Reproductive competition and sexual selection

Abstract

This paper traces the development of our understanding of the development of different approaches to estimating the strength of reproductive competition and sexual selection in the two sexes, based on measures of the operational sex ratio, the opportunity for sexual selection and contrasts in selection gradients between the sexes. It argues that different approaches provide complementary insights into the causes of sex differences in reproductive competition, the operation of sexual selection and the evolution of secondary sexual characters and that improvements in our understanding of the evolution of secondary sexual characters will require a more comprehensive understanding of the ways in which social and ecological conditions modify reproductive competition and development in females and males.

This article is part of the themed issue ‘Adult sex ratios and reproductive decisions: a critical re-examination of sex differences in human and animal societies’.

1. Introduction

The origin of sex differences in morphology and behaviour between the sexes has intrigued philosophers and scientists for more than a thousand years [1] but, like so many other important biological insights, the first recognition of a connection between the evolution of sex roles, sexual competition and sexual dimorphism can be traced back to the first half of the nineteenth century. Over 30 years before Charles Darwin published the The Descent of Man [2], John Hunter, the eminent surgeon, anatomist and classifier of monsters had argued that differences between the sexes were of two kinds: those involving the sexual organs themselves, which were evident from birth and did not change during an individual's lifetime, and those that only developed when individuals approached breeding age (such as differences in body size, plumage and the tendency to be fat) [2,3]. He termed the latter ‘secondary’ marks or characters of sex because they appeared after the primary sex differences [3] and argued that ‘secondary’ sexual characters were functionally related to fighting or display and that their extent varied with ecology.

The males of almost every class of animals are probably disposed to fight, being, as I have observed, stronger than the females; and many of these are parts destined solely for that purpose, as the spurs of the cock, and the horns of the bull …One of the most general marks (of sex) is the superior strength of make in the male; and another circumstance, perhaps equally so, is this strength being directed to one part more than another, which parts (sic) is that most immediately employed in fighting. This difference in external form is more particularly remarkable in the animals whose females are of a peaceable nature, as are the greatest number of those which feed on vegetables, and the marks to discriminate the sexes are in them very numerous. [2]

In The Origin of Species [4], Darwin was principally concerned with explaining the evolution of traits that increased the survival of individuals. However, he recognized that there was a wide variety of traits that were usually more highly developed in males and were unlikely to contribute to survival and one of the primary aims of the The Descent of Man [5] was to provide an explanation of the evolution of ‘secondary’ sexual characters—which he identified as those affecting reproductive competition or breeding success rather than the ‘act of reproduction’ itself. He distinguished sexual selection from natural selection on the basis of its operation through reproductive competition and suggested that this can take two separate forms:

Sexual selection depends on the success of certain individuals over others of the same sex in relation to the propagation of the species; whilst natural selection depends on the success of both sexes, at all ages, in relation to the general conditions of life. The sexual struggle is of two kinds; in the one it is between the individuals of the same sex, generally the males, in order to drive away or kill their rivals, the females remaining passive; whilst in the other, the struggle is likewise between the individuals of the same sex, in order to excite or charm those of the opposite sex, generally the females, which no longer remain passive, but select the more agreeable partners. (Chapter XXI; page 614).

He goes on to describe how males tend to compete more intensely for breeding opportunities and mating partners than females, while females are generally more selective of breeding partners than males. Much the same distinction between natural and sexual selection is still used today, though sexual selection is now commonly restricted to selection operating through differential success in mating and fertilisation [6–8].

Darwin appreciated that intense reproductive competition between males (and the secondary sexual characters associated with it) was connected to the evolution of polygynous breeding systems and that biases in the adult sex ratio (ASR) in favour of males would be likely to increase reproductive competition between males and to strengthen selection for traits associated with competitive activities in males to a greater extent than in females:

That some relation exists between polygamy and the development of secondary sexual characters, appears nearly certain; and this supports the view that a numerical preponderance of males would be eminently favourable to the action of sexual selection. (Chapter VIII; page 217).

While Darwin's statement of the principle is correct, he did not have access to accurate demographic figures and did not appreciate that the temporary aggregations of males around receptive females are often associated with a preponderance of females at the population level.

The accuracy of his observation of an association between the development of secondary sexual characters and polygynous mating systems has subsequently been confirmed by a range of comparative analyses demonstrating associations between the distribution of sexual dimorphism and promiscuous, polygynous or polygynandrous breeding systems, both in mammals [9–13] and in many other organisms [14]. So, too, has Hunter's suggestion of an association between the relative development of traits used to compete for breeding opportunities and a diet of vegetables for, in general, polygyny is more common in herbivorous animals where females are more often social than in carnivores or insectivores [13,15].

Darwin's first category of sexual selection (now often referred to as intrasexual selection) was more readily accepted as an important evolutionary mechanism than his second category, involving active choice of mates (now often referred to as intersexual selection). After his death (1882), interest in sexual selection as an evolutionary process waned until it was resurrected in the 1930s by Fisher [16] and Huxley [17,18]. Fisher's work provided the first formal model of the operation of selection through female mating preferences [16], while Huxley synthesized the increasing evidence of sex differences in aggressive behaviour and in the initiation and complexity of displays [17,19].

2. Reproductive competition and parental investment

While Fisher and Huxley re-invigorated interest in sexual selection and in the evolution of sex differences in morphology and behaviour, neither addressed the underlying question of why it should be that males were commonly both more aggressive and more highly ornamented than females, while females often appeared to be more selective in their choice of mating partners than males. Modern thinking on this topic is strongly influenced by the work of Bateman [20]. Working with fruit flies, Bateman demonstrated that the reproductive success of males varied more widely and increased more rapidly with the number of partners they bred with than did the reproductive success of females, and argued that sex differences in reproductive competition and in the intensity of selection for traits associated with breeding success were a consequence of the capacity of males to increase their breeding success by mating with multiple partners. Bateman's conclusions have sometimes been contested on methodological grounds [21], because female reproductive success can also increase with partner number [22] or because sex differences in reproductive variance can arise for stochastic reasons [23,24]. In addition, there are some species where sex differences in fitness do not depend principally on partner number in either sex [25]. However, the effects of partner number on female breeding success are typically small [26] and Bateman's fundamental insight that, where males do not care for their offspring, sex differences in the relationship between partner number and reproductive success (‘Bateman gradients’) commonly generate larger individual differences in breeding success, more frequent reproductive competition, more intense selection and a greater development of secondary sexual characters in males represents one of the cornerstones of modern evolutionary theory [6,7,27,28].

Bateman attributed the presence of sex differences in reproductive variance and Bateman gradients to sex differences in the size and rate of production of gametes and in the extent of parental care. Both points were subsequently developed in an influential review by Trivers [29] that explicitly attributed the greater variance in reproductive success, the greater intensity of breeding competition and the greater development of weaponry and ornamentation in males and the greater selectivity of breeding partners in females to the reduction or absence of parental investment by males [29]. The concept of parental investment (defined by Trivers [29] as ‘any investment in offspring that increases the offspring's chances of surviving (and hence reproductive success) at the cost of the parent's ability to invest in other offspring’, p.136) provided important insight into the reasons for sex differences in reproductive competition since it captured the trade-off between investment in current and future offspring and incorporated the effects both of anisogamy and of parental care [30–33]. Trivers argued that reproductive competition is more frequent and more intense in the sex that invests less because there are more individuals of that sex that are ready and competing to breed at the same time (typically males) than there are individuals of the sex that invests more (typically females). The ratio of the number of males and females competing for breeding opportunities at the same time (termed the ‘operational sex ratio’ (OSR)) determines average levels of competition for breeding partners and opportunities in the two sexes and (in many cases) the frequency and intensity of intrasexual conflict [30].

The Operational Sex Ratio is commonly related to sex differences in reproductive competition both between and within species and provides a general basis for explaining the usual direction and distribution of sex differences in reproductive competition, variation in breeding success and the relative development of secondary sexual characters across species [11,30,33–37]. However, measuring it is fraught with difficulty [37–39]. For example, in polygynous species, what categories of males should be included? All adults that are not currently involved in parental care? Only those that have established breeding territories or dominance over other group members? Only those that breed successfully? And, at what point in the breeding cycle should it be measured? As a result, the OSR is not well suited for comparative tests of the relationship between parental investment and different aspects of sexual selection.

One approach to assessing the direction of biases in the OSR and predicting which sex should, on average, compete most intensely for access to breeding partners is to estimate the relative length of time for which one breeding attempt prevents individuals of each sex from competing for further breeding attempts [38,39]. Measures of average ‘Time Out’ in the two sexes should provide an index of the OSR and indicate which sex should be more competitive. However, once again, it is not easy to decide what categories of individuals should be included and mean ‘Time Out’ does not necessarily indicate the potential fitness gains that males and females can make from winning contests over reproductive opportunities.

While biases in the OSR and sex differences in ‘Time Out’ predict the relative numbers of males and females competing for mating opportunities, they do not incorporate any estimate of the potential fitness gains that males and females can achieve by winning reproductive contests. One approach that has been used to predict whether successful males or successful females can gain more from winning reproductive contests is to compare the maximum observed rates at which successful males and successful females can complete successful breeding attempts with a partner (the potential reproductive rate or PRR) and to test whether sex differences in PRR are consistently related to the observed frequency or intensity of competition for breeding partners in the two sexes. This has the advantage that (in contrast to most other indices of sexual selection) it is possible to extract estimates of PRR in the two sexes from the empirical literature and to explore their relationship to observed variation in reproductive competition. For example, in some oviparous vertebrates where males are the principal or only care-giver, males compete intensely for females while, in others, females compete for males [40]. Comparisons of estimates of maximum PRR by males and females show that males have higher PRRs than females in species where they can care for multiple broods of eggs at the same time and compete more strongly for breeding opportunities (as in a number of fish and amphibians), while females compete more strongly and have higher PRRs than males in oviparous species where males care for a single clutch at a time [40], as in some mouthbrooding fish and a number of wading birds. Most species falling in the first category are ectotherms while most of those falling in the second category are homeotherms or mouthbrooding fish where neither parent can care for multiple clutches simultaneously and clutch size is relatively small [40,41].

A more sophisticated approach to predicting the intensity of reproductive competition in the two sexes is to combine estimates of the relative numbers of competing males and females based on measures of ‘Time Out’ with estimates of Bateman gradients in the two sexes. An index of this kind (the Jones Index) was first suggested by Jones in 2009 [42] and later developed and extended by Kokko et al. [43] in a theoretical model of sexual selection that neatly incorporates both the effects of parental investment (OSR) and those of Bateman gradients and provides the most precise basis yet for general predictions about sex differences in competition for breeding partners. Simulations of the extent to which different indices predict the relative intensity of sexual selection in the two sexes suggest that those that combine measures of the relative number of competing males and females with estimates of the relationship between mating frequency and fitness (i.e. Bateman gradients) are more accurate than those based either on the OSR or on Bateman gradients alone [8]. However, the extent to which they predict the distribution of sex differences in behaviour or in secondary sexual characters has yet to be tested.

Although estimates of the OSR and Bateman gradients (and indices that combine them) provide an indication of the likely intensity of breeding competition in males and females, they do not provide an index of the strength of sexual selection on particular traits nor do they necessarily predict the relative development of sex differences in aggressive behaviour, weaponry or ornamentation [37,44–47]. There are several reasons for this. Males can increase their fitness by tactics other than direct confrontation (including scramble competition, sperm competition and cryptic mating tactics) so that biases in the OSR do not necessarily intensify selection for competitive behaviour or traits associated with competitive success [48]. The fighting tactics of males also differ between species, leading to contrasts in selection on particular male traits and in the relative development of particular forms of weaponry in the two sexes [13,44,49]. Similarly, qualitative differences in female mating preferences can generate contrasts in the form and development of male displays and ornamentation [50,51].

The extent of sex differences in competitive behaviour and associated morphological traits is also influenced by the intensity of breeding competition among females and the relative development of behavioural and morphological traits that affect the ability of females to compete successfully [52–58]. Direct competition between females can be intense wherever females compete for limited resources necessary for breeding and is not confined to species where males are the principal care-givers [25,56,58]. It can often lead to the evolution of increases in body size in females and, less commonly, to the development of female weaponry—and both can reduce the extent of sex differences in size and weapon development [59]. Moreover, male mate choice is highly developed in some species and can lead to the evolution of ornaments in females [53,60,61]. As a result, while there is evidently some relationship between sex differences in breeding competition, weaponry and ornamentation and indices of the intensity of sexual selection, it is unsurprising that it is often inconsistent and that the distributions of sex differences in behaviour and morphology are not always closely related to general indices of reproductive competition [11,14,62].

A fundamental limitation of the approaches that I have outlined so far is that they take as their starting point a situation where the two sexes differ in the size of their gametes and in the extent of parental care and explore how these differences are likely to affect reproductive rates, reproductive competition and the intensity of sexual selection. However, the evolution of parental care and of reproductive competition are often likely to interact in the course of evolution and there are fundamental questions that need to be asked about these processes [29,63–66]. In particular, how does the evolution of reproductive competition affect the evolution of parental care, as well as vice versa? And what role has anisogamy played in the evolution of sex differences both in parental care and in reproductive competition? These questions have recently been the focus of extensive theoretical investigation that has clarified the assumptions on which traditional models of intrasexual competition, parental investment and the evolution of sex roles have been based and has demonstrated the potential complexity of predictions (see [65,66]). In particular, it has exposed the limitations of models that assume that selection will maximize the rate of reproduction in each sex and has emphasized the need to consider the effects of investment in parental care and sexual competition on lifetime reproductive success and reproductive value [67,68]. It has also underlined the need for models to comply with Fisher's argument that, in diploid species, the total number of offspring produced by each sex must be identical and the mean reproductive output of individuals consequently depends on the sex ratio [67,68]. Finally, it has demonstrated the potential effects of variation in the sex ratio at different stages of the lifespan and the need to think carefully about the categories of individuals that should be included in models of the evolution of interactions between parental care and reproductive competition [65].

Although theoretical research on the evolution of sex roles and reproductive competition emphasizes the limitations of simple models based on reproductive rates [65], it tends to support traditional interpretations of the evolution of sex roles and reproductive competition. For example, under many conditions, anisogamy appears likely to favour the greater probability or development of parental care by females and of increased investment in reproductive competition by males [64,68]. Similarly, sex differences in ‘Time Out’ and in the OSR would usually be expected to predict the direction of sex differences in reproductive competition unless ‘Time Out’ is very short in one sex, so that variation has little effect on sex differences in reproductive competition [65,69]. All other things being equal, sex differences in reproductive competition would often be expected to generate sex differences in investment in traits associated with competitive success [65].

3. Reproductive competition and the opportunity for sexual selection

While theoretical models of the evolution of sex differences in reproductive competition and parental care generate broad predictions that can be tested by comparative analyses [40,65], they do not incorporate empirical measures of variation in reproductive success or of the influence of factors that affect it. Building on Bateman's (1948) [20] observation that variance in breeding success is greater among male fruit flies than among females, Arnold, Wade, Shuster and their colleagues showed how empirical estimates of variation in reproductive success between individuals (the opportunity for selection) can be calculated and compared in males and females and partitioned into different multiplicative components of breeding success, including variation in mating success, in offspring production and in survival or longevity [28,70–73]. Variance in mating success (Is), which provides an index of the opportunity for sexual selection, can be partitioned into sub-components, such as the relative frequency of mating with partners of varying quality, the incidence of extra-pair paternities or the frequency of coerced matings [7,74–76]. Similarly, variance resulting from differences in offspring production can be partitioned into sub-components, including access to breeding opportunities, litter size, breeding frequency and offspring survival [45,77]. The information provided by measuring the opportunity for selection or the opportunity for sexual selection in each sex and partitioning it into different sub-components provides insights into the potential strength of selection operating through effects on different components of fitness at different stages of the life history and indicates the potential benefits of traits that affect reproductive success at each stage. As a result, it provides a basis for exploring the influence of specific phenotypic traits on different components of breeding success [7,8,45,78].

Applications of Arnold and Wade's framework to empirical data for a wide range of species generally confirms Bateman's original suggestion that variance in mating success within seasons is greater in males than females in many promiscuous and polygynous animals [7,45,74,78]. However, the application of this approach to assessing the strength of sexual selection in iteroparous species introduces complications because, in many promiscuous, polygynous or polygynandrous species, the effective breeding lifespans of males are much shorter than those of females, starting and ending earlier [13,79,80]. As a result, a large proportion of observed variance in male breeding success within years is a consequence of variation in male age and (standardized) variance in male breeding success within years is consequently often greater than (standardized) variance in male breeding success calculated across the lifespan [44,81]. Among females, individual differences in breeding success are often relatively small within seasons but are consistent across them, so that (standardized) variance in lifetime breeding success in females is often considerably greater than (standardized) variance within seasons. As a result, sex differences in the opportunity for selection and in the opportunity for sexual selection are often substantially smaller when calculated across the lifespan than when calculated within seasons and, in some cases, disappear altogether [44,82]. While effects of this kind may be most obvious in long-lived, iteroparous species, they may also be widespread in short-lived animals with seasonal life cycles if the duration of breeding activity in individual males is shorter than that of females.

Although estimates of the opportunity for sexual selection provide a general indication of the potential strength of sexual selection in the two sexes and the potential magnitude of sex differences, like indices of the relative intensity of reproductive competition, they do not necessarily reflect sex differences either in the relative intensity or frequency of contests for breeding opportunities or in the strength of selection for specific secondary sexual characters [7,37,44,45,78,83]. One reason for this is that a substantial proportion of variance in breeding success in both sexes (and especially in males) is likely to be of random origin [21,23,24,84]—though it is important not to overestimate this, for empirical studies show that individual differences in phenotype commonly make an important contribution to variation in male breeding success and survival [7,45]. More importantly, in different species, individuals maximize their breeding success by different strategies and tactics and success depends on different attributes [13,48], so that measures of the opportunity for sexual selection do not necessarily reflect the frequency of contests or the relative strength of selection operating on particular behavioural or morphological traits [37,44–46].

Here, too, reproductive competition between females complicates relationships between sex differences in the opportunity for sexual selection and sex differences in behaviour and morphology (see above). Not only is competition between females for mating partners more common than was previously supposed but females often compete to become sexually mature, to obtain access to resources necessary for breeding or to rear their offspring [57]. In some cases, intrasexual competition of this kind can lead to the evolution of secondary sexual characters in females that resemble those of males [57–59], so that sex differences in competition for mates do not necessarily reflect either the relative intensity of reproductive competition in the two sexes or the development of sex differences in behaviour or morphology. In others, it can lead to selection contrasting adaptations in the two sexes.

4. Reproductive competition and selection gradients

A third approach to investigating the evolution of sex differences is to measure and compare estimates of covariance (including selection differentials and selection gradients) between specific traits and mating success, providing opportunities to compare the effects of particular behavioural or morphological traits on components of reproductive success in the two sexes [7,37,44,45,47]. Differing relationships between breeding success and specific traits in the two sexes provide a basis for predictions about the form of sex differences that are likely to evolve. For example, in red deer, where body size is important in winning fights and relationships between size and breeding success among adults are stronger in males than in females [85], the weight of calves at birth is related to their size and fighting success as adults and records of individual life-histories show that birth weight is more closely correlated with measures of lifetime breeding success in males than in females [86]. As would be expected, male red deer calves are born heavier than females [87].

Comparisons of selection differentials and selection gradients operating through variation in breeding success on the same traits in the two sexes provide the most direct way to explore the evolutionary processes responsible for the evolution of sex differences [44,47]. Some advocates of this approach argue that it is the only useful way to answer general questions about sexual selection [37]—but comparisons of selection gradients have their own limitations [78]. Traits that enhance mating success often have countervailing effects on other components of fitness that need to be considered by attempts to understand their evolution in males and females [88–90]. Like other approaches to understanding the evolution of sex differences, the investigation of selection gradients also raises questions about the period over which variation in mating success should be measured and the categories of individuals that should be included. The causes of contrasts in selection gradients are often difficult to identify and this approach does not necessarily shed light on the reasons for variation in the intensity of reproductive competition in the two sexes. Moreover, very few studies have access to data that allow measurement of correlations between traits used in competitive encounters and components of the lifetime breeding success of individuals of both sexes in natural populations [91]. As a result, although this approach can be useful for dissecting the selection pressures maintaining sex differences in secondary sexual characters in particular species (see [7]) and may eventually provide a basis for generalizations, studies that have explored these relationships are still too rare for quantitative interspecific comparisons to be feasible. In conclusion, it is clear that multiple approaches have contributed to our understanding of sexual selection and are likely to continue to do so.

5. Variation in reproductive competition

Since Fisher and Huxley rekindled interest in it in the 1930s, the field of sexual selection has advanced substantially. All three of the approaches to investigating sexual selection described above have made complementary contributions to this process: exploration of relationships between parental investment, reproductive competition and sexual selection provide insight into the fundamental reasons for the contrasting roles of the two sexes and the relationship between sexual selection, sexual dimorphism and breeding systems; analysis of variation in the opportunity for selection identifies the potential strength of sexual selection at different stages of the life history of individuals of the two sexes and shows how this can vary with the nature of breeding systems; and comparisons of selection gradients reveal the relative strength of selection operating through mating success and other components of fitness on specific traits at different times and under different conditions. Asking questions about all three of these aspects of sexual selection requires different kinds of data and comparisons at different levels.

There have also been substantial advances in our understanding of the benefits and costs of specific traits and their contributions to breeding success in the two sexes, generated partly by improved techniques of measuring reproductive success in males and partly by increases in the number, duration and resolution of field studies that have tracked the development, behaviour and breeding success of individuals throughout their lives [92–94]. For example, studies of a wide range of animals, from beetles to mammals, have demonstrated the benefits of body size and weaponry to the mating success of individual males [49,95], and similar approaches have been used to investigate the benefits of presence and size of male weapons [90,96]. More recent studies of species where females compete for breeding opportunities have shown that similar traits affect the breeding success of individuals and that intense competition among females can lead to situations where females are larger and more aggressive than males [57,58,97].

There have also been improvements in our understanding of the costs of secondary sexual characters and of the mechanisms that constrain their development. For example, there is now substantial evidence that, in many sexually dimorphic animals, increased growth and size in males reduce their survival during periods of food shortage and that males are often more likely to die from starvation or associated causes such as increases in parasite load [98–100]. Some of the strongest effects of this kind have been documented in ungulates. For example, in red deer, rising population density and declining food availability lead to increase in the loss of male fetuses [101], to increased mortality in juvenile males relative to juvenile females [98], to increased rates of dispersal in adolescent males [102], to reductions in the relative longevity of males [81], and to strong biases towards females in the adult population [103]. Similar tendencies for adverse environmental conditions to affect male survival disproportionally have been documented in a substantial number of other sexually dimorphic animals, including humans [13,33]. Sex differences in growth can also affect the costs of producing and rearing members of the faster-growing sex to their parents. For example, in some mammals where males are the larger sex (including red deer), male offspring suckle more frequently than females and rearing sons depresses their mother's subsequent survival and, in some cases, their fecundity more than rearing daughters [13,87,104]. By contrast, there is, as yet, little evidence of relative increases in female mortality among juveniles and adolescents in species where females grow faster than males and are larger as adults, possibly because the evolution of increased growth and size in females is constrained by effects on fecundity rather than survival [13].

The costs of reproductive competition can have important consequences for the ASR and the Operational Sex Ratio with downstream effects on variance in breeding success, the frequency of fighting and the intensity of selection on traits used in contests. In a wide range of promiscuous, polygynous and polygynandrous species that breed seasonally, male biases in the OSR decline during the course of the breeding season as males become exhausted and leave the pool of individuals competing to mate, leading to reductions in the intensity of intrasexual competition between males as the season progresses [105]. Contrasts in environmental conditions that affect the relative condition and survival of adults and the ASR can also lead to similar differences between populations. For example, in some sexually dimorphic ungulates, increases in male dispersal, juvenile mortality and adult mortality all contribute substantially to variation in the ASR, which often becomes increasingly female biased as populations approach carrying capacity. These biases can reduce the intensity of direct competition between males, leading to reductions in the frequency and duration of pre-copulatory mate guarding by males [88,106,107] and to increases in the relative breeding success of young and old males [102], suggesting a relaxation of the effects of secondary sexual characters on male fitness. In some cases, these effects are sufficiently large to have practical implications for the management of populations. For example, in ungulates where populations are managed for sport, the income generated depends principally on the off-take of males—which declines as populations approach carrying capacity, male mortality and dispersal increase and the ASR becomes progressively biased towards females [103,108]. In such cases, the size of the female population that optimizes revenue can be substantially lower than the maximum population density. For example, in red deer, maximum sustainable yield of mature males from local populations is maximized when female numbers are restricted to around 50% of the level they reach if left unculled [103,108].

Contrasts in the susceptibility of the two sexes to starvation are not, of course, the only mechanisms leading to biases in the ASR. In some species, sex differences in rates of dispersal expose members of one sex to higher risks of mortality [13,109]. In addition, in some oviparous species, members of whichever sex is principally involved in guarding eggs or hatchlings are exposed to greater risks of predation, leading to ASRs biased towards the opposite sex and to increased variability in relative numbers of the caring sex [33,110,111]. And, in species where sex determination is determined by social or environmental conditions, biases in the ASR can be caused by changes in the primary sex ratio [112]. Nor is it the case that the only effect of changes in the ASR is to modify intrasexual competition for breeding partners. In many species, variation in the ASR is likely to affect the extent of mate choice as well as the frequency of extra-pair (or extra-group) matings and the intensity of sperm competition [13,36,113,114]. In addition, in bi-parental species, it may influence the extent or duration of care in the two sexes while, in uni-parental species where either sex may care, it may affect the proportion of males and females involved in care [110,111]. In both cases, changes are likely to affect relationships between the ASR and breeding competition in the two sexes.

So what directions might research on sexual selection now usefully take? It is important to appreciate that research on sexual selection has contrasting and divergent aims—and is likely to continue to do so. Theoretical research will continue to explore the coevolution of parental care, mating strategies and reproductive competition in the two sexes. As Fromhage and Jennions' contribution [65] to this issue shows, predictions about the relative intensity of reproductive competition in the two sexes depend on the assumptions that are made and can be affected by interactions between reproductive competition and parental care as well as by variation in the ASR. Subsequent generations of models will need to avoid the pitfalls he describes, to think carefully about the age categories to include, the periods over which reproductive success should be calculated and, where relevant, the need to comply with the Fisher condition that the total number of offspring produced by each sex must be equal and the mean reproductive output of individuals of both sexes must depend on the sex ratio. This could, in turn lead to the exploration of more specific models based on contrasting assumptions and to a closer integration of theoretical and empirical research, generating predictions that can be tested by comparative analyses based on the distribution of traits in contemporary species and on phylogenetic reconstructions within lineages. In addition, there are important areas of sexual selection that have still attracted relatively little attention from theoreticians. While a large proportion of theoretical studies have focused on the evolutionary causes and consequences of intrasexual competition between males and mate choice in females, recent studies have demonstrated the importance of reproductive competition between females and of mate choice in males and more extensive theoretical investigation of both of these processes is overdue.

Further studies of the extent of variation in fitness among individuals of both sexes are also needed. The analytical framework provided by the work of Arnold and his colleagues [70,71] provides a coherent framework for investigations of the components of variation in breeding success between individuals but relatively few empirical studies have yet been able to explore the extent and causes of variation in lifetime breeding success in both sexes in natural populations. Those that have done so have generated some surprising results. For example, studies of several promiscuous and polygynous species show that age can exert such a strong effect on the breeding success of males that (standardized) variance in lifetime breeding success is not consistently greater in males than females. In addition, since the effective breeding lifespans of males decline sharply as the degree of polygyny rises, (standardized) variance in male breeding success may not be consistently larger in polygynous species than in monogamous ones. We also need to understand considerably more about the extent of variation in lifetime breeding success in species with unusual breeding systems—including protandrous and protogenous fish, polyandrous birds, cooperative mammals and eusocial insects. Studies that have explored the life histories of individuals in these species have often generated unexpected results. For example, recent research on cooperative mammals, where a single female and a single male monopolize breeding in each group and their young are reared by all group members, shows that (standardized) variance in breeding success is commonly larger in females than males and that secondary sexual characters are often well developed in females.

In addition, we need to know more about the relative strength of selection on particular traits in species with contrasting breeding systems. There are still relatively few studies that have been able to compare the effects of the same phenotypic traits on lifetime breeding success and its components in both sexes and it is often difficult to isolate the effects of specific traits on specific components of fitness. Interspecific contrasts in the relative development of different secondary sexual characters suggest that there probably are large qualitative differences in the relative importance of different forms of weaponry and ornamentation and we lack a general framework for explaining the evolution and distribution of these differences. We also need to understand the mechanisms constraining the extent of reproductive competition and the development of secondary sexual characters in both sexes and the tactics that individuals use to mitigate the costs of competition. For example, while the presence of more than one male in breeding groups of social mammals reduces the capacity of dominant males to monopolize paternity, recent studies suggest that it can generate substantial benefits to the breeding tenure of males and enhance their lifetime reproductive output. In addition, we need to know more about the extent to which selection pressures change as individuals move from being dependent juveniles, to non-breeding adolescents and eventually to becoming breeders. The development of many secondary sexual characters (including the relative aggressiveness of individuals, their size and the development of weaponry and ornamentation) is often flexible and changes with social and ecological conditions—often to different extents in the two sexes. There is increasing evidence that some (and, possibly, many) of these adjustments in development are a consequence of changes in epigenetic mechanisms and may represent adaptive responses to changes in constraints and selection pressures. The recognition of changes in selection pressures and flexibility in adaptations as individuals progress through the successive stages of their life histories has important implications both for the categories of individuals that theoretical models of sexual selection need to consider and for their predictions.

Finally, we need to know the answers to related questions about the causes and consequences of variation in breeding success in our own species. It is ironic that, despite the difficulties of measuring individual variation in reproductive success and its components in animal populations under natural conditions, studies that have monitored the breeding success of substantial samples of individuals from birth to death across multiple generations are more abundant for nonhuman animals than for humans. As many of the contributions to this issue show, there are both striking similarities and striking contrasts between the causes and consequences of reproductive competition and variation in breeding success among males and females between humans and other animals. Understanding the causes of these differences offers both fundamental and practical insights into the biological and cultural mechanisms affecting reproduction, welfare and survival in human populations.

Data accessibility

This article uses no additional data.

Competing interests

I declare I have no competing interests.

Funding

This paper was written as a component of ERC grant no. 294494 to T.C.-B.