Polyandry, by elevating sexual conflict and selecting for reduced male care relative to monandry, may exacerbate the cost of sex and thereby seriously impact population fitness. On the other hand, polyandry has a number of possible population-level benefits over monandry, such as increased sexual selection leading to faster adaptation and a reduced mutation load. Here, we review existing information on how female fitness evolves under polyandry and how this influences population dynamics. In balance, it is far from clear whether polyandry has a net positive or negative effect on female fitness, but we also stress that its effects on individuals may not have visible demographic consequences. In populations that produce many more offspring than can possibly survive and breed, offspring gained or lost as a result of polyandry may not affect population size. Such ecological ‘masking’ of changes in population fitness could hide a response that only manifests under adverse environmental conditions (e.g. anthropogenic change). Surprisingly few studies have attempted to link mating system variation to population dynamics, and in general we urge researchers to consider the ecological consequences of evolutionary processes.
A large shift in our understanding of mating systems occurred in the 1970s when birds, the most conspicuously monogamous animals, were shown to be promiscuous. This finding was quickly realized to have important consequences for population management, as illustrated by an unsuccessful attempt to control a perceived agricultural pest, red-winged blackbirds, by vasectomizing males. Bray et al.  were surprised to find that females paired with sterilized males remained fertile, suggesting that females mate with extra-pair males. At the time, such findings were big news, because they showed that variation in mating systems was a significant and understudied predictor of population dynamics, which could be a key to the success of both pest management and conservation programmes. Since then, much theory has been developed to contrast the population-level consequences of monogamy and multiple mating [2–5], including factors causally underlying mate availability, such as the temporally varying number of males per female [6,7].
The study by Bray et al.  also illustrates that monandry was long taken as the null model. After Bateman  highlighted that fruitfly females benefit less (and sometimes not at all) from multiple mating than do males, the presumed ubiquity of monandry seemed so obvious as to be beneath empirical investigation. Lest the reader now feel condescending towards earlier researchers, it is prudent to note how strongly this belief continues to the present day. Monogamy is still a common assumption in evolutionary and ecological models, and the number of studies seeking to explain polyandry (reviewed in [9,10]) far outweighs the number investigating the evolution of monandry. Although this bias may simply reflect a greater number of polyandrous species, the popular research question ‘Why do females mate multiply?’ appears to carry with it the implicit assumption that monandry does not warrant special investigation.
This view of monandry as a null model for female behaviour may be incorrect: it is illuminating to ask why individuals of either sex would evolve behaviour any more sophisticated than mating every time a potential mate is encountered . Monandry requires a complete loss of female receptivity after one mating or, in the case of social and genetic monogamy, specific acceptance of mating advances by one male, but not others. The propensity of females to reject additional mates is also able to respond to selection [12,13]. In short, monandry is an evolved outcome that requires an explanation. We therefore consider polyandry a more suitable null model, at least in species where individuals typically encounter multiple potential mates.
Defining the scope of this review requires stating precisely what we mean by polyandry. A behavioural ecology textbook usually introduces the reader to monogamy, polygyny, polyandry and polygamy/polygynandry, depending on whether no-one, males only, females only or both sexes mate multiply [14,15]. Given that multiple mating by both sexes has proved to be commonplace [10,16], one might expect polygynandry to have swamped all other terminology. However, this is decidedly not the case: a Web of Science topic search for polygynandr* (conducted on 18 July 2012) yielded 138 hits, a meagre result compared with 1860 for polyandr* and 3243 for polygyn*. This appears to reflect confusion over whether ‘polyandry’ and ‘polygyny’ are defined via the ability of one sex to control access to mates via social pair bonds  ( recommends ‘promiscuity’ for cases where no such bonds exist), or if polyandry is taken to simply mean that a female mates multiply, whether or not the system is also polygynous. It appears that common usage has shifted away from defining polygyny and polyandry as mutually exclusive alternatives (as in [14,15,17]). Here, we define polyandry in its broad sense (as in [9,13,16,18]): populations in which females sometimes mate with more than one male, and/or have their eggs fertilized by more than one male, are polyandrous. Nevertheless, the diversity of polyandrous mating systems presents us with a clear challenge. As we will see below, it matters greatly whether multiple matings occur rapidly enough that sperm of different males compete over fertilization of the eggs.
With this review, we aim to identify routes by which polyandry can affect population fitness, with a focus on conservation biology. We also attempt to link individual-level effects of polyandry to their demographic consequences, although we conclude that existing data make it difficult to do so reliably. Polyandry is shown to have a great many potential positive and negative effects on populations, and we endeavour to synthesize these and highlight potentially fruitful avenues for future research.
2. Multiple mating modifies the cost of sex
Whether a female is monandrous or polyandrous depends on mate encounter rates as well as on the behaviour of males and females when encounters happen; the behaviours themselves evolve according to the net fitness consequences of polyandry for the individuals involved, and the cost they pay to either resist or to seek additional matings. Even if polyandry has a net negative effect on female inclusive fitness, it can persist if the costs of resistance would outweigh its benefits (‘convenience polyandry’; [19–21]). Note also that stochastic variation in mate availability might make some females monandrous and others polyandrous even if they use identical behavioural rules [11,22].
Whatever the selective forces (if any) behind polyandry, its consequences are profound. This is especially true when polyandry occurs within a breeding cycle, leading males to be uncertain about the parentage of their mates’ offspring, because uncertain paternity is a core reason why male parental care tends to be less common than female care [23,24]. Curiously, this holds true even in systems where females mate multiply in order to receive care from more than one male: care effort per male is still predicted to decline with polyandry . This means that polyandry influences a fundamental property of two-sex systems: the famous twofold cost of sex . Under specific conditions, asexual reproduction doubles the fitness of individual females and thus population fitness. However, the cost of sex is only exactly twofold when a suite of surprisingly stringent assumptions is met: specifically, the existence of males should have no effect on female reproductive output. For example, there should be no paternal care, no harmful male harassment of females, no mate or sperm limitation, no genetic benefits of sexual selection, and a male must be equivalent to a female in intraspecific competition for food and other resources .
Polyandry is expected to increase the cost of sex in the great majority of cases. By selecting against paternal investment, polyandry increases the cost of sex [23,24]. Further, it can favour male traits that displace females from their phenotypic optimum; below, we review such instances in the context of intra- and inter-locus sexual conflict, which can increase the cost of sex beyond the twofold baseline that arises in the total absence of paternal investment. Sexual conflict, of course, can also exist under monandry or even monogamy: if there is a biased operational sex ratio, selection may favour traits that improve competition for mates despite reducing productivity at the mated stage. In general, however, polyandry greatly increases the opportunity for sexual conflict , and thus we expect much of the cost of sex to be a direct consequence of polyandry.
3. Demographic consequences of inter-locus sexual conflict
Inter-locus sexual conflict occurs when the sexes have differing optimal outcomes in a male–female interaction . It may occur in all mating systems except those in which strict lifetime monogamy is the only option and the sex ratio is even . Because it can produce sperm competition, polyandry creates the potential for conflict over female-mating rate—for example, males may attempt to prevent their mates from mating with another male. Sexual conflict over female-mating rate may, therefore, be stronger in polyandrous than monandrous species. Conversely, sexual conflict over the mating rate may sometimes be lower in polyandrous species, because females are mating at a rate that is closer to the male optimum. Polyandry is similarly expected to affect sexual conflict over a range of other traits. For example, polyandry can select against male care [23,24,29] and promote the evolution of harmful competitive adaptations . A few studies, which we will now review, have addressed how inter-locus sexual conflict affects demographic parameters. However, we note that there is ample scope for additional research.
Studies of desertion in birds provide evidence that reduced male care can cause a population-wide decline in productivity. For example, penduline tits often desert their nest in the hope that their partner will stay and rear the eggs alone. Deserting is thought to be advantageous on average, even though the parent left behind often deserts as well (reviewed in ). This conflict leads to very many nests (30–40%) being deserted by both parents, and this wasted effort may substantially reduce the number of young produced each year. Although biparental desertion might be an extreme example, reduced productivity when males do not care for young generalizes to a very broad pattern: in a dataset of 980 bird species, female-only care led to a 20 per cent reduction of productivity compared with other types of care .
Male harassment and copulation attempts have been shown to depress female survival and fecundity in several diverse taxa (e.g. water striders , lizards , sea lions , Drosophila ). Experimental evolution studies manipulating the mating system have provided some of the strongest evidence that polyandry selects for both elevated male harm and costly female resistance to that harm [36–40]; for example, Crudgington et al.  showed that males evolved under polyandry courted females more frequently, and that females housed with these males produced fewer progeny than females interacting with males evolved under monogamy. Additionally, female dung flies evolved under polyandry were found to be less fecund than monogamous females when mating only once, implying that female adaptations to male harm have a cost that outweighs the potential genetic benefits of sexual selection .
In a twist to the conflict over mating rates, males may attempt to increase their fitness by imposing a suboptimally low future mating rate on their female mates . Adaptations such as mating plugs  and mate guarding are, therefore, potentially harmful to female fitness. Mate guarding appears particularly harmful to female reproduction when it makes females unable to escape senescent  or sperm-depleted males . However, mate guarding is a good example of a male-mating adaptation that can actually increase female fitness, e.g. if the mate performs vigilance duties or deters harassment from other males. In gammarids, guarded females had a shorter inter-moult duration suggesting a benefit of being guarded , and female crickets guarded by a male were less likely to be predated . Male attempts to increase fertilizations gained with a particular female can also occur after mating: in Drosophila and other insects, males transfer seminal fluid proteins that inhibit further matings and induce females to lay more eggs in the short term , at a cost to female lifetime reproductive success .
Another detrimental effect of polyandry is its positive effect on the incidence of sexually transmitted infections (STIs). These infections likely have population dynamic consequences, particularly when the parasite sterilizes, but does not kill the host, leaving it to mate and compete with uninfected individuals . Confounding this problem is the prediction that males should often be selected to invest less in avoiding disease or clearing infections than females . Lowered male immune defence should exacerbate sexual conflict over mating rates in the case of STIs, and turn males into easy targets for pathogens, in turn increasing the incidence of disease in females .
4. Does it matter?
There is abundant evidence that female lifetime productivity can be reduced by male adaptations to polyandry (for further examples see ). However, the extent to which such negative effects show up in measures of population performance (ultimately affecting extinction rate) is much less clear, principally because population dynamics are not solely driven by juvenile production. Demographically, male-induced harm may have no visible effects as long as the birth rate remains greater than the death rate: for example, many juveniles may experience density-dependent mortality, and adult populations are then expected to be of similar size in each generation regardless of variation in female fecundity. The relevant life-history stage determining population limitation is crucial here. However, changes in the environment (for example, anthropogenic impact) can adversely affect the birth and death rates, and this may push a population below the threshold of positive growth; populations with greater female survival and fecundity are expected to be more robust to such challenges. High levels of conflict may therefore weaken the capacity of populations to persist in a challenging environment.
It is curious that few studies of sexual conflict have considered the full life-history consequences, together with the relevant ecology, to determine how conflict impacts population growth. In a slightly different context, that of within-sex conflict for breeding opportunities, intraspecific conflict was estimated to have slowed down the recovery of Seychelles magpie robins from ‘Critically Endangered’ status by 33 per cent . Given that every year spent at a low population size elevates the species-wide vulnerability to environmental hazards , it appears possible that conflict makes populations more vulnerable whenever it slows down population growth. By extension, male–female conflict, too, may have real population consequences. Bearded vultures provide an example of such a conflict that is pertinent to conservation biology (the species is endangered in Europe; ). This species exhibits bi-parental care, which should lower the cost of sex (because a hypothetical asexual vulture would not raise its young as efficiently as a sexual pair). However, vulture pairs are sometimes joined by an additional male who was not able to find a female of his own. Joining a pair presumably affords males greater fitness than making no attempt to breed, but trios rear substantially fewer offspring than do pairs (likely as a result of antagonism between the males). This male-driven behaviour is a concern when designing conservation efforts.
Theoretical  and empirical  work suggests that inter-locus sexual conflict coupled with a male-biased adult sex ratio can be very damaging. Male–male competition for matings and fertilizations is essentially a zero-sum game, in which the limiting resource (females/eggs) can diminish as competition becomes more intense. Sexual selection may come to resemble a tragedy of the commons, where the harmful fallout of male–male competition increases female mortality, leading to even stronger competition (and, therefore more harm) for the remaining females, potentially driving females extinct unless their resistance compensates sufficiently . In principle, this might mean that removing males could in some cases improve the performance of endangered populations. Conservationists are understandably reluctant to consider such extreme measures without good scientific backing. Indeed, intuitive predictions have not always been supported by data. One might expect that males would have a strongly negative effect on population-wide reproductive output in the hihi, Notiomystis cincta, a bird with a fierce mating system in which males pin down females in face-to-face forced copulations. However, despite intense harassment, adult sex ratio had little or no effect on adult female survival or the number of fledglings per female . This highlights that the apparent severity of male harm can be misleading. Infanticide is another potential problem for conservation by male removal: in many mammals (e.g. bears, lions), removing paired males can cause unpaired males to kill offspring belonging to the removed males, meaning that removing males could actually negatively affect population growth. Such infanticide would also exacerbate the impact of hunting targeted at males.
Perhaps, the clearest cases of male harm with real-life conservation concern are found in heterospecific contexts. If costs of mating are low for males, males are expected to mate relatively indiscriminately, and the eagerness to mate can extend to heterospecific matings. In Skiffia bilineata, a goodeid fish native to Mexico, forced copulations by invasive guppies are a clear threat to native populations . Still, males are not always the sex that causes the majority of the damage during an invasion. While indiscriminate mating habits underlie the problem of feral ruddy ducks, Oxyura jamaicensis, hybridizing with the globally threatened white-headed duck Oxyura leucocephala , a genetic study has found that most crosses in the wild are between female ruddy ducks (the invasive species) and male white-headed ducks .
Another context where male harassment can really matter is when it selects for sexual segregation. Although sexual segregation is not always a consequence of sexual conflict over mating rates , females have been shown to accept suboptimal habitat, either to avoid harmful mating attempts by males  or because males are behaviourally dominant and force females into inferior habitat in the non-breeding season [62,63]. When sexual segregation is pronounced, conservation measures that protect only part of a species’ range might only protect one sex (for a case study on large pelagic sharks and the threat from fisheries see ).
Given that current methods for pest management, conservation and hunting/harvesting often alter the adult sex ratio, the links between male traits, male abundance, sexual segregation and population performance appear understudied . The most popular experimental approach is to contrast monogamy with multiple mating, although it would perhaps be more realistic to consider the scenario in which females always encounter multiple males, but the male : female ratio (or the frequency with which females encounter males) varies. In an intriguing experiment of the latter type, Edward et al.  found that Drosophila melanogaster females experienced a shorter lifespan when exposure to males was increased, but because they also laid eggs earlier in life, this rescheduling of their life history resulted in a net benefit if the population was assumed to be growing (life-history theory predicts that the importance of early-life reproduction is elevated in populations that are expanding in size; ).
The above arguments illustrate that it is difficult to predict a priori the extent to which inter-locus sexual conflict affects population persistence. Male harm may be substantial, but it could be largely mitigated by female counter-adaptations, or might be insufficient to have much impact on population productivity in populations that, despite conflict, still produce many more progeny than can possibly survive. Future studies could aim to quantify explicitly how male harm that has evolved in response to polyandry affects population persistence in a changing environment, and there is a clear need for more theory integrating the ecological and evolutionary consequences of sexual conflict.
5. Demographic consequences of intra-locus sexual conflict
Intra-locus sexual conflict refers to the situation where the optimal phenotype differs between the sexes, leading to an evolutionary compromise that is suboptimal for either sex . Intra-locus conflict might have important consequences for demography; for example, if there is sex-specific selection on a shared trait such as body size, and selection on males moves females off their phenotypic optimum because of cross-sex genetic correlations, female productivity and, therefore, population persistence may be reduced. To our knowledge, no studies have tested for a link between the degree of polyandry and the intensity of intra-locus conflict. However, it seems reasonable to assume that polyandry intensifies intra-locus conflict in many cases, because it imposes additional sex-specific selection pressures relative to monandry and monogamy, as described in §2.
The extent to which sex-specific selection affects population performance probably depends greatly on how much intra-locus conflict has been resolved through the evolution of sexual dimorphism, e.g. through sex linkage or sex-biased gene expression [68,69]. For example, many costly male sexually selected traits such as antlers are absent or reduced in females, which could improve female productivity and, therefore, population fitness. However, recent experiments with Gnatocerus cornutus beetles demonstrate ‘hidden’ intra-locus conflict that exists even when sexual trait expression is limited to males . Beetles were selected for large or small mandibles, a trait that is only well developed in males; as expected, males from the large-mandible lines were more successful at fighting and gaining mates. Even though females lack well-developed mandibles, female fitness was negatively correlated with male mandible size across selection lines because of correlated responses to selection in non-sex-specific morphological traits. This experiment illustrates that female trait evolution will be adversely affected by selection on male-specific traits whenever the latter are genetically correlated with traits expressed in females. Given the ubiquity of genetic correlations among traits, it seems likely that substantial intra-locus conflict will exist even in species with strong sexual dimorphism (though sexual dimorphism should certainly help; [71,72]).
In this context, it is good to remember that selection on males would still negatively impact population growth even if intra-locus conflict were completely resolved through the evolution of sex-limited gene expression. For example, if males grow faster for the duration of parental care, mothers may have to provision sons more (at a cost to daughters); this applies even if growth has evolved to follow precisely optimal sex-specific trajectories. In principle, mothers may then be selected to produce fewer sons, which would mitigate this conflict; however, full compensation of this type might be rare [73, pp. 22, 74]. Connallon et al.  also point out that genes that are only expressed in one sex (and, therefore subject to selection half as often) are expected to harbour more deleterious mutations, such that intra-locus conflict imposes a genetic load even after it has been resolved. Additionally, allowing males to approach their phenotypic optimum might amplify any negative consequences of males on population demography, for example, if the better-adapted males were more able to harass females or compete with them for food or territories. ‘Risky’ male-specific traits (e.g. singing, bright coloration, lowered investment in immunity) might also increase predation and parasite pressure on females, because the males represent easier targets that encourage population growth in natural enemies [5,51,75]. Finally, some of these male adaptations may reduce the density of males to the point where females become mate- or sperm-limited, compromising population fitness. The frequency of ‘evolutionary suicide’ caused by exaggerated sexual traits that harm the survival of the male is debatable [76,77], in part because selection for trait exaggeration should weaken as the operational sex ratio moves towards females. While sperm limitation does occur in nature , we are not aware of evidence showing this to be a consequence of too few males surviving their massive investment in sexual traits. In an intriguingly direct link to multiple mating, however, Charlat et al.  discuss the special case of Wolbachia-infected butterfly populations, where female-mating frequency increases as males become rarer, triggering a cycle of increasing depletion of male-mating resources and female promiscuity. The potential population dynamic consequences appear intriguing, as do cases of male scarcity owing to sex ratio distorting selfish genetic elements .
Lastly, many male- and female-imprinted genes are thought to reflect sexual conflict resulting from polyandry over provisioning of the developing foetus in placental mammals, in a similar fashion to conflict over parental care [81,82]. Loci of paternal origin in the offspring are predicted to draw more resources from the mother than are maternal loci, because males may not be the father of offspring produced later in the female's life. Such genomic tug-of-war has been proposed to be a causative agent of human diseases including preeclampsia and mental illness [83,84].
6. Does sexual selection oppose or augment natural selection?
If sexual selection is indeed stronger under polyandry, how does this impact population fitness? Above, we have mainly considered direct demographic effects of male and female trait divergence under polyandry, but long-term population performance also depends on population-level characteristics such as mutation load and adaptation rate. Most studies consider how these traits respond to sexual selection or the presence of sexually antagonistic alleles, which usually leaves the precise link to polyandry undefined. This is likely because the link between polyandry and the strength of sexual selection is understudied. Sexual selection may tend to increase with the degree of polyandry (e.g. by generating conflict over mating rate, and facilitating both pre- and post-copulatory sexual selection), but one could also conceivably make the opposite prediction: in the extreme case in which all females mate with all locally available males, pre-copulatory selection disappears. Intriguingly, we are not aware of any systematic comparative tests that compare the strength of pre-copulatory sexual selection with the degree of promiscuity, although it has been noted that extra-pair fertilizations in birds can either intensify or weaken sexual selection, depending on the covariance between within-pair and extra-pair success .
For the purposes of the following section, we tentatively assume that polyandry is positively correlated with the intensity of sexual selection across taxa and then review the population-level consequences of sexual selection, while reminding the reader that this link is relatively unexplored.
(a) Sexual selection and population fitness
Theoreticians have argued that sexual selection might have a number of beneficial effects on populations, such as accelerating the removal of deleterious alleles and the fixation of beneficial ones, because males carrying inferior genes should have reduced reproductive success [86–92]. Crucial to this argument is the assumption that genes affecting male reproductive success also have beneficial effects on traits that affect population fitness . For example, an allele conferring high attractiveness in males might also increase fecundity when expressed in females. If loci affecting male reproductive success tend not to affect any other traits, sexual selection might have little effect on population fitness. Conversely, if a sufficient number of alleles that increase male-mating success are detrimental when expressed in females (or cause the males to harm or compete with females), sexual selection would oppose natural selection and lower population fitness [69,93].
A number of recent studies have attempted to measure whether sexual selection has a positive, negative or negligible effect on female or population fitness, with mixed results (reviewed in [90,94]). Some of the strongest evidence comes from selection experiments in Drosophila spp. McGuigan et al.  studied mutation accumulation lines with and without the opportunity for sexual selection. Male mating success was higher in sexually selected lines, and the majority of de novo mutations affecting male mating success pleiotropically affected female productivity in the same direction, suggesting that sexual selection increases both sexual and non-sexual fitness. There was also a non-significant trend for sexual selection to reduce extinction rate. Morrow et al.  eliminated selection on one sex and measured the change in fitness of the other, and found that the fitness of both sexes declined when either males or females were prevented from responding to selection, suggesting that sexually concordant genetic variation outweighs antagonistic variation. However, the fitness of the unselected sex declined more rapidly, demonstrating the presence of significant amounts of antagonistic variation. Sharp & Agrawal  measured the selective consequences of eight deleterious alleles on viability and female fecundity as well as male mating success, and found that in seven out of eight cases natural and sexual selection were aligned (the eighth locus was under sexually antagonistic selection); this is in broad agreement with a large transcriptomic study indicating sexually antagonistic selection on around 8 per cent of all genes . Another study evolved maladapted D. melanogaster populations under enforced monogamy or promiscuity; sexual selection imposed a net cost on productivity, and sexual selection did not improve egg-to-adult viability . Studies of bulb mites Rhizoglyphus robini have also produced evidence that male–male competition is beneficial for populations; sexual selection was found to improve viability  and female fecundity , and to reduce the rate of extinction .
A few studies have examined whether sexual selection accelerates adaptation to a novel environment , which might have important implications for the long-term persistence of populations. In Callosobruchus maculatus beetles, populations with the opportunity for sexual selection adapted more rapidly to a novel food source than those without. However, two studies of Drosophila found no such evidence: Holland  found no difference in the rate at which polyandrous and monogamous populations adapted to thermal stress, whereas Rundle et al.  found no effect of sexual selection on adaptation to novel food. Bonduriansky  also raised the intriguing possibility that sexual selection increases the rate of adaptation and ecological diversification by pushing species off their naturally selected phenotypic optimum, allowing them to cross fitness valleys to reach higher peaks in the adaptive landscape.
(b) Post-copulatory sexual selection and population fitness
Some genetic elements adversely affecting population fitness might have no effect on male mating success, but might be a handicap in post-copulatory sexual selection, such that they could only be screened out by polyandrous females. A fascinating example is provided by studies of Drosophila pseudoobscura populations harbouring an X-linked selfish genetic element that distorts both Mendelian segregation and the sex ratio by killing the Y-bearing sperm of male carriers. Males with this element pass it on in 100 per cent of their gametes, but are very poor sperm competitors . Females mating with a carrier and non-carrier male therefore produce predominantly non-carrier offspring. Females in experimental populations containing the selfish genetic element evolved higher levels of polyandry , and populations prevented from being polyandrous were more likely to be driven extinct by a shortage of males caused by the spread of the selfish genetic element . Post-copulatory screening against carrier males is especially important because females appear to be unable to distinguish between carrier and non-carrier males prior to mating . Selfish genetic elements that harm male fertility may be common, suggesting a taxonomically widespread benefit of polyandry to population persistence . Accordingly, a recent study of the selfish t haplotype in Mus domesticus, another segregation distorter that works by impairing non-carrier sperm in males, found evidence that polyandry may explain the puzzlingly low frequency of the t allele in wild populations .
(c) Complexities and unresolved issues
Although these studies suggest that pre- and post-copulatory sexual selection can improve population fitness by improving mean genetic quality in at least some taxa, it is important to recall that demography matters too: male–male competition can also have strongly detrimental effects on female productivity that might more than cancel any population-level genetic benefits. After it was first established that male harm is widespread, several authors speculated that females mating to harmful males might recoup some or all of their lost direct fitness through indirect benefits such as ‘sexy sons’ [110–113]. The extent to which this actually occurs is debated, and likely varies among taxa [114–118]. However, even though heritable male reproductive success might partially or fully offset female direct fitness lost to inter-locus sexual conflict, we stress that indirect benefits from sexy sons do not diminish the costs of male harm to population viability. This is because male–male competition, by virtue of being a zero-sum game over paternity, tends to select for traits that have a negative impact on the rate of population growth. If male–male competition that harms females is to have a positive effect on population fitness, we must additionally assume a genetic correlation between naturally selected traits and sexually selected, harmful traits. This is not so far-fetched; for example, large male body size can simultaneously predict mating success and harm to females , but might also indicate ‘good genes’ that increase the viability of both male and female offspring. In short, the net effect of sexual selection on population fitness depends on the direct effects of male–male competition on female survival and fecundity, the indirect effects on offspring survival and fecundity, but not the indirect effects on offspring reproductive success.
Adding to this complexity, a recent study  highlighted that the relative amount of genetic variation with sexually antagonistic or sexually concordant fitness effects depends on the population's evolutionary history. Populations near their adaptive peak were found to harbour relatively more sexually antagonistic genetic variation than those adapting to a new environment, because loci with sexually concordant fitness effects are predicted to lose genetic variation in stable environments (in contrast to sexually antagonistic loci). Furthermore, sexually selected traits might interfere with the operation of natural selection. In D. melanogaster, highly-fecund females are more attractive to males and therefore experience more male harm, depressing their fitness and weakening natural selection on female fecundity .
To sum up, it is far from clear how polyandry affects population fitness via its putative effects on the strength of sexual selection. Solving this question requires not only that we know how selection operates on males and females and the extent to which gene expression is sex-limited , but also that we appreciate how the fitness of each sex affects population demography , and, in general, how polyandry influences the strength of pre- and post-copulatory sexual selection. There is tremendous potential for feedback between ecology and evolution that leaves much fertile ground for future empirical and theoretical research.
7. Inbreeding avoidance and genetic benefits of post-copulatory sexual selection
Inbreeding may reduce population viability by increasing the expression of deleterious recessive phenotypes and eroding the genetic variation required to respond to novel biotic and abiotic environmental challenges . In many species, individuals (especially females) avoid mating with close relatives, which increases the heterozygosity and potentially the fitness of their offspring relative to random mating . Polyandry gives the possibility of a second, post-copulatory round of sexual selection to screen out close relatives, as well as low-quality males [18,85].
In a number of species, males have reduced success in post-copulatory sexual selection if they are closely related to the female. For example, female Gryllus bimaculatus crickets mating with both an unrelated male and a brother store fewer sperm from the brother, greatly reducing his share of paternity . Similarly, red junglefowl Gallus gallus store fewer sperm when mating with their brothers . In Poecilia reticulata guppies, female ovarian fluid causes the sperm of close relatives to swim more slowly than sperm of unrelated males, lowering the number of eggs fertilized by related males following polyandry . Blue tits constrained to pair with close relatives by social monogamy seek extra-pair copulations with other males, reducing inbreeding depression in their offspring . Evidence for post-copulatory inbreeding avoidance has also been found in mammal, reptile, amphibian and plant species, although there are several counter-examples in which it was searched for but not detected (reviewed in ). Whether females evolve mechanisms that disfavour related males’ sperm likely depends on the relative costliness and frequency of inbreeding . In species where females often encounter their relatives and cannot avoid mating with them, e.g. in species that typically live in small and fragmented populations, post-copulatory avoidance of inbreeding may significantly enhance female and population fitness. However, species that have only recently been reduced to small population sizes by human activity are less likely to possess post-copulatory blocks against inbreeding, because these would not have been needed at their original population sizes; polyandry might, therefore, do relatively little to increase outbreeding.
Even when levels of inbreeding are held constant, potential sires may differ greatly in their genetic compatibility. The majority of loci have epistatic fitness effects (i.e. the relative fitness of different alleles depends on other loci), so the optimum male is one whose genes will interact favourably with those of the female when brought together in the offspring. Low fitness of hybrids between genetically divergent groups (‘outbreeding depression’) is thought to be predominantly caused by the break-up of co-adapted complexes of epistatically interacting loci, which can select for mate choice for individuals from the same genetic group [128,129]. The frequency of maladaptive hybridization can be increased by anthropogenic environmental change (e.g. habitat loss and artificial introductions), making genetic compatibility of potential importance to conservation.
Polyandry might facilitate post-copulatory sexual selection against incompatible males in some species , with beneficial consequences for population viability. For example, the endangered Gouldian finch Chloebia gouldiae occurs in two distinct races, which are easily distinguished by their head colour, which is either black or red (other colours are also rarely observed). Inter-morph matings produce poor quality offspring, many of which do not survive to maturity . Females frequently engage in extra-pair copulations, and there is a strong post-copulatory bias towards same-morph males; females paired with an opposite-morph male can, therefore, receive both paternal care and healthy offspring by being polyandrous . Polyandry coupled with the strong genetic incompatibility between the morphs means that population fitness is likely to be significantly higher than one would predict from the high number of mixed-morph breeding pairs observed in nature (20%; ). In Drosophila, populations of Drosophila yakuba experimentally evolved in sympatry with the closely related Drosophila santomea evolved post-copulatory mechanisms that reduced the proportion of their eggs fertilized by the other species in only ten generations (hybrids are largely sterile): sympatrically evolved D. yakuba that mated with a conspecific lost his sperm more quickly from storage and therefore remated more quickly than controls . Additionally, populations of Tribolium beetles that had recently experienced a genetic bottleneck evolved higher rates of polyandry, likely because polyandrous females had a greater proportion of their eggs fertilized by genetically compatible males . Many more studies have found greater offspring viability in polyandrous females, implying that genetic benefits of post-copulatory sexual selection may be widespread .
Post-copulatory sexual selection can also provide ‘good genes’ whenever success in sperm competition or cryptic female choice is genetically correlated with traits under natural selection . An interesting example of conservation importance is provided by studies of hybridization between farmed and wild fish, such as salmon and cod. Farmed fish escape in considerable numbers, and may potentially introduce their domesticated (maladapted) genes into wild populations . Salmon and cod are polyandrous, so the competitiveness of farmed males in both pre-and post-copulatory sexual selection is a key predictor of gene flow from farmed to wild stocks. In some cases farmed males are highly competitive , increasing the ecological impact of escapees, whereas in others they are less competitive . These results underpin the importance of mating system variation to conservation biology and local adaptation in general.
Lastly, polyandry may allow females who have mated but subsequently encounter a better-quality male to ‘trade-up’, especially in species with last-male sperm precedence . Post-copulatory sexual selection could, therefore, augment mate choice, amplifying any population-level benefits  of the latter. However, this also has the flipside that effective population sizes (see §9) can be reduced if the successful sires constitute a smaller subset of all males.
8. Evolutionary consequences of elevated within-family genetic diversity
Polyandry frequently causes mixed paternity within the clutches of individual females, meaning that similarly aged offspring may be half-siblings. This increase in within-family genetic diversity may have far-reaching evolutionary consequences, and have both positive and negative effects on population viability.
Sibling rivalry, which is ultimately caused by the difference in relatedness between an individual to itself versus its sibling, is predicted to be more intense in broods containing half-siblings . Conflicts among siblings may reduce productivity if they lead to the evolution of costly competitive traits such as begging . Conflicts among siblings can be intense, as exemplified by tadpoles of some toads and salamanders, which develop into cannibalistic morphs at high densities, potentially leading to population collapses . Studies suggest that cannibals are less likely to develop in highly related groups  and may refrain from eating close kin , suggesting a possible population-level cost of polyandry. Conflicts might also be more subtle; for example, the evolution of dispersal under offspring control depends on the degree of relatedness among siblings. Polyandry could select for either higher or lower dispersal distances depending on how it modifies sibling conflict relative to parent–offspring conflict . Dispersal ability is a key predictor of persistence, especially in fragmented populations or those affected by environmental change, highlighting another potentially important evolutionary consequence of polyandry for population fitness.
An alternative possibility is that increased within-family genetic diversity resulting from polyandry actually reduces competition among siblings by allowing them to occupy more distinct ecological niches. Elevated offspring performance in polyandrous broods is often assumed to result from ‘good genes’ gained during post-copulatory sexual selection, but recent work highlights that reduced competition may provide a widespread alternative possibility. Two studies of marine invertebrates found that polyandry elevated offspring performance even when pre- or post-copulatory sexual selection was experimentally precluded, suggesting that genetic diversity benefits families [146,147].
By increasing within-family genetic diversity, polyandry may also affect the efficacy of selection when selection is relatively soft and groups are composed of siblings . Polyandry should increase the mean variance in fitness among competing siblings relative to monandry, because some offspring patrilines are expected to be fitter than others. Increased variance in fitness makes selection more effective, leading to more rapid adaptation , at least in traits that affect competition among siblings (which may or may not also increase population persistence). In other words, polyandry improves the mean competitiveness of individuals that survive sibling competition because it pits strong competitors against weaker ones more often than does monandry, similar to ‘seeding’ in sports such as tennis. The population-level benefits of this process should be especially pronounced if low-quality individuals die early in development and then free up resources for their fitter siblings. Many birds produce an excess of offspring in every clutch, and may even expedite the deaths of the weakest by treating them roughly . In insects such as flies and butterflies, females often lay many eggs on a food resource, and the death of some individuals leaves more for the others. Broods with a high variance in fitness may lose the weaker individuals more quickly, increasing the number or quality of surviving offspring. Polyandry might therefore provide direct and indirect benefits, even when males that produce high-quality offspring are not more successful in sexual selection.
9. Effect of polyandry on the effective population size
The effective population size (Ne) is a major determinant of the fixation probabilities of both beneficial and detrimental alleles. As Ne increases, the relative importance of stochastic changes in allele frequency (genetic drift) declines, such that negatively selected alleles have a greater chance of drifting to fixation in smaller populations (reviewed in ). Conversely, beneficial mutations have a lower chance of being lost owing to drift before they can fix. Small populations are therefore expected to accumulate more deleterious mutations and fewer beneficial ones than larger populations. The decline in fitness may further reduce Ne (e.g. by reducing fecundity or survival), increasing the rate of fitness decline and trapping the population in an accelerating downward spiral (the ‘extinction vortex’ or ‘mutational meltdown’, [152,153]).
The same set of reasons that make it difficult to predict whether polyandry intensifies or weakens sexual selection also lead to different possible relationships between polyandry and Ne. Polyandry generally increases Ne relative to monogamy , but if it combines with paternity skew (e.g. because of last-male sperm precedence) towards specific males, Ne will be negatively affected. Polyandry can therefore help protect against extinction if a large number of males mating elevate Ne, but the opposite result is also possible.
10. How does it all add up?
Above, we have highlighted a plethora of ways in which polyandry could positively and negatively affect population fitness, and consequently rates of extinction. We also emphasize that most of the putative links between polyandry and population fitness often have no visible demographic effects. For example, male-induced reductions in female fecundity, or genetic benefits of post-copulatory sexual selection, will have negligible demographic consequences if females are still able to produce many more progeny than can survive at density-dependent equilibrium. Positive and negative effects of polyandry on demographic parameters may only become important once birth and death rates are modified by environmental change. This, of course, makes testing the ideas challenging: for example, a healthy population may show little evidence that male harm reduces population growth rates or densities, yet this harm could make the population more vulnerable if an adverse change in the environment erodes the ‘buffer’ of surplus young.
The indirect nature of this link might explain why polyandry and sexual selection have received mixed support as predictors of extinction risk. For example, sexually dichromatic birds were found to have a higher rate of local extinctions in a 21 year dataset of North American bird counts , and testis size predicts current perceived risk of extinction in a dataset of 1030 birds , but there is no such effect in a dataset comprising 1007 species of mammals (see also [94,157]). More large-scale ecological and comparative studies appear welcome in this context. For example, in antbirds, dichromatic species and those with presumed stronger sexual signalling (measured as song pitch) have larger ranges, supporting the idea that sexual selection and natural selection might act in synergy. However, cause and effect will often be difficult to disentangle. For example, extra-pair paternity in birds covaries with higher population-wide genetic variability . Does polyandry maintain much variability in this case (for why it can do so see ), or does the presence of high genetic variability increase the benefits for females of seeking extra-pair copulations? Another type of study that appears surprisingly rare, both in an ecological and evolutionary context, is the simple manipulation of adult male and female numbers. For example, in a correlative study, a high proportion of male frogs diminished froglet survival ; in an experimental study, lizard demography proceeded in a downwards spiral in enclosures that were initialized with a male-biased population . As a whole, it appears that too few studies consider the ecological consequences of sexual conflict in a manner fully informed by population ecology; much remains to be done.
We thank Jean Clobert and an anonymous reviewer for helpful comments on the manuscript.
One contribution of 14 to a Theme Issue ‘The polyandry revolution’.
- © 2013 The Author(s) Published by the Royal Society. All rights reserved.