L04-shaw_wiley_2010

L04-shaw_wiley_2010 - Shaw K L 8 Wiley C(2010 The genelic...

Info iconThis preview shows pages 1–10. Sign up to view the full content.

View Full Document Right Arrow Icon
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 2
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 4
Background image of page 5

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 6
Background image of page 7

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 8
Background image of page 9

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 10
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: Shaw, K. L. 8. Wiley, C. (2010) The genelic basis of behavior. In: Evolutionary Behavioral Ecology (eds, Weslneai, D.F & Fox, C.W.), Ch. 5:71-89. 5 KERRY L SHAW AND CHRIS WJLEY ehavioral ecologists are interested in behaviors expressed under particular ecological condiw " tions and the evolutionary causes and consequences : of such behavioral strategies. Evolution of behav- _-1 ior can occur only when there is a change in allele : frequencies. It therefore follows that there can be " no behavioral evolution without a genetic basis to _ behavioral variation. Thus the study of behavioral genetics is essential to our understanding of behav- ioral ecology and evolution because it informs us _ about the genetic and environmental contribu- tions to variation in individual behaviorai traits, and further may provide explanatory power for understanding major concepts in behavioral ecol- ogy. Recent advancements in molecular tools for studying genes have revolutionized the study of behavioral ecology; among a host of other uses, genetic tools have allowed us to infer paternity of offspring, indirectly assess dispersal patterns, and examine phylogenic relationships between taxa. Although these varied applications of genetic tools have had a profound impact on the study of hehav~ iotal ecology, this chapter will deal specifically with the insight that can be gained from more specific knowledge of the genetic basis of the actual behav- iors being studied. The idea that behavioral traits can have a strong genetic basis may be less intuitive than for other traits, such as morphological characters. This is perhaps illustrated most famously by the “nature- nurture” debate, in which “nurture” advocates Tl The Genetic Basis of Behavior espoused the view that human behavior is largely the effect of development, learning, and social environment (see Philosophical Psychology, 21[3], 2008) for a recent treatment of this debate). It is now realized that all phenotypes, including behav- iors, are affected by both genes and environment. That behaviors have a genetic basis, at least par- tially, is vital to our expanded appreciation of the importance of natural selection in generating bio- diversity le.g., through sexual or kin selection]. Many intricate and complex behaviors that serve as an endless source of fascination ,to the student of animal behavior defy easy comprehension as to how genes contribute to their existence (e.g., socile behavior in honey bees or sexual signaling in birds}. However, evidence is beginning to accumulate that even the most complex of behaviors can have genetic bases that can be dissected and understood both in terms of development and evolution. The study of behavioral genetics has a lengthy history (Greenspan 2008) and the importance of genetic insights to theories of behavioral evolution {e.g., Hamilton 1964) cannot be overemphasized. Yet, few genetic studies of specific traim are currently integrated into our understanding of the behavioral ecology of an organism, perhaps because optimal- ity models often omit explicit descriptions of the genetic basis of behavior, or because the study of behavioral genetics often has been undertaken in model laboratory organisms {Moore SC Boalte 1994; Boake et a1. 2002). For many years, simple 72 Foundations genetic assumptions (whether implicit or explicit) underlying behavioral “strategies” have been com- monplace in studies of behavioral ecology, and Studies of the selective forces that influence these behaviors have proceeded largely in ignorance of the underlying genetics. By and large, the approach termed the phenotypic gambit (Grafen 1984), in which simple genetic assumptions are made, have been successful in predicting conditions under which alternative strategies produce higher fimess payoffs. Although this simplification can be justi- fied in some traditional approaches to behavioral ecology, we can also see a steady increase in known instances in which these simplifications are likely misleading (e.g., Hadfield et al. 2,007). Furthermore, making simplified genetic assumptions greatly con- strains the scope of behavioral ecology and evolu- tion. This chapter attempts to provide examples of more complex questions of interest to behavioral ecologists that require details of, for example, the genetic origins of behavioral diversity or the evolu- tionary genetic processes underlying the fate of new behavioral genetic variation. Simple assumptions about the genetic inheritance of such behavioral variation would be counterproductive. Fortunately, the approaches used to study the evolution of other phenotypes can be readily applied to the genetics of behavior, using techniques that are increasingly tractable in natural populations. Like- wise, population and quantitative genetic models of behavioral trait evolution depend upon, and are improved by, knowledge of the genetic basis of behav- ioral variation (Moore 8:: Boake 1994). The aims of this chapter are to (1) discuss how genetic approaches to the study of behavior, particularly when knowledge of the underlying genes is lacking, can give us deeper insight into behavioral ecology and evolution and (2) discuss what might be possible in the study of behav— ioral ecology and evolution if we knew the actual genes underlying behavioral variation. Throughout, the focus is on the study of behavioral trait evolution rather than fhe employment of molecular methods for the study of behavioral processes. To launch a clear and meaningful discussion of the genetic basis of behavior, we must first clarify two classes of questions asked in die area of behav- ioral genetics. The first question asks, what is the genetic basis of the fully functional behavior (includ- ing those structures and molecular processes that combine to produce the behavioral output)? The second question, although fundamentally related to the first, represents an evolutionary perspective by asking, what is the genetic basis of behavioral variation? The scope of the latter question does not require knowledge of all genes that contribute to the structural development and expression of a behavior, but focuses on the underlying genetics cur- rently contributing to variation in a behavior. The content of this chapter largely deals with this second question, the genetic basis of behavioral variation, because this is of primary interest to behavioral ecologists studying the fitness of organisms in con- temporary or past populations. However, when considering the evolutionary origins of a behavior, the genetics of behavior and the genetics of behav- ioral variation are closely related because any of the genes contributing to the development and expres- sion of a behavior may vary in a given population and thereby contribute to behavioral variation. GENETICS OF BEHAVIOR WITHOUT THE GENES There are many reasons that motivate investiga- tions of the genetic basis of behavioral differences. Even in compelling behavioral systems in which lit- tle is known about the genome, we can gain imporw tant insights into questions of interest to behavioral ecologists, given some ability to perform husbandry or observe or construct pedigrees. Below, we dis- cuss a number of interesting and important ques- tions that may be addressed without knowledge of the Specific genes underlying a behavior. Our choice of examples is not exhaustive, but rather represents an introduction to some of the important roles that behavioral genetics can play in furthering Our understanding of behavioral ecology. Causes of Behavioral Variation Nongene‘tic Causes As discussed previously, the causes of behavioral variation are profoundly important because only variation in behavior due to variation in genes can evolve. However, the presence of behavioral varia- tion does not guarantee that there is a genetic basis to that variation. For example, hormones transferred to the egg or fetus during gestation may influence the neural development of the offspring, thereby influencing behavior through uongenetic means. In addition to such maternal effects, many behaviors have a strong propensity for cultural transmission The Genetic Basis of Behavior via learning from parents, other relatives, or nonre- ; 13th individuals. Finally, behavioral variation may be induced by an underlying variability in the physi- ‘ cal environment. Phenotypic plasticity, variation ‘ due to each of these differences in environment, is _ widespread and sometimes dramatic (see chapter 6). 7 Such is the case in many social hymenoptera in ‘ which different individuals perform different tasks _'such as nursing, foraging, and defense of the nest (Seeley 1 995), roles that are sometimes accompanied -‘ by considerable differences in morphology (e.g., as . - in the big-headed ant, Pbeidole megacepbala). This array of variable developmental outcomes, known as polyethism (behavioral variation) or more gener- ally polyphenism (phenotypes in general), is due not to genetic variation but to variation in developmen- tal or behavioral environment. Another conspicuous example of polyphenism appears in certain amphib— ians (Relyea 2004) in which apparently genetically identical individuals can develop into different " trophic morphs with distinctive morphologies and ' dietary behaviors. Behavioral phenotypes, like other ' phenotypes, are products of genes and environment, both of which can cause behavioral variation (Mac- kay 5c Anholt 2007). - Determining a Genetic Basis to Behavioral Variation The important goal of documenting a genetic basis to behavioral differences among popula— tions or species, thereby ruling out environmental differences as causes of the variation, is often first achieved through a “common garden” study, in which behavioral variants are reared in an identical environment. Behavioral differences that persist, or “breed true,” suggest a genetically heritable basis to the variation. For example, in a study by Sim- mons (2004),-comrnon garden breeding was con- ducted for several generations with the oceanic field cricket, Teleogryllus oceanicus,_ultimately showing that song differences among populations are due (at least partially) to genetic differences. Common garden experiments are not restricted to species amenable to laboratory Study. For example, even in Wild birds, cross-fostering chicks bemeen nests of closely related species has proven a useful tool for testing the genetic basis of behaviors ranging from mate preferences (Smeher et al. 2007) to foraging patterns (Slagsvold 5C Wiebe 2007). When trait variation exists within a single popu- lation, common garden studies are of limited utility 13 and further study must determine the degree to which variation is due to genes Versus environment. This is typically carried out by comparing the pbe— notypes of relatives to that of nonkin, something that is achievable when pedigrees are known. in such instances, a first goal is to determine whether the behavioral variation we see is based on alterna- tive alleles at a single locus (the realm of population genetics) or small contributions from variation at many loci (quantitative genetics). These alternatives have been studied using two very different research approaches in genetics (Greenspan 2004) that date back to the origins of the field of genetics itself. Although the research legacies of both approaches continue today, the consequences to modeling out- comes in behavioral evolution are significant (see below). If segregating variation within a family or population suggests polygenic control (i.e., varia— tion due to the contributions of many genes), a more formal treatment to estimate heritability can be undertaken to quantitatively attribute portions of phenotypic variance to genes and environment (Freeman Sc Heron 2004). Examples demonstrating heritability in behavioral variation have been pub- lished for a diversity of behaviors, from dispersal behavior (e.g., ballooning in spiders: Bonte 8: Lens 2007; migratory behavior in birds: Pulido 2007), to male signaling (e.g., cricket song: Sinmions 2004). Genetic Architecture of Behavioral Variation Genetic architecture refers to features characterizingr the relationship between genotype and phenotypic variation, such as the number of loei involved in trait variation, the number of alleles per loans, the allelic interactions both within and between loci, the amount an allele contributes to the phenotypic variation, the degree of pleiotropy (i.e., how many phenotypes a given gene can affect), as well as the: heritability of the trait. Below we discuss some of the ways in which an understanding of the genetic architecture of behavioral variation can yield insight into the evolutionary potential of the behavior. Single Genes versus Complex Genetic Architectures When an evolutionarily optimal endpoint lies out- side standing variation in a trait, we expect a popu— lation to move more slowly toward that optimum when potential variation is controlled by a single 74 Foundations locus than When variation is due to many genes. Under these conditions, traits with the potential for polygenic variation should evolve more quickly because there are more sources of new mutation. Conversely, traits determined by single genes may evolve more rapidly toward an adaptive optimum when the optimum is within standing variation in the trait. This is because favorable alleles are less likely to be concealed by variation at other loci, and thus can sweep to fixation more quickly. An understanding of the genetic architecture underly- ing phenotypic variation is therefore important for our understanding of the evolutionary potential of populations. A phenomenal example of this was reported recently in the oceanic field cricket mentioned above. The distribution of this species includes northern coastal Australia and many of the Pacific islands, extending more recently across the Hawai- ian archipelago. In Hawaii, I oceanicus is now in contact with the acoustic parasitoid tachinid fly, Ormt'a ocbracea, itself a recent invader from North America. Homing in on the song of male crickets, female 0. ocbmcea locate and larviposit on or near the host cricket. Larvae burrow into and eventually kill the cricket. On the island of Kauai, in fewer than 20 generations, a wing mutation rendering males mute, and thus well protected from the para- sitoid, swept to near fixation showing just how rap- idly a population can evolve under strong selective pressures (Zuk et al. 2006). Different methods have been developed to ana- lyze predictions of trait evolution when phenotypic variation is due to allelic variation at single or mul- tiple loci (Greenspan 2.004). Single-locus evolution is considerably easier to predict because models focus on genotypes and allele hequency change due to particular evolutionary forces and can be studied as deviations from the Hardy-Weinberg equilibrium. With complex genetic underpinnings to behavioral variation comes more complex mod- eling in which phenotypic changes are predicted, often in the absence of known genotypes. Classic quantitative genetic predictions of behavioral evo— [ution have been built around the "breeder’s” equa- tion (so-called due to its origin in the agricultural world), R = 1238, where R is the response to selec- tion, bl is the heritability of the trait and S is the selection differential on the trait (Freeman 65 Her- ron 2004; box 5.1). This model has been extremely useful because it offers a quantitative prediction of how much a population will respond as a function in traits to those within an epistatic network, are of two measurable features (heritability and selec- tion intensity), and is most effective when pheno- typic variation is due to many genes with. simple, additive effects. However, there are many different ways in which multiple genes can contribute to a phenotype that do not conform to simple, additive, quantita- tive generic models, and these may further compli- cate an estimation of evolutionary potential. One example is a threshold trait, in which quantitative variation at multiple loci may be phenotypically visible only when a certain quantity in expression is reached (Pulido 2007). Alternatively, complex genetic architectures may arise through epistatic interactions (in which one gene modifies the expres- sion of another) between the genes involved. Genes that operate within an epistatic network may be coadapted to function with other genes in the net- work, and expression in terms of timing and quan- tity of gene product will combine to produce a complex behavioral phenotype. Evolution of such phenorypes must proceed by mutual adjustments of multiple genetic factors. The evolution of all genes, from those contributing to additive variation further complicated if these genes have pleiotropic . effects on other aspects of fitness. The genes for- ager and period, known to have pleiotropic effects in Drosopbifa melanogaster, are discussed in a later section. Such complex genetic architectures can be the source of extensive constraint on behavioral evolution. Yet despite this importance of genetic architecture to trait evolution, our current under- standing of the genotype-phenotype relationship of quantitative traits is extremely limited, particu— larly in natural populations. What is clear is that the simplified views of both single gene control and additive quantitative genetics are frequently inap- ptopriate in the world of behavioral genetics. The diversity of genetic architectures of behav- ioral variation is illustrated nicely through the example of the genetics of animal migration. Within many species of winged insects there is variation in migratory behavior (Roff 6: Fairbairn 2007). in polymorphic populations, some individuals of a species undertake migratory flight, whereas oth- ers are sedentary. In some species, nonmigratory individuals are behaviorally, physiologically: and morphologically unable to fly because they lack fully developed wings and wing muscles to power flight. The genetic basis of this polymorphism difr fers in different insect species—both single gene BOX 5.1 A Brief Introduction to Quantitative Genetics )ason B. Wolfand Allen I. Moore Quantitative genetics provides a statistical description of the various influences on mea- surable traits (Falconer 5c Mackay 1996). We typically consider metric or quantitative (continuous) traits, but there can be exceptions. In behavior, most traits, or at least most influences on traits, are continuously distributed. Moreover; researchers that study behav- ior typically adopt a phenotypic approach, making quantitative genetics the most com- mon way to address how behavior is influenced by inheritance. In addition to its utility in providing a framework for understanding sources of trait variation, quantitative generics also provides a means of understanding phenotypic evolution, including factors such as the rate of, and constraints on, adaptive evolution. Thus, while” approaches such as opti- mality and game theory are useful for predicting expected outcomes of evolution, quanti- tative genetics provides insights into evolutionary processes and factors that limit optimal evolution or the attainment of an optimum. Using quantitative genetics to understand trait evolution involves defining the various factors that can influence a trait, and then to statistically partition the effects of these influ— ences. Influences can be very general or very specific. This is best expreSsed with linear equations. We begin with the most basic equation in quantitative genetics, which partitions the expression of a trait (2, which is the phenotypic value, or the trait value you measure on an individual) into a genetic (g, which reflects all of the genetic influences contributing to a trait) and an environmental component (2, the environmental deviation, which includes all of the environmental contributions influencing the expression of the trait): z=g+e (1) This equation describes all traits in all organisms because all trait variation must be ulti- mately attributable to genetic (there are no organisms that develop or live without genes) and environmental (there are no organisms living in a vacuum) influences. Therefore, all other models for partitioning trait variation are special cases of this simple equation. For example, one may wish to partition the genetic component into those genetic influences that are independent of all other influences (i.e., an additive component, a, which is also called the additive genetic value or breeding value; sec Arnold 1994a), and those influences that are dependent upon interaction between alleles at a locus (i.e., the dominance component, d, which is also called the dominance deviation) and between alleles at different loci, (the epistatic component, 1', sometimes called the interaction deviation or epistatic deviation): z=a+d+i+e (2) This partitioning is important because only additive generic influences are passed intact from generation to generation, which is why it is the variance of a that contributes to the evolutionary response to selection (this is the all important additive genetic variance) and to the resemblance of all types of relatives. Because the dominance and epistatic compo- nents depend on interactions with other alleles either at the same locus (dominance) or at other loci (epistasis), they are not heritable and do not contribute to the evolutionary response to selection. NevertheIeSS, the dominance and epistatic components in equation 2 are sometimes of interest because of their contribution to various evolutionary processes and the covariance between certain types of relatives. It is also possible to partition the environmental component into components attrib- utable to various environmental factors (e.g., one might want to attribute phenotypic (continued) 3'5 Box 5.1 (cant) variation to a particular environmental factor that is predictably shared by individuals in the population). In addition, because it is the additive genetic (breeding) value that is the heritable component of equation 2, it is common to write equation 2 as a partitioning of heritable and nonheritable components, z=a+€ where a now includes all nonheritable components contributing to trait expression, whether they be attributable to environmental or nonheritable genetic factors. We use this convention throughout. Equations 1-3 describe the influences on trait expression. However, it is populations, not individuals, that evolve and so we need to know how the identifiable genetic and environmental factors influence variation in a population. These same equations can be used to derive expressions that partition variation in trait expression into a set of vari- ance components. This partitioning is important for many reasons, the most significant of which is that it allows us to separate the heritable component(s) that contribute to trait evolution (the additive genetic variance) from those that do not. It also allows us to understand the contribution of components that contribute to the resemblance of relatives. For example, assuming that the terms are independent from each other, we can partition variation in the expression of the trait (i.e., the phenotypic variance, denoted P herein, also often denoted as VP) shown in equation 3 into additive genetic (denoted G herein, also often denoted as Va) and environmental (nonheritable) variances (denoted E herein, also often denoted as l:12): P=G+E (4) (Falconer 8c Mackay 1996). Note that we have switched from examining the compo- nents that influence the trait value of an individual (2) to describing components of trait variation in a population of such individuals. The additive genetic variance (G) term in equation 4 is the only heritable component of phenotypic variation, and it is, therefore, common to express the ratio of the additive genetic variance to the phenotypic variance (G/P), which is known better as narrow—sense heritability, 131 (see Falconer 6c Mackay 1996). Narrow-sense heritability expresses the proportion of variation in a trait among individuals that is attributable to the additive effects of alleles. Expressions similar to equation 4 could be derived for equations 1 and 2, and, assuming that the components are independent, one would simply separate variation in trait expression into a set of terms, each of which corresponds to the individual terms in the equation. The terms in, for example, equation 3 are defined as being independent, but this can be violated if there is a nonrandom association between genetic and environmental effects, the so—called genotype—environment correlation. The presence of a genotype-environment association (generally measured as a covariance, rather than as a correlation) adds a term to equation 4 that accounts for the association between the additive genetic (a) and envi- ronmental values (9): P=G+E+2cov{a,e) (5) A positive covariance between the additive genetic and environmental values inflates the phenotypic variance, whereas a negative covariance decreases the phenotypic vari- ance. A genotype-environment covariance also alters the estimation of genetic variances (continued) 76 empirically because it alters the resemblance of relatives. Genotype—environment covari- ances may arise when individuals “choose’1 or assort nonrandomly in specific environ- ments. Such covariances may be important for behavior (Falconer BC Mackay 1996) if, for example, individuals choose their soda! environment, but in general there is little empiri- cal work directed at detecting these in natural populations. It is also possible to include an interaction between the genetic terms and the envi- ronmental term, which is the so—called genotype-by—environment interaction (G x E, or sometimes GEI). The presence of G x E indicates that the effect of the genotype on the phenotype depends on the environment that the individual experiences and, likewise, that the effect of the environment on trait expression depends on the genotype of the individual (see chapter 6). It can also be thought of as a measure of the degree to which phenotypic plasticity varies with the genotype of the individual. The G x E is usually lumped into the environmental/nonheritable terms in equations 1—3 and will appear in the environmental term (E) in equation 4 because, like the dominance and epistasis terms, it is generally not heritable. The G x E can be partitioned into the interaction of various genetic values (a, d, or i) and the environment. G x E for behavior is not often investigated, but the evolution of almost all of the behaviors of interest to behavioral ecologists depends on G x E. For example, anytime a behavior is conditional on some cue, be it environmental or social, then this is a form of phenotypic plasticity (chapter 6), the evolution of which is dependent on the presence of and magnitude of G x E (Boake et al. 2002). The methods for obtaining these partitionings of phenotypic variances depend on the statistical tools of analysis of variance (hence the name, as R. A. Fisher invented ANOVA to allow the partitioning of different varianCe components leading to variation in phenotypes within a population) and regression (advanced by F. Galton to under- stand parent-offspring resemblance). Statistical approaches to determining quantita- tive genetic components typically require a breeding design in which relatedness among relatives is known—that is, parent-offspring, sibling, or other relatives created by a breeding program (Falconer 8C Mackay 1996). More recently, with the advent of cheap computing power, pedigree-based approaches are possible in which a mixture of related- nesses in unbalanced designs can be analyzed. This opens up the possibility of examining quantitative genetic influences in natural populations as long as pedigree information is available (Kruuk 2004). Understanding the relative importance of genetic, and especially heritable, variation and environmental or nonheritable variation is important for various problems of inter- est to behavioral ecologists. Many models of behavioral evolution make explicit genetic predictions, and so quantitative genetic approaches are required if one wishes to examine these problems theoretically or empirically. For example, for good genes models of indica- tor traits in Sexual selection to work in explaining mate choice for indirect genetic ben- efits (see chapter 24), male genetic quality must be heritable (in the narrow sense). Other examples include the fact that measures of genetic correlations provide insights into life history trade-offs, and limits to plasticity, and that understanding the heritability of traits provides clues as to the relative importance of environments (e.g., during learning or in parent-offspring interactions). In general, researchers interested in partitioning components of trait variation have an ultimate goal of understanding the evolutionary dynamics of trait means (either predicting them or reconstructing them). By separating the heritable from the nonheritable compo— nents of variation one can understand the relative evolutionary labiality of a trait because it is heritability that translates the change in trait mean within a generation (denoted s, the selection differential) caused by selection to an evolutionary change in the trait mean between generations (denoted as, to indicate a change in the mean phenotypic value, 2). This is expressed in the classic breeders’ equation that is central to quantitative genetics (see also chapter 3): (continued) 77 Box 5.1 (cont) A? = I115 (6) We can also view the additive variance as a measure of the (linear/additive) relationship between variation at the molecular level (allelic variation) and phenotypic variation. As such, it predicts how selection at the phenotypic level cascades down to produce changes at the molecular level, as in changes in allele frequencies that are generally regarded as the ultimate measure of evolution. This leads to an alternative view of the breeders' equa- tion (which is mathematically equivalent to equation 6), in which the linear relationship between a trait and fitness, given as a selection gradient (fl) is translated into a linear relationship between the genotype and fitness through the additive genetic variance, which then leads to the Change in the mean trait value: AE=G,B I (7) where B is equal to 5/!” (see chapter 3). _ The breeders’ equation (expressed in equations 6 and 7) can be used to understand the rate of adaptive evolution, which may be of interest to behavioral ecologists. However, because the breeders’ equation makes predictions only about the rate and not the direction of evolutionary response to selection (because'it always predicts that selection response is in the direction of selection at a rate defined by the heritability or additive genetic vari- ance), the breeders’ equation generally does not make predictions that conflict with the optimality models favored by behavioral ecologists. This contrasts with more complex quantitative genetic models concerned with multiple traits under selection, in which pre~ dictions of quantitative genetic theory may be at odds with predictions from optimality models. This suggests that behavioral ecologists should be concerned with multiple traits and adopt a multivariate view. We have provided only a brief introduction to univariate quantitative genetics, which is focused on the genetics and evolution of a single trait, whereas evolutionary quantitative genetics is ultimately a multivariate problem. Organisms are not a collection of independent traits. Instead, in almost any conceivable system, individuals express multiple traits that are generally correlated genetically and phenotypically with each other and rarely, if ever, have independent effects on fitness. When considering multiple traits, one would examine com- ponents contributing to variation in each as outlined above. However, because these traits may not be independent, we need to understand the genetic and environmental components contributing to the relationship between traits. That is, we need to understand how genetic and environmental sources of variation make multiple traits correlated with each other. The relationships between traits are of interest for various reasons, most notably because they link the evolution of one trait with selection acting on other traits. It is through this link— ing of the evolutionary dynamics of multiple traits that predictions of quantitative genetics theory can be at odds with that of optimality theory. This is because individual traits are not free to evolve to their optimal values, but rather, one must consider the evolution of sets of correlated traits wherein conflicting selection pressures (like trade-offs) may result in constraints on adaptive evolution (see Arnold 1994a). For example, if we consider a case of two correlated traits (which we will call trait 1 and trait 2.), we could express the phenotypic value of each of the two traits as in equation 3 and define components of variation for each as in equation 4, but if we wish to under- stand the evolution of either trait, we would have to add a term to account for the genetic relationship between traits to either expression for the breeders’ equation (equation 6 or 7). Taking the more traditional approach using equation 7, we would include a sec- ond term accounting for the genetic relationship between traits, measured by the additive (continued) 78 .11.‘lu7."-.s.--_I‘ :‘l Fl ji '3}. a a ":l .3 The Genetic Basis of Behavior genetic covariance (often standardized to be a genetic correlation) between the traits ( GP), to predict the evolutionary response to selection of trait 1 (A2,): 7 5E1 = Gllffll + G123: (Si where G,1 is the additive genetic variance of trait 1 and the subscripts on the betas indicate independent selection gradients on traits I ([3,) and 2 (Hz). The multivariate view of evolution provided by equation 8 implies that we cannot understand the evolution of a single trait in isolation and, as a result, each individual trait may not evolve in the direction predicted by selection or to the optimum predicted by an optimality model. It is ultimately the genetic relationship between traits, generally resulting from pleiotropic connections between them (owing to a shared genetic basis), that makes the viewpoint on evolution provided by quantitative genetics different from that provided by fitnessfselection-based approaches such as optimality theory that are cen- tral to behavioral ecology. Because it is unarguably biological reality that individuals are not collections of independent traits, but are collections of suites of traits linked through developmental and physiological processes, one cannot lose sight of this difference, even if one does not adopt a quantitative genetic perspective when studying behavior. Although the two-trait view of quantitative genetics is a step closer to reality, actual biological systems are likely to involve large numbers of correlated traits under com- plex patterns of linear and nonlinear selection. Genetic and phenotypic correlations between traits complicate selection and evolutionary dynamics in a variety of ways and, as a result, a reader wishing to eventually implement quantitative genetic techniques should ultimately become familiar with multivariate models. Arnold (1994a) provides a good introduction to multivariate quantitative genetics as well as a number of other important issues in quantitative genetics theory. Blows (2007; see also accompanying commentaries and response) also does a particularly good job of making it clear why it is critical to understand selection and genetics on sets of correlated traits and how one’s understanding of trait evolution can differ significantly from that expected from a single trait perspective. (typical for the Holometabola) and polygenic control (typical for hemimetabolous insects) have been documented (reviewed in Roff 5c Fairbairn 2007). Evolutionary trajectories and equilibrium frequencies of migratory and sedentary strategies have been described using single-locus and poly- gcnic genetic models. Furthermore, the idea that a threshold architecture underlies bird migratory behavior, thereby maintaining genetic variation in nonmigratory populations until threshold expres« 51011 is reached, may explain the rapid response to selection for migratory behavior (Pulido 2007). By this model, standing variation in genes underlying migratory behavior exists in the nonmigratory pop» Plafion below threshold values. Migratory behavior 1'8 phenotypically expressed only when recombina- tion brings together suitable allelic variants, and under selective conditions in which migration becomes advantageous, the population can respond quickly to this change in selective pressure because genetic variation has been maintained. Genomic Distribution ofLocr' The genomes of most, but not all, animal taxa are subdivided into sex chromosomes and autosomes, and there is a growing interest in the types of genes that tend to congregate on these two typm of chro- mosomes. Below we discuss two topics of research of particular relevance to behavioral ecologists: (1) genes underlying sex-specific traits and (2) genes involved in sexual selection. The location of sex-determining genes impacts the evolutionary potential of sex-specific traits. For example, many traits, including behaviors, affect the fitness of the two sexes differently. When a behavior BO Foundations increases the fitness of one sex but reduces the fitness of the other, it leads to sexual antagonism (Arnqvist 5C Rowe 2005). One way organisms may avoid such sexual conflicts is through sex-specific expression of genes (Elleg'ren EC Parsh 2.007). This allows both sexes to evolve their respective optimal trait val- ues (Lande 1980). Despite the fact that individuals of opposite sex possess almost identical genomes, sexual dimorphism (differences between the sexes in morphology and behavior) is widespread, implying that sex-specific gene expression is common. Sex- specific gene expression generally requires a physi- cal linkage between genes determining sex and the genes regulating sex-specific expression of sexually antagonistic traits. Consequently, genes initiating the cascade of developmental processes leading to sexual dimorphism are typically located on the sex chromosomes (Fisher 1930), for example, as is found in chicken (Kaiser BC Ellegren 2006). It is important to appreciate, however, that once the eas- cade is initiated, the regulated genes need not be on the sex chromosomes (e.g., in humans and Drum- phila most are distributed on autosomes; reviewed by Fairbairn ‘56 Roff 2006). Even in taxa that lack sex chromosomes, the evolution of new, autosomal, sex-determining genes are facilitated if they are physically linked to genes under sexually antagonis- tic selection (van Doorn 56 Kirkpatrick 2007). Besides being an important site for genes coding for sex—specific behaviors, the sex chromosomes are thought to be a key location for genes under sexual selection (box 5.2). Theoretical models suggest that evolution of ornaments in response to sexual selec— tion is most likely if genes coding for the ornaments and preferences for the ornaments are located on the sex chromosomes. In particular, Z-ljnkage of female preference (in taxa with female heterog- amety) is especially conducive to a Fisherian run- away, whereas X—linlcage of female preference (in taxa with male heterogarnety) coupled with auto— somal inheritance of male displays favors sexual selection based on “good genes” (Reeve Sc Pfennlng 2002; Albert 8: Otto 2005). Overall, the ZW sex chromosomal system in taxa with female heterog- amety (e.g., birds and Lepidoptera) is thought to be especially conducive to the evolution of elaborate male ornaments and courtship behaviors (Reeve EC Pfennig 2002.), and this is supported by compara- tive analyses of several taxonomic groups (Reeve 8C Pfennig 2002). However, significant correlations between sex-chromosome system and ornamentau tion are not ubiquitous (see Mank et al. 2.006), and this may reflect the facts that male ornamentation can evolve through processes other than female choice (e.g., male-male competition) and that genes underlying sexually selected traits are not always linked to sex chromosomes (Mank et al. 2006). Furthermore, male ornaments that are condition dependent are expected to be influenced by a large number of loci underlying general condition, and these are not expected to be located on sex chromo- somes (Rowe Be Houle 1996). In Drosophila (an XY sex-determination sys- tern) most genes underlying sexually selected traits also have pleiotropic effects on other aspects of fit- ness; such genes do not tend to occur more often than chance on the sex chromosomes (Fitzpatrick, 2004). These results find support in crickets (Shaw et al. 2007; Shaw Sc Lesnick 2009) but contrast with data from birds in which genes underlying mate choice and ornamentation are located on sex chromosomes (St-Ether et al. 2007). With the recent proliferation of information on the genomic distri- bution of ioai, it is becoming increasingly apparent that genes on the sex chromosomes do much more than simply determine sex. However, what should also be clear from the studies mentioned above is that the issue of the importance of sex chromosomes for sexual selection is far from resolved. Because the genomic location of loci has important implica- tions for sex-specific behaviors and behaviors tied to mate choice, it is a topic that will continue to be of key interest to behavioral ecologists. Effect Sizes of Alleles Since the modern synthesis, the nan-Darwinian view that species differences arise from small effect substitutions at multiple loci has dominated theo- retical treatments of adaptive evolution (e.g., Lande 1980). The possibility that allelic substitutions of large effect cause adaptive species differences has been suggested by recent experimental work using quantitative trait locus (QTL) mapping approaches (see Broman 2001 for an accessible review of QTL methods). Behavior obviously plays a central role in the adaptive evolutionary process, but very little is known about the distribution of allelic effect sizes underlying behavioral variation or their con- tributions to fitness. Recent empirical results show that the dichotomy of small versus large effect sizes underlying adaptive evolution is overly simple. Fur— thermore, theory suggests that the size distribution of phenotypic effects of allelic substitutions that Box 5.2 Diversity ofSex—Determining Mechanisms Daniel A. Warner and Fredric ). Janzen When a given trait affects the reproductive fitness of males and females differently, sexual selection should generate phenotypic divergence between the sexes. Indeed, difv ferences between the sexes are among the most spectacular sources of phenotypic varia— tion within populations. How this variation between the sexes, in terms of underlying genes, arises or is maintained is of primary interest to behavioral ecologists, and is often impacted by sex—linked genes. The remarkable diversity of sex-determining mechanisms (SDMs) in animals has important consequences for primary sex ratios and inheritance of sex-linked traits, and hence has fundamental implications for the evolution of behav- ioral variation. Whether an embryo develops into a male or female is the product of genes and the environment, and the degree to which these factors contribute to sex determination varies dramatically among taxa. Biologists have traditionally classified SDMs under two main categories: one” in which sex is determined entirely by genotypic factors (e.g., sex chro- mosomes) passed from parents to offspring (i.e., genotypic sex determination, GSD), and the other in which sex is determined by environmental factors during development (i.e., environmental sex determination, ESD). Even within these categories, we see remarkable diversity in patterns of sex determination, and elements of both genotypic and environ- mental influences on sexual differentiation in single species or populations are common (e.g., Kozielska et al. 2006; Radder et al. 2008). GSD typically involves sex chromosomes, which contain genes (e.g., SRY in mammals) that initiate the developmental cascade toward male or female gonadal development and further shape secondary sexual traits. In many raxa (e.g., all mammals, some reptiles, some amphibians, some fish, some invertebrates), males are the heterogametic sex con- taining both X and Y sex chromosomes, whereas females are homogametic (XX). The opposite pattern occurs in some other taxa (e.g., all birds, some reptiles, some amphibians, some fish, some invertebrates) in which females are heterogametic (ZW) and males are homogametic (22). The diversity of GSD mechanisms does” not stop here. In some inver- tebrates, sex determination depends on the ratio of the X chromosomes to autosomcs. In other systems, multiple sex-determining genes, epigenetic factors, or sex-determining factors on autosomes may influence offspring sex (Kozielska er al. 2006). Moreover, GSD systems can involve sexes differing in overall ploidy in the absence of heteromorphic sex chromosomes. This occurs in haplodiploid insects in which females arise from fertilized eggs (diploid), but males arise from unfertilized eggs (haploid). Under ESD, heteromorphic sex chromosomes are absent, and thus the sex of an individual cannot be predicted from its genotype. Instead, sex is determined by envi- ronmental factors during development. This mechanism evolved multiple times inde- pendently in diverse taxa, and the environmental variables involved vary broadly (e.g., temperature, photoperiod, salinity, maternal nutrition, mate availability). However, the influence of such environmental factors is not independent of genes. That is, environ- mental conditions interact with genes that influence molecular pathways involved in sex determination. Perhaps BSD is most widely recognized in the form of temperature- dependent sex determination (TSD), which has been well studied in reptiles, but it also occurs in a diversity of other taxa (e.g., fish, invertebrates). TSD exhibits an assortment of patterns and varies among taxonomic levels (figure I). For example, the sensitivity of sex determination to temperature, and even the direction of the temperature effect, varies among species, populations, and even individuals within populations (e.g., Ewart et al. 2005). (continued) BOX 5.2 front.) (3) Females produced at one extreme Some turtles —__........_> uatara Some lizards lb) Females produced at both extremes Variation In pivotal Variliom in slope temperatures ol reaclion norm Sex ratio (‘1’: male) Geographic variation in reaction norms of Ottawa serpentine Crocodilians Some lizards Some turtles Some turfles Some leards ..-.—..—..._._.....9 Sex rello (it. male) -fl5fl mm (c) Both sexes produced at all temperatures Extreme incubation temperatures override genotypic sex determination 75‘ E fliesilzards 212W female heterogamely 3'3 Some turtles x Pagans vimpr ‘3 —-W ——> 3 xxrxv male hstarngamety x % Qanotypic sex determination Easy-am dummy; (D Figure I Diversity of sex-determining patterns in reptiles. All graphs show sex ratio ("A male) as a function of increasing egg incubation temperature (it-axes). The three major patterns of sex determination with respect to incubation temperature are shown to the left of the arrows. Patterns to the right of the arrows are variants of those patterns. (21) Patterns of TSD in which males and females are produced at opposite temperature extremes. Pivotal temperature (i.e., temperature that produces 1:1 sex ratio) varies considerably among species, populations, and individuals. Additionally, considerable diversity occurs in the transitional range of temperatures (i.e., range of temperatures that yield mixed sex ratios) among species. (b) Pattern of TSD in which females are produced at both temperature extremes, and males at intermediate temperatures. In some species, intermediate temperatures produce mixed-sex ratios, and other species show geographic variation in shapes of reaction norms (tag, the common snapping turtle, Chelydm serpentine; Ewe-rt et a]. 2005). (c) Sex ratio is not influenced by incubation temperature (genotypic sex determination). Recent studies on two distantly related Australian lizards with heteromorphic sex chromosomes demonstrate that extreme incubation temperatures can override genotypic sex determination. Such temperature effects reverse genotypic females to phenotypic males (Radder et a1. 2008) and vice versa (Quinn ct al. 2007). Differences in patterns of inheritance and influences on primary sex ratiOS between GSD and ESD generate fundamentally different consequences for behavioral variation. For example, if genes responsible for sexually selected traits reside on sex chromo- somes, then sexual dimorphism is probable (Rice 1984). Whether alleles are linked to the Y or W chromosomes influences the inheritance of traits in sons or daughters, respectively (Reeve 8c Pfennig 2002). Expression of these sex-linked genes generates (continued) 82 The Genetic Basis of Behavior sexual dimorphism, including sex-specific behaviors that are fundamental to fitness in each sex. 011 the other hand, pure ESD lacks sex chromosomes,'so such inheritance is unlikely, yet sexual dimorphism is common in ESD species. In these systems, environ- mental factors trigger expression of sex«detcrmiuing genes or genes controlling sex ste- roid hormones that elicit sex-specific phenotypes. Under these situations, mismatches sometimes occur between gonadal sex and behavioral sex (c.g., Gutzkc dc Crews 1988), which likely reduces individual fitness. As a consequence, selection should favor an ESD system that enables each sex to develop in its respective optimal environment (e.g., Warner 5C Shine 2003). Under certain conditions, overproducing one sex enhances maternal fitness, and SDMs can influence the potential for adaptive maternal control over offspring sex ratios. The random nature of chromosome segregation during meiosis could constrain sex ratio bias in GSD systems, thereby limiting sex allocation patterns. However, a substantial liter- ature suggests otherwise, because many species with heterornorphic sex chromosomes skew offspring sex ratios in a presumably adaptive direction (reviewed in Cockburn et al. 2002; see also box 26.3). Mechanisms that enable sex ratio biases under GSD are largely unknown, but multiple hypotheses have been proposed (Pike 5c Ferris 2003). On the other hand, haplodlploidy provides females with considerable control over brood sex ratios, which renders this SDM especially conducive to maternal sex ratio adjustment. Indeed, empirical work on these systems strongly conforms to predictions from sex allocation theory (reviewed in Odo 6c Hunter 2002). In contrast, species with ESD can influence offspring sex ratios via oviposition behaviors. For example, selection should favor genes that enable females to detect, and subsequently select, nest sites that yield desirable clutch sex ratios under the prevailing conditions. Indeed, numerous studies of TSD reptiles show that nesting behavior is nonrandom with respect to temperature, and that this behavior is repeatable and heritable (Janzen 5C Morjan 2001). The impressive diversity of SDMs has critical evolutionary consequences for behavioral variation. This brief overview reflects the traditional view of SDMs by focusing on GSD versus ESD systems. However, recent evidence suggesm that both G51) and ESD might co—occur within populations (Quinn et al. 2007; Raddcr at al. 2003). Such complexity, should it be more than transient, would have manifold ramifications for behavioral evolu- tion. Investigating the proximate causes of SDMs and their ultimate outcomes will likely reveal greater diversity among taxa and will provide critical insights into ecological and evolutionary mechanisms driving behavioral variation. 33 C move natural populations toward adaptive targets declines exponentially (Orr 2005). These issues are important to the question of whether population evolution is mutation limited (i.e., selection pres- sures cxist but populations are unable to respond ' due to limited genetic variation) or selection limited 0-6., ample genetic variation exists but populations do not evolve substantially due to the absence of 1 selective forces), and the potential for gradual vor- Sus punctuated change in behavior. One area relevant to behavioral ecology that has been studied with these questions in mind is _. the genetic basis of song differences between closely related populations and species of insects. Many Species of insects communicate acoustically in a variety of contexts. One of the most conspicuous contexts is male-female communication, in which males (and sometimes females) sing to attract pro- spective mates, a functional context suggesting the action of sexual selection on mate attraction. Such songs are frequently quite stereotyped in that the variability within species is usually much less than that between species. In cases in which it is possible to cross different song variants (whether within or between species), we have the opportunity to esti- mate the size of allelic effects that contribute to variation in acoustic behavior, affording tests of and potential insights into the past action of sexual selection (e.g., Shaw et al. 2007; Shaw 8c Lcsnicl: 2009i g4 Foundations In a recent review, Gleason (2.005) summarizes details of genetic architecture from quantita— tive genetic (and QTL) studies of courtship song in Drosopbiln, noting that a range of estimated effect sizes from small to large underlie the inter— specific differences of various traits of the Draw- philo courtship songs. However, large effect size estimates must be interpreted with caution for at least two reasons. First, effect sizes can be over- estimated when sample sizes are low (Broman 2.001). Second, without identifying the molecu- lar cause for the behavioral difference, it is dif- ficult to know whether one or many sequential substitutions are responsible for the difference. In a recent groundbreaking study on the evolution and development of trichome pattern, McGregor er al. (2007) found that a large difference in pat- tern between D. melanogaster and D. simulans, previously attributed to a single gene (shaver:- baby), is due to three independent changes in that gene. Thus, three separate allelic substitutions have accumulated to produce the large difference, a change originally thought to be due to a single substitution of large effect. With estimates of small effect, one is on some- what more stable ground. Finding small effect sizes is significant because such estimates are unlikely to be biased downward; small estimates that are potentially biased upward are still small. Shaw et a1. (2007) found that multiple QTL underlie dif- ferences in pulse rate of species of the Hawaiian cricket genus Laupala. Effect sizes were generally small and at the lower limit of the power to detect them based on the sample size of the study. Thus, in both the Drosophila studies and the Laupala study, at least some effect size estimates for the differences between species are of similar magni- tude to the variation found within species, sug- gesting that standing variation could be the source of interspecific differences (i.e., the evolution of behavior in populations is not mutation limited). Interestingly, in the case of Drosopbila, genes for song differences identified through mutation stud- ies, and genes implicated through QTL studies of song in natural populations are not the same, by and large (Gleason 2005). Despite nearly 20 candidate genes identified for some involvement in courtship song through mutation studies, we are still ignorant about what genes contribute to natural variation in courtship song in this model organism. Correlated Evolution In animal signaling systems, such as in sexual sig- naling, behaviors frequently evolve in a correlated fashion. Under most models of sexual selection, the evolution of sexual signals requires a concomi- tant evolution of response to that signal. As a con— sequence, coevolving traits such as male signals and female preferences for those signals often are hypothesized to show genetic correlations within populations (i.e., a presence of allele A at one locus corresponds to a higher than random chance that the individual also contains allele B at a sec- ond locus). Genetic correlations may arise either through pleiotropy or linkage disequilibrium (due either to physical linkage or nonrandom associa- tion of alleles at unlinked loci). For example, link- age disequilibrium is expected to arise through assortative mating whenever there is genetic varia- tion for a male trait and a female preference for that trait (Lande 1981), owing to the fact that offspring inherit both maternally derived genes for the prefer- ence and paternally derived genes for the preferred trait. This linkage disequilibrium is the basis for Fisher’s (1930) runaway hypothesis of sexual selec- tion, in which females preferring ornamented males produce sons with improved success at attracting mates that likewise carry the preference genes for those traits. Thus, enhanced mating success of these sons incidentally increases the frequency of the preference genes that they carry. In practice, assessing genetic correlations between traits is anal- ogous to measuring heritability of a single trait, and is carried out through comparing phenotypes of relatives. To detect genetic correlations caused by assortative mating and linkage disequilibrium, it is important that individuals are allowed to exercise mate choice. Conversely, to test for physical linkage or pleiotropy, random mating should be enforced for a number of generations prior to assessment. Whether we should expect physical linkage between loci underlying male-female processes such as sexual selection is currently unclear. In some cases, physical linkage may facilitate coevolution, when new mutations influence a trait in a favorable direction (e.g., a mutation for large ornaments in a locus tightly linked with another containing alleles for preferring large ornaments). However, equally likely is the reverse scenario, in which the new mutation reduces ornament size—a mutation that is unlikely to spread if positioned near a gene for What is clear is that there are a number of spe— cific cases in which physical linkage between genes gene flow. The main force preventing the divergence ‘ of two populations that continue to exchange genes ' Between loci affecting male traits and female pref- erences (e.g., when these loci are located near one 'nother on a given chromosome). Given the current ontention concerning the pervasiveness of sympat» speciation in nature, substantial insight may e gained by asking whether the necessary condi— _'ons (e.g., physical linkage between male traits and male preferences) are widespread. GENETICS WITH THE GENES 1-} lots may have arisen from simpler origins (Robinson et al. 2005). Knowledge of the specific genes partic- '_'ipatiug in behavioral variation would enable tests L" to determine the sources of that variation, such as '_ single gene versus complex genetic architectures, "and the modes of gene action and interaction. Ulti- - mately, knowledge of the genes enables insights _ into, and estimates of, behavioral variation that 3 can be acted upon by selection. In a few notable examples, the genes underlying behavioral varia~ _ tion have been discovered, but this accomplishment is relatively rare and there remains much potential along this avenue of research. Approaches and strat- '- figies to identifying genes are discussed more fully in chapter 28. Here, we discuss a selection of topics about which we might make significant strides with The Genetic Basis of Behavior 85 specific knowledge of the genes involved, providing examples when they exist. The Genetic Basis of Sexual Signals and Preferences As mentioned earlier, linkage of genes underlying- cocvolving sexual traits may be ascertained through assessing genetic correlations . between the traits, without knowledge of the actual genes involved. However, with knowledge of the genes, we can further discern whether genetic correlations arise through linkage disequilibrium, physical linkage, or pleiotropy. The last scenario asserts that the same genes underlie multiple components of mate choice. Models of sexual selection normally assume sepa- rate loci code for male traits and female preferences, although this need not be the case. It is possible that similar neural mechanisms control signal production and reception, a situation termed genetic coupling. The desafl gene in Drosopbila melanogdster is one such example, in which a known inversion contrib- utes both to the production of pheromones used In sexual signaling and to the detection of these phero- mones (Marcillac et al. 2005). Genetic coupling has also been invoked to explain tight linkage betwer :1 female wing coloration and male preferences for this coloration in Heliconins butterflies (Kronforst et al. 2006), and song and acoustic preference in Leupold crickets (Shaw 55 Lesnick 2009). However, in other systems in which genetic coupling was initially sus- pected, more detailed assessment revealed separate neural pathways for signal and preference (reviewed by Butlin 8c Ritchie 1989). Butlin and Ritchie (1989) argue that separate neural pathways may be similarly affected by common genes, and thus such a finding is not inconsistent with generic coupling. Due to the incipient stage of such research, we currently have little idea how widESpread such situations may be, or the evolutionary implications of such genetic cou- pling. Pleiotropy is often considered a hindrance to evolution, constraining certain traits from reaching their optimum values because fitness of the underly- ing gene depends on the sum of its effects across all traits (Fitzpatrick 2004]. In contrast, when coevolv- ing traits such as those involved in communication between the sexes have similar underlying genetic bases, evolution may proceed more rapidly. Knowl- edge about the genes affecting male traits and female preferences will allow for more informed theoretical models of evolutionary processes. 86 Foundations The Genetic'Basis ofSocial Behavior That social behavior has a biological basis and can evolve has fueled further study into the genetic basis of the constituent behavioral components. In recent years, genomic approaches have enabled much program in understanding the genetic and molecu- lar bases of social behavior (Robinson et al. 2.005). A wide taxonomic diversity of organisms exhibit social behavior, in both simple and complex forms. One of the most common ways that organ- isms engage iti social behavior is in the context of sexual reproduction. It could be argued that all sexual organisms are social in this sense, because some coordination betvveen male and female is necessary to achieve fertilization. Although many species have promiscuous mating systems, some species have evolved toward pairvbonding and monogamy, extending social interactions beyond single acts of sexual reproduction. A well-studied example of this occurs in the prairie vole (Micrones— ocbrogaster), a species in which monogamy is influ- enced by the neuroendocrine gene vasopressin 1a receptor {V1 aR; Young 6:: Hammock 2007). VI aR expression was significantly higher in the ventral forebrain of the monogamous M. ochrogaster than in its close relative, the promiscuous meadow vole, M. pennsylyanicus (Insel et al. 1994). In an excit— ing study in which VIaR levels were experimen- tally manipulated, Lim et al. {2004) showed that social behavior could be dramatically influenced by increasing die levels of this gene expressed in the ventral forebrain. These results provide an oppor- tunity to associate naturally occurring behavioral variation with molecular causes and potential insights into the molecular mechanisms that medi- ate observed variation in the level of male prairie vole pair~bonding {Ophir et al. 2008). The evolution of social behavior may well be caused, initially, by selective pressures to deal with the demands of reproduction. The communal rear- ing of offsPring occurs in a variety of animals from arthropods to birds and mammals, illustrafing'one way in which the evolution of social life has become more elaborate. The social hymenopteta provide one of the best models for understanding the com- plexities of social life and its genetic underpinnings, largely due to the intense study of the honeybee, Apis melifem. In the honeybee, a conspicuous part of social life involves foraging for the hive, but this behavior shows age-related expression. Young bees typically stay in the hive performing nurse activities to developing embryos, transitioning to foraging only later in life (Seeley 1995). Noticing the similar- ity between the behavioral effects of the gene for-agar (for) in D. melanogaster and the behavioral transi- tion experienced by aging honeybeEs, Ben-Shahar ct al. (2002.) examined gene expression of Amfor (for’s homolog in honeybee). A striking correlation between age (and worker task) and Amfor expres« sion suggested that the onset of foraging behavior involves the expression of a cyclic GIVE-dependent protein kinase, the gene product of for {Osborne er al. 1997). Based on this and other examples, Rob— inson et al. (2005 ) assert that an emerging theme in the genetics of social evolution is that the genes used by solitary organisms to perform vital biological functions are also used in similar contexts in social organisms, albeit with evolutionary “retooling.” Pleiotropy and Behavioral Evolution With the identification of actual genes underlying behavioral variation comes the potential to exam- ine the degree to which pleiorropic gene action constrains or promotes the evolution of behavior. Pleiotropic genes are genes that affect multiple phe- notypes (Hall 1994) and so evolution in one trait can have additional consequences in some other vital area of the biology of the organism (Green- span 2004). Thus, pleioo'opy at a locus can both constrain and enhance evolution above and beyond a locus devoted to a single phenotype. For example, pleiotropy can constrain evolution if a new muta- tion that increases fitness through its effect on one trait simultaneously decreases fimess through its effect on the pleiotropic trait. Given that most new mutations are deleterious, such fitness trade-offs in pleiotropic genes are more likely than a mutation that results in fitness gains for both traits. In a recent review summarizing evidence for this idea, Fitzpat- rick (2004) discussed (putatively) sexually selected genes in D. melanogaster and their pleiotropic effects on nonsexually selected phenotypes, building a powerful case for constrained evolution on these genes beyond the force of sexual selection. Indeed, Greenspan (2004) suggests that pleiorropic action of genes affecting behavior may be widespread. Although it is perhaps too soon to say this is dem- onstrated, two notable examples of the pleiotropic effects of behavioral genes support the idea. Perhaps not surprisingly, these cases come from the genetic model D. melanogaster, albeit through studies of natural variation. Before we discuss thme in more detail, we stress that more examples illustrating .fimctional consequences of natural allelic variation at behavioral loci are needed to critically evaluate the role of pleiottopy in behavioral evolution. Extensive work has been conducted on the genetic basis of circadian rhythms, perhaps because daily clocks are so pervasive among life forms and because early work identified circadian mutants in D. melanogaster that shortened, lengthened, or abolished altogether, the circadian period (see Tau- ber 6t Kyriacou 2008 for a recent review). These circadian period mutants were subsequently cloned and characterized and were shown to be allelic variants at the same locus, named period (per). Of interest to behavioral ecologists, researchers have since found allelic variants of per in European, African, and Australian populations of D. melano- gaster, stemming from variable numbers of repeats in the threonine-glycine repeat region that is char- _ acteristic of this gene (see Kyriacou et al. 2008 for a recent review). In Europe and Northern Africa, major allelic variants that differ in their ability to buffer the expression of the circadian period from fluctuating temperatures are distributed in a latitu- . dinal cline. The allele with the most robust temper- ature compensation is found at highest frequency in high latitudes and greatest temperature variability, ‘ making a strong case for the action of natural selec- ' tion in maintaining this variability. In addition to having an effect on locomotor _ and eclosion circadian rhythms, the per mutants also affect a short-period rhythm in Drosopbila, the courtship song. This short-period, behavioral effect parallels that of circadian rhythm, with short and long period mutants causing a shortening and lengthening the interpulse interval of the courtship song. The role of per in song variation among these mutants was solidified in an exciting transgenic study. Mutant D. melanognster with arrhythmic song were transformed with the per homolog from D. simulrms, restoring rhythmic song in these flies with the periodicity of D. simulans (Wheeler er al. 1991). Since these pioneering studies, many addi— tional song genes have been identified, document- ing the effect of additional loci on courtship song Variation due to induced mutations in Drosopbila (See Gleason 2005 for a review). Many of these are Similarly pleiotropic in their effects. The gene forager (for), dlSCUSSEd above in rela- tion to social behavior, has also been shown to The Genetic Basis of Behavior 37 exhibit pleiotropic effects in D. melmzogasrer. ini— tial behavioral observations on flies collected from orchard populations identified twu forms of larval behavior, the sitter and rover phenotypes, which exhibit relatively short and long food trails, respec— tively (Soltolowski 1980). Molecular investigations revealed that for codes for a cyclic wadependent protein lcinase (PKG) and that forR is responsible For higher levels of PKG activity { Osborne et al. 1997). Thus the behavioral variation traces its cause to a single gene, and evidence suggests that variation in the system is maintained by frequency dependent selection (Fitzpatrick er al. 2007). Recent work has now identified a second behavioral phenotype affected by this polymorphism. Rover flies show superior short-term learning and poorer long-term memory of odors compared to sitter flies (Mety et al. 2007), and localized PKG expression to the mushroom bodies, suggesting a role for for in olfa c- toty learning. These intriguing results raise the pos- sibility of an adaptively marched, functional link between for’s role in feeding locomotion and a role in olfactory learning of a food source. If true, this would represent a case in which pleiotropy might facilitate adaptive evolution. The Genetic Basis ofthe Mutation/Selection Balance One of the basic results of population genetics theory is that heritable (additive) genetic varia- tion underlying phenotypic targets of directional selection should be eliminated as advantageous alleles increase in frequency and populations evolve {Freeman 3: Herron 2004; chapter 3 of this vol- ume). In general, we can predict the frequency of an allele in a given future generation (11') based on the frequency of the allele in the previous genera- tion (17; for a 1 locus, 2 allele trait) when we know the fitness of the three genotypes in the population. An advantageous allele will increase in frequency according to its current frequency in the population and its fitness in heterozygotic and homozygotic states relative to the mean fitness of the population according to the following equation: I" = (102%. + crew...) I (122 W... +2qu.. + qZWM) in which p’ is the predicted frequency of allele A in the next generation, 33 and q are the current fre- quencies of alleles A and a, respectively, and WM, WM, and W“ are the relative fitnesses of the three 33 Fou ndations possible genotypes. Note that the denominator equates to the average fitness of the population (see Freeman SC Herron 2004: 193, for a basic treat- ment). If the fitness of A is higher than average when it is matched with itself and the alternative allele, we expect its frequency to rise. It might therefore be puzzling why lower fitness phenotypes persiSt in natural populations. One such example is that of behaviors associated with human mental disorders. For disorders that demonstrably affect the fimms of their bearers, such as schizo- phrenia, Keller and Miller (2006) argue that allelic variants causing the disorder persist due to a “muta- tion—selection" balance. Under the mutation-selec- tion balance, selection removes deleterious alleles that contribute to lowering fimess, but allelic varia« tion is nonetheless maintained by continual renewal through mutation. Thus the frequency of such del- eterious alleles is expected to be a function of the mutation rate and the severity of selection removing such alleles from the population. Keller and Miller (2006) favor this evolutionary explanation over alternative hypotheses that might cause the main- tenance of genetic variation, such as the persistence of ancestral polymorphisms under nonequilibrium conditions, or a past selective environment in which such traits were neutral or even favored. Knowledge of the specific genes that lower fitness can lead to estimates of mutation rates and selec- tion intensities against deleterious alleles, which in turn can facilitate predictions of the expected fre— quency of the behavioral phenotype. The mutation- selection balance is most likely to be a significant contributor to the maintenance of genetic varia- tion when traits have a quantitative genetic basis, because the potential source of genetic variation is complex. For example, in the case of schizophre- nia, many candidate genes have been suggested to be involved in the occurrence of this disorder (Sullivan 2008), in keeping with the hypothesis that the genetic underpinnings of variation in the popu— lation are quantitative. Molecular Tests and Mechanisms of Natural Selection Knowing the genes that affect behavioral compo- nents of fitness can enhance the power to detect selection, especially at the molecular level. A topic on which great progress has been made in the last decade is the function and evolution of reproduc- tive proteins in Drosopbila and other animals (see also chapter 23). For example, male accessory gland products have been shown to directly affect aspects of female reproductive behavior. In particular, the transfer of accessory gland proteins (Acp’s) from male to female has been shown to cause behavioral changes between virgin and mated females. Com- pared to virgins, mated Drosopbila melanogaster females show lowered receptivity to remating, have increased egg production, increased rates of ovula- tion and oviposition, and an increased rate of food intake (reviewed in Ram 8C Wolfner 2007). In some cases, specific differences in behavior have been traced to the presence of specific acces- sory gland proteins. The accessory gland protein known as sex peptide (SP) contributes via oogenesis to the elevated egg—laying rate in mated females and subsequent latency to rcmating (reviewed in Ram 85 Wolfner 2007). Interestingly, a sex peptide recep» tor (SPR) has recently been identified (Yapici er al. 2003) that plays a role in mediating the behavioral effects of SP in females. Although SP (and SPR) is apparently highly conserved across Drosophila, other Acp’s such as the prohormone ovulin have evolved rapidly (Haerty er al. 2007). Ovulin stimu- lates ovulation in mated female D. malanogaster (Herndon 8:: Wolfner 1995) and shows a high rate of amino acid substitution indicative of positive selection (Tsaur et al. 1993; see Jensen et al. 2007 for a recent review of molecular tests of selection). The Genetic Basis of Convergent versus Parallel Evolution The study of evolution by natural selection can be approached from many other angles in addition to the molecular investigations discussed above. One of these is the study of convergent evolution, in which similar phenotypes have evolved repeatedly in the history of a group. Frequently it is claimed that con- vergent evolution is a signature of common selection pressures in different organisms because the same solution to a biological “problem” has been reached independently on multiple occasions. Observations of repeated outcomes in behavioral states are rela- tively easy to observa, but beg the question as to whether they are due to the same (parallel evolution) or different (convergent evolution) genes. Although both parallel and convergent evolution may signal the action of selection causing evolutionary outcomes, parallel evolution may also signal genetic constraint, in other words, that there are evolutionary limits to the phenotypic options available to a group of closer :4 as; glared organisms (Arendt Sc Reznick 2008). Only if mm knowledge of the changes in the genes involved dim we be able to conClusively distinguish between arallel and convergent evolution. ' It is certainly the case that many genetic pathways 6 phenotypic change are possible (e.g., as appears to 'e the case for Drosophfla courtship song; Gleason 005). Recently, however, extensive evidence has ccumulated demonstrating parallel evolution due 0 single genes. Particularly intriguing is the case of elanocortin-l receptor (Mair), which affects pig- mentation in the mouse (Mus musculus) and appar- endy a variety of other animals (Fitzpatrick et al. 005; Arendt 5C Reznick 2.008). In both the lesser now goose (Amer c. camlescens) and arctic skuas Stercorarius parasitism), McIr correlates perfectly with the melanin polymorphisms affecting mare hoice (Mandy et al. 2004), providing a rare link etween genes and mating'behavior in the wild. YNTH ESlS AND CONCLUSIONS sBoake ct al. (2002; see also Andersson 8C Simmons 32006) contrast the complementary benefits of top- . clown and bottom-up genetic approaches to study -.the evolution of behavior. A top-down approach _--emphasizes the genetic causes of the evolutionary process through the study of phenotypic patterns. To the extent that we can describe accurately the §:_"genetic architecture of behavioral traits, we will suc- Ceed in revealing the role that genetic architecture 5 plays in the force of evolution. In other words, how '- does genetic architecture of behavior constrain or enable evolutionary change, in the context of par- * ticular processes? In a bottom-up approach, the -_ effects of the genes underlying trait variation are ' under study, with the goal of elucidating the genetic mechanisms governing the expression of behavior or behavioral variation. This approach is vital to our understanding of behavioral evolution as well, iu'part because it is needed to identify genes contrib» uting to the development and expression of behav- ior, at the genie, physiological, and behavioral levels. Not only will we gain a specific understanding of the genes underlying behavior, but we will also be able to assess the degree to which genes are evo- lutlonariiy cohopted from other roles. Although the top-dowu role ofgenetics in behavioral ecology may seem more directly relevant to behavioral ecology today, the integration of bottom-up strategies will ultimately fold back into our understanding of the The Genetic Basis of Behavior 89 process of behavioral evolution. The genetic basis of behavior is a vital component to understanding the evolutionary process of behavioral change. SUGGESTIONS FOR FURTHER READING For the importance of distinguishing between genetic and environmental influences on behavior, see Mackay and Anholt (2.007). Grafen (1984) dis- cusses the traditional perspective on genetics and behavior in behavioral ecology, and Moore and Boake (1994) discuss alternative contributions from evolutionary genetics. For a comparison of different genetic traditions in the study of beha v- ioral genetics see Greenspan (2004). Recently, a synergy can be detected between the fields of behav— ioral genetics and behavioral ecology as techniques become more accessible in the study of naturally occurring behavior (e.g., Boake et al. 2002; Robin— son et al. 2005). Although no definitive text exists, the insights to be gained from genetic approaches to behavior, behavioral ecology, and behavioral evolution are apparent from a Wide range of studies (Boake et al. 2002; Fitzpatrick et al. 2005). Boake CRB, Arnold S], Breden F, Meffert LM, Ritchie MG, Taylor E], Wolf )3, SC Moore A] (2002) Genetic tools for studying adaptation and the evolution of behavior. Am Nat 160: 5143—5159. ' Fitzpatrick M], Ben-Shahar Y, Smid, HM, Vet LEM, Robinson, GE, 85 Sokolowski MB (2005) Candidate genes for behavioural ecol- ogy. Trends Ecol Evol 20: 96—104. Grafen A (1984) Natural selection, kin selection, and group selection. Pp 62—84 in Krebs JR SC Davies NB (eds) Behavioral Ecology: An Evolutionary Approach. Sinauer Associates, Sunderland, MA. Greenspan R] (2004) Quantitative and single—gene perspectives on the study of behavior. Annu Rev Neurosci 27: 79—105. Mackay TFC BC Anholt RRH (2007) Ain’t mis- behavin? Genotype-environment interactions and the genetics of behavior. Trends in Genet— ics 23: 311—314. Moore A] 8C Boake CRB (1994) Optimality and evolutionary genetics: complementary procedures for evolutionary analysis in behavioral ecology. Trends Ecol Evol 9: 69—72. Robinson GE, Groainger CM, 86 Whitfield CW (2005) Sociogenon-u'cs: social life in molecular terms. Nat Rev Genet 6: 257—270. ...
View Full Document

This note was uploaded on 08/27/2011 for the course BIONB 2210 taught by Professor Seeley during the Fall '10 term at Cornell.

Page1 / 10

L04-shaw_wiley_2010 - Shaw K L 8 Wiley C(2010 The genelic...

This preview shows document pages 1 - 10. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online