition over territories. Behav. Ecol. Sociobiol. 22, 79-84. Jennings, M. J., and Philipp, D. P. (1992). Female choice and male competition in longear sunfish. Behav. Ecol. 3, 84-94. Johnson, L. S., and Kermott, L. H. (1990). Possible causes of temtory takeovers in a northtemperate population of house wrens. Auk 107, 781-784. Kaufmann, J. H. (1983). On the definitions and functions of dominance and territoriality. Biol. Rev. Cambridge Philos. Soc. 58, 1-20. Kiester, A. R. (1979). Conspecifics as cues: A mechanism for habitat selection in a Panamanian grass anole (Anolis auratus). Behav. Ecol. Sociobiol. 5, 323-331. Klomp, H. (1980). Fluctuations and stability in great tit populations. Ardea 68, 205-224. Klopfer, P. H., and Ganzhorn, J. V. (1985). Habitat selection: Behavioral aspects. I n “Habitat Selection in Birds” (M. L. Cody, ed.), pp. 435-453. Academic Press, Orlando, FL. Klopfer, P. H., and Hailman, J. P. (1965). Habitat selection in birds. I n “Advances in the Study of Behavior” (D. S. Lehrman, R. A. Hinde, and E. Shaw, eds.), Vol. 1, pp. 279-303. Academic Press, New York. Kluyver, H. N., and Tinbergen, L. (1953). Temtory and the regulation of density in titmice. Arch. Ne‘erl. Zool. 10, 265-287. Koenig, W. D. (1990). Territory size and duration in the white-tailed skimmer Plathemis lydia (Odonata: Lihellulidae). J . Anim. Ecol. 59, 317-333. Kramer, H. G., and Lemon, R. E. (1983). Dynamics ofterritorial singing between neighboring song sparrows. Behauiour 85, 198-223. Krebs, J. R. (1970). Regulation of numbers in the great tit (Aves: Passeriformes). J. Zool. 162, 317-333. Krebs, J. R. (1971). Territory and breeding density in the great tit, Parus major. Ecology 52, 1-22. Krebs, J. R. (1977). The significance of song repertoires: The Beau Geste hypothesis. Anim. Behav. 25, 475-478. Krebs, J. R. (1982). Territorial defence in the great tit (Parus major): Do residents always win? Behav. Ecol. Sociobiol. 11, 185-194.

228

JUDY STAMPS

Krebs. J . R.. and Pemns. C. M. (1978). Behaviour and population regulation in the great tit (Parus major). I n “Population Control by Social Behaviour” (F. J. Ebling and D. M. Stoddard, eds.). pp. 23-47. Institute of Biology, London. Kroodsma, D. E., and Byers, B. E. (1991). The function(s) of bird song. Am. Zoo/. 31, 318-328. Lack, D. (1937). The psychological factor in bird distribution. Br. Birds 31, 130-136. Lack. D. (1948). Notes on the ecology of the robin. Ibis 90, 252-279. Lank. D. B.. and Smith, C. M. (1992). Females prefer larger leks: Field experiments with ruffs (Philomachus pugnax). Behau. Ecol. Sociobiol. 30, 323-399. LaPrade, H. R.. and Graves, H. B. (1982). Polygamy and female-female aggression in redwinged blackbirds (Agelaius phoeniceus). Am. Nut. l20, 135-138. Lindstrom, K. (1992). The effect of resource holding potential, nest size and information about resource quality on the outcome of intruder-owner conflicts in the sand goby. Behau. Ecol. Sociobiol. 30, 53-58. MacLean. S. F.. Jr.. and Seastedt, T. R. (1979). Avian territoriality: Sufficient resources or interference competition. Am. Nar. 114, 308-312. Marden. J . H . . and Waage. J. K. (1990). Escalated damselfly territorial contests are energetic wars of attrition. Anim. Behau. 39, 954-959. Martin, T. E. (1988).On the advantage of being different: Nest predation and the coexistence of bird species. Proc. Natl. Acad. Sci. U . S . A .85, 2196-2199. May. D. J. (1949). Studies on a community of willow-warblers. Ibis 91, 24-54. Maynard Smith. J . (1974). “Models in Ecology.“ Cambridge Univ. Press, Cambridge. Maynard Smith, J. (1978). Optimization theory inevolution. Annu. Reu. Ecol. Syst. 9,31-56. Maynard Smith, J . (1982). “Evolution and the Theory of Games.” Cambridge Univ. Press, Cambridge. Maynard Smith. J., and Parker, G. A. (1976). The logic of asymmetric contests. Anim. Behau. 24, 159-179. McCleery, R. H., and Perrins. C. M. (1985). Territory size. reproductive success and population dynamics in the great tit Parus major. Spmp. Br. Ecol. Soc. 25, 353-374. Milinski. M.. and Parker, G. A. (1991). Competition for resources. In “Behavioural Ecology: An Evolutionary Approach” ( J . R. Krebs and N. B. Davies. eds.). 3rd ed., pp. 137-168. Sinauer, Sunderland, MA. Mitchell. W. A.. and Valone. T. J. (1990).Commentary: The optimization research program. Q . Rev. B i d . 65, 43-52. Mallei-, A . P. (1987). Intruders and defenders on avian breeding territories: The effect of sperm competition. Oikos 48, 47-54. Mdler. A . P. (1990). Effects of parasitism by a hematophagous mite on reproduction in the barn swallow. Eco/ogy 71, 2345-2357. Mbller, A. P. (1992a). Frequency of female copulations with multiple males and sexual selection. A m . Nut. 139, 1089-1101. Mdler. A. P. (1992b). Relative size of avian breeding territories and the risk of cuckoldry. Anim. Behciu. 43, 860-861. Moins. D. W . (1987). Spatial scale and the cost of density-dependent habitat selection. Euo/. E c o ~ .1, 379-388. Morris. D. W. (1989). Density-dependent habitat selection: Testing the theory with fitness data. Euol. Ecol. 3, 80-94. Morton. E. S. ( 1982). Grading. discreetness. redundancy and motivational-structural rules. In “Acoustic Communication in Birds” (D. E. Kroodsma and E. H. Miller, eds.), Vol. pp. 183-212. Academic Press, New York. Morton. E. S.. Forman, L., and Braun, M. (1990). Extrapair fertilizations and the evolution of colonial breeding in purple martins. Auk 107, 275-283.

TERRITORIAL BEHAVIOR

229

Mountjoy, D. J., and Lemon, R. E. (1991). Song as attractant for male and female European starlings, and the influence of song complexity on their response. Behau. Ecol. Sociobiol. 28,97-100. Myers, J. P., Connors, P. G., and Pitelka, F. A. (1979). Territoriality in nonbreeding shorebirds. Stud. Avian Biol. 2, 231-246. Nice, M. M. (1937). Studies in the life history of the song sparrow. I. Trans. Linn. SOC. N . Y. 4, 1-247. Noble, G . K. (1939). The role of dominance in the social life of birds. Auk 56, 263-273. Noe, R. (1990). A veto game played by baboons: A challenge to the use of the Prisoner’s Dilemma as a paradigm for reciprocity and cooperation. Anim. Behav. 39, 78-90. Nolan, V. (1978). “Ecology and Behavior of the Prairie Warbler Dendroica discolor,” Ornithol. Monogr. 26. Am. Omithol. Union, Lawrence, KS. Ono, K. (1981). The effects of habitat structure on the social behavior and spacing patterns of juvenile lizards (Anolis aeneus). Ph.D. Dissertation, University of California, Davis. Orians, G. H. (1991). Preface. Am. Nut. 137, sl-s4. Orians, G. H., and Wittenberger, J. F. (1991). Spatial and temporal scales in habitat selection. Am. Nat. 137, s29-s49. Parker, G. A., and Sutherland, W. J. (1986). Ideal free distributions when individuals differ in competitive ability: Phenotype-limited ideal free models. Anim. Behuu. 34,1222-1242. Patterson, I . J. (1980). Territorial behavior and the limitation of population density. Ardea 68,53-62. Phillips, R. P. (1971). ‘The relationship between social behavior and the use of space in the benthic fish Chasmodes bosquiunus Lackpide (Teleostei, Blennidae). 11. The effect of prior residency on social and enclosure behavior. 2. Tierpsychol. 29, 389-408. Pianka, E. R. (1988). “Evolutionary Ecology.” Harper & Row, New York. Pitelka, F. A. (1959). Numbers, breeding schedule, and territoriality in pectoral sandpipers of northern Alaska. Condor 61, 233-262. Pleszczynska, W. K. (1978). Microgeographic prediction of polygyny in the lark bunting. Science N . Y. 201,935-937. Pulliam, R. H., and Danielson, B. J. (1991). Sources, sinksand habitat selection: A landscape perspective on population dynamics. A m . Nut. 137, s50-s66. Pulliam, R. H., and Parker, T. H. (1979). Population regulation of sparrows. Fortschr. Zool. 25, 137-147. Raimondi, P. T. (1988). Settlement cues and determination of the vertical limit of an intertidal barnacle. Ecology 69, 400-407. Ray, C..Gilpin, M.,andsmith, A. T. (1991).Theeffect ofconspecificattractionon metapopulation dynamics. In “Metapopulation Dynamics: Empirical and Theoretical Investigations” (M. Gilpin and I. Hanski, eds.), pp. 123-134. Academic Press, London. Reid, M. L., and Weatherhead, P. J. (1988). Topographical constraints on competition for territories. Uikos 51, 115-1 17. Riechert, S. E. (1982). Spider interaction strategies: Communication versus coercion. In “Spider Communication: Mechanisms and Ecological Significance” (P. N. Witt and J. Romer, eds.), pp. 281-313. Princeton Univ. Press, Princeton, N.J. Riechert, S. E., and Tracy, C. R. (1975). Thermal balance and prey availability: Bases for a model relating web-site characteristics to spider reproductive success. Ecology 56, 265-284. Robinson, S . K. (1985). Fighting and assessment in the yellow-rumped cacique (Cacicus cela). Behav. Ecol. Sociobiol. 18, 39-44. Rodenhouse, N. L., and Holmes, R. T. (1992). Results of experimental and natural food reductions for breeding black-throated blue warblers. Ecology 73, 357-372. Rohwer, S., and Roskaft, E. (1989). Results of dyeing male yellow-headed blackbirds solid

230

JUDY STAMPS

black: Implications for the arbitrary identity badge hypothesis. Behav. 0 0 1 . Sociohid. 25, 39-48. Rosenberg. R. H.. and Enquist. M. f 1991). Contest behaviour in Weidemeyer's admiral butterfly Linzenitis tcwidrmeserii (Nymphalidae):The effect of size and residency. Anim. Brhuv. 42, 805-81 I . Rosenrweig. M. L. ( 1985). Some theoretical aspects of habitat selection. I n "Habitat Selection in Birds" (M.L. Cody, ed.), pp. 517-540. Academic Press, London. Rosenzweig. M. L. (1991). Habitat selection and population interactions: The search for mechanism. A m . N u t . 137. sS-s?8. Rowland. W. J. f 1988). Aggression versus courtship in threespine sticklebacks and the role of habituation to neighbors. Anim. Beliau. 36, 348-357. Ruby. D. E. (19861. Selection of home range site by females of the lizard .S~,eloporusjarroui. J . Herpetol. 20, 466-469. Sale. P. F. (1968). lnfluence of cover availability on depth preference of the juvenile maini. Acunthurits rriosregtcs sanduicensic. Cop& 68, 802-807. Searcy. W. A , , and Andersson. M. (1986).Sexual selection and the evolution of song. Annu. Rev. C O ~ .Sy.rt. . 17, 507-533. Schlurer, D., and Repasky. R . R. (1991). Worldwide limitation of finch densities by food and other factors. Erolog?: 72, 1763-1774. Schoener. T. W. (1983).Simple models of optimal territory size: A reconciliation. Am. Nut. 121, 608-619. Schoener. T. W. (1987). Time budgets and territory size: Some simultaneous optimization models for energy maximizers. Am. N o r . 27, 259-291. Shelford, V. E. (1913). The reactions of certain animals to gradients of evaporating power and air. A study in experimental ecology. Biol. Bit//. (Woods Hole. Muss.) 25, 79-120. Shields, W. M . , Crook. J . R.. Hebblethwaite, M. L.. and Wiles-Ehmann. S. S. (1988). Ideal free coloniality in the swallows. I n "The Evolution of Social Behavior" (C. N . Siobodchikoff. ed.). pp. 189-228. Academic Press. San Diego. CA. Shutler. D.. and Weatherhead. P. J. (1992). Surplus territory contenders in male red-winged blackbirds: Where are the desperados'? Bchur;. Ecol. Sociohiol. 31, 97-106. Slagsvold. T. (lY86).Nest site settlement by the pied flycatcher: Does the female choose her mate for the quality of his house or himself? Ornis Scand. 17, 210-220. Smith. A . T . , and Peacock. M . M. (1990). Conspecific attraction and the determination of metapopulation colonization rates. Conserv. B i d . 4, 320-323. Smith. J . N . M.. and Arcese. P. (1989). How fit are floaters? Consequences of alternate territorial behaviors in a nonmigratory sparrow. A m . N o t . 133, 830-845. Smith. R. J . F. (1986). Evolution ofalarm signals: Role ofbenefits ofretaininggroup members or territorial neighbors. Am. N u t . 128, 604-609. Smith, S . M. (1988). Extra-pair copulations in black-capped chickadees: The role of the female. Brkuuiotrr 107. 15-23, Smith. T. M . , and Shugart. H. H . (19871. Territory size variation in the ovenbird: The role of habitat structure. Eculogy 68, 695-704. Stamps. J . A . (1976). Egg retention, rainfall and egg laying in a tropical lizard Anolis ameirs. Copeiu, pp. 759-764. Stamps, J. A. (1978). A field study of the ontogeny of social behavior in the lizard Anolis aeneus. Behnuioirr 66, 1-31, Stamps, J. A. (1983a). Territoriality and the defense of predator-refuges in juvenile lizards. Anini. Behau. 31. 857-870. Stamps. J . A. (l983b). The relationship between ontogenetic habitat shifts, competition and predator avoidance in a juvenile lizard (AnolIs uenetrs ). Behau. E d . Sociohiol. 12, 19-33.

TERRITORIAL BEHAVIOR

23 1

Stamps, J. A. (1983~).Sexual selection. In “Lizard Ecology: Studies on a Model Organism” (R. B. Huey, E. R. Pianka, and T. W. Schoener, eds.), pp. 169-204. Harvard Univ. Press, Cambridge, MA. Stamps, J. A. (1984a). Rank-dependent compromises between growth and predator protection in lizard dominance hierarchies. Anim. Behuu. 32, 1101-I 107. Stamps, J. A. (1984b). Growth costs of territorial overlap: Experiments with juvenile lizards (Anolis aeneus). Behau. Ecol. Sociobiol. 15, 115-1 19. Stamps, J. A . (1987a). Conspecifics as cues to territory quality: A preference for previously used territories by juvenile lizards (Anolis aeneus). Am. Nut. 129,629-642. Stamps, J. A. (1987b). The effect of familiarity with a neighborhood on territory acquisition. Behuu. Ecol. Sociobiol. 21, 273-277. Stamps, J. A. (1988a). Conspecific attraction and territorial aggregation: A field experiment. A m . Nat. 131, 329-347. Stamps, J. A. (1988b). The effect of body size on habitat and territory choice in juvenile lizards. Herpetologica 44,369-376. Stamps, J. A. (1990). The effect of contender pressure on territory size and overlap in seasonally territorial species. Am. Nar. 135, 614-632. Stamps, J. A. (1991). The effects of conspecifics on habitat selection in territorial species. Behav. Ecol. Sociohiol. 28, 29-36. Stamps, J. A. (1992). Simultaneous versus sequential settlement in territorial species. A m . Nut. 139, 1070-1088. Stamps, J. A., and Buechner, M. (1985). The territorial defense hypothesis and the ecology of insular vertebrates. Q. Rev. Biol. 60, 155-181. Stamps, J. A . , and Eason, P. K. (1989). Relationships between spacing behavior and growth rates: A field study of lizard feeding territories. Behau. Ecol. Sociohiol. 25, 99-107. Stamps, .I. A., and Krishan, V. V. (1990). The effect of settlement tactics on territory sizes. Am. Nat. 135, 527-546. Stamps, J. A., and Tanaka, S. K. (1981a). The relationship between food and social behavior in juvenile lizards (Anolis aeneus). Copeiu, 422-434. Stamps, J. A., and Tanaka, S. K. (1981b). The influence of food and water on growth rates in a tropical lizard (Anolis aeneus). Ecology 62, 33-40. Stamps, J. A., and Tollestrup, K. (1984). Prospective resource defense in a territorial species. Am. Nut. U3, 99- 114. Stamps, J. A., Tanaka, S. K., and Krishnan, V. V. (1981). The relationship between selectivity and food abundance in a juvenile lizard. Ecology 64, 1079-1092. Stamps, J. A., Buechner, M., and Krishnan, V. V. (1987). The effects of habitat geometry on territorial defense costs: Intruder pressure in bounded habitats. A m . Zoo/. 27,307325. Stefanski, R. A. (1967). Utilization of the breeding territory in the black-capped chickadee. Condor 69, 259-267. Stephens, D. W., and Krebs, J. R. (1986). “Foraging Theory.” Princeton Univ. Press, Princeton, NJ. Stutchbury, B. J. (1991). Floater behavior and territory acquisition in male purple martins. Anim. Behau. 42,435-443. Sutherland, W. J., and Parker, G. A. (1992). The relationship between continuous input and interference models of ideal free distributions with unequal competitors. Anim. Behau. 44, 345-356. Svardson, G. (1949). Competition and habitat selection in birds. Oikos 1, 157-174. Tavistock, M. (1931). The food-shortage theory. Ibis 13, 351-354. Taylor, R. J. (1988). Territory size and location in animals with refuges: Influence of predation risk. Euol. Ecol. 2, 95-101.

232

JUDY STAMPS

Tejedo, M. (1993). Do male natterjack toads join larger breeding choruses to increase mating success'? Copeia, pp. 75-80. Tinbergen, N. (1952). On the significance of territory in the herring gull. Ibis 94, 158-159. Townshend. D. J . (1985). Decisions for a lifetime: Establishment of spatial defense and movement patterns by juvenile grey plovers. J . Anim. Ecol. 54, 267-274. Trivers. R. L. (1985). "Social Evolution." Benjaminlcummings, Menlo Park, CA. Ursin. E. (1979). Principles of growth in fishes. S y m p . Zoo/. Soc. London 44, 63-87. Valladares, G . , and Lawton. J . H. (1991). Host-plant selection in the holly leaf-miner: Does mother know best? J . Anim. Ecol. 60,227-240. van den Assem, J. (1967). Territory in the three-spined stickleback (Gasterosteusaculeatus). Behauiour. Suppl. 16, 1-164. van Buskirk. J . (1986). Establishment and organization of territories in the dragonfly Sympetrim rirbiciindulum (Odonata: Libellulidae). Anim. Behau. 34, 1781-1790. van Rhijn, J. G., and Vodegel, R. (1980). Being honest about one's intentions: An evolutionary stable strategy for animal contests. J . Tlieor. B i d . 85, 623-641. Verner. J. (1992). Data needs for conservation biology-Have we avoided critical research. Condor 94, 301-303. Vines. G. (1981). A socio-ecology study of magpies Pica pica. Ibis 123, 190-202. Walters. J. R.. Copeyon, C. K., and Carter. J. H.. Ill (1992). Test of the ecological basis of cooperative breeding in red-cockaded woodpeckers. Auk 109, 90-97. Walton. R . . and Nolan. V.. Jr. (1986). Imperfect information and the persistence of pretenders: Male prairie warblers contesting for territory. Am. Nut. 128, 427-432. Waser. P. M.. and Wiley. R. H. (1979). Mechanisms and evolution of spacing in animals. In "Handbook of Behavioral Neurobiology" (P. Marler and J. G. Vandenbergh, eds.). Vol. 3. pp. 159-223. Plenum. New York. Wesdowski. T.. Tomidoje. L., and Stawarczyk, T. (1987). Why low numbers of Purus mujor in Bidowieta Forest-Removal experiments. Acta Ornithol. 23, 303-316. Wiens. J . A.. Addicott. J. F., Case, T. J., and Diamond, J. (1986).Overview: The importance of spatiai and temporal scale in ecological investigations. I n "Community Ecology" ( J . Diamond and T. J . Case. eds.), pp. 154-172. Harper & Row, New York. Wiley. R. H.. and Wiley. M. S. (1980). Spacing and timing of nesting ecology of a tropical blackbird: Comparison of populations in different environments. E d . Monogr. 50, 153-178. Williams. G . C.. and Nesse. R. M. (1991). The dawn of Darwinian medicine. Q . Reu. B i d . 66, 1-22. Wingfield, J. C.. Hegner. R. E., Duffy, A. M., and Ball, G. F. (1990). The "challenge hvpothesis": Theoretical implications for patterns of testosterone secretion, mating systems and breeding strategies. Am. Nut. 136, 829-846. Yasukawa. K . (1990). Does the "teer" vocalization deter prospecting female red-winged blackbirds'? Belzau. Ecol. Soc.iobiol. 26, 421-426. Yasukawa, K., and Searcy. W. A. (1985). Song repertoires and density assessment in redwinged blackbirds: Further tests of the Beau Geste hypothesis. Behau. Ecol. Sociohiol. 16, 171-175. Ydenberg, R. C.. Giraldeau, L. A.. and Falls. J. B. (1988). Neighbors. strangers and the asymmetric war of attrition. Anim. Beliau. 36, 343-347.

.4DVANCES IN THE STUDY OF BEHAVIOR, VOL. 23

Communication Behavior and Sensory Mechanisms in Weakly Electric Fishes BERNDKRAMER ZOOLOGISCHES INSTITUT DER UNIVERSITAT D-93040 REGENSBURG, GERMANY

I. INTRODUCTION Animals are the most complex organisms we know. Most of them show high mobility, sensitivity for the most diverse stimuli, and the ability to process information and learn. Animals are also capable of using resources that are unpredictable or difficult to find or obtain; they can camouflage or disguise themselves; and they can attack or flee. Such a mode of life disperses the individuals of a population. Sophisticated orientation and communication mechanisms have evolved in many species, enabling individuals to find resources or mates even at exceedingly low densities. It is only in recent years that the power of these orientation and communication mechanisms has begun to be unraveled. Among the most famous examples are the dance language and orientation of bees (von Frisch, 1967); the homing of pigeons over hundreds of kilometers in unfamiliar territory (Keeton, 1979); the often intercontinental migrations of birds from their summer to their winter quarters, and vice versa (Alerstam, 1990); the marine migrations of salmon for several years, followed by their homing to the small streams they were born in by olfactory cues (Hasler and Scholz, 1983); the barn owl’s precise localization and flash capture of a rustling mouse in total darkness (Payne, 1962; Konishi, 1993); and the ability of bats to detect and catch flying nocturnal insects by echolocation (Griffin, 1958; reviews by Neuweiler, 1984, 1993; Suga, 1990).

11. WEAKLYELECTRIC FISHES

Less well known are electric fishes, although strongly electric fish (the electric eel, the electric ray, and the electric catfish) played an important 233

Copyright D 1994 by Academic Precs, Inc All nghts of reproduction In dny form reserved

234

BERND KKAMER

role i n the advent of both the physical study of electricity and the science of neurobiology in the eighteenth and nineteenth centuries (e.g., Wu, 1984).Some of the tr~etrkljelectric fishes that are considered in this chapter were known by a few eighteenth-century biologists like Carl von Linne; however. it was only in 1951 that H. W. Lissmann of the University of Cambridge discovered these fishes’ regular, albeit weak, electric organ discharges (EODs), aided by electronic measuring devices. Teleost weakly electric fishes comprise the African elephant fishes (Mormyriformes: approximately 200 species) and the South American knifefishes (Gymnotiformes; perhaps 70 species), which are only distantly related. [A recent addition to this list are three “squeakers,” members of the exclusively African catfish family Mochokidae, whose behavior is so little known that they are not dealt with in this review (Hagedorn et al., 19901.1 Weakly electric fishes are both electrogenic and electroreceptive. Their electric system consists of a motor part, the usually myogenic electric organ. which is controlled by a pacemaker nucleus in the hindbrain; and a sensory part, the cutaneous electroreceptors, the afferences of which are connected to huge, specialized brain areas [for the motor part, see reviews by Bennett (1971a): for the sensory part: Bennett (1971b), Szabo t 1974). and Szabo and Fessard (1974): the more recent developments are traced by several reviews in the volume edited by Bullock and Heiligenberg (1986) and more briefly in Kramer (1990b)l. The electric system of both groups of nocturnal fishes is adapted to two functions: active, EOD-dependent electrolocation (Lissmann and Machin, 1958; review by Bastian, 1986) and communication (reviews by Kramer, 1990a.b). Although their electric systems are superficially similar (Finger ef al., 1986). great differences between African (Mormyriformes) and South American weakly electric fishes (Gymnotiformes) are found on a behavioral level. Even within these two groups the diversity is great; therefore. no single species can serve as a “model” electric fish. For a consideration of evolutionary questions. see Finger ef al. (1986) and various chapters in Bullock and Heiligenberg (19861, or Kramer (1990b), and the papers cited therein.

111.

PULSEA N D WAVEFISHES

Weakiy electric fishes discharge their electric organs in a pulse- or in a wakelike fashion (”buzzers” or “hummers,” named after the sound of amplified EODs when fed into a loudspeaker; Fig. 1). Whether a species is a buzzer or a hummer does not appear to be correlated with ecology

COMMUNICATION BEHAVIOR IN WEAKLY ELECTRIC FISHES

Time (ms)

235

F r e q u e n c y (kHz1

FIG. I . Pulse (top) and wave (bottom) discharges of two species of weakly electric fishes. A pulse discharge is of short duration (left) and has a broad amplitude spectrum (right);

made audible it resembles a click. A wave discharge consists of pulses repeated at an extremely stable frequency; its amplitude spectrum therefore consists of spectral lines, representing the fundamental frequency and its overtones (harmonics). Fed into a loudspeaker, a wave discharge generates a humming sound. Top: Gnathonernus petersii (Mormyridae); bottom: Eigenrnannia lineata (Sternopygidae, Gymnotiformes.) From Kramer (1990b).

nor with an adaptation to a special mode of life, but is strongly linked to phylogeny. There are representatives of each discharge type on both of the continents in which they are found, Africa and South America. They are therefore remarkable examples of convergent evolution. Discharging the electric organ in a pulse- or in a wavelike fashion has consequences for both the forms the signaling can take and the processing of electrosensory information. In most species, an EOD pulse is of short duration and constant waveform, often shorter than a nerve action potential, and of relatively low repetition rate. Therefore, several pulse fishes may all discharge simultaneously, with relatively little risk of temporal coincidences of their discharges (time sharing). On the contrary, a wave EOD is “on” all the time, and the EOD of an individual will be superimposed on that of another if sufficiently close, so that neither can receive its own EOD in “pure” form; therefore, the individuals of a community of wave fishes rarely discharge at the same frequency (frequency sharing). We may add that in pulse fishes (especially in mormyrids) the discharge rate tends to be low and unstable, whereas the wave fishes’ discharge frequency may be very high (depending on the species, up to 1800 Hz),

236

BERND KRAMER

but certainly very stable, probably more so than any other biological signal source. While in pulse fishes the EOD amplitude is usually high (in a few species so high that the discharge is felt by the human hand touching a moist fish, sometimes even causing discomfort), the EOD amplitude of wave fishes is generally much weaker. The amplitude spectrum of a single pulse EOD shows a continuous distribution of frequencies, rising from dc to a broad peak region before ieveiing off at still higher frequencies (Fig. 1). An amplitude spectrum of a wave discharge, on the contrary, shows energy only at specific points, the fundamental frequency and its harmonics or overtones (which are integer multiples of the fundamental), with no energy in between (nor at dc). Therefore, a pulse EOD’s artificial acoustic representation is broadband, sounding click-like to the human ear, whereas that of a wave EOD is “harmonic” like the sound of a flute or similar musical instrument, with a characteristic timbre depending on the number and relative intensity of overtones. Both pulse and wave fishes have been shown to be exceedingly sensitive to the fine detail of their discharge, which usually varies among the individuals of a population. Some forms and mechanisms of communication will be discussed for both the African and the South American weakly electric fishes.

Iv. THEINTERDISCHARGE INTERVAL CODE

IN THE

MORMYRIDAE

All elephant fishes (Mormyridae) tested have been found to discharge their electric organs in a pulselike fashion; their monospecific relative, Gymnarchits niloriclrs, is the only known African wave fish. So little is known about its communication behavior that it is not dealt with here (but see the review by Kramer, 1990b. and the papers cited therein). A pulse fish’s EOD activity has two aspects: the waveform of its EOD and the sequence of interdischarge intervals. Although the EOD waveform is fixed for a certain individual (there are exceptions, see Section VI), the pattern of interdischarge intervals is highly variable from moment to moment, and thus can encode the slightest change of state of excitement, and also specific “messages” addressed to conspecifics dwing social interactions (Fig. 2). Therefore, mormyrids have an interdischarge time intervui (IDI) code of communication. It is one thing to encode a message and another to decode it. On a very technical level we may ask, for example: Are mormyrids able to “measure” the variation of inter-EOD time intervals, as generated by conspecifics in various ethological contexts? The only species tested,

COMMUNICATION BEHAVIOR IN WEAKLY ELECTRIC FISHES

237

o.isec

Gnathonernus petersii FIG. 2. The electric organ discharge of a mormynd has a species-characteristicwaveform (top) that is extremely stable for an individual.The sequence of pulse intervals is, however, variable and reflects the state of excitement of a fish in a species-characteristic manner (Gnathonernus petersir). Note difference in time bars by a factor of 1O00. From Kramer (1985b).

Pollimyrus isidori,is an “expert” in the precise measurement of interpulse time intervals: trained, food-rewarded individuals discriminate a change of as little as 2% in a train of pulses spaced by 50 ms, and 3% at 100 ms (Kramer and Heinrich, 1990). This discrimination performance was achieved with either the rewarded or the unrewarded pulse train presented one at a time (well separated from the next presentation by random intervals varying between 30 s and 3 min); that is, without direct comparison of stimuli, in a similar way to the capacity of certain humans who are endowed with “absolute pitch.” An ID1 code of communication was already suggested by one of the first experimental studies using pairs of resting mormyrids (Moller and Bauer, 1973; see also Moller et al., 1989). Since then, the ID1 code has been demonstrated for several behavioral functions, including aggressive signaling (Bauer, 1972; Kramer, 1974, 1979; Bell et al., 1974; Kramer and Bauer, 1976; Bratton and Kramer, 1989); “threat” signals from fish attacked by an aggressive conspecific, shown also during escape behavior (Kramer, 1976); group cohesion (Moller, 1976; Serrier and Moller, 1981; Moller et al., 1982; Moller and Serrier, 1986; Graff, 1986); the specific

238

BERND KRAMER

EOD latency response, a form of entrained electrical signaling (Bauer and Kramer. 1974; Kramer, 1974; Russell et nl., 1974): species recognition (Kramer and Lucker, 1990); and electrical signaling during courtship and spawning behavior (Bratton and Kramer, 1989; Crawford, 1991). Because of its presence in all the species used in these studies, and almost all other mormyrids tested, it is likely that an ID1 code of communication, in a species-characteristic form. is present in all members of the Morniyridae (i.e.. it is a symplesiomorphy or shared trait in all members of a taxon in the sense of Hennig, 1966). As an example, the electrical signaling in the courtship and spawning of the small mormyrid P. isidori will be described. It became possible to study the reproductive behavior of mormyrids in captivity after Birkholz’s (1969) and Kirschbaum’s (1975, 1987) breeding successes in aquaria. Electrical communication in mormyrids has also been reviewed by Moller ( 1980a), Hopkins (1986), and Kramer (199C)b).

V.

ELECTRICAL SIGNALING I N THE COURTSHIP A N D SPAWNING OF A MORMYRID FISH

During its diurnal resting behavior at the bottom of a shallow West African river or stream, in the shelter of rocky crevices or tree roots, the mean discharge rate of a P. isidori is low (< 10 pulses per second, or pps) but varies from moment to moment. Every 2-3 s there is a brief, sharp EOD rate acceleration. Attacks on conspecifics. which are more common during the night, are also accompanied by sharp EOD rate accelerations but are followed by a high discharge rate display at or beyond 100 pps. These electrical displays are an obligatory component of the motor pattern “attack” and have signal value (Kramer, 1978, 1979; Bratton and Kramer, 1989). Totally different is the signaling of a female with a ripe ovary (mormyrids have only one gonad on their left side). After dark she signals her readiness to spawn by ticking away regularly like a clock, at a very low discharge rate of only 6-8 pps. In the life of a P. isidori, a low rate of EODs generated at constant intervals occurs only during the time of reproduction, and only at night when courtship or spawning will occur. Although not yet observed in the wild, but very likely from aquarium observations, the female appears to wander about to choose among the territorial males that try to attract her attention with their long-range advertisement calls. These songs, which only the males produce, take on the form of deep grunts, growls, and moans (Crawford et al., 1986) and are especially intense and of long duration on contact with a female emit-

239

COMMUNICATION BEHAVIOR IN WEAKLY ELECTRIC FISHES

a d 300

YET-TP-------------1 200

100

10

20

30

LO

50

60

70

80

90

100

4

4

Spawning

Spawning

b v -FW 0

r

r

10

20

30

LO

50

60

70

80

90

I s 1 100

FIG. 3. Electric communication in a mating pair of Pollimyrus isidori during their nocturnal spawning (a, male; b, female). The ordinates are the interdischarge time intervals (ms); neighboring points are connected by lines to show a trend. The abscissas are time (s). The record begins with the female just returning (r) to her waiting site (FW, female wait phase). During this time, the male performs housekeeping activities (ET, egg transport; TP, temtory patrolling) while displaying a “high sporadic rate sequence” of EOD activity, including many bursts of high discharge rate. As soon as the female arrives at the spawning site (FS, female spawning site wait for male) the male switches to the “medium uniform rate” (MUR) of much lower rate, as constantly displayed by the female as her “readiness-to-spawn” signal. During close contact, especially vent-to-vent coupling (VV) and oviposition (OP), short discharge breaks (DBR) or longer discharge arrests (DAR) are observed. After the female quiver (QR), she returns to her home region, followed by the male’s switching back to his MUR discharge pattern. From Bratton and Kramer (1989).

ting her electrical readiness-to-spawn display (Bratton and Kramer, 1989; Crawford, 1991). As in many other fishes, at this point spawning is still no more than a possibility. Before spawning occurs, there will be much more singing by the male, which he does almost continuously, while also attacking the female severely, as if to drive her away. This behavior may seem odd at first but is probably necessary because of the females’ eagerness to eat any eggs or larvae that a male may be guarding. (A male normally guards the fry from several spawnings and is extremely aggressive toward any conspecific approaching its nest, including females ready to spawn). Perhaps the long courtship phase (of about 3 hr on a spawning night), in addition to allowing time for the synchronization of the reproductive physi-

240

BERND KRAMER

ology of the mates, also serves to test the “intentions” of a female-will she be an egg producer or an egg eater? Spawning becomes much more likely when the engagement phase begins (about 1 hr after dark). In spite of still being severely attacked, the female rapidly descends in the water column and stops swimming on the bottom close to the nest. Instead of dealing the female a deadly blow, as one might expect from his previous aggressive behavior, the male immediately switches from the rather high and variable discharge rate he usually displays when guarding the eggs (ET-TP in Fig. 3) to the same EOD pattern displayed by a female who is ready to spawn (see Fig. 3, bottom): a very \ow EOD rate at almost constant inter-EOD intervals (6-8 pps). He tries to back up from behind and to position himself alongside the female, who quickly turns through 180”,and this leads to rapid head-to-tail circling of both fish for a few seconds (Fig. 4), ended by the female’s hasty escape to her hiding place high up in the water column where she is relatively safe. The male who had stopped singing immediately switches back to his high and variable EOD rate (Fig. 3, at times designated by “r” in the female record) and directs dangerous “courtship attacks” at the female while intensely singing. The female repeats her brief visits to the male’s territory once or twice a minute despite the male’s aggression, whose singing rapidly wanes to almost nothing. The female’s shyness also wanes and she allows the male to position himself along her side. The male pitches head downward and rolls on his side 90”while engaging the female’s anal fin with his own. While so coupled tightly vent-to-vent, the female is pushed upward, and both fish perform a complete, slow somersault rotation. Only then do the fish disengage and the female rapidly swims away. After repeating this rather elaborate exercise for about 2 hr, usually at a rate of once or twice per minute, spawning may begin (Fig. 4G). The spawning behavior resembles the courtship behavior in every aspect, including that of electrical communication, except that the “somersault rotation” is skipped. On the female’s arrival at the spawning site the male lines up in parallel and quickly rotates to one side, stimulating the female’s anal fin region with his own in a quivering action. After she releases a few eggs the female disappears while the male fertilizes the eggs (his sperm is immobile because the spermatozoa lack flagella). The male transports the eggs in his mouth to the nest, where he sticks them into plant material (Java moss in the laboratory setting). As she did during the preceding courtship period, the female continues to visit the spawning site once or twice per minute until all eggs are shed (usually up to 200 in one night). This can take 4 hr. Only then does the female stop visiting the spawning site, whereupon she also changes her EOD activity:

COMMUNICATION BEHAVIOR IN WEAKLY ELECTRIC FISHES

24 1

FIG. 4. The nocturnal courtship and spawning behavior in Pollimyrus isidori, redrawn from infrared video recordings. See text; H, the nest with eggs; A, head-to-tail circling; B, male arriving alongside of the stationary female; C-D, vent-to-vent coupling; E-F, rotation; G , oviposition. From Bratton and Kramer (1989).

she begins to regularly alternate between a high (about 70 pps) and a low EOD rate (about 5 pps), at two changes per second. The male, who did not sing for most of the courtship and the whole spawning period, starts to sing again intensely. He constantly patrols his territory, searching for any stray eggs that he picks up and puts into his nest. Why does the male sing at all? He could probably just as well signal his state of being a breeding male with a territory and a nest by using a purely electrical display. There may be two reasons why sound is involved: (1) It appears that the useful range of the male’s song is greater than that of his weak electrical discharge so that the probability of encountering a suitable mate is enhanced by the use of sound. (2) EODs may not be so well suited to inform females about the “quality” of a mate, compared to a courtship song that occurs only in males and probably is more costly (in terms of both energy expenditure and predation risk). The frequency of the song (especially the moan part), its intensity, and its duration may give females a better indication about a male’s age, reproductive state, health, and ability to successfully defend and care for the brood than any electric organ discharge display, since EODs are also part of ordinary life

242

BERND KRAMER

and therefore subject to many other selection pressures (e.g., arising from their electrolocation and social cohesion functions). It seems to be advantageous for the male to stop singing at the earliest moment (which he actually does after establishing a more than passing contact with a female ready to spawn), because of the acoustically competent predators that abound in tropical fresh waters. After all, nocturnal electrocommunication in mormyrids probably has evolved in response to pressure from visual predators (like the tigerfish, Hydrocynus forskalii, Characiformes); there is no point in exchanging these for acoustic ones (e.g.. Clarias catfishes).

VI.

INDIVIDUAL DISCRIMINATION I N A MORMYRID FISH

Does a mating pair of elephant fishes know each other individually? In the case of P . isidori we cannot help thinking that they must, otherwise the hundreds of short separations occurring during a spawning night (see previous section) would greatly increase the risk of strange females successfully stealing eggs or fry from the nest of an unaware male. [Egg eating may increase a female’s individual fitness considerably; see the review by FitzGerald ( 1992).] A damselfish male recognizes its territorial neighbors individually by their vocalizations (Myrberg and Riggio, 1983, probably an example of the “dear enemy effect” (Fisher, 1954; discussjon in McGregor, 1991). It is not yet known whether territorial P . isidori males also recognize each other by the small individual variations of the acoustic properties of their songs. Although it has been said that some mormyrids show individuality by their “personal fingerprint” ID1 patterns (Bauer, 1974; D. Malcolm, cited in Moller, 1980b: not particularly supported by Teyssedre ef ai., 1987), to date there is little compelling evidence that this actually plays a role in communication, given the volatility of these patterns (see Kramer, 1990b). In the case of P . isidori females that are ready to spawn, an individual-specific EOD pattern is even more difficult to imagine because of the simplicity and monotony of the pattern, leaving little room for mark:, of individuality (see previous section), By contrasl. the EOD waveform shows great intraspecific variability in P. isidori (Lucker and Kramer, 1981). and sexually dimorphic EOD waveforms as well a5 amplitude spectra, assumed to be the basis of mate recognition, were suggested by Westby and Kirschbaum (1982). However, these latter claims had to be substantially reduced (Bratton and Kramer, 1988), mainly because a method of sexing nonbreeding fish employed by Westby and Kirschbaum was found to be unreliable by Bratton and

COMMUNICATION BEHAVIOR IN WEAKLY ELECTRIC FISHES

243

Kramer in their population of the same species, and also because of an unnaturally high water conductivity. Mate recognition occurs in part by acoustic rather than electric signals in this species, as well as through cues from the ecological and behavioral context (Crawford et al., 1986; Bratton and Kramer, 1989; Crawford, 1991); see previous section. Instead of a clear-cut sexual dimorphism like, for example, the antlers of red deer stags or the peacock’s tail, what Bratton and Kramer (1988) found in the two sexes of their laboratory population of P . isidori (consisting of both breeding and nonbreeding fish kept in water of a natural, low conductivity of 100 pS/cm) was broad overlapping of the distributions for the EOD waveform parameter, PIIP2 amplitude ratio (Fig. 5A and Table I). [In natural habitats of P. isidori in the Cornoe and Bandama rivers in the Ivory Coast, water conductivity was 90-95 pS/cm during the beginning of the rainy season of late April and early May of 1990 and 1991 (B. Kramer, personal observation).] The means of the P1/P2 amplitude ratio are statistically significantly different for the two sexes (Bratton and Kramer, 1988),partially supporting and making more specific an initial similar claim by Westby and Kirschbaum (1982). In males the amplitude of the first head-positive peak is lower than the second, whereas females tend to have a relatively higher P1 phase (in males, the range of the ratio is from 0.04 to 0.94; in females, 0.37 to 3.33; i.e., the females overlapped two-thirds of the male range). Because of this overlap only a minority of individuals could be reliably sexed by using the EOD waveform; no statistically significant difference emerged in any of several other EOD waveform measures, including duration data and spectral amplitude properties (Bratton and Kramer, 1988). In a group of six females and eight males that were breeding in the laboratory in water of low conductivity ( 3 5 k S.E. 6 pS/cm), the amplitude of the P1 phase of their EOD showed less overlap compared to Bratton and Kramer’s fish (Fig. 5B), but still five or six individuals, that is, more than one-third, were in the overlap region (Crawford, 1992). Only when P1 amplitude was plotted against a second parameter, total EOD duration (itself not significantly different between the sexes), did the score combinations for female and male EODs not overlap. However, there is no evidence for the assumption that fish actually perform such a two-parameter analysis, and a two-parameter separation may arise from chance alone considering the small sample size of fish used. Although a larger number of individuals would have been desirable for this hypothesis, this result may indicate that in P . isidori that are breeding the EOD waveform could be used as a cue for mate recognition, but for some unknown reason it is not used (as shown in the next paragraph).

A

B

Duration (Dl) ( p s )

FIG.5. (A) Oscillograms of the electric organ discharges of two Pollirnyrus isidori, showing the great interindividual variability (at a natural 100 p S k m water conductivity). Individuals were selected to show a difference in EOD waveform for the two sexes: males tend to have a lower PllP2 amplitude ratio than females, but many individuals of the laboratory population had values lying in the overlap region, and hence could not be reliably sexed on the basis of the EOD waveform. Head-positivity is up; 2 MHz digitization. From Bratton and Kramer (1988). (9)Plot of all 14 individuals breeding in the laboratory ( 3 5 + 6 pS/cm), as represented by their scores on two EOD variables: the amplitude of the PI phase (as percentage of the total peak-to-peak amplitude of an EOD) and total EOD duration (each individual is represented by six points. i.e., measurements, taken on the same day). Note that the two sexes differ significantly in PI, although 5 individuals are in the overlap region. There is no significant difference for the duration of the EOD; however, in a two-parameter plot such as this one. the area for the males does not overlap with that for the females in these 14 fish. Reproduced with permission from Crawford (1992) and Company of Biologists Ltd.

245

COMMUNICATION BEHAVIOR IN WEAKLY ELECTRIC FISHES

TABLE I EOD WAVEFORM VARIABILITY AT 100 pSIcm I N Pollimyrus isidori (GIVENAS RANGESOR MEANS? STANDARD DEVIATION)“

Males ( N = 10) P1/P2 ratio

0.04-0.94 0.49 f 0.26 -19 13 23.1-37.8 28.6 f 5.3 P1-N separation (ps) 17.8-28.9 Mean Pl-N separationc 23.0 3.6 Pl-P2 separation (ps) 33.8-60.9 Mean Pl-P2 separationc 47.0 9.5 Peak amplitude frequency (kHz) 8.0-20.0 Mean peak ampl. frequency 13.4 k 4.0

Mean PUP2 ratiob Mean 100 (Pl-P2)/Nb N duration (ps) Mean N durationc

*

* *

(kHz)”

Females (N

=

14)

0.37-3.33 1.17 0.82 - 4 ? 13 17.3-31.6 25.8 ? 3.9 15.8-25.8 20.5 2.7 28.9-52.4 41.8 f 7.4 10.5-25.0 16.4 4.4

*

*

*

Mann-Whitney U 30.5 31.0 52.0 45.0 51.0 42.5

’For definition of P1, P2, and N, see Fig. SA. N-wave duration was measured as the time between zero-crossings. Peak amplitude frequencies were determined from amplitude spectra (as in Fig. 1, top). Note that there is a statistically significant difference of the mean P ratio between the sexes (second row). All other waveform parameters do not differ significantly between the sexes (Bratton and Kramer, 1988). Differences significant at p < 0.025 (Mann-Whitney U-test, two-tailed). Difference not significant ( p > 0.10).

Because females with male-typical EOD waveform spawned repeatedly without any problem (Bratton and Kramer, 1989), and as also shown by specifically designed experiments that used artificially generated male and female EOD waveforms that were interchanged in combination with different ID1 patterns (Fig. 6; Crawford, 1991), the EOD waveform cannot be a salient cue in the mate formation of P. isidori. Why then is there a difference in EOD waveform for the two sexes of P. isidori, even if only of a statistical (but significant) nature for the population at large, although-perhaps-nearly dimorphic for breeding individuals? Two possible reasons come to mind: sexual selection (perhaps in its incipient phase; reviews by Wilson, 1975; Maynard Smith, 1991), which could act through hormonal mechanisms, or the different waveforms could be hormonal “by-products.” Bratton and Kramer (1988)argue that the latter explanation is the more parsimonious one. Testosterone has been shown to have an anabolic, strengthening effect on the electric organ in several mormyrids and causes the EOD waveform to change when administered to females (Freedman etal., 1989; Bass, 1986; Landsman and Moller, 1988; Landsman et al., 1990).

246

sf

BERND KRAMER

g 125

- -

VSPl

I I

0'

__

Gizr-l

I

90

TIMEINSECONDS 6o

3O

z O L Or

-

-iGzF-7

3o

I

I

TIME IN SECONDS 6o

90

FIG. 6. Acoustic responses (R)of a resident male Pollirnyrtrs isidori to three different combinations of EOD playback via electrodes (S). Acoustic responses are shown as spectrographic charts with frequency (kHz) on the ordinate: grunts are the vertical spikes of broad frequency composition. The sequence of pulse intervals (SPl) is given below in each panel. with e;ich dot being an individual interdischarge interval (ms). Top panel: a female's readiness-to-spawn discharge pattern of low and constant rate evokes a high rate of male grunting. even when combined with a male EOD waveform (middle panel). However, when this male EOD waveform was combined with a discharge pattern recorded from a male that displayed a "high sporadic rate" (Bratton and Krarner, 1989). this evoked less than half the grunt responses in the two other panels. This shows that in order to evoke a high rate of grunting in a nesting. territory-defending male who is ready to court, the interdischarge interval pattern, and not the EOD waveform. of a female is important. Reproduccd with permission from Crawford (1991) and S. Karger AG. Basel.

COMMUNICATION BEHAVIOR IN WEAKLY ELECTRIC FISHES

247

Even if the difference in EOD waveform between males and females was due “only” to side effects of their hormonal physiology, the presumed result certainly had consequences in evolutionary terms: the enhanced degree of intraspecific variability has, apparently, made possible individual recognition of conspecifics by their EOD waveforms. As shown by Graff and Kramer (1989, 1992), trained, food-rewarded P . isidori discriminate artificial playbacks of their own species’ EODs recorded from different individuals, for example, those shown in Fig. 5A (and also more similar ones; similar results were obtained in Gnathonemus petersii). (For methods of signal generation, see Kramer and Weymann, 1987.) The ability to recognize its mate individually on a spawning night is of prime importance for a male’s reproductive success (see preceding section); a female’s EOD waveform might be the only cue available to a male for discriminating among different females. For territorial boundary disputes among neighboring males, however, and from a female’s point of view, the males’ songs might be an additional (or even better?) source of information allowing the discrimination between individual males. There are reports of sexual dimorphisms in EOD waveform for several other mormyrid species, sometimes combined with claims of mate recognition by EOD waveform, the ID1 pattern being irrelevant (e.g., review by Hopkins, 1988). In the present author’s opinion (e.g., Kramer, 1985a, 1990b),most of these reports suffer from unresolved questions of systematics, lack an intraspecific variability analysis, or offer only insufficiently controlled behavioral experiments because of difficult conditions in the wild. Also, the confounding effect of water conductivity on EOD waveform has only rarely been controlled for (Bratton and Kramer, 1988; Kramer and Kuhn, 1993). An interesting new finding in G . petersii, obtained on the day of their importation by a commercial dealer to New York, NY, is that of Landsman (1993). He observed EODs of statistically longer duration in males compared to females only in fish imported in June, but not May or October. May was the prerainy season, June the rainy breeding season, and October the postrainy season in the Nigerian origin of the fish in 1988. However, commercially imported fish coming from origins impossible to verify (except perhaps in a very gross manner), after a journey the duration and stress of which (Landsman et al., 1987; Landsman, 1991) cannot be assessed by a temperate zone customer, only allow limited conclusions concerning questions of intraspecific variability and sexual dimorphism. If these doubts may all be disregarded, then this result may indicate a seasonal EOD difference in the two sexes of a mormyrid. This could also explain the finding of no EOD sex difference in G . petersii by Kramer and Westby (1989, who studied a nonbreeding laboratory population.

248

BERND KRAMER

We now have in G. petersii nearly all possible suggestions: male EODs were found to be ( 1 ) of shorter duration than those of females (Landsman er a / . , 1987); (2) of longer duration compared to those of females for fish imported during the rainy season, although neither the rain nor the precise origin of the fish were confirmed by the personal presence of the scientist (Landsman, 1993); and (3) in spite of a considerable interindividual variability in both sexes, male EODs were not systematically different from female EODs (Kramer and Westby, 1985). This confusing situation should, perhaps, lead us to f a ) consider the question still open and (b) choose a different strategy of studying the question before publishing still another possibility. The groundwork necessary for establishing variability data for the EOD of each species will be laid only by studying breeding laboratory populations (the only example being P . isidori), by carefully controlled ethological experiments, and by systematically/taxonomically oriented field studies, such as those by Crawford and Hopkins (1989) or Moller and Brown 1990). In the first of these two studies, EODs slightly, but systematically, different from those of Morrnyrirs rume led to the discovery of a new sibling species. Mormyrus subundulatus, which has subsequently been defined also morphologically, The opposite case is given by the second study (Moller and Brown, 1990). In a collection of mormyrids clearly determined as Mormyrops curviceps on anatomical grounds, all caught at the same place in West Africa, two markedly different EODs were observed. An individual had either short EODs of about 0.5 ms duration or long EODs of about 1.6 ms duration and, most unusually, a reversed polarity of phases (the first main phase being head-negative); this was unrelated to sex or developmental stage. It is as yet unclear whether these totally different EOD waveforms betray the existence of a pair of sibling species that are morphologically so similar that they passed undetected until now, or two EOD morphs within a single species. Much work will be necessary to clarify the systematic status of these morphs and the behavioral significance of their markedly distinct EOD waveforms. A recently collected population of Marcusenius macrolepidotus from the Sabi river, Eastern Transvaal, South Africa, does show markedly distinct EOD durations between mature males (1.3-1.4 ms) and females (about 0.6 ms). Fish were collected by the scientists themselves (from the 22nd to 2Sth of September. 1993, that is, during the spring season shortly before raining), and studied for their EOD waveform immediately after capture in the original river water (Kramer and Skelton, in preparation). I feel this case is a safe example for sexual dimorphism of EOD waveform in a mormyrid.

COMMUNICATION BEHAVIOR IN WEAKLY ELECTRIC FISHES

249

VII. CONSTANCY OF THE MORMYRID EOD WAVEFORMI N A VARIABL~E ENVIRONMENT BY IMPEDANCE MATCHING In tropical African rivers and streams, water conductivity varies both geographically and seasonally, from a high of about 150 pWcm (or a resistivity of 7 kR . cm) to a low of about 5 pS/cm (or a resistivity of 200 kfl - cm, which is almost the equivalent of deionized water). Variations in conductivity may severely affect the EOD waveform as the following examples show. It has long been known that the second, head-negative phase of the discharge of G . petersii (Fig. 2 , left) decreases in amplitude and increases in duration when the resistance of the medium passes beyond a certain threshold, as is the case, for example, in deionized water (Harder et al., 1964;Bell er al., 1976),but also in tropical fresh waters of low conductivity. These biophysical observations confirmed the theory of the electrophysiology of the mormyrid electric organ, which states that the second phase of the discharge is electrically evoked by the current associated with the first, head-positive phase (Bennett, 1971a). However, a conductivity-dependent change of EOD waveform not only is of biophysical and electrophysiological interest but also has implications for electrocommunication. This was studied in P . isidori and PerrocephaZus bouei, the EOD waveforms of which were shown to depend strongly on water conductivity within the ecologically relevant range (Bratton and Kramer, 1988); Figs. 7B and 7E show this phenomenon in two other species. In the case of P . isidori with its great intraspecific EOD waveform variability, this dependence on an unstable ecological factor appeared to make the EOD waveform still more unreliable as an indicator of sex or of individual identity during communication. However, this conclusion of Bratton and Kramer (1988) was premature for the long term. It is true that we do not find in mormyrids the kind of “anatomical” impedance matching that we find in strongly electric fish like the marine electric ray, as compared with, for example, the freshwater electric eel. The eel’s organ has a low-current-high-voltage output and consists of a few columns that are very long (each composed of about 6000 electrocytes arranged in series). Thus the eel is very well adapted to generate dangerous shocks in a medium of very high resistance. By contrast, the ray’s electric organ discharges into a medium of very low resistance and consequently has a high-current-low-voltage output, generated by many short columns arranged in parallel. Therefore, the ray is also well adapted to its entirely different habitat and effectively immobilizes its prey by electroshocks. The electric organs of the weakly electric mormyrids, however, are unlike those of both fishes: four short columns, located

250

BERND KRAMER

*

3.5

u m

/-&

*.

0 20

60

T i m e !hl 100 I40

FIG. 7. Electric organ discharge waveforms of two mormyrids under high and low water conductivity. Left: Campylumurmyrus tamandua ( N = 7): right: C. rhynchopizonrs ( N = 3). Oscillograms were recorded at 2 MHz digitization rate. (A.D with *) Both species' EODs were recorded at 15OpS/cm conductivity before the experiment. When fish were transferred into water of only 10 pS/cm. the head-negative N phase was almost abolished (B,E). However. after about 2 days in the low-conductivity water the original waveform had almost recovered (A.D. no asterisks). (C.F) The time course of change for the PIN amplitude ratio for both species: after an initial steep rise associated with the transfer of the fish into water of low conductivity (at 0 hr) found in most fish, PIN ratios receded to values close to normal after about 2 days. The waveform differences as shown in A and D were permanent (>3 months): that is. the waveform recovery was incomplete. and conductivity had a graded, permanent effect on the EOD waveform at least up to 70 pS/cm. From Kramer and Kuhn (1993).

in the caudal peduncle of the tail fin, are composed of a rather low and constant number of electrocytes (depending on the species, about 100). In spite of the anatomically fixed situation, the electric organs of two mormyrid species (Cnmpylomormyrustamandua and Cnmpylomormyrirs rhvnchophorrrs) studied were clearly able to match their biophysical properties as a current and voltage source to the experimentally imposed impedance changes of the surrounding medium (Fig. 7; Kramer and Kuhn, 1993).The matching process required about 2 days when the water conductivity change was very strong. After adaptation to the new conductivity, the original waveform was largely but not quite restored.

COMMUNICATION BEHAVIOR IN WEAKLY ELECTRIC FISHES

25 1

The matching process itself is as yet unknown but might be triggered by a hormonally mediated, osmotic stress reaction (Mazeaud and Mazeaud, 1981; reviews in Rankin and Jensen, 1993), which could, in turn, initiate several cytological and biochemical responses in the electrocytes (e.g., the synthesis of membrane channel proteins; reviewed in Mills and Zakon, 1991). The synthesis of sodium channel proteins and their processing to the mature form requires 24 hr in the electric eel’s electrocytes (Thornhill and Levinson, 1987), which agrees well with the time course of the behaviorally observed impedance matching in the electric organ of Campylomormyrus. Because of the ability of the electric organ to adapt to water of a wide range of conductivities within about 2 days, we now believe that the potential information content of the EOD waveform as a mark of species or individual identity is not destroyed (although affected) by the ecologically caused variations of the physical properties of the communication channel (sensu Shannon and Weaver, 1949), that is, the water the fish live in. It has been said that this variation was unimportant because fish that interact are swimming in the same water (Crawford, 1992), dismissing the arguments of Bratton and Kramer (1988) as to the confounding effect of conductivity. The observation that fish usually swim in water certainly is correct, but who knows whether 5 min ago it was the same water? There may be extreme variations between confluent tropical streams and rivers, an exemplum maximum being the Solimoes (Amazon) with its 70 pS/cm white water and the Rio Negro with its 15 pSlcm black water at their confluence near Manaus, Amazon (B . Kramer, personal observation), which mix only after several kilometers. Recent evidence also shows that the other function of the electric system of mormyrids, active electrolocation, critically depends on the speciescharacteristic EOD waveform because the relevant electroreceptors, the mormyromasts with their anatomically and functionally specialized A and B receptor cells (Bell, 1990; Bell et al., 1989), are sensitive even to slight EOD waveform distortions (von der Emde and Bleckmann, 1992), such as occur during the active electrolocation of objects with a capacitive impedance component (von der Emde, 1990). It is not yet clear whether this sensitivity for EOD waveform changes, as locally induced by objects with a capacitive impedance component, is related to the remarkable ability of P . isidori to discriminate the individually variable EOD waveform of conspecifics (acting globally on a fish’s receptor population). Therefore, the recently discovered M . subundulatus (Crawford and Hopkins, 1989) might discriminate the EODs of members of its own species from those of its sympatric sibling species, M . rume, although their EOD waveforms differ only slightly.

252 VIII.

BERND KRAMER

ELECTRICAL SIGNALING

IN

GYMNOTIFORM PULSE

SPECIES

Weakly electric pulse fishes within the South American order Gymnotiformes comprise the families Hypopomidae, Rhamphichthyidae, and Gymnotidae [For a brief systematic overview, including treatment of the monospecific electric eel of the family Electrophoridae, which is both strongly and weakly electric, and not dealt with here, see the review by Kramer (1990b).] The discharge activity of these pulse fishes is in some ways similar to, but also markedly different from, that of mormyrids. It is similar in that most species (see the following for exceptions) increase their discharge rate when disturbed by, for example, a vibratory, an acoustic, or an electrical stimulus. In addition, in the few species investigated, an attack on a conspecific is usually, but not always, accompanied by an EOD display reminiscent of that shown by an aggressive mormyrid: a sharp increase of EOD rate followed by a slower decrease (SID; e.g., in Gymnotus carapo; Black-Cleworth, 1970; Westby, 1975a). During a decrease the discharge rate “dies away” from the peak rate (up to 250 pps in G. curapo) to the resting discharge rate level (around 40 pps), following a time course resembling an exponential decay (unlike mormyrids). The signaling in gymnotiform pulse species differs from that of mormyrids mainly in three ways. First, gymnotiform EOD pulses tend to be of longer duration (most species’ EODs are within 1-4.4 ms) compared to those of most mormyrids. Second, the SID display is only statistically associated with attack behavior and, in addition, is rather unspecific, accompanying various other behaviors or behavioral states as well, for example. in G. carapo, predatory attacks, being prodded by a stick, or being attacked by a conspecific. Mormyrids, however, broadcast specific EOD displays as an obligatory part of a variety of different behaviors or behavioral states. Third, whereas a mormyrid’s EOD activity very often displays a rhythm, that is, a pattern of intervals of different lengths in a characteristic sequence that accompanies specific behaviors (see Section IV), in gymnotiform pulse species there is no such pattern [except the tonic change during discharge rate increase or decrease, which may, however, be very rapid; e.g., in the “decrement burst” displayed by Hypopomus occidentalis, consisting of a few EODs of very high rate interspersed into the normal, on-going activity; Hagedorn (19SS)l. In gymnotiform pulse species. inter-EOD intervals are usually of almost equal duration, varying only statistically about the mean (except when involved in some dramatic activity like attack). Usually this variation is quite narrow; for example, during “quiet” discharge periods of resting G. carapo, the standard deviation may be almost as low as 1% of the mean of lo00 inter-EOD intervals.

COMMUNICATION BEHAVIOR IN WEAKLY ELECTRIC FISHES

253

Therefore, the term “buzzer” is much more appropriate for most gymnotiform pulse species, which also tend to discharge at higher rates, than for mormyrids (up to about 65 pps at rest during the day compared to less than 10 pps in mormyrids). By and large, EOD displays in gymnotiform pulse species like G. carapo seem to be limited to discharge rate increases, decreases, and brief stops or “breaks”; otherwise, the mean inter-EOD interval duration (or its reciprocal, the discharge rate) and its standard deviation characterize the EOD activity sufficiently. However, detailed EOD interaction maneuvers and sensitivity to the phase of a discharge cycle have also been reported (Westby, 1975b, 1979). The ethological significance is still unclear but a function relating to sensory physiology seems well founded. Some gymnotiform pulse species (within the genera ffypopygus,Steatogenys, Rhamphichthys, and Hypopomus) are unlike other pulse species because they do not respond to any ordinary form of stimulation by a discharge rate change. With extreme regularity and at rather high rates for pulse species, they drone on like a quartz clock; the standard deviation of 1000inter-EOD intervals is only 0.3% of the mean. Only specific electrical stimulation will make them slightly shift their rate. These fishes’ electrical behavior resembles that of gymnotiform wave species (see the following), except that they discharge in the form of pulses. Apart from their electrical “jamming avoidance” behavior to artificial stimulation (Gottschalk and Scheich, 1979; Scheich et al., 1977; Heiligenberg, 1974, 1977; Heiligenberg et al., 1978a), the signaling of these fishes during different behaviors is unstudied. Reproductive behavior in gymnotiform pulse species has not yet been studied in great detail either. Both Hypopomus occidentalis and Hypopomus pinnicaudatcw emit EODs resembling single-cycle sinusoids, with the male EODs being of longer duration compared to those of females (Hagedorn and Carr, 1985;Hopkins et a [ . , 1990).When female ff. occidentalis were stimulated with trains of single-cycle sinusoidal pulses of male duration they gave more discharge rate responses (“decrement bursts”; Hagedorn, 1988) compared to under stimulation with pulses of shorter (female) duration (Shumway and Zelick, 1988). These authors suggest that females discriminate male from female EODs (or artificial pulses of appropriate duration) by their difference in spectral amplitudes: male EODs tend to have their spectral amplitude peak at lower frequencies than do females (826.2k200.4 Hz versus 984.4297.6 Hz; Hagedorn and Carr, 1985). Another interesting finding is the observation that the EOD duration in male H . occidentalis is quite plastic: winning or losing a territorial contest in a small aquarium is sufficient to increase or reduce, respectively, a

754

BERND KRAMER

male’s EOD duration and amplitude within 2 days; a loser’s EOD hence becomes more femalelike (Hagedorn and Zelick, 1989). Individual recognition by EOD waveform has recently been shown in territorial G. carcipo. The individuals differ somewhat in the waveform of their EOD pulse, similar to the variability observed in the mormyrid P. isidori, which also recognizes conspecifics individually by their EOD (see Section VI). When the prerecorded EOD pulse of a G. carupo, designated as “neighbor” who was removed for an experiment, was played back from an experimental fish’s “incorrect” side, the fish attacked the dipole model used for playback significantly more often compared to a playback from the fish’s “correct” side, that is, that neighbor’s usual position with regard to the experimental fish (McGregor and Westby, 1992). The sensory mechanism of EOD waveform discrimination is still unknown, but the authors suggest a hypothetical scanning mechanism as presented in an earlier paper (Hopkins and Westby, 1986), although a simpler mechanism, such as the one proposed for H . occidenralis (see the foregoing), can also be imagined. It should be noted that in the mormyrid P. isidori, which also discriminates its conspecifics’ individual EOD waveforms. a scan sampling mechanism can be excluded because of an EOD rate that is inherently too variable and unrelated to the stimulus pulse rate (Graff and Kramer, 1989, 1992).

1X. ELECTRICAL SIGNALING I N GYMNOTIFORM WAVESPECIES A.

FREQUENCY MODULATIONSA N D FREQUENCY SENSITIVITY

Gymnotiform wave species comprise the families Sternopygidae (at least 11 species) and Apteronotidae (at least 25 species). The Sternopygidae with their myogenic electric organs usually discharge at lower frequencies than the Apteronotidae. which possess electric organs of neural origin (Bennett, 1971a; Bass. 1986). Although a few sternopygids discharge at frequencies far beyond those of any ordinary nerve or muscle (greater than 800 Hz), certain apternotids go still higher ( u p to 1800 Hz; Kramer, 1990b). The special adaptations making possible such high discharge frequencies are unstudied (except the observation that electric, i.e., fast. synapses abound in both the sensory and the motor pathways: Szabo, 1967: Waxman et al., 1972). Also the constancy of the discharge frequencies is unrivaled. In Apteronotits albifrons, the standard deviation of the mean EOD cycle, measured over a thousand cycles of about 1 ms duration, may be as low as 0.012% (or 0.14 ps) and was actually limited by the accuracy of the measurement (Bullock, 1970). Still lower values were observed in Eigenmannia (Kramer, 1987).

COMMUNICATION BEHAVIOR IN WEAKLY ELECTRIC FISHES

255

These fishes only rarely modulate their frequency during the day but may do so frequently at night when fighting or courting. Eigenmannia males, for example, give series of brief interruptions (“chirps”) at a rate between one per minute and five per second. A female will only spawn when a male has chirped at her site for at least one hour (Hagedorn and Heiligenberg, 1985). In response to mild attacks by the male she will raise her EOD frequency by a few to several hertz over a period of tens of seconds (“long rises”). Hopkins (1974) has studied these and other EOD frequency modulations. No detailed ethological study is yet available correlating discharge frequency displays with behavior, comparable to the studies in Gymnotus or the mormyrids. A frequency modulation that can be evoked by artificial stimulation is the so-called “jamming avoidance response” (JAR). This response was discovered by Watanabe and Takeda (1963) and given its present name by Bullock et al. (1972a,b). The neural mechanisms of perception and motor control are reviewed by Heiligenberg (1988,1991). The JAR usually is an EOD frequency shift away from the frequency of a stimulus signal (e.g., a sine wave), if above threshold. The absolute detection threshold of a signal and the threshold for the JAR are identical, provided the stimulus frequency is sufficiently close to that of the fish (within about +20 Hz difference; reviewed in Kramer and Kaunzinger, 1991). Traditionally the JAR has been seen as a fish’s attempt to escape from a condition in which its active electrolocation performance is impaired. However, as argued by Kramer (1987), the experimental evidence is not particularly strong, given that stimulus intensities capable of interfering with a fish’s electrolocation performance are unrealistically high (as determined by Heiligenberg, 1977).Another wave species from the same family that does not possess a JAR, Sternopygus, shows only an impairment of electrolocation at a stimulus strength 50 times that of its own near-field EOD intensity (Matsubara and Heiligenberg, 1978). This shows that electrolocation of an object, as sensed by a small, local population of electroreceptors, is not necessarily “jammed” by another fish’s EOD acting globally on all the electroreceptors of a fish. An observation that strongly suggests a function of the JAR in the context of social communication was the discovery of the sexual dimorphism of the JAR (Kramer, 1987). Adult males are almost unresponsive to jamming stimuli, whereas females only lower their discharge frequencies to stimuli of higher frequency than their own discharge and will not respond to stimuli that should evoke a frequency increase. Juveniles show a high degree of interindividual variability and may respond in both directions, although asymmetrically (i.e., stronger for an identical frequency difference of one sign compared to that of the opposite sign).

256

BERND KRAMER

’The hypothesis of a communication function of the JAR implies that, instead of trying to minimize the “jamming” or disrupting effect of another fish’s signal on its own EOD, a fish would be trying to maximize (or optimize) the modulating effect of that signal on its own EOD in order to improve its own resolution for analyzing other fishes’ signals. Information of interest for a fish displaying a JAR would be the frequencies of these signals and their waveforms, because both contain relevant information as to the age and sex of the sender (see next section). Also, by better distinguishing the various signals that a fish receives, it may obtain a better idea of the spatial structure of the group of which it forms a part. The sensory mechanism for EOD waveform analysis, as proposed by Kramer and Otto (1991), needs the JAR if the frequencies of two fish are too close (see next section). How good is a fish’s frequency and intensity discrimination? Trained Eigenmannia received a food reward when they detected an alternation in a sequence of sine wave bursts (repeated at 2 per second, with 150 ms silence between the bursts), either in frequency or in intensity. Close to a fish’s discharge frequency, and at a 30-dB sensation level, fish discriminated frequency differences as small as 0.52 Hz and intensity differences as small as 0.56 dB (Kramer and Kaunzinger, 1991). A frequency difference threshold of about 0.5 H z corresponds well to the minimum frequency difference (0.6 Hz) necessary to evoke only JARS of one sign and not the other about a fish’s “decision” point. which usually is close to 0 Hz difference (stimuli were not frequency-clamped to a fish’s EOD; Kramer, 1987). At stimulus frequencies higher or lower than a fish’s own discharge frequency, the frequency discrimination of Eigenmannia (and also intensity discrimination) declined steeply (Fig. 8. left). However, compared to other acoustico-lateral senses in lower vertebrates for which difference thresholds are known {i.e., sensitivity to water surface waves and in audition), the electrosensory frequency difference threshold of Eigenmannia is exceptionally high and in the range of the frequency difference thresholds for hearing in the most sensitive vertebrates with a cochlea (e.g., the human; Fig. 8, right). This is especially true for frequencies close to or above a fish‘s EOD frequency. The frequency modulations occurring in social behavior are all considerably above the detection limit of Eigenmannia (Hopkins, 1974; Hagedorn and Heiligenberg, 1985; see also Kramer, 1987). The superior electrosensory acuity of Eigenmannia is the result of an ingenious simultaneous comparison of a fish’s own signal with the stimulus, using the physics of the beating superimposition signal (for clear descriptions of the physics, see Scheich, 1977; Heiligenberg et ul., 1987b;

257

COMMUNICATION BEHAVIOR IN WEAKLY ELECTRIC FISHES

;jE

51

0.1 3

* Fish I

\ 100

1000

Stimulus frequency (Hz)

’

0.00ll

’

’

.,

I

.

-x- Eigenmannia

-+-Goldfish *Man

1

100

1000

Stimulus frequency (Hz)

FIG. 8. Electrosensory frequency discrimination (left) in the wave fish Eigenmannia, as compared with auditory frequency discrimination in the goldfish and the human (right). Left: Close to a fish’s discharge frequency, lowest frequency difference thresholds were found, steeply rising to both lower and higher frequencies. However, expressed as a fraction of the respective stimulus frequency (Weber-Fechner ratio, Af/J right), these thresholds did not rise much above a fish’s ( N = 3 ) discharge frequency. The electrosensory frequency discrimination of Eigenmannia is exceedingly acute as compared with the hearing difference thresholds of even one of the most sensitive mammals, the human; the goldfish is much less sensitive. For the human the most “flattering” data that could be found in the literature were used (Wier et a / . , 1977), although not shared by other authorities (e.g., Zwicker, 1982). The goldfish data are from Fay (1970), as also given in Fay (1988). From Kramer and Kaunzinger (1991).

for natural EOD beats, see Kramer and Otto, 1991). The fish’s electrosensory system is specialized to detect minute modulations of its own EOD by a stimulus signal in both amplitude and phase. In an especially clear way, this has recently been demonstrated in Sternopygus, which is much less sensitive to stimuli of exactly its own frequency, or integer multiples thereof, compared to stimulus frequencies only slightly different from one of these frequencies (Fleishman et al., 1992). This is because stimuli of exactly the frequency of one of the harmonics of the EOD (including its fundamental) do not beat against a fish’s EOD. Because Sternopygus does not possess a JAR, frequency identity of a stimulus wave with its EOD (or one of its higher harmonics) may be maintained for a short period of time without the need for frequency-clamping. A similar result was obtained in Eigenmannia using a frequency-clamped stimulus (Kaunzinger and Kramer, 1993). Unlike other vertebrates, Eigenmanniu and similar fish need not deal with stimuli as they occur, varying over many orders of magnitude in both

2.58

BERND KRAMER

frequency and intensity. but face a small range of variation only of the superimposition signal, which allows these fish to specialize in a high difference sensitivity.

B.

SENSIT~VITY FOR EOD WAVEFORMS

Eigenmanni~r has a sexually dimorphic EOD waveform: females and juveniles show a rather sinusoidal waveform with the positive half-waves being only slightly shorter than the negative ones, whereas adult males have much more asymmetric waveforms with longer negative half-waves (see Fig. 10).Fish discriminate these waveforms in playback experiments, with neither intensity nor frequency being factors (Kramer and Zupanc, 1986: Kramer and Otto, 1988; for method, see Kramer and Weymann, 1987). The natural EOD, in addition to varying in waveform, also varies in the composition of spectral amplitudes, that is, in its harmonic content (the harmonic content of a voice or a musical instrument determines its timbre). Male EODs have much stronger higher harmonics than the EODs of females o r juveniles (Kramer, 1985a); therefore, fish could in theory, and more conventionally, discriminate the sexually dimorphic EODs by their difference in harmonic content rather than in waveform. (For the human, an audio playback of the female EOD has a “dull” timbre, not unlike a flute, whereas male EODs have a more brilliant quality, similar to that of a violin, the sound of which is also rich in overtones.) Only recently has it become clear that Eigenrnannia is able to discriminate !artificially generated) signals of different waveforms that have identical amplitude spectra, that is, no difference in harmonic content. These signals, when made audible, are therefore indistinguishable for the human. To test for a “pure” waveform (time domain) sensitivity (as opposed to sensitivity for the harmonic content of a signal, which is in the frequency domain), phase differences between the harmonics of a signal were introduced. This changes a signal’s waveform but does not affect its amplitude spectrum (Fig. 9) and is not detected by the auditory system of the human. Trained Eigenmannin clearly detected a difference of 90” in phase relationship between the two harmonics of a pair of signals as shown in Fig. 9, and much smaller phase differences as well. The threshold phase shift is below 22”. causing a signal to change its waveform only slightly (Kramer and Teubl, in press). A mechanism for this new sensory capacity was proposed by Kramer and Otto (1991). The electroreceptors on a fish’s right and left body sides face opposite polarities of a distant stimulus source (Fig. 10). Because eiectroreceptors are polarity-sensitive, the stimulus signal is added to or subtracted from the fish’s own EOD, according

259

COMMUNICATION BEHAVIOR IN WEAKLY ELECTRIC FISHES

. 0

2

L

6

8

10

0

2

L

6

8

10

Time [ m s l

1

2

1

2

'

.

.

.

3

L

5

3

L

5

Frequency [kHz1

FIG. 9. Artificially generated signal waveforms (left) that have an identical amplitude spectrum (right). These electrical signals, played at intensities and frequencies typical for Eigenmanniu, are discriminated by trained, food-rewarded fish, although acoustical presentations are indistinguishable for the human. The signals are composed of only two sine waves, the fundamental frequency, f,,and its harmonic of two times that frequency, fi. Though the phase difference, relative to the amplitude peaks of both sine wave components, is 0" for the upper waveform, it is a maximum 90" ( d 2 ) for the lower waveform. From Kramer and Otto (1991).

to a receptor's location. Therefore, a fish's right and left populations of receptors receive different superimposition signals, with phase differences between the zero-crossings (Fig. 11). A sensory circuit has been found in a specialized part of the Eigenrnunniu midbrain (the torus semicircularis) performing a left-right (or front-rear) comparison of afferences from the T-electroreceptors, which preserve precise temporal information about such phase differences (Can, 1990). This comparison should enable a fish to reconstruct the form of a wave signal modulating its own EOD; the ability to do so, even in the absence of any spectral amplitude cues, has been proven (see earlier). A fundamental drawback in communication for all wave fishes, as compared with pulse fishes, is the impossibility of receiving their conspecifics' EOD waves in pure form (nor their own, for that matter). This drawback has elegantly been turned into an advantage. First, by scanning through another fish's EOD waveform by beat analysis, fish gain time: a fish may choose whichever duration of a beat cycle it finds convenient by changing its frequency (e.g., by displaying a JAR; see previous section). A JAR

260

BERND KRAMER

A

Ti m e [msl Frti. 10. Schematic representation of the polarity (left) of electric organ discharges of the wave type. as experienced by a fish's electroreceptors. and their afferent responses (Treceptors. right). (A) An Eigenrnunnia electric organ (bar) generates a dipole field that has the same polarity for its left and right side T-electroreceptors in the fish's skin, which therefore fire in synchrony (shown for a female EOD waveform: R, right; L, left side electroreceptors). ( 6 )However. for the EOD of a distant fish (as an example, a male EOD waveform i s chosen). the electroreceptors of the right and left body side fire in alternation. because they "see" the signal with opposite polarities. A male EOD would be represented by a short-long pattern of difference intervals for the T-afferences (D), whereas a female EOD would be represented by almost equal difference intervals. In reality, both fields (A and 6 ) are superimposed. but the basic idea still holds (see Fig. I I ) , and a neural circuit for the comparison of rightileft time disparities has been proposed (Carr. 1990).From Krarner and Otto (1991).

usualiy leads to beat cycles between 113 and 117 s, or about 100 times an EOD cycle (of about 2.5 ms). Second, for the reconstruction of a stimulus waveform modulating a fish's own discharge from its T-receptor afferences there has probably been a preexisting central neural circuit that only had to change its function within the acoustico-lateral senses. A circuit for the detection of right1

COMMUNICATION BEHAVIOR IN WEAKLY ELECTRIC FISHES

r

0

1

1

6

I

I

12

I

I

18

I

I

24

I

26 I

I

30

Imsl

FIG. 1 1 . An Eigenmunniu female’s own EOD (of 400 Hz and 100%amplitude)is superimposed by a conspecific’s EOD (of 450 Hz and 30% amplitude). Left: the conspecific is another female; right: the conspecific is a male. Because of the frequency difference of 50 Hz chosen in this electronically generated example, a full beat period is only 20 ms (centered in both cases). Top: the superimposition is either additive (solid line) or subtractive (dotted line), as experienced by the fish’s left and right side electroreceptors (see Fig. lo), depending on the fishes’ relative position. Note that in both cases (left and right graphs) small time disparities about the zero-crossings occur. Bottom: these zero-crossing time disparities are plotted on an expanded scale (as phase in microseconds on the ordinate). The time functions of these right-left side phase differences correspond to the waveforms superimposing a fish’s own signal (which serves as a “carrier” signal), that is, to a female EOD (left) and to a male EOD (right). The reconstruction of the waveform modulating a fish’s own EOD seems to occur centrally from T-receptor afferences (in the midbrain; Carr, 1990). From Kramer and Otto (1991).

left side time disparities of sound signals is probably present in all vertebrates and underlies their directional hearing. The human, and still more so the barn owl, detects time disparities with an exceedingly high precision in the microsecond range; in the barn owl’s brain the neural circuit has been described in great detail (see Section I). The other class of high-frequency tuberous electroreceptors, the amplitude-sensitive P-receptors, only increase or decrease their probability of firing with the amplitude of the beat envelope. Because no detailed phase relationship to the stimulus is known, it is difficult to imagine that

262

BERND KRAMER

P-receptors could provide the information that the brain needs to reconstruct the waveform of a stimulus modulating a fish’s EOD.

X. THE“So WHAT?” QUESTION All the foregoing research on (nearly) invisible creatures that live in some of the most remote habitats and communicate in a way that cannot be felt nor observed by the human (except in a very indirect way) may be very fine-but what is it good for? Different people may find different answers to this question. Without going into detail I will indicate a few points that I personally find of interest. Weakly electric fishes are a good example for demonstrating that in order to make progress in answering a specifically behavioral question, like that of electrocommunication or electrolocation, more general biological and evolutionary problems have to be solved first. However, without the initial behavioral work (e.g., that of Lissmann, 1958; Lissmann and Machin, 1958) we would not have seen the problem, or much less of it. Therefore, the continuous interplay of both is necessary for making progress. One of the most difficult problems for Darwin’s theory of evolution (Darwin, 1859), even in its modern form, has been the question of how transitional traits can evolve and exist when they do not convey a selective advantage to their bearer. Among many other examples, Darwin considered the problem of the evolution of strongly electric fishes (he did not know of weakly electric ones). Today we know that weak electric organs need not be useless; on the contrary, in terms of species numbers, weakly electric teleosts have been much more successful than their strongly electric cousins. A weak electric organ that forms part of a sophisticated intelligence system represents an adaptive peak totally different from a strong organ that functions as an electric “club” (although admittedly an “intelligent” one, simultaneously acting in a volume of water surrounding a fish). This may teach us to be careful before declaring any organ or structure “useless.” A new sensory-motor system specifically dedicated to the electric modality opens up a whole new world and represents an enormous challenge for the zoologist. It also offers a host of research opportunities. Two examples may illustrate this point. (1) The question of how sensory systems deal with the stimulation due to an animal’s own action, or reafference, has been studied especially successfully in electric fishes, more so than in any other system I know of (Bell and Szabo, 1986; Bell, 1986, 1989). (2) Vertebrate hair cells, like the sensory cells in our inner ear, are difficult to access and to study. Electric fishes have very similar cells, the

COMMUNICATION BEHAVIOR IN WEAKLY ELECTRIC FISHES

263

electroreceptors, exposed in their skin. A number of electrophysiological and behavioral studies have profited from this situation and shown an exquisite sensitivity to the fine detail of stimuli, phase sensitivity, etc., that could not have been imagined previously. Its role in behavior is only beginning to be studied. XI. SUMMARY Teleost freshwater fishes of the orders Mormyriformes (the elephantfishes plus Gymnurchus from Africa) and Gymnotiformes (the knifefishes from South America) are both electrogenic and electroreceptive. These fishes’ electric system has a motor part, the electric organ, and a sensory part, the cutaneous electroreceptors that project to large, specialized brain areas. The electric systems of both groups of fishes, although evolved independently, are adapted to the same two functions, nocturnal electrolocation and communication. Weakly electric fishes discharge their electric organs in a pulselike or in a wavelike fashion (“buzzers” and “hummers,” respectively). Whether a species is a hummer or a buzzer does not appear to be correlated with ecology but is strongly linked to phylogeny. There are representatives of both discharge types on both continents where these fishes are found. The elephantfishes (Mormyridae, about 200 species) are, apparently, all pulse fishes, whereas the related, monospecific Gymnurchus (Gymnarchidae) is the only known African wave fish. There are five families of South American knifefishes, with the majority of the 70 or so species being hummers, usually discharging at extremely constant frequencies (about 50-1800 Hz). The sensory mechanisms of social communication, as studied by behavioral means, are reviewed in this chapter with the question of mechanisms of reproductive isolation in mind. The chapter focuses on the electric organ discharge as the basic communication unit, and on the frequency, repetition rate, or temporal patterns of discharges. In both wave and pulse fishes the frequencies or repetition rates of discharges are not usually species-specific but are species-characteristic, because of more or less broad overlap between two or more species (depending on the local community of species). Electrosensory discrimination thresholds for frequency and intensity are unusually low in a wave fish, lower by far than those for other acoustico-lateral senses of aquatic lower vertebrates, rivaling the discrimination thresholds for audition in the most sensitive mammals (e.g., the human). A similar conclusion applies for the pulse rate sensitivity of a mormyrid. Species specificity becomes apparent when more information about the

264

BERND KRAMER

discharge activity is considered. In the case of pulse fishes, especially mormyrids, this comprises temporal patterns of discharges, which also vary greatly according to behavioral context (like aggression, escape, courtship, feeding, etc.); these fishes have an interdischarge interval code of communication. In the case of wave fishes, various types of frequency modulations and brief, repetitive discharge stops occur. Also, wave fishes may engage in “phase coupling” and “jamming avoidance,” maneuvers that involve precise interaction with another fish’s discharges. The degree of species specificity of a fish’s discharge activity is usually enhanced by features of the waveform of a single discharge; this is true in both pulse and wave fishes. Usually there is considerable intraspecific variability of discharge waveforms, and there are also examples of sexual dimorphism. At least a few species can discriminate the individually variable pulse or wave discharge waveforms of their species. In a wave fish, a sensory mechanism based on the temporal analysis of beat patterns can explain the observed results. This new sensory capacity detects the phase modulation within a beat, which always occurs when the wave discharges of two fish mix in the water. In pulse fishes, several hypothetical sensory mechanisms for the discrimination of intraspecific pulse waveforms have been proposed but it is not yet clear which is generally involved. In any case, the sensitivity of weakly electric fishes to the fine detail of their discharges shows that the electrosensory world is much more colorful than could be imagined until recently.

References Alerstam. T. (1990). “Bird Migration.” Cambridge Univ. Press, Cambridge. Bass, A. H. (1986). Electric organs revisited: Evolution of a vertebrate communication and orientation organ. In “Electroreception” (T. H. Bullock and W. Heiligenberg, eds.), pp. 13-70. Wiley, New York. Bastian. J. (1986). Electrolocation: Behavior, anatomy and physiology. In “Electroreception” (T. H. Bullock and W. Heiligenberg, eds.), pp. 577-612. Wiley, New York. Bauer, R. (1972). High electrical discharge frequency during aggressive behaviour in a mormyrid fish. Gnorhonemus petersii. Experientio 28, 669-670. Bauer. R . (1974). Electric organ discharge activity of resting and stimulated Gnathonemus petersii (Mormyridae). Behouiour 50, 306-323. Bauer. R. and Kramer, B. (1974). Agonistic behaviour in mormyrid fish: Latency relationship between the electric discharges of Gnathonernus petersii and Mormyrus rume. Experienria 30, 51-52. Bell. C. C. (1986). Electroreception in mormyrid fish. Central physiology. I n “Electroreception” (T. H. Bullock and W. Heiligenberg, eds.). pp. 423-452. Wiley, New York. Bell, C. C. (1989). Sensory coding and corollary discharge effects in mormyrid electric fish. J . EXP.B i d . 146, 229-253. Bell. C. C. (1990). Mormyromast electroreceptor organs and their afferent fibers in mormyrid

COMMUNICATION BEHAVIOR IN WEAKLY ELECTRIC FISHES

265

fish: 111. Physiological differences between two morphological types of fibers. J. Neurophysiol. 63, 319-332. Bell, C. C., and Szabo, T. (1986). Electroreception in mormyrid fish. Central anatomy. I n “Electroreception” (T. H. Bullock and W. Heiligenberg, eds.), pp. 375-421. Wiley, New York. Bell, C . C., Myers, J. P., and Russell, C. J. (1974). Electric organ discharge patterns dunng dominance related behavior displays in Gnathonemus petersii. J. Comp. Physiol. A 92,201-228. Bell, C. C., Bradbury, J., and Russell, C. J. (1976). The electric organ of a mormyrid as a current and voltage source. J . Comp. Physiol. A 110, 65-88. Bell, C. C., Zakon, H., and Finger, T. E. (1989). Mormyromast electroreceptor organs and their afferent fibers in mormyrid fish: I. morphology. J . Comp. Neural. 286, 391-407. Bennett, M. V. L. (1971a). Electric organs. I n “Fish Physiology” (W. S. Hoar and D. J. Randall, eds.), Vol. 5 , pp. 347-491. Academic Press, London and New York. Bennett, M. V. L. (1971b). Electroreception. I n “Fish Physiology” (W. S. Hoar and D. J. Randall, eds.), Vol. 5 , pp. 493-574. Academic Press, London and New York. Birkholz, J. (1969). Zufdlige Nachzucht bei Petrocephalus bovei. Aquarium 3,201-203. Black-Cleworth, P. (1970). The role of electrical discharges in the non-reproductive social behaviour of Gymnotus carapo (Gymnotidae, Pisces). Anim. Behav. Monogr. 3(1), 1-77. Bratton, B. O., and Kramer, B. (1988). Intraspecific variability of the pulse-type discharges of the African electric fishes, Pollimyrus isidori and Petrocephalus bovei (Mormyridae, Teleostei), and their dependence on water conductivity. Exp. Biol. 47, 227-238. Bratton, B. O., and Kramer, B. (1989). Patterns of the electric organ discharge during courtship and spawning in the mormyrid Pollimyrus isidori. Behav. Ecol. Sociobiol. 24, 349-368. Bullock, T. H. (1970). The reliability of neurons. J . Gen. Physiol. 55, 565-584. Bullock, T. H., and Heiligenberg, W., eds. (1986). “Electroreception.” Wiley, New York. Bullock, T. H., Hamstra, R. H., and Scheich, H. (1972a). The jamming avoidance response of high frequency electric fish. I. General features. J . Comp. Physiol. A 77, 1-22. Bullock, T. H., Harnstra, R. H., and Scheich, H. (1972b). The jamming avoidance response of high frequency electric fish. 11. Quantitative aspects. J. Comp. Physiol. A 77,23-48. Carr, C. E. (1990). Neuroethology of electric fish. Principles of coding and processing sensory information. BioScience 40, 259-267. Crawford, J. D. (1991). Sex recognition by electric cues in a sound-producing morrnyrid fish, Pollimyrus isidori. Brain Behav. Evol. 38, 20-38. Crawford, J. D. (1992). Individual and sex specificity in the electric organ discharges of breeding mormyrid fish (Pollimyrus isidori). J . Exp. Biol. 164, 79-102. Crawford, J. D., and Hopkins, C. D. (1989). Detection of a previously unrecognized morrnyrid fish (Mormyrus subundulatus) by electric discharge characters. Cybium 13(4), 3 19-326. Crawford, J. D., Hagedorn, M., and Hopkins, C. D. (1986). Acoustic communication in an electric fish, Pollimyrus isidori (Mormyridae). J . Comp. Physiol., A 159, 297-310. Darwin, C. R. (1859). “On the Origin of Species by Means of Natural Selection, or the Preservation of Favoured Races in the Struggle for Life. Cambridge Univ. Press, New York (reprint, 1975). Fay, R. R. (1970). Auditory frequency discrimination in the goldfish (Carassius auratus). 1. Comp. Physiol. Psychol. 73, 175-180. Fay, R. R. (1988). “Hearing in Vertebrates: A Psychophysics Databook.” Hill-Fay Associates, Winnetka, IL.

266

BERND KRAMER

Finger, T. E., Bell. C. C.. and Cam, C. E. (1986). Comparisons among electroreceptive teleosts: Why are electrosensory systems so similar? In “Electroreception” (T. H. Bullock and W. Heiligenberg, eds.). pp. 465-481. Wiley, New York. Fisher, R. A. (1954). Evolution and bird sociality. I n “Evolution as a Process” (J. Huxley, A . C. Hardy. and E. B. Ford, eds.), pp. 71-83. Allen & Unwin. London. FitzGerald. G. J. (1992). Filial cannibalism in fishes: Why do parents eat their offspring? Trends Ecol. Euol. 7 , 7-10. Fleishman. L. J . . Zakon. H. H., and Lemon, W. C. (1992). Communication in the weakly electric fish Sfernopygus macrurus. 11. Behavioral test of conspecific EOD detection ability. J . Cornp. Physiol. A 170, 349-356. Freedman, E. C i . , Olyarchuk, J., Marchaterre, M. A., and Bass, A. H. (1989). A temporal analysis of testosterone-induced changes in electric organs and electric organ discharges of mormyrid fishes. J . Neurobiol. 20, 619-634. Gottschalk. B. and Scheich, H. (1979). Phase sensitivity and phase coupling: Common mechanisms for communication behaviors in gymnotid wave and pulse species. Behau. E d . Sociobiol. 4, 395-408. Graff, C . (1986). Signaux electriques et comportement social du Poisson a faibles decharges, Marcusenius rnacrolepidotus (Mormyridae, Teleostei). Ph.D. Thesis, Univ. Paris-Sud Centre d’Orsay, Orsay. Graff, C., and Kramer, B. (1989). Conditioned discrimination of the E.O.D. waveform by Pollimyrus isidori and Gnathonernus petersii. In “Neural Mechanisms of Behavior” (J. Erber, R. Menzel, H. J. Huger, and D. Todt, eds.), p. 94. Thieme, Stuttgart. Graff, C. and Kramer, B. (1992). Trained weakly-electric fishes Pollimyrus isidori and Gnathonernus petersii (Mormyridae, Telostei) discriminate between waveforms of electric pulse discharges. Ethology 90, 279-292. Griffin, D. R . (1958). “Listening in the Dark.” Yale Univ. Press, Princeton, N.J. Hagedorn, M. (1988).Ecology and behavior of a pulse-type electric fish, Hypopornus occidentalis (Gymnotifonnes, Hypopomidae), in a fresh water stream in Panama. Copeia 1988(2), 324-335. Hagedorn. M., and Cam, C. E. (1985). Single electrocytes produce a sexually dimorphic signal in South American electric fish, Hypopornus occidentalis (Gymnotiformes, Hypopomidae). J. Comp. Physiol. A 156, 51 1-523. Hagedorn, M.. and Heiligenberg, W. (1985).Court and spark: Electric signals in the courtship and mating of gymnotoid fish. Anim. Behau. 33, 254-265. Hagedorn, M.. and Zelick. R. (1989). Relative dominance among males is expressed in the electric organ discharge characteristics of a weakly electric fish. Anim. Behau. 38, 520-525. Hagedorn, M.. Womble, M., and Finger, T. E. (1990). Synodontid catfish: A new group of weakly electric fish. Behavior and anatomy. Brain Behau. Euol. 35, 268-277. Harder, W., Schief. A., and Uhlemann, H. (1964). Zur Funktion des elektrischen Organs von Gnathonernus petersii (Gthr. 1862). J . Comp. Physiol. A 48, 302-311. Hasler. A. C., and Scholz. A. T. (1983). “Olfactory Imprinting and Homing in Salmon.” Springer-Verlag, Berlin. Heiligenberg, W. (1974). Electrolocation and jamming avoidance in a Hypopygus (Rhamphichthyidae, Gymnotoidei), an electric fish with pulse-type discharge. J. Comp. Physiol. A 91, 223-240. Heiligenberg, W. ( 1977). “Principles of Electrolocation and Jamming Avoidance in Electric Fish. Studies of Brain Function,” Vol. 1. Springer-Verlag, Berlin and New York. Heiligenberg, W. ( 1988). Neural mechanisms of perception and motor control in a weakly

COMMUNICATION BEHAVIOR IN WEAKLY ELECTRIC FISHES

267

electric fish. In “Advances in the Study of Behavior” (P. Slater et al., eds.), vol. 18, pp. 73-98. Academic Press, San Diego, CA. Heiligenberg, W. (1991). “Neural Nets in Electric Fish. MIT Press, Cambridge, MA. Heiligenberg, W., Baker, C., and Bastian, J. (1978a). The jamming avoidance response in gymnotid pulse species: A mechanism to minimize the probability of pulse-train coincidence. J . Comp. Physiol. A 124, 211-224. Heiligenberg, W., Baker, C.,and Matsubara, J. (1978b). The jamming avoidance response in Eigenmannia revisited: The structure of a neural democracy. J. Comp. Physiol. A 127,267-286. Hennig, W. (1966). “Phylogenetic Systematics.” Univ. of Illinois Press, Urbana. Hopkins, C. D. (1974). Electric communication: Functions in the social behavior of Eigenmannia uirescens. Behaviour 50, 270-305. Hopkins, C. D. (1986). Behavior of Mormyridae. In “Electroreception” (T. H. Bullock and W. Heiligenberg, eds.), pp. 527-576. Wiley, New York. Hopkins, C. D. (1988). Neuroethology of electric communication. Annu. Rev. Neurosci. 11,497-535. Hopkins, C. D., and Westby, G. W. M. (1986). Time domain processing of electric organ discharge waveforms by pulse-type electric fish. Brain Behau. Euol. 29, 77-104. Hopkins, C. D., Comfort, N. C., Bastian, J., and Bass, A. H. (1990). Functional analysis of sexual dimorphism in an electric fish. Hypopomuspinnicaudatus, order Gymnotiformes. Brain Behau. Euol. 35, 350-367. Kaunzinger, I., and Kramer, B. (1993). Schwebungen (“Jamming”) sind notwendig fur die Detektion sozialer Signale bei dem schwachelektrischen Messerfrsch Eigenmannia. Verh. Dtsch. Zool. Ges. 86.1, 216. Keeton, W. T. (1979). Avian orientation and navigation. Annu. Rev. Physiol. 41, 353-366. Kirschbaum, F. (1975). Environmental factors control the periodical reproduction of tropical electric fish. Experientia 31, 1159-1 160. Kirschbaum, F. (1987). Reproduction and development of the weakly electric fish, P o l h y r u s isidori (Mormyridae, Teleostei), in captivity. Enuiron. Biol. Fishes 20, 11-31. Konishi, M. (1993). Listening with two ears. Sci. Am. 268(4), 34-41. Kramer, B. (1974). Electric organ discharge interaction during interspecific agonistic behaviour in freely swimmingmormyrid fish. A method to evaluate two (or more) simultaneous time series of events with a digital analyser. J . Comp. Physiol. A 93, 203-235. Kramer, B. (1976). Flight-associateddischarge pattern inaweakly electric fish, Gnathonemus perersii (Mormyridea, Teleostei). Behauiour 59, 88-95. Kramer, B. (1978). Spontaneous discharge rhythms and social signalling in the weakly electric fish Pollimyrus isidori (Cuvier et Valenciennes) (Mormyridae, Teleostei). Behau. Ecol. Sociobiol. 4, 61-74. Kramer, B. (1979). Electric and motor responses of the weakly electric fish, Gnathonemus petersii (Mormyridae), to play-back of social signals. Behau. Ecol. Sociobiol. 6,67-79. Kramer, B. (1985a). Jamming avoidance in the electric fish Eigenmannia: Harmonic analysis of sexually dimorphic waves. J . Exp. B i d . 119, 41-69. Kramer, B. (1985b). Kommunikation mit elektrischen Signalen bei Fischen. In “Verhaltensbiologie” (D. Franck, ed.), pp. 273-277. Deutscher Taschenbuch-Verlag/Thieme, Miinchen and Stuttgart. Kramer, B. (1987). The sexually dimorphic jamming avoidance response in the electric fish Eigenmannia (Teleostei, Gymnotiformes). J . Exp. Biol. 130, 39-62. Kramer, B. (1990a). Sexual signals in electric fishes. Trends Ecol. Euol. 5, 247-250. Kramer, B. (199Ob).“Electrocommunication in Teleost Fishes: Behavior and Experiments.” Springer-Verlag, Berlin.

268

BERND KRAMER

Kramer, B., and Bauer, R. (1976). Agonistic behaviour and electric signalling in a mormyrid fish, Gnathonernus petersii. Behav. Ecol. Sociobiol. 1, 45-61. Kramer, B., and Heinrich, U. (1990). Discrimination ofinter-pulse intervals. I n “Electrocommunication in Teleost Fishes: Behavior and Experiments” (B. Kramer, ed.), pp. 170-177. Springer-Verlag, Berlin. Kramer, B., and Kaunzinger, I. (1991). Electrosensory frequency and intensity discrimination in the wave-type electric fish Eigenmannia. J . Exp. Biol. 161, 43-59. Kramer. B., and Kuhn, B. (1993). Electric signalling and impedance matching in a variable environment: The electric organ of a mormyrid fish actively adapts to changes in water conductivity. Naturwissenschaften 80, 43-46. Kramer. B., and Lucker, H. (1990). Species recognition by EOD interval pattern. I n “Electrocommunication in Teleost Fishes: Behavior and Experiments” (B. Kramer, ed.), pp. 157-170. Springer-Verlag, Berlin. Kramer. B.. and Otto, B. (1988). Female discharges are more electrifying: Spontaneous preference in the electric fish, Eigenmannia. Behau. Ecol. Sociobiol. 23, 55-60. Kramer, B., and Otto, B. ( 1 9 9 1 ) . Waveform discrimination in the electric fish Eigenrnannia: Sensitivity for the phase differences between the spectral components of a stimulus wave. J. Exp. Biol. 159, 1-22. Kramer, B.. and Teubl, H. (1994). Spectral phase sensitivity in a wave-type electric fish. Naturwissenschaften. Kramer. B., and Westby. G. W. M. (1985). N o sex difference in the waveform of the pulse type electric fish, Gnathonemus petersii (Mormyridae). Experientia 41, 1530-1531. Kramer, B., and Weymann, D. (1987). A microprocessor system for the digital synthesis of pulsed or continuous discharges of electric fish (or animal vocalizations). Behau. Brain Res. 23, 167-174. Kramer, B., and Zupanc, G. K. H. (1986). Conditioned discrimination of electric waves differing only in form and harmonic content in the electric fish, Eigenmannia. Naturwissenschaften 73,679-680. Landsman, R. E. (1991). Captivity affects behavioral physiology: Plasticity in signaling sexual identity. Experientia 47, 31-38. Landsman, R. E. (1993). Sex differences inexternal morphology and electric organdischarges in imported Gnathonernus petersii (Mormyriformes). Anim. Behuv. 46, 417-429. Landsman, R . E., and Moiler, P. (1988). Testosterone changes the electric organ discharge and external morphology of the mormyrid fish, Gnathonernus petersii (Mormyriformes). E.rperienria 44, 900-903. Landsman, R. E.. Jou, S. H., and Moller, P. (1987). Stress obscures signalling of sexual identity in Gnathonernus petersii (Mormyriformes). In “Reproductive Physiology of Fish” (D. R. Idler. L. W. Crim, and J. M. Walsh. eds.), p. 307. Memorial University of Newfoundland. St. John’s. Landsman, R. E., Harding,C. F., Moller, P.,andThomas. P. (1990).Theeffectsofandrogens and estrogen on the external morphology and electric organ discharge waveform of Gnathonernus petersii (Mormyridae, Teleostei). Horm. Behau. 24, 532-553. Lissmann, H. W. (1951). Continuous electrical signals from the tail of a fish, Gvmnarchus niloticus Cuv. Nulure (London) 167, 201-202. Lissmann, H. W. (1958). On the function and evolution of electric organs in fish. J . Exp. B i d . 35, 156-191. Lissmann. H. W . , and Machin, K. E. (1958). The mechanism of object location in Gvmnarchus nilnticus and similar fish. J . Exp. Bin/. 35, 451-486. Lucker, H.. and Kramer. B. (1981). Deveiopment of a sex difference in the preferred

COMMUNICATION BEHAVIOR IN WEAKLY ELECTRIC FISHES

269

latency response in the weakly electric fish, Pollimyrus isidori (Cuvier et Valenciennes) (Mormyridae, Teleostei). Behav. Ecol. Sociobiol. 9, 103-109. Matsubara, J., and Heiligenberg, W. (1978). How well do electric fish electrolocate under jamming? J . Comp. Physiol., A 125,285-290. Maynard Smith, J. (1991). Theories of sexual selection. Trends Ecol. Evol. 6 , 146-151. Mazeaud, M. M., and Mazeaud, F. (1981). Adrenergic responses to stress in fish. In “Stress and Fish” (A. D. Picketing, ed.), pp. 49-75. Academic Press, New York. McGregor, P. K. (1991). The singer and the song: On the receiving end of bird song. Biol. Rev. 66,57-81. McGregor, P. K., and Westby, G. W. M. (1992). Discriminationofindividually characteristic electric organ discharges by a weakly electric fish. Anim. Behav. 43, 977-986. Mills, A., and Zakon, H. H. (1991). Chronic androgen treatment increases action potential duration in the electric organ of Sternopygus. J . Neurosci. 11, 2349-2361. Moller, P. (1976). Electric signals and schoolingbehavior in a weakly electric fish, Marcusenius cyprinoides (Mormyriformes). Science 193,697-699. Moller, P. (1980a). Electroreception and the behaviour of mormyrid electric fish. Trends Neurosci. 3, 105-109. Moller, P. (1980b). Electroperception. Oceanus 23, 42-54. Moller, P., and Bauer, R. (1973). ‘Communication’ in weakly electric fish, Gnathonemus petersii (Mormyridae). 11. Interaction of electric organ discharge activities of two fish. Anim. Behav. 21,501-512. Moller, P., and Brown, B. (1990). Meristic characters and electric organ discharge of Mormyrops curviceps Roman (Teleostei: Mormyridae) from the Moa River, Sierra Leone, West Africa. Copeia 1990(4), 1031-1040. Moller, P., and Semer, J. (1986). Species recognition in mormyrid weakly electric fish. Anim. Behav. 34, 333-339. Moller, P., Semer, J., Squire, A., and Boudinot, M. (1982). Social spacing in the mormyrid fish Gnathonernus petersii (Pisces): A multisensory approach. Anim. Behav. 30, 641-650. Moller, P., Semer, J., and Bowling, D. (1989). Electric organ discharge displays during social encounter in the weakly electric fish Brienomyrus niger L. (Mormyridae). Ethology 82, 177-191. Myrberg, A. A., and Riggio, R. J. (1985). Acoustically mediated individual recognition by a coral reef fish (Pomacentrus partitus). Anim. Behav. 33, 41 1-416. Neuweiler, G. (1984). Foraging, echolocation and audition in bats. Naturwissenschafren 71, 446-455. Neuweiler, G . (1993). “Biologie der Fledermause.” Thieme, Stuttgart. Payne, R. (1962). How the barn owl locates prey by hearing. Living Bird 1, 151-159. Rankin, J. C., and Jensen, F. B., eds. (1993). “Fish Ecophysiology.” Chapman & Hall, London. Russell, C. J., Myers, J. P., and Bell, C. C. (1974). The echo response in Gnathonemus petersii (Mormyridae). J . Comp. Physiol. A 92, 181-200. Scheich, H. (1977). Neural basis of communication in the high frequency electric fish, Eigenmannia virescens (jamming avoidance response). I. Open loop experiments and the time domain concept of signal analysis. J . Comp. Physiol., A 113, 181-206. Scheich, H., Gottschalk, B., and Nickel, B. (1977). The jamming avoidance response in Rhamphichthys rostratus: An alternative principle of time domain analysis in electric fish. Exp. Brain Res. 28, 229-233. Semer, J., and Moller, P. (1981). Social behavior in mormyrid fish (Mormyriformes,Pisces):

270

BERND KRAMER

short and long-term changes associated with repeated interactions. In “Sensory Physiology of Aquatic Lower Vertebrates” (T. Szabo and G. Czkh. eds.), Vol. 31, pp. 221-233. Pergamon Press/Akadkmiai Kiado, Budapest. Shannon, C. E.. and Weaver, W. (1949). “The Mathematical Theory of Communication.” Cniv. of Illinois Press, Urbana. Shumway, C. A , , and Zetick, R. D. (1988). Sex recognition and neuronal coding of electric organ discharge waveform in the pulse-type weakly electric fish, Hypopornus occidenlu1i.s. J . Comp. Physiol. A 163,465-478. Suga, N. (1990). Biosonar and neural computation in bats. Sci. Am. 262(6), 34-41. Szabo. T. (1%7). Activity of peripheral and central neurons involved in electroreception. In “Lateral Line Detectors” (P. Cahn, ed.), pp. 295-311. Indiana Univ. Press, Bloomington. Szabo. T. (1974). Anatomy of the specialized lateral line organs of electroreception. In “Handbook of Sensory Physiology” (A. Fessard, ed.). Vol. 111, Part 3, pp. 13-58. Springer-Verlag. Berlin and New York. Szabo, T., and Fessard, A. (1974). Physiology of electroreceptors. I n “Handbook of Sensory Physiology” (A. Fessard, ed.), Vol. 111. Part 3, pp. 59-124. Springer-Verlag, Berlin and New York. Teyssedre. C., Boudinot, M., and Minisclou, C. (1987). Categorisation of interpulse intervals and stochastic analysis of discharge patterns in resting weak-electric fish (Gnuthonemus prtersii). Behuviour 102, 264-282. Thornhill, W. B.. and Levinson, S. R. (1987). Biosynthesis of electroplax sodium channels in Electrophorus electrocytes and Xenopus oocytes. Biochemistry 26, 4381-4388. von dcr Emde. G. (1990). Discrimination of objects through electrolocation in the weakly electric fish, Gnathonemus petersii. J . Comp. Physiol. A 167, 413-421. von der Emde, G.. and Bleckmann, H. (1992). Extreme phase-sensitivity of afferents which innervate mormyromast electroreceptors. Natunvissenschaften 79, 13 I- 133. von Frisch. K. (1967). “The Dance Language and Orientation of Bees.” Harvard Univ. Press (Belknap), Cambridge, MA. Watanabe, A,. and Takeda, K. (1963). The change of discharge frequency by a.c. stimulus in a weak electric fish. J . Exp. B i d . 40, 57-66. Waxman, S. G., Pappas, G. D., and Bennett, M. V. L. (1972). Morphological correlates of functional differentiation of nodes of Ranvier along single fibers in the neurogenic electric organ of the knife fish, Stemarchus. J. Cell Biol. 53, 210-224. Westby, G. W. M. (1975a). Comparative studies of the aggressive behaviouroftwogymnotid electric fish (Gymnotus carupo and Hypopomus artedi). Anim. Behau. 23, 192-213. Westby. G. W. M. (1975b). Has the latency dependent response of Gymnotus carupo to discharge-triggered stimuli a bearing on electric fish communication? J. Comp. Physiol. A 96%307-341. Westby. G. W. M. ( 1979). Electrical communication and jamming avoidance between resting Gymnotus corapo. Behav. Ecol. Sociobiol. 4, 381-393. Westby. G. W. M.. and Kirschbaum, F. 11982). Sex differences in the waveform of the pulse-type electric fish, Pollimyrus isidori (Mormyridae). J . Comp. Physiol., A 145, 399-403. Wier, C. C.,Jesteadt, W.. and Green. D. (1977). Frequency discrimination as a function of frequency and sensation level. J . Acoust. SOC. A m . 61, 178-184. Wilson. E. 0.(1975). “Sociobiology: The New Synthesis.” Harvard Univ. Press, Cambridge, MA. Wu, C. H . (1984). Electric fish and the discovery of animal electricity. A m . Sci. 72,598-607. Zwicker, E. (1982). “Psychoakustik.” Springer-Verlag, Berlin and New York.

Index

A

independent offspring inclusion, 57,79 kidnapping, 57,79 mixed species broods, 58-59 nest takeover, 60-62,78-79 young dumping, 57,79 Altruism coefficient of relatedness, 117-1 18 conditional and TFT,116 evolution, 115 inclusive fitness, 104 kin-groups, 103, 116-117 pleiotropic effects, 116 Amphiprion spp., see Anemone fish Anemone fish nest takeover, 49-51 satellite males, 49-51,75 Angelfish, interception and female theft, 6 Anolis aeneus, see Lizards Antarctic plunder fish, nest takeover, 60 Apeltes quadracus, see Sticklebacks Apistogramma borellii kidnapping, 57 satellite males, 49 Arctic graylings, nest takeover, 3 Artificial selection, 140 traits, 136 Assessment environmental and social factors in habitat and territory selection, 174 direct, 187-188, 222 indirect cues, 188-189,222 ultimate and proximate issues, 187-188 broad-sense individual and trait-group selection economy of explanation, 120-124 h ypothesis-generatingpotential, 126-129 modeling simplicity, 124-126

Abudefduf saxatilis, see Damselfish Acanthochromis polyacanthus, see Damselfish Adaptation and Natural Selection, 104 Aggressive behavior, see also Parasitic behavior; Territorial behavior elephant nose fish, cannibalism, 239 sticklebacks courtship, 149, 154-155 evolution, 138-139 hormonal influences, 159-164 juveniles, 148-151, 153-155, 160, 165 genetic correlations with dominance ability, 157, 164 with female aggression, 115, 164 with territorial aggression, 155, 159, 164 levels, 141, 149 male dominance ability, 149, 151-155 sexual maturity, 160-161, 165 territoriality, 139, 149-151, 153-155 genetic correlations with androgen production, 161-163 with breeding coloration, 161, 163 with dominance ability, 157 with courtship aggression, 155 Allele frequency, 109 Alloparental care benefits, 76 broodcare helpers, see Broodcare helpers egg stealing, 60-62, 79 heterospecific broods, 58-59 intraspecific adoptions, 56-58 family conflux, 57, 79 farming out, 57

271

272

INDEX

B Babblers, grey-crowned, I14 Bagrtcs meridionalis. see Catfish Bass, fertilization stealing, 9 Behavior, see also specific types control, genetic correlations as tool for study, 135-138 evolution, broad-sense individual and trait-group selection assessment economy of explanation, 120- 124 hypothesis-generating potential, 126-129 modeling simplicity, 124-126 roadmap analogy, 119-120 Behavioral ecology, 102, 110. 113, 129 game theory and, I18 proximate mechanisms versus ultimate factors, 217.222 Behavioral traits, polygenic control, 135 Behavior-genetic studies, 140 Between-trait-group fitness, 115-1 16 Betia spp.. see Bornean fighting fish Biology, evolutionary, see Evolutionary biology Blackbirds, red-winged, 210, 220 Blennies flexibility in reproductive behavior, 77 joint defense, 5 I nest takeover, 4 nuptial humps, 30 satellite males. 49. 51 Bluebirds, 218 Bodianrrs diplotaenia. see Wrasses Bornean fighting fish, broodcare helpers, 64-65 Bourgeois males, 2, 10. 72 female choice, 4 I , 74 versus parasitic males, 29-41 gonadbody weight ratio. 32-33 numbers, 29 pair spawning equivalents, 34-35 reproductive codeffort, 29-33 behavioral, 29, 31, 33 morphological, 30-31 physiological, 30 reproductive success, 33-37 spawning origins, 37-41 satellite reciprocity. 74

Broad-sense individual selection assessment economy of explanation. 120-124 hypothesis-generating potential, 126-129 modeling simplicity, 124-126 fitness, 113-1 18 historical perspective, 102-107 mathematics, 107-1 10 trait-group selectionist perspective, 111-113 average effect of competing alleles, 112 Broodcare helpers, 62-70.79 breederhelper reciprocity, 69-70 cooperation versus cheating, 68,70 costhenefits, 65-66 egg cannibalism, 67-68 egg cleaning, 68 egg dumping, 67 evolutionary background, 65 growth, 66, 70 kin selection, 69 paying for staying, 67, 70, 76, 80 predation pressure. 76 cheating and, 68 sexual maturity, 68 size and. 67 reproductive success, 66 simultaneous parasitic spawning, 67 size dependency, 65 territory defense, 65 territory inheritance, 66-67 Bullheads, nest takeover. 4. 60-61

z.

L

Campylomormyrus fhynchvphorus spp., see Elephant nose fish Canthigasfer rostrate, satellite males, 49 Canthigasfer ualenrini, female theft, 6 Catfish, interspecific broodcare, cichlidcatfish association, 46, 5 5 , 58-59 Coralliocetus angelica, nest takeover, 4 Chaefodon c a p i s t r a m , interception and female theft, 6 Chain pickerel. interspecific broodcare, 59 Chromis spp., see Damselfish

213

INDEX

Chubs, see also Cyprinids interspecific egg dumping, 45 interception and female theft, 6 nest takeover, 4 satellite males, 48 Cichlasoma spp., see Cichlids Cichlids broodcare helpers, 62-69,76 color patterns, 36,40 farming out, 45,57 female choice, 41 fertilization stealing, 9 flexibility in reproductive behavior, 77 genetic predisposition, 40 interspecific broodcare, 57 cichlidkatfish association, 29,46,58-59 interspecific spawning, 37 intraspecific broodcare, 54-55 joint defense, 51 kidnapping, 54, 5?,62 mating activities, 42 nest takeover, 4,60 reproductive strategy, 40-41 reproductive success, 33, 36-37 satellite males, 48-51 territorial coloration, 30 Cirrhilabrus ternminckii, see Wrasses Clepticus parrae, see Wrasses Colisa chuna, see Honey gouramis Coloration, fish, 28 territorial, 4 , 7 , 3 0 Commensalism, 76 Communication mechanisms, weakly electric fish electrical signalling, see Electrical signalling impedance matching, 248-25 I individual discrimination, 242-248 interdischarge time interval code, 236-237 Contests, territorial game theoretical models, 206,216 owners and intruders, 209-21 1 persistence, 209-21 1 social interactions. 208,216 space acquisition and, 21 1-217 spacing patterns and, 206,209-211 spatial relationships, 208-209, 216 Coots, 210 Coralliocetus angelica, see Blennies

Coris julius, see Wrasses Cottus gobio, see Bullheads Crytocara, interspecific broodcare, 55 Ctenochaetus striatus, see Surgeonfish Cyprinidon spp., see Pupfish Cyprinids, see also Chubs; Minnows; Shiners breeding tubercles, 30 egg dumping interspecific, 55 predisposition, 46 interspecific broodcare, 55 joint nest building, 52 satellite males, 49 spawning associations, 46

D Damselfish, 16,242 nest takeover and fertilization stealing, 9 independent offspring inclusion, 57-58 territorial coloration, 30 Darters coloration, 28, 30 joint courtship, 52 nest takeover, 4,60-61 Darwin, Charles, 102-103,262 Dichromatism, 30 Drosophila melanogaster, 146

E Ecology, see Behavioral ecology; Evolutionary ecology Egg dumping, 67 interspecific, 55 predisposition, 45-46 Eigenmannia spp., see Knife fish Electrical signalling elephant nose fish, courtship and spawning, 238-242 knife fish pulse species, 251-253 wave species EOD waveform sensitivity, 258 frequency modulations and sensitivity, 254-257

274

INDEX

Electric organ discharge, 234-235. 236 water conductivity and. 249-251 Electrophoretic analysis. 22 Elephant nose fish, African acoustic responses. 246 cannibalism, 239 courtship and spawning. 238-242 electrocommunication, 239, 242. 263 impedance matching. 248-25 1 individual discrimination, 242-248 interdischarge time interval code, 236-238, 263 male singing, 240-242 mate recognition. 243. 245, 247 sexual dimorphism. 242-243.247,263 EOD. see Electric organ discharge Erefmodus cyunosfictus.see Cichlids Erirwyon sucettci. see Suckers Esox niger. see Chain pickerel Etheosroma spp.. see Darters Erroplf1.s muculurtrs. see Cichlids Evolution. multivariate, see Multivariate evolution Evolutionary biology. 129 Evolutionary ecology. 102. 110. 129 Extrapair paternity. birds. 218, 221

F Fallfish reproductive success, 35 satellite males. tolerance of, 50 Female mimicry, 2. 28-29. 79 Females. fish choice ofmales. 41, 43. 71-72. 74 parasitic behavior, 43-46 broodcare, 43-44 egg dumping, 45-46 Female theft. 2.6-7.78 Fertilization coercive. 2 disruptive selection. 23 internal. 29 kleptogamic, 2. 10. 23 mate monopolization. 1-2 parasitic. 2 predictability of egg release. 71 resource monopolization, 1-2

stealing, 2, 9, 16, 72 satellite males, 50 territorial coloration and, 7 by subordinate male, 22-23 success. 71-72 pair spawning equivalents, 34 weight ratios, 36 Fish. weakly electric, 233-264 buzzers, 234,263 communication mechanisms, see Communication mechanisms electric organ discharge, 234-235,263 intraspecific variability, 242, 244-245, 247 electrocommunication, 234,262-263 electrogenic, 234, 263 electrolocation. 234, 262-263 capacitive impedance component, 25 1 electroreceptive. 234, 263 electrosensory information, 235, 263 Gymnotiformes. see Knife fish, South American hummers, 234 Mormyriformes, see Elephant nose fish, African PUP2 amplitude ratio, 243 pulse, 234, 263 amplitude, 236 wave, 234, 263 amplitude, 236 Fisher Exact Probability Test, 45 Fitness, social environment inclusive, 104. 114-1 16 neighborhood-modulated, 114-1 16 translatability, 117 within- and between-trait-group, 115-1 16

G Gumbusia ufjnis, 29 Gumbusiu holbrooki, 29 Game theory, 118-1 19 Garpike. interspecific broodcare, 45-46.59 Gasterosteus ucirleurus, see Sticklebacks Genes estimating variances and covariances, 146 linkage disequilibrium. 135

275

INDEX

pleiotropic gene action, 135 as replicators, 105, 112-113 Genetic correlations estimation from selection designs, 141-147, 164 artificial selection, 146 correlated responses of traits, 145, 154- 159 cross-covariance, 142 double selection experiment, 142-145, 154 heritability and, 155 large sampling error, 144 selection responses, 142, 149-154 standard error, 144 linkage disequilibrium and, 146 negative, 136 phenotypic evolution and, I36 physiological basis, 164 pleiotropic action of genes, 136 quantification, 137 Genetic coupling, 147 Genetic drift, 144-146, 157 drift effect, 145, 158 quantitative genetic analysis, 158 Genetic relatedness, 114 Genetics, quantitative, see Quantitative genetics Gnathonemus petersii, see Elephant nose fish Goby, freshwater, nest takeover, 4,60 Graylings, Arctic, see Arctic graylings Great tit, 189, 205 Grosbeaks, black-headed, 221 Group selection, see Trait-group selection Gymnarchus niloticus, see Elephant nose fish Gymnofus carapo, see Knife fish

avian, 184, 186 definition, 174-175 individual fitness and food, 183, 186 and population density, 181-182, 185, 205 intrinsic quality, 180-186 conspecific effects, 183, 185 environmental factors, 182-183, 186, 222 food, 182, 186 limiting factors, 183 reproductive success, 184 models, 182 Allee-type ideal free distribution, 181 Fretwell and Lucas ideal, 181 multidimensional niche, 180 negative effects of conspecifics, 181, 185 positive effects of conspecifics, 181, 185 conspecific attraction, 2 17-2 18 conspecific cuing, 217 reduced predation, 218 temtory function and, 180 theory, 182 Halichoeres maculipinna, see Wrasses Haplochromis spp., see Cichlids Harpagifer bispinis, see Antarctic plunder fish Hemilepidotus hemilepidotus, see Cichlids Heritability estimation, 142, 144-145 realized, 151 Hermaphrodites, fertilization stealing, 9 Hemng gulls, 185 Holacanthus passer, see Angelfish Honey gouramis, female theft, 7 Hybopsis biguttata, see Cyprinids Hypopornus occidentalis spp., see Knife fish

H I

Habitats, disturbed, 188-189 Habitat selection, 174 assessment of environmental and social factors direct, 187-188, 222 indirect cues, 188-189, 222 ultimate and proximate issues, 187- 188

IDI, see Interdischarge time interval code Inclusive fitness, 104, 114-1 16 Individual selection, see Broad-sense individual selection Interception, 2, 6-7 Interdischarge time interval code, 236-238, 263

276

INDEX

Intrademic selection, see Trait-group select ion

J Jamming avoidance response, 255-259.263 Julidochromis spp., see Cichlids

K Kin selection, 114 Kleptogamy, 10, 15 internal fertilization, 29 parasitic males, number of, 29 satellite males, 48 Knife fish, South American, 234 electical signalling, 25 1-257 electroreceptors. 258-261, 263 electrosensory frequency difference threshold, 256-257, 263 EOD duration, 253 frequency modulations, 255,263 waveform sensitivity. 258-261 individual recognition, 253-254 inter-EOD intervals, 252 jamming avoidance response, 255-259, 263 sexual dimorphism, 255-256.258. 263 spawning, 255 superimposirion signal, 256-258

L Lactoriiifornasini, satellite males, 49 Lamprologus spp., see Cichlids Larus argenratus. see Herring gulls Lepisosfeus O S S F U S , see Garpike Lepomis spp., see Sunfish Levels-of-selection, 129 debate. 102, 107 controversy, 110 history, 103-107 kinship and, I13 models. 113 Linkage disequilibrium, 135-136, 146 Lizards. 174, 189-204, 220. 222 contests, territorial costs, 216-217

dominance relationships, 214-215 ownerlintruder, 212 direct assessment, 198-200 environmental factors, 198 food supply, 199-200 habitat selection, 191-193 criteria, 191-192 environmental factors, 192 predation, 192-193, 198 proximate factors, 192 indirect assessment, 200-204 conspecific attraction, 202-203, 207-208 conspecific cuing, 200-202,206-208 juvenile growth rates, 194-195 dominance relationships, 215 prey abundance and, 195 social status. 197 territory size, 197 persistence and space acquisition, 215 social interactions and space acquisition, 211-217 spacing patterns, 195 and competitors, 1%-197 and fitness components, 196-197 territory function, 193-195 environmental factors, 193-194 food supply, 194, 197

M Males, fish, see also Bourgeois males; Parasitic males; Satellite males accessory males, 16 female mimicry. 28-29, 72, 79 initial phase males, 6 kleptogamic, 72 mimetic, 28 sexual maturity, 23 Marine electric ray, 249 Martins, purple, see Purple martins Mate monopolization, 1-2 female theft, 2, 6-7, 78 interception, 2, 6-7, 78 nest takeover, 2-4,78 piracy, 2, 5, 78 Mice, 207-208 Micrometrus minimus, see Surfperch Micropterus salmoides, see Sunfish

INDEX

Minnows, see also Cyprinids conditional reproductive behavior, 38 nest takeover, 60-61 satellite males, 48, 74 Mormyrops curviceps, see Elephant nose fish Mormyrus rume spp., see Elephant nose fish Moxostoma spp., see Suckers Multivariate evolution, 136-137, 139, 164 Mus musculus, see Mice Mutualism, 76, 78

N Natural selection, see also Broad-sense individual selection; Trait-group selection fitness, 113-1 10, 175 game theory role, 118-1 19 historical perspective, 102- 107 flexible partitioning, 1 1 1 mathematics, equivalence between models, 107-1 10 Neetroplus nemutopus, see Cichlids Neighborhood-modulated fitness, 114-1 16 New group selection, see Trait-group selection Nocomis spp., see Cyprinids Notemigonus crysoleucas, see Shiners Notropis spp., see Shiners

0

Occam’s razor, 104, 120 Oncorhynchus spp., see Salmon Ophiodon elongatus, nest takeover, 60 Orientation mechanisms, 233

P Padogobius martensi, see Goby Para blennius sanguinoientus, see Blennies Parasitic behavior female fish, egg dumping, 45-46.55.67 male fish interception and female theft, 2, 6-7,78

277

nest takeover, 3-4,9,60-62 piracy, 5 sneaking, 9- 10, 36 Parasitic males, 2, 10 versus bourgeois males, 29-41 gonadbody weight ratio, 32-33 numbers, 29 pair spawning equivalents, 34-35 reproductive cost/effort, 29-33 behavioral, 29,31,33 gonadal investment, 32, 75 morphological, 30-3 1 physiological, 30 reduction in growth, 32 reproductive success, 22-23,33-37 spawning origins, 37-41 reproductive tactics disruptive selection, 37 genetic predisposition, 37 phenotypic, 37 size dependency, 39 resource competition, 73 satellites, see Satellite males Parrot fish interception and female theft, 6 nest takeover and fertilization stealing, 9 territorial coloration, 30 Parus major, see Great tit Pelvicachromis pulcher, see Cichlids Perissodus microlepis, see Cichlids Petrocephalus bovei, see Elephant nose fish Pimephales promelas, see Minnows Piracy, 2 , 5 Plainfin midshipman, gonadbody weight ratio, 75 Pleiotropy, 135-136 Plunder fish, Antarctic, see Antarctic plunder fish Poecilia latipinna, reproductive strategy, 41 Pollimyrus isidori, see Elephant nose fish Polycentrus schomburgkii, reproductive strategy, 41 Pomastostomus temporalis, see Babblers Pomatoschistus minutus, see Goby Porichthys notatus, see Plainfish midshipman Predation dilution, 76, 79 Prisoner’s Dilemma, 116-1 18 Pseudocrenilabrus spp., see Cichlids

278

INDEX

Pupfish flexibility in reproductive behavior, 77 interception and female theft, 6 joint defense. 5 1 nest takeover. 3 reproductive strategy, 41 satellite males, 49. 5 I , 74 Purple martins. 211, 221

Q Quantitative genetics, 136, 138, 140. 164 behavioral traits and, 165-166 genetic drift and. 158 theory. 154

R Reproductive behavior. fish alternative mating tactics, 9-10, 15-16 alloparental care, see Alloparental care conditional. 40.73-74 courtship electrical signalling. 238-242 joint, 52-54 environmental influence, 38 genetic influence. 38 increase sperm production, 75 interception and female theft. 2, 6-7. 78 joint defense. 5 1-52 joint nestbuilding, 52-54 lifetime strategy, 39, 72-73 mate aquisition, 71 mate recognition, 243. 245. 247 piracy, 2, 5, 78 raising versus producing offspring, 47 spawning, see Spawning sperm competition, see Sperm competition success rates, 72 Resource monopolization, 1

S Salmon alternative mating tactics, 16. 22-23 disruptive selection. 23.73

fertilization success, 73 growth patterns. 39 kleptogamy. 16 lifetime reproductive strategy, 39 morphological structures, 30 reproductive strategy, 41 reproductive success and female proximity. 35-36 reproductive tactics, 37-38 satellite males, 48 stream-resident. 23 Salnio salar. see Salmon Saluelinrrs spp., see Trout Snrofherodon spp., see Cichlids Safaria pauo, see Blennies Satellite males. 3 age and, 39 benefits of, 50 paying for staying, 52 reproductive success. 36-37 spawning, origins. 37 subtenitories, 49 tolerance of, 48-50,75,79 Satellite threshold model. 71 Selective mechanisms economy of explanation, 120. 122-124, 130 hybridization, 124 individual versus trait-group, 123-124 within- and between-group, 124 hypothesis-generating potential, 120, 126- I30 individual versus trait-group, 127- 129 partitioning of fitness, 127 modeling simplicity, 120, 124-126, 130 individual versus trait-group, 125-126 neighborhood-modulated fitness, 126 Selfish gene theory, 105, 1 I2 Senioritus afromaculatus,see Chub Sernorilus corporalis, see Minnows Serranorhrornis robusrus, see Cichlids Serranus spp., see Bass Shiners, 30; see nlso Cyprinids egg dumping. 45 interspecific broodcare, 59 joint spawning. 52 Sneakers, 3,8. 33,72 age and, 39 Social behavior, evolution, 11 1 Sociobiology, I13

279

11VDEX

Sparisoma radians, see Parrot fish Sparrows, song, 211 Spawning electrical signalling, 238-242 group, 71-72 joint, 52-54 parasitic, 10, 16 patterns, 74 simultaneous parasitic, 16 spawning breaks, 5 success, 5-6 pair spawning equivalents, 34 Sperm competition alternative mating tactics, 9-10, 15-16 bourgeois and parasitic males numbers, 29 reproductive costs, 29-33 behavioral, 29,31,33 morphological, 30-3 I physiological, 30 spawning origins, 37-41 reproductive success, 33-37 female choice, 41-43, 71-72 female mimicry, 28-29 fertilization stealing, 7-9 forced copulation, 29 kleptogamy, 1 6 , 2 2 4 3 Sticklebacks aggressive behavior, see Aggressive behavior, sticklebacks anadromous, 165 egg number and reproductive success, 62 egg raiding, 62 female choice, 61-62 fertilization stealing, 7 freshwater, 165 genetic correlations, 147-164 aggressiveness and clutch maturation, 161 common genetic influences, 155 control lines, 146-149 dominance ability and breeding coloration, 158 genetic drift and, 157-158 juvenile aggressiveness and dominance ability, 157, 164 juvenile and female aggressiveness, 155, 164 juvenile and territorial aggressiveness, 155, 159, 164

proximate model, 162 territorial aggressiveness and androgen production, 161 I63 territorial aggressiveness and breeding coloration, 161, 163 territorial aggressiveness and dominance ability, 157 territorial and courtship aggressiveness, 155 life cycle, 139-140 phenotypic correlation, 156, 159 nuptial coloration, 139 reproductive success, 36 Streakers, 72 Suckers breeding tubercles, 30 broodcare, 45-46 interspecific broodcare, 59 joint courtship, 52 joint spawning, 53 nest takeover and fertilization stealing, 9 reproductive success, 35 satellite males, 52-53 trio spawning, fertilization success, 54 Sunfish egg dumping, 45-46 female choice, 41 female mimicry, 39,40 flexibility in reproductive behavior, 77 gonadbody weight ratio, 32-33 interspecific broodcare, 59 nest takeover and fertilization stealing, 479 piracy, 5 physiological reproductive costs, 30 reproductive strategy, 41 reproductive success, 36 satellite males, 48, 50, 74 territorial coloration, 30 Surfperch, dwarf, satellite males, 49 Surgeonfish, nest takeover and fertilization stealing, 9 Symphodus spp., see Wrasses Synodonris muliipunciaius, see Catfish

-

T Territorial advertisement, 206 auditory signals, 219-220

280 birds. 207 female choice, 220 Territorial behavior, 173-223 aggregation, 219 animal contests, I74 communication signals, 174 cost/benefit , 179- 180 economic model, 187, 222 environmental factors. 185 assessment, 187 female choice, 184, 219-220 food. 176 patterns, 204-217 aversive. 205 contests. 206 reproductive success, 175. 179, 218 social groups, 178 social interactions, 208 spacing patterns. 175. 185 assessment. 187 lifetime reproductive success, 179 parasites and. 178 Territoriality. definition, 174 Territory function, 1 7 6 1 8 0 comparative approach, 176 economic approach, 176, 178-179 size and predation. 177 habitat selection and, 180 T M , see Tit-for-tat rule Thulussoma spp.. see Wrasses Thymallus amicus, see Arcric graylings Tilupia rendalli. see Cichlids Tit-for-tat rule. 116 Tits. great, see Great tit Trait-group selection assessment economy of explanation. 120-124 hypothesis-generating potential, 126- 129 modeling simplicity, 124-126 broad-sense individual selectionist perspective, 110-1 I 1 fitness. 113-1 18 historical perspective, 102- 107 mathematics. 107-1 10 models, 105, 107 Traits artificial selection. 136. 140 behavioral. polygenic control, 135

INDEX

genetic correlation, 136 interdependence of, 136 Tripterygion tripteronorus. see Sunfish Tropheus irsacae, see Cichlids Trout alternative mating tactics, 16, 23 joint spawning, 52 reproductive strategy, 41

V Vehicles of selection, 112 individual versus trait-group, 113 nested hierarchy, 113

W

Within-trait-group fitness, 115-1 16 Woodpeckers, redcockaded, 184 Wrasses, 10,29 behavioral reproductive costs, 33 female choice, 41-43 female mimicry, 28 female theft, 6 fertilization success, 33-35 joint defense, 51-52 nest takeover, 3-4.60 physiological reproductive costs, 30-32 piracy, 5 reproductive strategy, 40-41 satellite males, 48-51, 74-75 subordinate males, 3, 5 , 6 , 48,77 temtonal coloration, 30

X XenorilapiaPauipinni. farming out. 45, 57 Xiphophorus nigrensis, 36,39 female choice, 4 1

Z Zacco temmincki, see Chub Zebrasoma spp., see Surgeonfish

Contents of Previous Volumes

Volume 13

Cooperation-A Biologist’s Dilemma JERRAM L . BROWN Determinants of Infant Perception GERALD TERKEWITZ, DAVID J. LEWKOWICZ, AND JUDITH M. GARDNER Observations of the Evolution and Behavioral Significance of “Sexual Skin” in Female Primates A. F . DIXSON Techniques for the Analysis of Social Structure in Animal Societies MARY CORLISS PEARL AND STEVEN ROBERT SCHULMAN

Plasticity and Adaptive Radiation of Dermapteran Parental Behavior: Results and Perspectives MICHEL VANCASSEL Social Organization of Raiding and Emigrations in Army Ants HOWARD TOPOFF Learning and Cognition in the Everyday Life of Human Infants HANUS PAPOUSEK AND MECHTHILD PAPOUSEK Ethology and Ecology of Sleep in Monkeys and Apes JAMES R. ANDERSON

Thermal Constraints and Influences on Communication DELBERT D. THIESSEN

Volume 15

Genes and Behavior: An Evolutionary Perspective ALBERT0 OLIVER10

Sex Differences in Social Play: The Socialization of Sex Roles MICHAEL J. MEANEY, JANE STEWART, AND WILLIAM W. BEATTY

Suckling Isn’t Feeding, or Is It? A Search for Developmental Continuities W. G. HALL AND CHRISTINA L. WILLIAMS Volume 14

Group Mating in the Domestic Rat as a Context for Sexual Selection: Consequences for the Analysis of Sexual Behavior and Neuroendocrine Responses MARTHA K. MCCLINTOCK

On the Functions of Play and Its Role in Behavioral Development PAUL MARTIN AND T. M. C A R 0 Sensory Factors in the Behavioral Ontogeny of Altricial Birds S. N. KHAYUTIN Food Storage by Birds and Mammals DAVID F. SHERRY Vocal Affect Signaling: A Comparative Approach KLAUS R. SCHERER 28 1

282

CONTENTS OF PREVIOUS VOLUMES

A Response-Competition Model Designed to Account for the Aversion to Feed on Conspecific Flesh W. J . CARR AND DARLENE F. KENNEDY

Volume 16 Sensory Organization of Alimentary Behavior in the Kitten K. V . SHULEIKINA-TURPAEVA Individual Odors among Mammals: Origins and Functions ZULEYMA TANG HALPIN The Physiology and Ecology of Puberty Modulation by Primer Pheromones JOHN G. VANDENBERGH AND DAVID M. COPPOLA Relationships between Social Organization and Behavioral Endocrinology in a Monogamous Mammal C. SUE CARTER. LOWELL L. GETZ, A N D MARTHA COHEN-PARSONS Lateralization of Learning in Chicks L. J. ROGERS Circannual Rhythms in the Control of Avian Migrations EBERHARD GWINNER The Economics of Fleeing from Predators R. C. YDENBERG AND 1.. M. DILL Social Ecology and Behavior of Coyotes MARC BEKOFF AND MICHAEL C. WELLS

Behavioral Ecology: Theory into Practice NEIL B. METCALFE AND PAT MONAGHAN The Dwarf Mongoose: A Study of Behavior and Social Structure in Relation to Ecology in a Small, Social Carnivore 0. ANNE E. RASA Ontogenetic Development of Behavior: The Cricket Visual World RAYJOND CAMPAN, GUY BEUGNON, AND MICHEL LAMBIN

Volume 18 Song Learning in Zebra Finches (Taeniopygia guttata): Progress and Prospects PETER J. B. SLATER, LUCY A. EALES, AND N. S. CLAYTON Behavioral Aspects of Sperm Competition in Birds T. R. BIRKHEAD Neural Mechanisms of Perception and Motor Control in a Weakly Electric Fish WALTER HEILIGENBERG Behavioral Adaptations of Aquatic Life in Insects: An Example A N N CLOAREC The Circadian Organization of Behavior: Timekeeping in the Tsetse Fly, A Model System JOHN BRADY

Volume 17 Volume 19 Receptive Competencies of Language-Trained Animals LOUIS M. HERMAN Self-Generated Experience and the Development of Lateralized Neurobehavioral Organization in Infants GEORGE F. MICHEL

Polyterritorial Polygyny in the Pied Flycatcher P. V. ALATALO AND A. LUNDBERG Kin Recognition: Problems. Prospects. and the Evolution of Discrimination Systems C. J. BARNARD

CONTENTS OF PREVIOUS VOLUMES

Maternal Responsiveness in Humans: Emotional, Cognitive, and Biological Factors CARL M. CORTER AND ALISON S. FLEMING The Evolution of Courtship Behavior in Newts and Salamanders T. R. HALLIDAY Ethopharmacology: A Biological Approach to the Study of Drug-Induced Changes in Behavior A. K. DIXON, H. U. FISCH, AND K. H. MCALLISTER Additive and Interactive Effects of Genotype and Maternal Environment PIERRE L. ROUBERTOUX, MARIKA NOSTEN-BERTRAND, AND MICHELE CARLIER Mode Selection and Mode Switching in Foraging Animals GENE S. HELFMAN Cricket Neuroethology: Neuronal Basis of Intraspecific Acoustic Communication FRANZ HUBER Some Cognitive Capacities of an African Grey Parrot (Psittacus erithacus) IRENE MAXINE PEPPERBERG Volume 20

283

Lekking in Birds and Mammals: Behavioral and Evolutionary Issues R. HAVEN WILEY

Volume 21

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

Social Behavior and Organization in the Macropodoidea PETER J. JARMAN The t Complex: A Story of Genes, Behavior, and Populations SARAH LENINGTON The Ergonomics of Worker Behavior in Social Hymenoptera PAUL SCHMID-HEMPEL “Microsmatic Humans” Revisited: The Generation and Perception of Chemical Signals BENOIST SCHAAL AND RICHARD H. PORTER

Volume 22

Male Aggression and Sexual Coercion of Females in Nonhuman Primates and Other Mammals: Evidence and Theoretical Implications BARBARA B. SMUTS AND ROBERT W. SMUTS Parasites and the Evolution of Host Social Behavior ANDERS PAPE MOLLER, REIJA DUFVA, AND KLAS ALLANDER

284

CONTENTS OF PREVIOUS VOLUMES

The Evolution of Behavioral Phenotypes: Lessons teamed from Divergent Spider Populations SUSAN E. RIECHERT Proximate and Developmental Aspects of Antipredator Behavior E. CURIO

Newborn Lambs and Their Dams: The Interaction That Leads to Sucking MARGARET A. VINCE The Ontogeny of Social Displays: Form Development, Form Fixation, and Change in Context T. G . G . CROOTHUIS