Sexual selection in hermaphrodites, sperm and broadcast spawners, plants and fungi

Abstract

Darwin was the first to recognize that sexual selection is a strong evolutionary force. Exaggerated traits allow same-sex individuals to compete over access to mates and provide a mechanism by which mates are selected. It is relatively easy to appreciate how inter- and intrasexual selection work in organisms with the sensory capabilities to perceive physical or behavioural traits that signal mate quality or mate compatibility, and to assess the relative quality of competitors. It is therefore not surprising that most studies of sexual selection have focused on animals with separate sexes and obvious adaptations that function in the context of reproductive competition. Yet, many sexual organisms are both male and female at the same time, often lack sexual dimorphism and never come into direct contact at mating. How does sexual selection act in such species, and what can we learn from them? Here, we address these questions by exploring the potential for sexual selection in simultaneous hermaphrodites, sperm- and broadcast spawners, plants and fungi. Our review reveals a range of mechanisms of sexual selection, operating primarily after gametes have been released, which are common in many of these groups and also quite possibly in more familiar (internally fertilizing and sexually dimorphic) organisms.

This article is part of the themed issue ‘Weird sex: the underappreciated diversity of sexual reproduction’.

1. Introduction

In The descent of man, and selection in relation to sex Darwin [1] wrote that ‘In the lowest classes the two sexes are not rarely united in the same individual, and therefore secondary sexual characters cannot be developed. […] Moreover it is almost certain that these animals have too imperfect senses and much too low mental powers to feel mutual rivalry, or to appreciate each other's beauty or other attractions’ (p. 321). Three implications are imbedded in these statements (first used by Michiels [2] in his chapter on mating conflicts and sperm competition in hermaphrodites). First, simultaneous hermaphrodites are not subjected to sexual selection. Second, sexual selection requires the ability to appreciate ‘beauty or other attractions’ and to gauge a rival's competitive ability. And third, anything that is not an animal can safely be ignored with respect to the study of sexual selection. Darwin was probably correct in asserting that not all organisms have the mental powers to appreciate ‘beauty or other attractions’, at least not the way we would perceive such characteristics, or feel rivalry towards a potential competitor. Yet, as we outline here, studying exactly the kind of organisms Darwin implicitly and explicitly ignored provides fascinating insights into sexual selection.

In Darwin's defence, it is not surprising that he disregarded the ‘lower’ organisms and plants when devising his theory of sexual selection. After all, he only started to think about the ways organisms find sexual partners because his theory of evolution by natural selection could not account for the many traits that seemed to lower the survival of the organism that carried them. Hence, from the beginning the study of sexual selection has been biased towards the usual suspects: sexually dimorphic, internally fertilizing animals. Moreover, it took the widespread use of DNA fingerprinting technologies to unequivocally determine paternity before the active role of females in mating was truly appreciated (nicely summarized in Birkhead's book [3]). And lo and behold, females turned out to be active participants during mate choice, and furthermore were shown to pursue matings with multiple males within a single reproductive episode (polyandry). Hence, not only is there the need to distinguish between a female's ‘genetic’ and ‘social’ mate, we have also come to realize that sexual selection does not necessarily end with copulation. Post-copulatory sexual selection, which encompasses sperm competition (competition among sperm for fertilization [4]) and cryptic female choice [5] (where females influence the outcome of such contests [6]), provides a plausible explanation for many apparently bizarre reproductive traits such as giant sperm (more than 5 cm) in the fruit fly Drosophila bifurca [7] and the convoluted reproductive tracts of female waterfowl [8]. Thus, the realization that polyandry is widespread takes us back to where sexual selection begins, with competition among gametes over fertilization [9].

Because of polyandry, we adopt Jennions & Kokko's [10] definition of sexual selection, which recognizes that mating does not guarantee parentage when females mate with multiple males. Thus, ‘sexual selection favours traits that improve the likelihood of fertilization given limited access to opposite sex gametes due to competition with members of the same sex’. The definition thus acknowledges that traits subject to sexual selection may not be as visually spectacular as the elaborate displays of, say, a male peacock, but this does not make them any less important to an individual's reproductive success. In fact, as we argue in §6, we may miss a suite of more subtle mechanisms that profoundly affect sexual selection if we stick to studying the usual suspects.

The organisms that are the topic of our review are often dismissed as ‘unusual’, simply because they have been less studied and their sex life happens to be rather different from ours. Unlike the more conventional study organisms, all or most selection acts after the gametes have been released. They include many simultaneous hermaphroditic animals in which mating partners interact directly, spermcasters and broadcast spawners as well as wind-pollinated plants that never meet during reproduction, plants that require a third ‘force’ (i.e. pollinators), and the immensely diverse fungi, including yeasts, moulds and mushrooms. In this prospective review of sexual selection in these groups, we first discuss the sexual arena of simultaneous hermaphrodites before moving on to systems in which sexual partners have no direct interaction during reproduction (spermcasters, broadcast spawners and wind-pollinated plants). From there, it is a relatively small step to move to plants that require a third force during reproduction (i.e. pollinators in the form of insects or birds). We end our taxonomic treatment of sexual selection by covering the fungi, a group displaying a bewildering array of breeding systems and reproductive characteristics. Our review highlights the enormous diversity of breeding systems displayed by these less-studied organisms, but also remarkable convergence in the ways that selection targets gametes when they compete (or are selected) for fertilization. We conclude by arguing that the lessons learned from moving out of one's comfort zone and venturing into the lesser-known realms of sexual selection can provide critical insights into the mechanisms of (gamete-level) sexual selection in more familiar taxonomic groups.

2. When you are both male and female

Here, we will mainly discuss simultaneous hermaphrodites (hereafter simply referred to as hermaphrodites) in fungi and mobile animals in which fertilization requires direct contact between mating partners. While some aspects of hermaphroditic fungi and mobile animals are also applicable to sessile hermaphrodites, including angiosperms, we discuss those in §§3–5 as these groups have some unique properties due to the fact they never interact directly. By definition, hermaphrodites produce both small, more numerous gametes (sperm), and large, less numerous gametes (ova). If we then assume that ova, in general, are limiting while sperm are abundant, it follows that a hermaphrodite might prefer to donate sperm to a particular mating partner, while not necessarily wanting to receive sperm from that same partner [5]. Such preference for the male role can explain why the marine flatworm Macrostomum lignano appears to remove sperm by sucking their female genital pore after mating [11]. While M. lignano only lays one or two eggs per day, they mate, on average, 24 times in 4 h [11]. Such a high mating rate is clearly not necessary to fertilize the small number of eggs, so their main goal appears to be to donate sperm and to remove sperm from their mating partner. Presumably more sperm are removed from less-desirable mating partners, as ‘sucking’ takes place straight after mating [11]. Hence, the level of sperm sucking is probably directly related to the relative desirability of the partner as male-mate. But where exactly does this preference for the male role come from? Once an individual has received sperm after the first mating in the female role, in many instances reproductive output cannot be increased by remating shortly after as female, whereas remating in the male role can increase reproductive output. Basically, the Bateman gradient (i.e. the statistical relationship between mating success and reproductive success) is steeper for the male role than for the female role [12], resulting in more ‘male–male’ competition [13,14]. Such male–male competition is very obvious in another marine flatworm, Pseudoceros bifurcus, where individuals ‘duel’ with their penises, both trying to traumatically inseminate their partner while attempting to avoid being inseminated [15]. When the latter fails, sperm from multiple partners are likely to compete for fertilization, setting the scene for sperm competition, cryptic female choice and conflicts over the fate of received ejaculates.

The high mating rates of hermaphrodites and their frequent failure to avoid being inseminated selects for traits that maximize fertilization success during sperm competition. At the same time, selection acts to increase the ‘female's’ control over which sperm fertilizes the eggs. Using comparative analyses of the flatworm genus Macrostomum, Schärer et al. [16] mapped sperm morphology to mating biology to determine the extent to which the evolution of sperm morphology is driven by sperm competition and cryptic ‘female’ choice. The answer: a lot. Sperm morphology is most complex when competition for fertilization is fierce. Competition is fierce when sperm compete directly within the reproductive tract over fertilization. But sperm may also have to combat ‘female’ countermeasures (cryptic female choice) aimed at preventing fertilization by particular males, or to prevent polyspermy. Interestingly, with the evolution of traumatic insemination, sperm morphology loses its complexity and becomes adapted to rapidly move through the body towards the ‘female’ reproductive organs [16]. Traumatic insemination most likely evolved to circumvent cryptic female choice, but also allows an individual to inseminate while lowering the risk of being inseminated.

While sperm competition and cryptic female choice are not unique to hermaphrodites, the ability to influence the sex role of one's mating partner (in other words change the allocation from male function to female function or vice versa) is a unique feature in these systems. Given that competition over the male role is fierce, anything an individual can do to reduce its mate's investment into male function can lower competition. Moreover, it pays to increase a partner's investment into female function, as that will increase the number of eggs that can be fertilized provided the organism stores sperm for later fertilization [17]. ‘Love darts’, found in many species of land snails, are thought to manipulate the reproductive physiology of the recipient, either by increasing the fertilization success of the sperm that is transferred at the same time as the love dart, or by functioning as an anti-aphrodisiac (see Schärer et al. [14] for a recent review). Likewise the seminal fluid of the freshwater snail Lymnaea stagnalis contains proteins that reduce both sperm transfer and fertilization success of a subsequent mating by the inseminated partner while at the same time manipulating female investment of that partner [18], similar to the anti-aphrodisiac found in seminal fluid of, for example, Drosophila [19].

Mating in one sex role can have significant effects on subsequent matings in the other sex role in mushroom-forming fungi. In these fungi, mating occurs between two haploid mycelia by exchange of nuclei (figure 1). In the male role, a mycelium donates a nucleus but no cytoplasm (hence the analogy to ‘males’), which is incorporated into the female mycelium (containing the cytoplasm). These investments reflect those of true (anisogamous) gametes (female gametes contribute the cytoplasm and male gametes do not) and are expected to have similar effects with respect to operational sex ratio and opportunities for multiple matings [20]. The male nuclei migrate and divide through the female mycelium, such that the latter becomes completely fertilized, but the nuclei within the mycelium do not fuse. The fertilized mycelium now consists of cells that each contains two nuclei, the originally ‘female’ one and the newly received ‘male’ one [21]. When this mycelium now meets an unfertilized haploid mycelium, the latter can become fertilized by either of the two nuclei from the former but the former can no longer accept a nucleus (i.e. mate in the female role) [22]. The first mating, in the female role, thus reduces further male-mating success, as it introduces direct competition in later mating opportunities. Similar trade-offs between sex roles in which previous matings in one sex role affect the other sex role negatively are likely in other hermaphrodites too. It does seem probable that such trade-offs are more probable when mating partners interact directly; when mating partners do not interact directly, there seems less scope for manipulation of any kind. An exception might be the gametophytes of several species of homosporous ferns. Without even getting into direct contact, gametophytes can influence the sex role of prospective mating partners. Spores in these species by default develop into either females or hermaphrodites and as they develop secrete hormones into their environment. Spores that receive these hormones do not develop into females or hermaphrodites, but into males. The obvious difference between the ferns and mushroom-forming fungi is that both the male and female roles are likely to benefit from the production of and response to the hormones excreted by the fern spores [23].

Figure 1.

Generalized fungal life cycles. (a) Single-celled yeasts grow by mitotic cell division as haploid (white in figure) or diploid (grey) cells. Timing of mating and meiosis differ between species, e.g. budding yeast mates as soon as possible and is predominantly diploid, whereas fission yeast starts meiosis directly after mating. (b) Filamentous ascomycete fungi grow as haploid mycelia on which fruiting bodies are formed that each are fertilized independently and within which meiosis immediately leads to haploid offspring again. (c) Mushroom-forming fungi initially grow as haploid mycelia (solid colour) with one nucleus per cell and mating occurs when two mycelia meet by reciprocal exchange of nuclei. These nuclei remain separate in the mycelium (called a dikaryon; striped) that can form fruiting bodies that produce haploid spores, but that can still fertilize a third haploid mycelium. Adapted from [20].

3. When partners do not interact directly

Sex poses unique evolutionary challenges to species where mating partners do not interact directly at mating. In many marine and freshwater animals, for example, externally shed sperm and eggs must navigate an often-turbulent three-dimensional environment (the ocean, a lake or a river) to locate and fertilize gametes from the opposite sex. This reproductive system—known as broadcast spawning—is prevalent among many sessile and sedentary marine invertebrates such as corals, echinoderms, molluscs and ascidians [24]. In other sessile marine invertebrates, mating can involve the release of sperm that are dispersed to fertilize eggs that have been retained by conspecifics—a system known as spermcasting [25], while in wind-pollinated plants (approx. 10% of all angiosperms [26]) externally shed pollen grains are similarly dispersed passively to be captured by female floral structures. A challenge common to all such species is that turbulent diffusion can very quickly reduce the concentration of gametes to the point where fertilization rates are limited [24,27,28]. Gamete limitation may be further compounded by low population density, the distribution of the sexes across space (e.g. distance between males and females), the presence of gamete predators and other factors. Given these uncertainties surrounding fertilization, one might be forgiven for assuming that fertilization is determined entirely by chance in these systems. As we show in §§4 and 5, however, an increasing body of evidence from these systems now challenges this assumption by demonstrating a range of putative mechanisms of sexual selection moderated exclusively by interactions operating after gamete release.

A number of studies of externally fertilizing taxa suggest that gametic incompatibilities (i.e. the inability of male gametes to penetrate and fertilize an ovum) can severely limit fertilization success (reviewed by Kosman & Levitan [29]). We should therefore expect selection to favour mechanisms that maximize the likelihood of fusion between compatible gametes. Arguably the best-characterized mechanisms in this regard are the self-incompatibility systems in flowering plants, which prevent fertilization from self-pollen and thus facilitate inbreeding avoidance [30,31]. Plants display multi-step mechanisms that favour conspecific interactions between male and female gametophytes [32] For instance, in plants with dry stigmas, the ability of pollen grains to penetrate the stigma surface relies on cross-linking the pollen cell wall with the lipids, carbohydrates and proteins of the stigma [33]. Cross-linking facilitates access of fluids to the pollen grain, thus activating the pollen grain and triggering its growth. Hydration of pollen in plants with dry stigmas appears to be highly species-specific [34,35], suggesting that divergence in the cues used might favour the rapid evolution of some male–female reproductive lipids, carbohydrates or proteins in plants [30]. Once pollen grains penetrate the stigma water gradients [36], chemical signals can orient their growth towards the egg sac [37], a process that is sensitive to species identity [38,39]. Finally, chemical signals guide pollen tubes to the micropyle, the opening of the ovule, where pollen enter, fertilize the egg and produce the embryo and the endosperm [40,41]. Variation in chemical gradients next to the micropyle involving metabolites like γ-aminobutyric acid might determine conspecific reproductive interactions. It would be interesting to see whether the chemical gradients from stigma to micropyle have a common genetic basis and whether there is scope for sexual selection.

In many ways, spermcasting is more comparable with wind pollination in terrestrial plants or water pollination in aquatic plants and mosses and ferns [42] than it is to broadcast spawning in marine invertebrates. Although all of these groups exhibit remote mating and lack any form of precopulatory mate screening, spermcasters and wind-pollinators share the distinction that eggs/ovules are retained for fertilization in close proximity to the (maternal) recipient. These attributes create opportunities for differential sperm utilization prior to fertilization [43,44] that are absent in broadcast spawners. In colonial, hermaphroditic spermcasting bryozoans and ascidians, for example, sperm have been shown to trigger differential maternal investment in eggs that favour fertilizations between compatible individuals [45–47]. Such self-incompatibility mechanisms parallel the outcrossing strategies typical of many flowering plants (see above). By contrast, patterns of maternal investment are fixed prior to gamete release in broadcast spawners, thus offering them no opportunities to invest differentially in reproduction. In broadcast spawners, therefore, the opportunities for mate choice and/or mating competition are constrained to interactions involving gametes or their associated accessory fluids [48].

Chemical signalling between eggs and sperm may provide a mechanism for avoiding gametic incompatibilities in broadcast spawning taxa. Sperm chemotaxis has been widely documented in broadcast spawning marine invertebrates [49], but the phenomenon occurs widely among metazoans [50], including humans and other mammals [51]. In addition to increasing the likelihood of fertilization overall, the chemoattractants involved in sperm chemotaxis may also fulfil a sexually selected function by moderating intraspecific sperm selection in favour of genetically compatible mates (reviewed in [48]). Two recent studies on a broadcast spawning mussel, Mytilus galloprovincialis, point to such a function by showing that chemoattractants from different female egg clutches elicit differential sperm chemotactic responses according to the specific combination of chemoattractants and sperm that were tested. In short, chemoattractants preferentially attract sperm from compatible males [52] and these gamete ‘preferences’ reflect (i) changes in sperm swimming behaviour and (ii) patterns of offspring survival [53]. The inevitable conclusion from these studies is that the behaviour of sperm is fine-tuned to respond differentially to the chemical signals emitted from different egg clutches, thus facilitating mate choice for genetically compatible individuals.

Once an externally shed sperm cell or pollen grain has successfully ‘navigated’ (or passively found) its way to an egg (or stigma), fertilization can then be contingent on reproductive proteins that ultimately moderate the fusion of gametes [54]. Broadcast spawning marine invertebrates are established models for understanding the role of gamete reproductive proteins (GRPs) in fertilization [30,55]. One of the best-characterized GRP systems occurs in echinoderms and involves the sperm-binding-protein Bindin and its corresponding egg binding receptor protein-1 (ERB1). In the sea urchin genus Echinometra, Bindin genotypes are polymorphic within species [56] and eggs from different females exhibit strong but differential affinities for sperm on the basis of their Bindin genotype when they compete to fertilize eggs [57]. Further evidence that intraspecific variation in GRPs moderates reproductive success has been reported in other sea urchin species [58–60], suggesting that sperm competition may act as a driving force for GRP divergence [54].

Interestingly, there is little (if any) evidence to suggest that sexual selection targets universally preferred or competitive traits in weakly mobile or sessile organisms with ‘remote’ fertilization [48]. This contrasts with the high number of studies on species with behavioural components of sexual selection, which have identified cellular equivalents of the ‘peacock's tail’ (sensu [7]) in studies of post-ejaculatory sexual selection or traits that confer a general advantage in sperm competition (reviewed in [61]). There is also evidence for relationships between various ejaculate traits (e.g. sperm velocity, size) and competitive fertilization success in some externally fertilizing vertebrates (e.g. salmon [62]), but to our knowledge similar evidence does not exist for the organisms considered here. One possible explanation for this disparity in patterns displayed by conventional species and those covered in this review, is that selection for gametic incompatibility avoidance may trump selection for intrinsically ‘good’ gamete traits in species where there is no opportunity for pre-ejaculatory mate assessment. Alternatively, the theoretical predictions for evolutionary responses to sexual selection by broadcast spawning (and presumably wind-pollinating) taxa may differ fundamentally from those generated for internal fertilizers [63], for example, because of complex evolutionary dynamics associated with changes in sperm abundance (box 1). Because of these unique selection pressures, the evolutionary responses to sexual selection in non-gonochoristic taxa may differ fundamentally from those reported for more familiar species, thus rendering any such adaptations more difficult to observe or anticipate (see box 1 and [71]). Finally, the relative paucity of studies that have actually tested for such patterns suggests that we are not looking hard enough [48].

Box 1. The sperm environment and complex (co)evolutionary dynamics in broadcast spawning invertebrates.

Sperm limitation, resulting in incomplete fertilization due to insufficient sperm, is thought to be a common phenomenon in the marine environment [27,28]. However, the reverse situation—sperm excess—may also occur when multiple individuals spawn within dense breeding aggregations [64,65]. The latter phenomenon—sperm excess—can lead to polyspermy, where more than one sperm enters the egg, typically leading to developmental failure and therefore constituting a high reproductive cost to females (e.g. [66]). Female reproductive success can therefore be limited at both ends of the sperm concentration continuum, with incomplete fertilization under sperm limitation and a high prevalence of embryo failure under saturating sperm concentrations [63].

The sperm environment continuum can generate complex coevolutionary dynamics between males and females. One central theoretical prediction is that under high sperm concentrations (and therefore high sperm competition) males will face selection to release even more sperm despite the increased risks of polyspermy to females [63]. For their part, females will be under increased selection to block polyspermy under conditions of sperm excess, for example, by evolving highly polymorphic sperm–egg receptor proteins that reduce fertilization rates [67,68] or through the facultative adjustment of egg size so that eggs are smaller targets when sperm concentrations are high [69]. Sperm competition can therefore be an important driver of sexual conflict (i.e. conflicting evolutionary interests between the sexes over reproduction) in broadcast spawners. Interestingly, by incorporating the effects of sperm concentration into classic game-theoretic models of sperm competition, the evolutionary predictions for sperm release strategies in broadcast spawners become radically different than those generated for ‘conventional’ organisms [63,70]. This may account for the somewhat paradoxical patterns of sperm release strategies under sperm competition (e.g. release fewer sperm when sperm competition intensifies [63]) and explain why evolutionary responses to sexual selection may not be immediately ‘visible’ to researchers working on more familiar taxa.

4. When a third force is needed

The sole purpose of flowers is sex. It therefore seems rather surprising that the potential for sexual selection in plants has mostly been ignored [23]. Apart from their assumed lack of ‘sentience’, the difficulty in applying sexual selection concepts to plants due to their putative lack of clear sexual dimorphism probably goes a long way to explaining why plants are not more often discussed in this particular context. A difficulty with applying ‘sexual dimorphism’ to plants [72] is that close to 90% of angiosperms are hermaphrodites [73] (note that the botanical definition of a hermaphrodite is more strict than in animals, as the same plant that carries both male and female flowers is not considered to be hermaphroditic). Yet, flowers and their structures, including stigmas, petals and styles, facilitate both intrasexual competition and female choice in angiosperms [23,74]. While wind-pollinated plants face similar constraints as spermcasters (see §3), many angiosperms require a third force as pollinator: an animal. As a result, sexual selection may shape the evolution of petals, nectar and other gifts aimed at pollinators, and traits that increase pollen export and acquisition [75,76]. Animal pollination removes some of the constraints that spermcasters face, as pollen can move further than sperm. Assuming close coevolution between plant and pollinator, the deposition of pollen will be more targeted than in spermcasters or wind-pollinated plants, where the movement of male gametes is dependent on the whims of currents or wind. (It is thus curious that animal pollination has not evolved in sessile spermcasters [77].) Because pollinators often deposit pollen from multiple flowers on the stigma, traits such as long styles, increased stigma surface and receptivity, and ovule abortion, may have evolved in response to female choice for the best ‘male’ [78,79]. The strength with which sexual selection acts on flowers and their structures is largely dependent on whether these structures evolved due to selection on male or female function. As in all simultaneous hermaphrodites, selection acts more strongly on male than female function [80,81] because resources required to produce seeds and fruits limit female fecundity rather than male availability (see §2).

A unique feature of plants that potentially affects the way sexual selection acts is the fact that gametes in the plants are multicellular structures that contain both tissue and the sperm and/or the egg. However, it is not clear whether such extra complexity provides more or fewer opportunities for sexual selection relative to the naked sperm and egg cells in other organisms. On the one hand, plant gametophytes are exposed to unpredictable environments, particularly when the sporophytic phase is short such as in mosses (note that the same applies to filamentous ascomycete fungi, see §5), which may provide stronger links between gametophyte viability and reproductive success. On the other hand, multicellular gametophytes might create fitness trade-offs if many of the genes they express are also expressed in other plant organs, but in an antagonistic way.

Being sessile, the evolution of flower morphology in hermaphrodites is clearly linked to the way in which pollen is dispersed [82]. In wind-pollinated plants, petals might be of little use compared with the size of the surface of the stigma and their extended periods of receptivity to pollen [78,83]. Pollinators, unlike the wind or current, clearly offer a sophisticated way to remove and transfer pollen among plants, and therefore increase reproductive success via both male and female functions. From the production of pheromone-like molecules (in some instances, mimicking the sex pheromone of the pollinator [84]) to the coevolution of flower spurs and bird beak lengths, plants have evolved traits that facilitate pollen removal and transfer by animals. Darwin's [85] famous prediction of the existence of a pollinator whose tongue should be over 20 cm long to reach the nectar of Angraecum sesquipedale is perhaps one of the most startling examples of such floral trait evolution (the actual pollinator that fits Darwin's description, Xanthopan morgani, was discovered 20 years after his death [86]). Similar examples exist in hummingbird-pollinated flowers, and in striking fashion, in the genus Aquilegia, where the length of nectar spur has tracked serial shifts to pollinators with longer tongues [87].

Mechanisms of pollen removal are as varied as the organisms that practise it. Bees, which pollinate most flowering plants, have specialized mouthparts for sucking nectar, and complex leg structures, like brushes and combs, that help to move pollen grains to their pollen baskets. Because pollen serves as food for developing bee larvae, bees remove most of the pollen from the ‘mating pool’, potentially forcing bee-pollinated plants to invest even more in male function. Other pollinators mainly collect pollen passively. Hummingbirds and moths have elongated and curved beaks or tongues that match spur lengths, and many birds and bats capture pollen on their foreheads and faces as they reach for nectar rewards [88]. Variation in nectar reward (and the incidental collection of pollen, with the exception of bees) can lead to variation in reproductive success within a population. In protandrous species (male flowers arise before female flowers), male–male competition will be severe when pollinators are limiting and plants are likely to invest more heavily in the production of nectar. And while a single pollinator visit can result in fertilization of all ova (thus maximizing female fecundity), male reproductive success continues to increase with increasing visitations (provided there are still ova to be fertilized) [89]. While it is tempting to assume that many flower traits that affect pollen removal are driven by sexual selection, other selective explanations, such as selfing-avoidance, can also play a role and should not be dismissed out of hand [90].

While sexual selection might play a role in maintaining variation in floral traits, whether or not it played a role in the evolution of novel flower structures is unknown. For instance, the evolution of carpels (the vessel of the ovules), and the corresponding seclusion of ovules and seeds, is thought to have occurred in response to multiple selective pressures (e.g. including pollen capture and guidance, protection from predators and seed dispersal [91,92]), some of which theoretically might have increased reproductive success via male or female function. In cycads and many relatives, which display ancestral forms of pollination, sticky exudates from leaves or cones producing ovules trap pollen grains. Insects using pollen as a food source find the exudates, and thus act as pollen vectors. However, while engaging in such behaviour they also consume ovules [93], therefore creating a reproductive cost to females. One way to circumvent the cost of egg predation would be to protect them (e.g. via the evolution of a carpel), which would increase reproductive success via female function, but would also maintain increased reproductive success via male function. This exercise does not presume that the evolution of carpels happened under these conditions, but it does illustrate that sexual selection and its interaction with natural selection can be a powerful force in not only creating trait variation within populations but also perhaps in driving trait innovation in nature. As such, some flower traits might have evolved once as a major evolutionary step in response to reproductive costs.

We should not ignore female function. Contrary to the general rule, pollen is often limiting in oaks and other masting plants that produce seeds synchronously once a year. Furthermore, masting species often abort many of their seeds, even when experimental pollen is abundant, presumably due to resource limitation [24]. Selective abortion of ovules can be a mechanism to maximize reproductive success via female function as well as a means to favour pollen with the best (compatible) genotypes [94]. Indeed, there is evidence to suggest that selective abortion creates non-random variation in reproductive success between the sexes (e.g. [95]). Similar mechanisms are easy to envision, and perhaps the most common is pollen competition where the speed of pollen tube growth is put to the test [96–98]. Pollen grains that grow faster might reach an ovule first and have an effect on reproductive success via female function. Pollen selection can then be intensified by style length [99,100]. Fertilization is affected both by the location where the pollen lands, and the speed with which pollen grows after landing. Flower morphology also has an effect; stigma surfaces that expand along the style provide an advantage for slow growers (and presumably inferior pollen) that land closer to the ovary relative to fast growers (and presumably superior pollen) that land on the tip of the style [101]. Evidence for the effects of gametic selection on the fitness of the diploid phase of the plant is equivocal, at least for traits like germination and seed mass [102]. However, recent theoretical work suggests that females should evolve systems to increase pollen gene expression when these genes are also important in the diploid phase [103]. While such female choice should be applicable to all species with a long diploid phase, is seems that plant morphology is particularly suited to challenge male gametes prior to fertilization.

Given the complexity of fertilization in angiosperms, genes that are likely to play a major role in sexual selection are those that increase the viability of pollen so that pollen can actually be dispersed, survive arrival at the stigma and endure until fertilization. Genes regulating elongation of pollen tube growth [104], and those that mask the effects of deleterious mutations (e.g. genes that are effectively diploid in the pollen tube such as flavonoids that speed up pollen tube growth [105]), are likely to play a major role in pollen endurance and viability, and ultimately in male-mating success [106]. In addition, changes to pollen coat, and to processes of communication between the pollen tube and style that can also affect reproductive compatibility between the sexes (e.g. the ability to dissolve the stigma surface, or to find the micropyle of the ovule), can all contribute to the evolution of molecular traits that favour reproductive success via male or female function. An unexplored aspect is whether there is additive genetic variation in pollen traits for being removed by a given animal, and thus for the possible evolution of pollen adhesion to the body of the pollinator. Similarly, genes making pollen dispersal by wind more efficient have yet to be identified, but obvious candidates are those responsible for pollen viability [107], defence against predators [108] and pollen clumping [109]. Identifying the genetic basis of pollen dispersal by both animals and wind in natural populations, and how this affects variation in reproductive success, will be an exciting area for future research.

5. Fungi do it too

Fungi are an immensely diverse group. They differ, for example, in ploidy level, breeding system, levels of sexual versus asexual reproduction, relative gamete sizes (isogamy versus anisogamy) and size at time of mating [110]. A variety of mechanisms for gamete selection that act during mating in fungi show similarities with those described earlier for plants and spermcasters. Similar to broadcast spawning animals (see §3), fungi react to chemical cues produced by potential mates [111], and it is these cues that in some species have been shown to be used in mate selection [112–115]. Such cues can attract potential mates over distances of decimetres in water fungi [116] to just millimetres in yeasts (see further), but can also be used at the moment of physical gamete interaction when membrane-bound compounds affect final mate selection [117]. Further selection can take place at the moment of fusion, similarly to pollen competition in vascular plants, while selective abortion of certain genotypes in polyandrous matings selects against embryos sired by certain mating partners. The exact opportunities for sexual selection depend strongly on the life cycle and mechanisms by which mating takes place. Even though there is huge variety in these life cycles and mechanisms, for our purpose here we can make some generalizations.

Most fungi are haploid for most of their life cycle with an asexual and a sexual phase. When conditions are favourable, the organism remains asexual. Filamentous species grow in size by extension of the hyphal network (somatic growth), while single-celled species increase in numbers via cell divisions. Many fungi also produce easily dispersing asexual spores for clonal propagation (remember that powder on mouldy bread). A change in environmental conditions often triggers sexual reproduction and mating. Mating consists of the fusion of haploid cells (figure 1a) and/or haploid multicellular individuals to form a diploid structure or individual (figure 1b). The resulting diploid individual will now undergo meiosis, producing meiotic haploid offspring in the form of spores. These haploid spores are resistant to environmental conditions and remain dormant until they are dispersed to a new environment or until the local environment has improved [118]. After germination, the haploid individual starts again with its asexual phase.

Some species are predominantly diploid. In such species, once a meiotic spore germinates, it will almost immediately mate with another individual and continue asexual growth as a diploid or dikaryon (see below). The consequences of mating for haploid and diploid species thus differ greatly: for haploids the partner will mainly affect the haploid offspring, but for the diploids, there are additionally direct consequences to the viability of the fusing individual if the partner has low compatibility or ‘bad genes’ [119]. Budding yeast, for example, is haploid when it initially germinates from a spore, but after mating it will become diploid and will continue growth by many asexual divisions before undergoing meiosis [110]. If a haploid cell fuses with a haploid cell of low compatibility, growth can be severely hampered, resulting in a direct reduction in fitness (figure 1a) [114,119]. In species with a mainly diploid life cycle, it is thus important to mate with an individual that is compatible in the current environment. In species that are primarily haploid, the diploid stage after mating will be very brief and yield genetically diverse haploid offspring. Haploid species tend to mate only when environmental conditions have become detrimental so that they produce offspring that, due to genetic diversity, might be better adapted to the new environment. Because the future environment is unknown, mating with a locally well-functioning partner is thus less relevant for predominantly haploids. We therefore expect that sexual selection for the ‘right’ partner will be stronger in diploid than in haploid species.

Yeasts mate by fusion of equally sized cells (i.e. isogamy), which means there are no sexes. In anisogamous species, individual sperm cells are cheaper to produce than eggs and with equal investment many more sperm are present for each egg, over which the sperm usually have to compete [120]. In isogamous yeast, there is no such bias. Each yeast cell can only mate once, because the zygote formed after mating will not be able to fuse again. Nevertheless, two factors suggest that there is still scope for competition for mates. First, in nature many cells will remain unfertilized at low densities because they cannot find each other—similar to sperm or pollen limitation—but also under high densities due to their inability to move if all local cells have already mated [121]. Second, even if multiple mates are present, the quality of the mates can differ greatly [122,123]. Sexual selection thus does not strictly require anisogamy, an observation previously made by Andersson [124, p. 3].

The best-studied isogamous species is baker's yeast Saccharomyces cerevisiae, which uses chemical communication to locate and select its mate. To find a mate, cells secrete small protein pheromones to which a potential mate can react by chemotropism. During chemotropism, each cell protrudes into the direction of the pheromone gradient produced by the other cell and when both meet mating takes place. Even though chemotropism occurs over very small distances (less than 1 mm), signalling is essential [111]. When given the choice, budding yeast will be choosy and mate with the partner that produces the largest amount of pheromones [112,115]. Because producing pheromones is costly, the signal can be used as an honest trait to distinguish between cells that differ in quality in the local environment [125]. The signal is furthermore used to discriminate conspecific from non-specific cells [113].

Over longer distances, water-soluble pheromones are used by multicellular aquatic fungi of the basal chytrid clade—the only fungi in which the gametes are equipped with flagella [110]. Species of the genus Allomyces are anisogamous with small motile male cells and larger less-motile female cells [126]. The male gametes swarm towards the female gametes, efficiently guided by pheromones produced by the latter [116], similar to gamete chemotactic signalling in broadcast spawners (see §3). On perception of the pheromone, male gametes can clump together and move as a group towards the pheromone source [126]. Grouping might improve swimming ability towards the mate, reminiscent of the so-called sperm ‘cooperation’ observed in some animals in which related sperm together compete against non-related sperm [127].

Aquatic species of the Oomycetes—not actually fungi but a group with similar attributes that belongs to a clade containing the brown algae [128]—also use pheromones to find each other over large distances. In the genus Achlya, species with separate males and females exist, as well as self-compatible and self-incompatible hermaphroditic species. Males and females communicate with each other through diffusible sex-specific pheromones and male individuals grow towards the source of the attractant to initiate mating. There is ample natural genetic variation in both the amount of pheromones produced between individuals that perform the same-sex role and the strength of reaction to the pheromones [129]. Unfortunately, the role of pheromones in mate choice and gamete competition has not been studied explicitly.

Whether pheromone production in the above-described examples is driven by sexual selection is not obvious, because sexual and natural selection might work in concert. Under low densities, natural selection will select pheromone preference to improve mate finding, which under higher densities facilitates sexual selection—even runaway selection [115]—for the higher pheromone-producing mates. Similar interactions between natural and sexual selections are expected to occur in all species where gametes find each other using secreted signals, single-celled organisms such as ciliates or algae [130], but also in externally fertilizing animals.

Sexual selection is not limited to pre-fusion events. The mushroom-forming fungi, which were mentioned briefly in §2, compete and select at the nuclear level over whom to mate with [22,118]. These fungi grow initially as a haploid mycelium containing a single nucleus per cell (called a monokaryon) and can mate with another mycelium by donation of nuclei in the male, and reception of nuclei in the female role (figure 1c) [21,131]. The female that was initially haploid becomes fertilized by nuclei that are actively transported through the cytoplasm of the entire mycelium. After mating, the two nuclei remain separate within the receiving female mycelium (a mycelium with two different types of nuclei is called a dikaryon), which cannot become fertilized again, but each of the nuclei is still able to fertilize another monokaryon. This monokaryon now has the opportunity to select between the two male nuclei. These so-called dikaryon–monokaryon matings are expected to occur often in nature [132]. Choice experiments showed that there is consistent preference for one of the two nuclei [22]. The vast variability and redundancy of pheromone genes in these fungi [111] suggests that choice might be based on compatibility between the pheromones produced by the male nuclei (in these fungi pheromones are not used for extracellular communication), and receptors in the receiving mycelium. During these matings, there also appear to be direct interactions between the competing nuclei, in which some fertilizing male nuclei have competitive dominance over others [22]. It is unknown if dikaryons that consist of the preferred nucleus type perform better than those that carry the alternative nucleus, but preference for improved compatibility at the dikaryon level is not unlikely because the two nuclei are ‘stuck together’ for what could be a very long time—mushrooms can become very old and very large [133]. More on sex and sexual selection in mushroom-forming fungi can be found in the paper by Vreeburg et al. [134].

Sexual selection can even continue after mating. Importantly, there is evidence of ‘polyandry’ in fungi, suggesting that there is scope for zygote-level sexual selection in these systems. In multicellular ascomycete fungi such as Penicillium, a haploid mycelium produces a fruiting body that after fertilization and a very brief diploid phase goes through meiosis and produces many haploid offspring. A different male can fertilize each fruiting body. Laboratory experiments have shown that fertilization by multiple males is possible [135], and that in crosses with conspecific and a closely related species, females selectively abort fruiting bodies fertilized by the wrong species [136]. Selection by the female between different conspecific males should be expected when multiple fertilizations occur, similar to selective abortion in plants. Owing to their cryptic lifestyle, no direct measures of mating success in fungi have been performed in nature, but due to high densities for many species [137] we expect that polyandry is likely to occur. Similar mating systems are found in bryophytes and some seaweed groups. These are predominantly haploid and the sporophyte grows on the female gametophyte, thus giving large control to the female over the nutrients provided. In the red seaweed Chondrus crispus, levels of multiple mating are very high with a bias towards some paternal genotypes. However, it is not clear if this is due to female preference or the clustering of male gametes [138]. It has been shown that female investment by sporophytes of the moss Sphagnum lescurii is selectively directed to more compatible males, using diploid heterozygosity as measure of compatibility [139].

6. Concluding remarks

Our review highlights a range of putative mechanisms of sexual selection operating in taxonomic groups that fall outside the ‘usual suspects’ in sexual selection research. Our summary of such mechanisms probably only scratches the surface, but nevertheless reveals a rich diversity of processes that may be targeted by sexual selection. Despite this diversity, however, we can see similarities among many of the systems we reviewed and make links to processes found in organisms that are more often studied in the context of sexual selection.

Let us start with organisms that function as males and females as we know it, albeit simultaneously. In the case of hermaphroditic fungi, mate selection can act at the level of the nucleus, where the sex role is determined by which nucleus retains its cytoplasm (the male nucleus is donated, whereas the female nucleus receives a nucleus from the other individual's mycelium). The mechanism by which nuclei are selected as mating partner is very similar to that found in broadcast spawners, via pheromones secreted by the male nuclei [111]. In fact, chemical signalling between gametes appears to be universal among sessile or weakly mobile broadcast spawning invertebrates, thus providing the potential for sexual selection to target post-ejaculatory processes that ultimately facilitate fertilization by preferred or compatible gametes. Remote communication between male and female gametes, or cellular-level interactions involving gamete recognition proteins, may therefore be the precursors to more derived mechanisms of mate choice in the highly mobile, sexually dimorphic organisms that are more familiar to most students of sexual selection [140].

Hermaphrodites nicely illustrate how selection acts differently on males and females. While it has been argued that, in general, simultaneous hermaphrodites prefer mating as a male, it seems that sex role preferences are not necessarily fixed, but are both context- and condition-dependent [14]. For example, when all your ova have been fertilized, the only sensible investment is in male function (unless more ova can be produced, which seems less likely due to the necessary investment in the developing zygote). By contrast, if the number of potential males is low, investing heavily in male function is unprofitable. Thus, even if male function usually benefits more from an additional mating, the actual fitness benefits will depend on a range of factors such as mating history, mate availability, fecundity, survival and sex-allocation decisions made by the mating partner (reviewed in [14]). Because it seems unlikely that selection can optimize both male and female functions in the same individual, simultaneous hermaphrodites need to trade-off male and female roles and investment into each role. Hermaphroditic plants are no exception. The need to invest in both male and female functions within the same flower leads to the coevolution of genes that are expressed in different flower structures and pollen, but selection on female function can, potentially, be antagonistic to selection on male function [141,142]. Pollen grains express over 60% of all genes present in the diploid organism, so flower structures and pollen grains are likely to share a large fraction of the additive genetic variation on which both sexual and natural selection can act. Thus, selection on the gametic haploid, the sporophytic diploid phase, or both, will affect flower traits. If selection on pollen (male function) and pistil structures (female function) both lead to an increase in reproductive success via male function, there is no conflict between male and female functions [31]. If, on the other hand, there is selection for genes expressed in the diploid phase that have a negative effect when expressed in pollen (or vice versa), conflict between female and male function arises [103]. Clearly, hermaphrodites are an excellent model system to decipher the exact conditions that favour investment in one sex over the other [5] in the absence of sexual dimorphism.

We suspect that the organisms considered in our review are often overlooked in the context of sexual selection, because in these organisms selection mostly acts after copulation or the release of gametes. In this sense, sexual selection is ‘cryptic’, favouring subtle variations in reproductive physiology that may fail to attract the attention of researchers more commonly familiar with overt sexual dimorphism and obvious adaptations that function in the context of reproductive competition. This apparent bias in the study of sexual selection has its roots with Darwin [1], who linked sexual selection to cognitive ability (§1) and overlooked the possibility that females often mate with more than one male [3]. As Darwin stated, ‘It is shown by various facts, given hereafter, and by the results fairly attributable to sexual selection, that the female, through comparatively passive, generally exerts some choice and accepts one male in preference to others. Or she may accept, as appearances would sometimes lead us to believe, not the male which is the most attractive to her, but the one which is the least distasteful’ (pp. 272–273). Darwin's views on female sexual behaviour, possibly influenced by Victorian prudery [143], almost certainly delayed the onset of studies on post-ejaculatory competition by denying the now irrefutable evidence that polyandry is widespread among most sexually reproducing organisms (see [144] and accompanying articles).

Can we apply what we learn from the organisms discussed here to more familiar groups? Parker [9] has recently argued that competition between ejaculates in sessile and relatively mobile organisms played a major role in the cascade of evolutionary events that culminated in pre-ejaculatory (i.e. Darwinian) sexual selection (i.e. mobility, internal fertilization and ultimately behavioural components of mate choice and mating competition; see also [140]). As such, the mechanisms of pre- and post-ejaculatory sexual selection observed in extant multicellular, mobile and internally fertilizing organisms have related evolutionary origins to sessile or relatively mobile groups in the extant plants, marine invertebrates and other groups considered in this review. The logical conclusion, therefore, is that studies of organisms that have retained strategies of external fertilization, along with features such as reduced mobility, lack of sexual dimorphism and high reproductive investment in gametes, have much to teach us about sexual selection in taxa with the opposite and often studied characteristics. So what can we learn from the less-studied organisms considered in our review?

Let us focus on one specific putative ‘mechanism’ of mate choice that is common to many of the organisms we have covered: chemically moderated gamete interactions that differentiate among prospective mates after gamete release. Such reproductive secretions, often referred to as ovarian fluid, female reproductive fluids (FRFs), chemoattractants and pheromones (depending on their known or presumed function or origin) occur ubiquitously across the animal, plant and fungi kingdoms (see §§2–5). Some of these secretions may represent evolutionary adaptations that initially facilitated syngamy in primitive unicellular organisms. We see, for example, that single-celled organisms such as baker's yeast use chemical attractants not only to locate mating partners, but also use them to select among possible mates, hinting at an evolutionary origin of gamete attraction and mate choice [125]. Such signalling mechanisms may have been adapted to signal both compatibility and mate quality in anisogamous species, as we still see in extant external fertilizers and possibly also internally fertilizing species.

As we have highlighted in our review, female-derived reproductive ‘secretions’ are known to play a role in moderating both conspecific and intraspecific fertilization dynamics in a number of species. The fact that processes such as sperm chemotaxis also occur in vertebrates with high mobility and complex behavioural mechanisms of mate choice seems, on the face of it, somewhat paradoxical. Indeed, observations of sperm chemotaxis in internal fertilizers, including mammals [51], were initially discarded because it was thought that copulation (leading to internal fertilization) would negate the need for sperm attractants (reviewed in [145]). Yet the unequivocal evidence for sperm chemotaxis in internal fertilizers [51,145] raises an inevitable question: what biological functions are served by sperm guidance mechanisms in species with advanced capabilities of species recognition, sophisticated modes of mate choice, and reproductive systems that guarantee the close proximity of sperm and eggs at ejaculation? The short answer to this question is that we do not know. This is true even in the vast majority of broadcast spawners, where sperm chemotaxis has been described extensively but evolutionary explanations for the phenomenon are rarely sought (but see [48,146]). We suspect that sperm chemotaxis, and similar chemically moderated processes that influence gamete behaviour (see §§3 and 5), may serve as moderators of mate choice at the gamete level, in both internally and externally fertilizing species. We eagerly await further insights from plants, fungi and broadcast spawners, some of which are already uncovering putative mechanisms of gamete choice moderated by female reproductive secretions [147]. In this way, we anticipate that that such studies will stimulate research on ‘the usual organisms’ to better understand the adaptive function of female reproductive secretion in these species.

Competing interests

We declare we have no competing interests.

Funding

Our work is supported by the Australian Research Council (FT120100120 to M.B.; DP150103266 to J.P.E.; DP140103774 and DP120104559 to D.O).