Geographic Variation in Behavior of a Primate Taxon: Stress Responses as a Proximate Mechanism in the Evolution of Social Behavior

Temperament is a complex behavioral trait that describes characteristic patterns of response to environmental, particularly social, conditions and perturbations. Disparities in the tendency to approach or avoid novelty or readiness to engage in aggressive interactions have been documented in comparisons between species (Christian 1970), subspecies (Gonzalez et al. 1981), populations within species (Champoux et al. 1994), inbred lines of laboratory animals (Scott and Fuller 1965), domesticated versus wild populations (Price 1984), and individuals within a species (Benus et al. 1992). Differences in physiological stress response systems (Selye 1937) are commonly identified as an important proximate mechanism underlying these temperament differences (Huntingford and Turner 1987, Kagan et al. 1988). Social systems of animals are perceived as emerging from relationships between individuals (Hinde 1983). Individual interactions, in turn, are hypothesized to reflect individual behavioral strategies which maximize inclusive fitness (Silk 1987). Selection on a physiological system, which can dramatically affect the pattern and outcomes of individual interactions, could produce evolutionary change in social organization and social behavior. Many workers explicitly suggest that temperament differences among primate species are adaptive in many instances, yet admit that the specific ecological and social selection pressures to which the neuroendocrine system is responding are often unclear (Thierry 1985, Clarke et al. 1988, Richard et al. 1989). Species-level comparisons have not offered many testable comparative models, probably because of confounding effects such as large phylogentic distances or uncertain phylogeny, inadequate knowledge of ecological and social conditions in the wild, drift, and convergent evolution. In short, little progress has been made toward understanding the evolution of stress-response patterns in primates. In this chapter I suggest that comparisons of geographically and genetically separated primate populations or subspecies may be an alternative and more successful approach to addressing the evolution of stress responses and the disparate social behaviors that result. Population and geographic comparisons are likely to be profitable for three reasons: (1) comparisons are less likely to be confounded by phylogenetic disparities (Arnold 1992), (2) the factors imposing different selective regimes among localities can perhaps be more readily identified, (3) hypothesis testing may be facilitated because populations suitable for testing a model will be easier to identify than new species.

The Use of Behavioral Ecotypes in the Study of Evolutionary Processes

Ecotypic variation refers to differences in traits between populations that reflect adaptation to different selection pressures. The origin of the term lies in Turesson’s (1922) observation that population differences in plant growth forms often breed true in a controlled environment. From experiments, Turesson (1922) concluded that much between-population variation in phenotypic traits reflects genotypic adaptation to local conditions. He described the phenomenon as ecotypic variation and the species population exhibiting a particular variant as an ecotype. The meaning of adaptation must be examined if the phenomenon of ecotypic variation is to be understood. Readers should refer to Reeve and Sherman (1993) for an in-depth analysis of the problems surrounding various definitions of this term. Briefly, Antonovics (1987) divided evolutionary studies into two distinct classes: those that consider the influences of past events (phylogenies) and those that consider why certain traits predominate over others in an ongoing selection process. Tinbergen (1963) proposed a similar subdivision. Most definitions of adaptation incorporate these two elements in that they require a history of selective modification of a trait. For instance, Harvey and Pagel (1991) express adaptations in terms of traits derived in their phylogenetic group that have current utilitarian function. However, ecotypic variation refers to trait differences that reflect adaptations to local conditions at one point in time. Phylogenetic constraints need not be examined except in terms of how they might limit adaptation. Therefore, when I refer to adaptation in this chapter, I will be limiting it to “a phenotypic variant that results in the highest fitness among a specified set of variants in a given environment” (Reeve and Sherman 1993). This definition is history-free. It is based on extant competing phenotypes, and thus fitness is of significance only in reference to current alternatives. Although the definition I borrow from Reeve and Sherman (1993) for adaptation does not specify that there be an underlying genetic mechanism, optimization models assume that there is (Charlesworth 1977). Minimally, we wish to know whether sufficient genetic complexity (variability) exists for a predicted optimal solution to be reached.

Thoughts on Geographic Variation in Behavior

In the past, behavior was assumed to be largely invariant within species, particularly those elements of behavior used as criteria of mate choice or in species recognition (see Magurran this volume, Verrell this volume). As is obvious from this volume, geographic variation could well be the common condition rather than the exception, and this applies to the full spectrum of behavioral phenotypes. Not only must students of behavior avoid typological thinking (Mayr 1963), but those wishing to infer similarity of behavior among populations must demonstrate the similarity just as surely as those interested in exploring population differentiation must demonstrate the differences. Behavior is as much a phenotype as is morphology; it is the expression of the combined effects of genotype and environment. Like other traits, behavior varies geographically because it is subject to geographically varying conditions and, hence, to natural selection, gene flow, and genetic drift. The chapters in this book provide examples of this variation, of the underlying genetic bases for the differences, and in many cases, the causes of the geographic variation. The study of geographic variation in behavior is in very early stages and lags well behind research on geographic variation in other kinds of traits (Endler 1977, 1986, 1995). Consequently, we cannot answer with assurance many of the questions we would like to be able to answer. However, we can take a first step using the insights offered by the research presented in this book. Before doing so, we briefly address some of the methodological issues that emerged over the course of the research because many are specific to the study of behavior or of geographic variation. We hope this will help others avoid problems encountered in these early studies. Many of the methodological issues discussed in the chapters in this book are related to the difficulty of working with behavioral characteristics that are extremely labile and responsive to environmental conditions. The remainder are issues related to the interpretation of data collected to assess patterns and causes of geographic variation. We will examine them in turn.

Variation in Advertisement Calls of Anurans across Zonal Interactions: The Evolution and Breakdown of Homogamy

The allopatric mode of speciation has become a dominant paradigm (sensu Kuhn 1970) in evolutionary biology over the last 50 years (Mayr 1942, 1992). In this model, the geographic range of a species is fragmented, the previously dedifferentiating effect of gene flow is interrupted, and the now separated populations diverge. If there is enough genetic differentiation during this period of isolation, then the disjunct daughter populations may become separate biological species (sensu Mayr 1942, 1992). This level of divergence is achieved by the development of sufficient genetic incompatibility, as reflected in an absolute infertility or sterility of hybrids or by sufficient reduction in the absolute or relative levels of adaptedness of hybrids so that none survives to maturity when in competition with parental individuals. Full and complete allopatric speciation, then, is marked by the acquisition of those properties needed for extensive and continuing coexistence. Broad overlap of geographic ranges (sympatry) can then develop without any significant interactions between individuals of the derived populations. In other situations, however, these essential properties may not have been acquired before the extrinsic barriers were removed. Thus, a critical stage is reached when the geographic ranges of previously separated daughter populations expand, and contact is established between individuals of the different genetic systems. Here the ecological compatibility, the specificity of mate choice, and the relative fitness of hybrids (if produced) are tested, and the following four outcomes may be envisaged. (1) If there is a cost to inbreeding, based on extrinsic and/or intrinsic factors, then the two lineages may diverge further in sympatry such that the attributes essential for stable coexistence arise, or are enhanced, through the direct action of natural selection within the context of the interaction (for references and recent commentaries, see Howard 1993, Littlejohn 1993, Butlin 1995). (2) If there is an ecological gradient and no cost to interbreeding because the hybrid progeny are as fit as, or fitter than, those of the parental taxa in part of the gradient, a geographically restricted and persistent hybrid zone may form (Moore 1977, Moore and Buchanan 1985, Littlejohn 1988, Hewitt 1989, Harrison 1993, Howard 1993).

The Causes and Consequences of Geographic Variation in Antipredator Behavior: Perspectives from Fish Populations

Predators are extremely effective agents of selection. After all, if an individual member of a prey species does not survive long enough to reproduce, it will have lost its chance (kin selection considerations apart) to bequeath its genes to future generations. It is not surprising, therefore, that many cases of population difference have been attributed to geographic variation in risk. These population differences can take a variety of forms and may, for example, involve modifications to morphology or to life-history traits. The correlation between armor and predation in the three-spined stickleback, Gasterosteus aculeatus, is one case that has been well documented (see Reimchen 1994 for a review and discussion), while another is the association between reproductive allotment and risk (Reznick and Endler 1982) and male color pattern and risk (Endler 1980) in the Trinidadian guppy, Poecilia reticulata. However, such adaptations can be futile if they are not accompanied by effective antipredator behavior. For instance, a cryptic color pattern confers no advantage if its holder chooses the “wrong” background or behaves in a conspicuous manner. Behavior is also flexible in a way that life histories or morphology may not be, and it allows moment-to-moment changes in response as risk increases or decreases. Because it is such an important weapon in the evolutionary arms race, antipredator behavior provides important insights into the causes and consequences of natural selection. Some of the best examples of geographically variable antipredator responses occur in populations of freshwater fish (see, e.g., Bell and Foster 1994). The predation regime of these populations is relatively easy to classify—at least in terms of the presence and absence of predatory species—and the distribution of key predators can explain much of the documented variation in antipredator behavior (see p. 140). Covariance in predation regime and antipredator responses is compelling evidence for natural selection. Moreover, because predation regimes can change (or be manipulated) over relatively short periods of time, there is an opportunity to record heritable changes in antipredator responses—in other words, to watch evolution in action.

Ecology, Phenotype, and Character Evolution in Bluegill Sunfish: A Population Comparative Approach

Evolutionary ecology explores the intimate relationships between the mechanisms responsible for the production and maintenance of organismal form and the ecological function of the structures and behaviors that compose form (Arnold 1983). The analysis of diversity from this perspective is founded on the premise that variation in measured phenotypes can reflect the results of the process of natural selection (Williams 1966, 1992; Gould and Vrba 1982). However, because the fitness consequences of any particular phenotype are the result of complex interactions among an individual’s genotype, morphology, behavior, and the environment within which it must function (Gould and Lewontin 1978, Endler 1986), a phenotype best suited for one set of environmental conditions may not perform best in another (e.g., Endler 1983, Rausher 1984, Ehlinger and Wilson 1988, Schluter 1993). When making comparisons among populations, phenotypic variation due to underlying genetic differences that may reflect evolutionary responses to different environments must be distinguished from phenotypic variation that results from phenotypic plasticity and/or genotype–environment interactions (Stearns 1989). For example, regional environmental variation can result in different selective regimes among populations and produce “site-dependent” fitnesses for phenotypes (e.g., Reznick et al. 1990, Robinson and Wilson 1994; chapters in this volume). Likewise, varying social and trophic conditions on a local scale can result in “situation-dependent” performances and payoffs for different phenotypes within populations (Maynard-Smith 1974, Ehlinger 1990, Krebs and Kacelnik 1991). Both phenomena may influence patterns of geographic variation and must be considered when studying phenotypic differences among populations. My aim in this chapter is to illustrate how population comparisons of bluegill sunfish (Lepomis macrochirus) can be used to study the evolution of behavioural and morphological variation. Critical features that shape bluegill trophic ecology (e.g., temperature, depth, substrate type, prey types, productivity, and predator abundance) vary among lakes in combination with forces that influence reproductive ecology (e.g., availability of spawning habitat and age or size structure of the population). Population comparisons provide unique opportunities for discerning the roles of sexual and trophic selection in bluegill phenotypic evolution.

Geographic Variations on Methodological Themes in Comparative Ethology: A Natricine Snake Perspective

The most distinctive and characteristic emphasis of early ethology was also what set it off from other post-Darwinian studies of animal behavior. This was the view that behavior varied among species in the same way as did morphological characters and that behavioral differences were as much a product of the evolutionary drama as were the characters that could be measured in museum collections (Tinbergen 1960, Lorenz 1981, Burghardt 1985, Burghardt and Gittleman 1990, Gittleman and Decker 1994). The logical extensions of this view were that behavioral phenotypes could be used in reconstructing phylogenetic histories, that the evolution of behavioral phenotypes could be studied in the same way as the evolution of other classes of traits, and that many of the behavioral differences among taxa reflected underlying genetic differentiation at both the species (Hinde and Tinbergen 1958) and population (Foster and Cameron 1996) levels. Behavior may also initiate evolutionary changes in other attributes of organisms (Mayr 1960, 1965, Wcislo 1989, Gittleman et al. 1996). Although the role of genes in behavioral determination remained controversial for years (see Gottlieb 1992, de Queiroz and Wimberger 1993 for current critiques), many behavior patterns have proven heritable (Mousseau and Roff 1987; papers in Boake 1994b). Indeed, some complex, “species-typical” behavior patterns are performed normally without opportunity for learning (Lorenz 1965). Such behavior patterns can be expressed early or late in development (Lorenz, 1981). At the other extreme, many complex behavioral phenotypes are learned with only slight, if any, genetically based predisposition to perform particular behavior patterns. Between these extremes is a diversity of interactions between genes and environment, including imprinting and complex developmental trajectories produced by interactions between neural development and experience. Many of the currently interesting and controversial questions in the nature–nurture debate do not center around species-typical behavior patterns. Instead, they concern the nature of genetic differences among individuals and populations in the performance of particular behavior patterns and in the ability to modify their performance with experience. Thus the problem must be conceptualized as one in which the interactions of specific genetic constitutions with specific environmental contexts need to be evaluated.

The Evolution of Behavioral Norms of Reaction as a Problem in Ecological Genetics: Theory, Methods, and Data

Behavior, like other phenotypic traits, varies as a function of genes and environment. Variation occurs at all demographic levels, within individuals over time, between individuals, and between populations and species. Whether variation is important will depend on the behavior and its context. For example, whether a bird scratches its head by extending a leg above or below the adjacent wing may not have profound fitness consequences, although species differences in this character may shed light on phylogenetic relationships (e.g., Wallace 1963; Simmons 1964). In contrast, other behaviors, such as the instantaneous decision to migrate or not, may affect fitness directly by altering the schedule of fecundity or mortality (Dingle et al. 1982). Such strategic behaviors (Maynard Smith 1982), which often depend for their expression on the assessment of local cues (Moran 1992), are complicated and important evolutionary traits. The phenotypic variability that defines them, however, has hindered our ability to treat them with formal evolutionary–genetic analyses that are central to the complete understanding of any putative adaptation. Much of the evolutionarily important variation observed in strategic behavior probably stems from differences among individuals due to genotype–environment interactions. To illustrate this in the most general terms, consider that behavioral distinctions among individuals may be based on (1) differences in the environmental conditions they experience, (2) differences in genetic elements that code for specific tactics or predispositions, or (3) differences in the genotype–environment interaction, manifested through developmental or facultative pathways, that is, “norms of reaction” (Schmalhausen 1949). Norms of reaction are functions that describe how a genotype is translated into a phenotype by the environment. They are becoming widely employed as a paradigm in evolutionary studies of physiological and life-history traits (e.g., Dingle 1992; reviewed by Stearns 1989), but are not yet used widely in studies of behavioral traits (but see Thompson this volume). Because much of the variation that behaviorists observe within populations and species is likely the result of a complex combination of individual differences in genetic code and differences in environment, norms of reaction need to be explored as a method for understanding the sources and structure of behavioral variation.

Geographic Variation in Sexual Behavior: Sex, Signals, and Speciation

The chapters in this volume share the theme that our understanding of pattern, process, and consequence in the study of behavioral evolution can be advanced by examining differences among conspecific populations. Traditionally, biologists have sought such understanding by comparing different species. Although differences among species are usually greater than differences among conspecific populations, so many factors can vary interspecifically that determining selection pressures driving behavioral divergence may be difficult. As Arnold (1992) has argued, confounding variables often are less prevalent in intraspecific studies; in addition, relatively small evolutionary changes may be perceptible. From a historical perspective, there appear to be good reasons for believing that sexual behavior should show little variation among conspecific populations. First, early species concepts largely were typological, and accorded intraspecific variation with no reality, let alone importance (Mayr 1976). Partially due to such thinking, early ethologists argued that certain behavior patterns should be largely invariant. For Lorenz (1970), courtship behavior was a prime example of such a “fixed action pattern” (or FAP), a predictable and stereotyped sequence of actions that was elicited by a specific releasing stimulus. Later work revealed that such sequences are more variable than once thought, leading to the suggestion that the FAP be replaced by the MAP, or “modal action pattern.” This stresses the average or modal nature of much behavior (Barlow 1977). The second reason for expecting little intraspecific variation in sexual behavior derives in part from the modern synthesis. The “founding fathers” of modern evolutionary biology placed great emphasis on the role of species differences in preventing interspecific mating and wastage of reproductive effort in the production of unfit hybrid offspring (e.g., Dobzhansky 1937). Indeed, Tinbergen (1953) stated that one of the functions of mating behavior is to ensure such reproductive isolation among species. Presumably, selection would strongly favor the production of unambiguous signals, leading to invariance. The earliest evidence demonstrating the existence of intraspecific variation in sexual behavior patterns came from studies that were firmly rooted in concepts that characterized the early stages of the modern synthesis.

Geographic Variation in Animal Communication Systems

Intraspecific communication is fundamental to most social behavior. It is also a special problem in animal behavior because it necessarily involves the interaction of two systems within a species, a sender and a receiver (Walker 1957, Blair 1964, Capranica 1966, Schneider 1974, Hoy et al. 1977, Hopkins and Bass 1981, Gerhardt 1988, Brenowitz 1994). Sender and receiver components are almost always separable morphologically, physiologically, and behaviorally. Each may be under different mechanistic and developmental control, and, especially in those cases in which the senders and receivers are segregated by sex, the impact of selection pressures and constraints can be very different (Brenowitz 1986, Ryan 1986, 1988; Wilczynski 1986, Endler 1983, 1993). The presence of two different but necessarily interacting components make the evolution of communication systems a particularly challenging problem in behavioral biology. In any communication system, the interaction between senders and receivers dictates some degree of matching such that the signal emitted by one member of the communicating pair is effectively received, recognized, and assessed by the other member (Blair 1964, Gerhardt 1982, 1988; Capranica and Moffat 1983, Littlejohn 1988, Ryan 1988, 1991; Endler 1993). Effective coupling of senders and receivers is crucial when communication underlies mate choice. Communication systems that accurately discriminate between heterospecifics and conspecifics, while effectively linking conspecifics to each other, are important for ensuring mating with genetically compatible conspecifics. As such, communication systems can be integral parts of speciation and the maintenance of species isolation (Blair 1958, Mayr 1963, Paterson 1985, 1993; Littlejohn 1981, 1988; Butlin 1987, Coyne and Orr 1989, Claridge 1993, Moore 1993, Wood 1993). The natural variation among and within species in both signals and receivers provides a means for examining the factors contributing to the evolution of communication systems (Templeton 1981, Ryan and Keddy-Hector 1992, Paterson 1993). Among the different levels of variation observed, geographic variation provides the best material for disentangling the myriad factors shaping the evolution and divergence of communication systems and for testing fundamental ideas about the evolution of behavior (Endler 1983, Baker and Cunningham 1985, Nevo and Capranica 1985, Ryan and Wilczynski 1991, Loftus-Hills and Littlejohn 1992).