Seminal fluid and accessory male investment in sperm competition

Abstract

Sperm production and allocation strategies have been a central concern of sperm competition research for the past 50 years. But during the ‘sexual cascade’ there may be strong selection for alternative routes to maximizing male fitness. Especially with the evolution of internal fertilization, a common and by now well-studied example is the accessory ejaculate investment represented by seminal fluid, the complex mixture of proteins, peptides and other components transferred to females together with sperm. How seminal fluid investment should covary with sperm investment probably depends on the mechanism of seminal fluid action. If seminal fluid components boost male paternity success by directly enhancing sperm function or use, we might often expect a positive correlation between the two forms of male investment, whereas trade-offs seem more likely if seminal fluid acts independently of sperm. This is largely borne out by a broad taxonomic survey to establish the prevailing patterns of seminal fluid production and allocation during animal evolution, in light of which I discuss the gaps that remain in our understanding of this key ejaculate component and its relationship to sperm investment, before outlining promising approaches for examining seminal fluid-mediated sperm competitiveness in the post-genomic era.

This article is part of the theme issue ‘Fifty years of sperm competition’.

1. Introduction

Sperm are the fundamental ingredient of any ejaculate, and their number, morphology and many other features have been profoundly influenced by sperm competition, the selective pressure that arises from competition between the ejaculates of rival males over fertilization of a given set of ova [1–4]. Nevertheless, there is a limit to how sperm alone might influence paternity outcomes, creating strong selection on males to find additional means to enhance fertilization success. At early stages of what has been termed the ‘sexual cascade’ of evolutionary transitions in reproductive strategy, this might involve enhanced mobility or traits we would now recognize to be under pre-mating sexual selection [5]. Especially with the evolution of internal fertilization, however, a further male trait that can come to play a potentially decisive role for male reproductive success is the complex mixture of proteins, peptides and other ejaculate substances known collectively as seminal fluid [6–8].

The widespread importance of seminal fluid to paternity outcomes is hardly surprising, and indeed was recognized from the outset [1]. Ejaculate substances are singularly well-positioned to influence both the post-ejaculation performance of own (and rival) sperm as well as the reproductive behaviour and physiology of mating partners, and the ways they might do so are many and varied. This therefore creates multiple potential routes for seminal fluid action. In fact, many features of seminal fluid—such as its complexity, multi-functionality, redundancy and species-specificity—are probably best understood as accumulated signatures of historical selection for novel seminal fluid functions exploiting these different routes, evolving principally through arms races among males over sperm competitive ability and/or between males and females to resolve associated sexual conflicts over mating and fertilization.

Here I argue that investment in seminal fluid is thus a common and uniquely influential means through which males can enhance their reproductive success. A thorough review of the underlying seminal fluid biochemistry and function is certainly beyond the scope of what I can cover, but a good overview is provided by Poiani [7] and a comprehensive survey of accessory gland proteins in Drosophila—by far the most important model system for evolutionary studies of seminal fluid to date—is the topic of a dedicated review by Wigby and colleagues in this issue [9]. I also leave aside the wider evolutionary significance of seminal fluid, for example in mediating non-genetic paternal effects [10,11] or for speciation [12,13]. Instead, I wish to explore how seminal fluid-mediated fitness effects interact with sperm investment on both evolutionary and ecological timescales. The relationships between sperm and seminal fluid investment are complex, depending crucially on the mechanisms of seminal fluid action and ultimately on the particularities of each species' reproductive ecology. I therefore begin by considering how we might incorporate such information to predict optimal male strategies for ejaculate production and allocation. Against this background, I then survey how male investment strategies manifest as variation in seminal fluid traits at three distinct (nested) biological levels: as variation in average seminal fluid production among species; as phenotypically plastic (as well as genetic) variation in seminal fluid production within species; and as differential allocation of seminal fluid reserves to a particular mating. Finally, I offer a perspective on how seminal fluid studies at all of these levels are being transformed by genomics-enabled approaches.

2. Theoretical predictions

The ‘sperm competition games’ developed by Parker and colleagues (e.g. [14–17]) have provided a powerful conceptual framework for studying the differential production and allocation of sperm according to the level of sperm competition expected to be faced over fertilization (i.e. its ‘risk’ and ‘intensity’—see [16]); as well as other relevant factors such as male condition; female mating status and reproductive value; and trade-offs with other sexually selected traits (reviews in [6,18,19]). Broadly speaking, these models assume one particular life-history trade-off between finding mates and gaining fertilizations, with investment in sperm leading to fitness returns according to a ‘raffle principle’, i.e. the greater the proportion of sperm from a particular male being present at the time of fertilization, the greater (all else being equal) his share of paternity [16]. However, when thinking about seminal fluid, this framework cannot simply be adopted wholesale. The sperm and seminal fluid fractions of the ejaculate differ in several key respects, not least of which is that—with a few important exceptions [20]—the sperm of most taxa represent a numerous but relatively uniform component that justifies the ‘raffle principle’ assumptions. By contrast, seminal fluid can comprise tens, hundreds or even thousands of different individual components (e.g. [21–26]), each potentially with their own specific functional and fitness relevance to sperm competition, with further complications arising from functional redundancies among seminal fluid components (e.g. [27,28]) and because a single seminal fluid component can have more than one function (e.g. [29]). In other words, when optimizing their production and allocation of this latter fraction of the ejaculate, the ‘seminal fluid games’ males play are inherently multidimensional.

When considering seminal fluid production and allocation strategies, an important distinction might be drawn between those seminal fluid components that directly boost sperm fertilizing ability versus those that function more or less independently of sperm. At least some components that fall into the former category probably need to be produced or allocated in direct proportion to the number of sperm contained in each ejaculate, which may create a natural tendency for seminal fluid investment to be positively correlated with sperm investment. Most theoretical studies to date have instead focused on the latter category, and specifically on a few key functional classes of seminal fluid. An initial model, for example, examined the fecundity-stimulating effects of seminal fluid, and predicted that optimal male investment depends on both the ‘fairness’ of the raffle determining paternity success and the male's own status as the favoured or disfavoured male in this contest (because favoured males anyway gain a higher share of paternity and so have more to gain from higher female fecundity) [30]. Once seminal fluid was modelled to influence both female fecundity and the fairness of the sperm competition raffle, favoured males were still predicted to invest more than disfavoured males in seminal fluid, but greater sperm investment might be expected for either favoured or disfavoured males depending on precisely how sperm and seminal fluid affect sperm competition outcomes [30].

A simplifying assumption of the above model [30] was to treat female behaviour as fixed. In reality, however, both the probability of female remating and the benefits (or costs) they will gain (or pay) by doing so may themselves coevolve in response to male sperm and seminal fluid investment strategies, with direct consequences for realized sperm competition levels. A subsequent model suggests that incorporating such complex co-evolutionary dynamics may be crucial; depending on the relative magnitude of male-mediated fecundity stimulation relative to overall female fecundity, this model found that allowing female coevolution could lead to a range of outcomes spanning from inter-sexual conflict to cooperation [31]. Male allocation to ejaculate-mediated fecundity stimulation was predicted to be highest when such ejaculate components have strong effects on female fecundity and when female remating probabilities are low (the latter presumably selecting conversely for low sperm investment [16]). Similar coevolutionary feedbacks between female behaviour and male sperm competition strategies probably occur also for ejaculate-mediated effects on female fertility and mortality [32].

Another important class of seminal fluid adaptations are those such as mating plugs that seek to prevent or delay female remating (see §3b below). Plugs moreover involve an additional layer of complexity, in that male fitness is determined both by the plug's ability to prevent female remating and by the ability of males to remove such plugs deposited by their rivals [33]. This creates complex dynamics even before taking female interests and potential evolutionary responses into account, but one model addressing the evolution of plug versus sperm investment suggests plug investment is increasingly disfavoured with high average sperm competition intensity (because when females typically mate with many males, this selects instead for high male mating capacity and larger ejaculate size, rather than for plug investment), but can stably evolve either when males are sperm-limited (i.e. when males have limited capacity to increase fitness through higher mating frequency) or typically out-number females (because, likewise, there are fewer mating opportunities per male) [33].

Subsequent work reinforces the notion that exact predictions depend crucially on seminal fluid's mechanism of action. A population genetic model sought to predict the optimal partitioning of seminal fluid investment between sperm competition avoidance (which would include seminal fluid-derived plugs) and sperm competition ‘defence’ and ‘offence’ (i.e. securing paternity when the first or second to mate with a doubly-mated female, respectively), and also to incorporate plasticity in seminal fluid allocation [34], though it did not address co-evolution with sperm investment strategies. As might intuitively be expected, investment in avoidance traits such as plugs was increasingly favoured (i) as they themselves become more efficient, and (ii) as offence mechanisms become relatively more efficient than defence mechanisms, but investment in offence and defence will often be equal. Plasticity in seminal fluid allocation, whereby males could allocate to different seminal fluid functions depending on their position in the mating sequence, could instead favour unequal investment in offence and defence. The influence of plasticity was also notable in that it led to multiple potential evolutionary outcomes exhibiting different seminal fluid compositions [34].

Finally, a model specific to hermaphrodites also speaks to the balance of investment in sperm versus seminal fluid [35]. Many patterns of selection on seminal fluid are likely to be preserved across sexual systems, though hermaphroditism does open up new potential targets for seminal fluid action [36,37]. This model, however, sought to predict optimal sex allocation patterns where male allocation could be divided between sperm production and investment in pre- and post-copulatory traits (the latter of which could but need not be seminal fluid) that affected the likelihood of mating and fertilization, respectively. The relative balance of allocation to sperm versus the post-copulatory trait was predominantly affected by the shape of the function describing how investment in this trait affected the efficiency with which sperm allocation was converted into fertilization gains [35]. Both sperm and post-copulatory trait investment (and indeed total male allocation) increased with increasing numbers of mates [35], in line with existing sex allocation theory [38].

Overall, the picture emerging from these few dedicated attempts to model seminal fluid investment is a complex one, depending on both the precise function of seminal fluid being modelled and the nature of correlated and co-evolutionary responses considered. Provisionally, then, we can conclude that sperm and seminal fluid investment will often be correlated (either positively or negatively), because the fitness benefits to be gained from investing in one often depend on the level of investment in the other, but that the precise relationship between them probably varies between animal groups. More tentatively, I would suggest that positive correlations will be most commonly seen in animal groups for which seminal fluid primarily functions to boost sperm function, whereas trade-offs between sperm and seminal fluid investment will most likely occur in taxa where seminal fluid has strong, independent effects that prevent or delay female remating, because this then immediately feeds back on sperm competition levels and would tend to relax selection on sperm investment. It is against this background that I will now explore the empirical data on investment patterns.

3. Seminal fluid production patterns across species

(a). Accessory reproductive glands

The size and number of accessory reproductive glands that form part of the male reproductive tract is highly variable among taxa. Even within mammals, for example, different species exhibit different combinations of, and relative contributions from, various glands including the seminal vesicles, multiple lobes of the prostate and bulbourethral glands. This diversity has often been used for taxonomic purposes, but a few studies also address the selective pressures acting on accessory reproductive gland size, and by extension investment in seminal fluid production as a whole. The majority of such morphological studies (e.g. in mammals [39,40], fishes [41] and some insects [42,43]) support a positive correlation between the two aspects of ejaculate investment, i.e. between sperm and seminal fluid production (table 1). However, this is not a universal pattern and there are still only relatively few taxa in which the question has even been addressed.

Table 1.

Interspecific tests for correlated evolution of male seminal fluid and sperm traits. (n represents the number of species (or in one case genera [39]) included in each study. The primate study listed as ‘n/a’ for the sperm trait tested instead for an association between seminal vesicle size and mating system; the strong association between mating system and relative testis size in primates is well established [44]. As in subsequent surveys of plasticity and allocation patterns, I hope the selected studies are representative, but they should not be treated as a comprehensive list.)

taxon	seminal fluid trait	sperm trait	n	relationship
primates	seminal vesicle size	n/a	27	positive—larger seminal vesicles found in species with multimale-multifemale and dispersed mating systems cf. monogamy, polygyny [39]
rodents	mass of three accessory reproductive glands: seminal vesicles, anterior prostate and ventral prostate	testis mass	13–29	positive for seminal vesicles and anterior prostate, uncorrelated for ventral prostate; relative testis mass also correlated with copulatory plug length (formed by seminal vesicle and anterior prostate secretions) [40]
goby fishes	seminal vesicle somatic index	gonado-somatic index	8	positive—mating system (n = 12 species) best predicted relative seminal vesicle size; no formal analysis for gonado-somatic index (n = 8 species) presented, calculated here as r = 0.96, p < 0.001 [41]
moths	accessory gland volume	testis volume	130	positive [42]
bushcrickets	spermatophylax mass	ampulla mass, sperm number	31–43	positive [43,45]—but no relationship between spermatophylax mass and testis mass [46]; ampulla may also contain non-sperm components
fungus-growing ants	relative accessory gland size	relative accessory testis size	11	negative—species in which queens evolved multiple mating have reduced accessory gland size but increased relative accessory testis size [47]
Drosophila	accessory gland length	testis length	22	uncorrelated [48], but see the electronic supplementary material, figure S1

One exception is the potential evolutionary trade-off between sperm and seminal fluid investment found among attine fungus-growing ants, which appears to be driven by variation in queen mating frequency [47]. Specifically, the evolution of multiple mating by queens is accompanied by both reduced relative accessory gland size and increased relative accessory testis size (the latter being where sperm are stored prior to ejaculation; like most social Hymenoptera, males do not have continuous sperm production [47]). Although data for the 11 included species showed a strong negative correlation between relative accessory testis and relative accessory gland size, a direct trade-off was not confirmed in a phylogenetically-controlled analysis (probably owing to low power). These data suggest opposing directional sexual selection on seminal fluid and sperm investment among fungus-growing ants, but it is important to stress the likely co-evolutionary (and sexually antagonistic) nature of this relationship: males are not simply responding to queen mating frequency as an externally imposed factor determining their optimal ejaculate investment strategy, but rather accessory gland products may themselves function in enforcing monogamy among singly-mating species. It is only after males lost the ability to control female mating behaviour that selection on accessory gland and accessory testis size was reversed [47]. Attempting to control female remating is probably a common function of seminal fluid, but is always vulnerable to the possibility that females (or rival males) might evolve means to evade or overcome such manipulation.

A possible further example occurs in Drosophila. Although overall there appears to be no clear interspecific relationship between testis and accessory gland investment [48], there is an intriguing trend suggesting the ratio of accessory gland to testis length across eight species could be negatively correlated with female remating propensities (electronic supplementary material, figure S1). This is consistent with the idea that accessory investment indirectly selects for reduced sperm investment via its suppressive effects on female remating [49], but there could certainly be other explanations for such a pattern, and too few data are available currently to draw any firm conclusions.

Only few studies to date have begun to examine the differential investment in or composition of seminal fluid in different taxa at the level of gene or protein expression (e.g. [50]), but such studies will certainly be important if we wish to move beyond gross morphological measures of seminal fluid investment and instead take into account specific functions of seminal fluid components. Experimental evolution approaches that seek to isolate the effect of sperm competition level on seminal fluid investment may be especially valuable here. For example, enforced monogamy led to a decline in the expression of many seminal fluid genes in Drosophila [51,52] as well as that of two major seminal fluid components, SVS1 and SVS2, in house mice [53]. In both cases, this was accompanied by corresponding reductions in sperm competitiveness, for which these expression changes may be at least partially responsible [52,54]. Rapid divergence might also be detectable through intraspecific, inter-population comparisons, the degree and direction of which is probably shaped by sperm competition levels. Accordingly, variation in seminal fluid proteomes among populations of the seed beetle Callosobruchus maculatus was associated with divergence in sperm competitive ability [55]. Similarly, natural populations of the great pond snail Lymnaea stagnalis differ in their expression of key seminal fluid genes, and corresponding functional variation could be established through artificial injection experiments [56].

(b). Plugs and other mechanisms to prevent remating

Mating plugs formed from seminal fluid secretions are taxonomically widespread, being found for example in several mammals [40,57], reptiles [58], insects [59], spiders [60] and nematodes [61]. They can sometimes be highly effective at preventing remating (e.g. [57,60,62]), with good evidence in some cases that plug efficacy varies with plug size [60,62,63]. Comparative patterns in butterflies suggest a potential trade-off between plug and spermatophore size (reviewed in [64]) and indicate that the degree of development of the plug is negatively correlated across species with female mating frequency [65]. Among rodents, plug size appears to correlate with the length of the female reproductive tract [40,66], but plugs are also relatively larger in species with higher levels of sperm competition [40]. However, plugs can also play multiple roles besides preventing or delaying remating (reviews in [40,65]) and these too may be highly relevant to sperm competition outcomes. In house mice, for example, the plug potentially influences paternity success via effects on both delaying mating by rival males and stimulating sperm transport [67–69]. Surprisingly, in one study, smaller plugs appeared to persist for longer in the female reproductive tract [68], but this could here reflect differences between mouse strains in not just plug size but also its protein composition [70]. There is clearly much we still need to understand about plugs in the context of sperm competition and associated sexual conflicts over their retention or removal (e.g. [71]). It may be naive to expect consistent patterns of investment across diverse taxa, particularly because many other (non-plug) seminal fluid proteins are also known to influence female remating propensities [64,72].

(c). Spermatophores and nuptial gifts

Taxa in which ejaculates are delivered as an external spermatophore make interesting case studies for studying seminal fluid investment patterns, because in these species seminal fluid can potentially influence female physiology and behaviour both via the ‘usual’ route when seminal fluid components are taken up into the female reproductive tract together with sperm, as well as by virtue of the fact that they may be ingested by the female as a nuptial gift. (Note that others have argued that all seminal fluid-mediated effects could fall under the definition of ‘nuptial gifts’, in which case what I refer to here would be just one—endogenous, oral—sub-category [73].) One advantage of such study systems is that is possible to directly recover and quantify the seminal fluid contribution to the ejaculate [6]. Among bushcrickets for example, comparative analyses do not appear to support a strong correlation between testis mass and spermatophylax mass [46] but do suggest that spermatophylax mass correlates positively with ampulla mass [43,45], which may be because depositing a larger spermatophylax enables more sperm to be transferred from the ampulla to the female tract before the nuptial gift is consumed (the ‘ejaculate protection’ hypothesis) [45]. Recent studies using proteomics and associated methodologies have begun to dissect the molecular underpinnings of the spermatophore in bushcrickets [74], as well as in butterflies [75,76], moths [77] and fireflies [78] with internally-deposited spermatophores.

An even starker separation between sperm and accessory male substances occurs in taxa where these accessory substances are delivered not as seminal fluid but via some other route, as for example on ‘love darts’ of some land snails [79]. Here again, the scope for trade-offs with sperm investment would seem to be greatest, but I am not aware of relevant comparative studies directly addressing this point (but see [80]).

4. Plasticity in seminal fluid production

Although overall seminal fluid production patterns appear to reflect the taxon-specific significance of this ejaculate component, and are often correlated with sperm production, this does not preclude that different seminal fluid production strategies might be favoured by different males of the same species or under different environmental conditions. Indeed, over the past several years, evidence for the ability to adjust seminal fluid production parameters according to the prevailing circumstances has begun to accumulate across diverse taxa (table 2).

Table 2.

Differential seminal fluid production based on (a) sperm competition level variation between alternative male morphs, tactics or roles and (b) socio-environmental cues of sperm competition level. (Associated sperm trait plasticity is based on the same study unless additional sources are cited.)

	species	contrast (a) or cue (b)	plastic seminal fluid trait	source	associated sperm trait plasticity?
(a)	bank vole, Myodes glareolus	dominant versus subordinate males	seminal vesicle size (but not copulatory plug size)—larger in dominant males	[81]	yes—absolute (but not relative) testis size larger and sperm numbers per ejaculate greater in dominant males
	house mouse, Mus musculus domesticus	dominant versus subordinate males	seminal vesicle size (larger in dominant males), protein concentration (higher in subordinates) and seminal fluid protein abundance (consistent differences in multiple proteins according to social status)	[82]	yes—dominant males have greater epididymal sperm reserves (despite similar testis size to subordinates)
	plainfin midshipman fish, Porichthys notatus	guarder versus sneaker alternative reproductive tactics	accessory glands—lobules and total accessory gland mass (but not nodes) relatively larger in guarders	[83]	yes—evidence for trade-off: sneaker males have relatively larger testes and much larger ejaculates, but guarders have higher sperm velocity owing to lobule secretions
	black goby, Gobius niger	guarder versus sneaker alternative reproductive tactics	accessory glands—seminal vesicles and mesorchial glands larger in guarder males	[84]	yes—evidence for trade-off: sneaker males have larger testes; seminal fluid enhances the velocity and fertilization ability of guarder sperm [85]
	grass goby, Zosterisessor ophiocephalus	large (guarder) versus small (sneaker) males	seminal vesicles—larger in larger (guarder) males	[86]	yes—evidence for trade-off, because testes larger in smaller (sneaker) males
	Chinook salmon, Oncorhynchus tswatchysha	jack versus hooknose alternative reproductive tactics	seminal fluid proteome	[87]	yes—jacks (sneakers) have relatively larger testes and faster sperm [88]; seminal fluid can affect own and rival sperm velocity [89]
	ant, Cardiocondyla obscurior	winged versus wingless males	seminal fluid proteome—56 out of 920 spots differed in intensity	[90]	yes [91].
(b)	house mouse, Mus musculus domesticus	number of rival males	seminal fluid composition, seminal vesicle size—larger with more rivals (but see [92])	[93]	yes—daily sperm production rate and epididymal sperm numbers increase with more rivals [92]
	bank vole, Myodes glareolus	number of rival male odours	seminal vesicle size	[94]	no—testis size, epididymal sperm counts, daily sperm production and sperm motility all unaffected
	fruit fly, Drosophila melanogaster	sperm competition intensity cues	seminal fluid gene expression (2 out of 3 genes investigated)	[95]	n/a
		population density during larval development	sex peptide and ovulin expression	[96]	n/a
		larval density, presence/absence of adult males during larval development	accessory gland size—larger in response to competitive cues	[97]	no—testis size unaffected or reduced by same cues
		socio-sexual environment	seminal fluid gene expression 2, 26 and 50 h after exposure to rivals (complex responses across genes and replicates)	[98]	n/a
	Australian field cricket, Teleogryllus oceanicus	calls or encounters with rival males (sperm competition risk and intensity, respectively)	seminal fluid gene expression (elevated under sperm competition risk; reduced under high sperm competition intensity), accessory gland mass	[99,100]	yes—increased sperm viability under sperm competition risk [101]; reduced under sperm competition intensity [102]
	flatworm, Macrostomum lignano	social group size	seminal fluid gene expression—some inconsistencies across studies, but many plastic genes	[103–105]	yes—testis size [106], testicular stem cell activity [107], sperm production rate [108] and spermatogenesis speed [109] all also vary with social group size
	great pond snail, Lymnaea stagnalis	social group size	seminal fluid gene expression—expression increases from social group size of 1 to 2 but no further increase from 2 to 5	[110]	n/a
	dart-shooting land snails (four species)	population density	mucus gland (and dart) morphology—some evidence for increased mucus production with increasing population density (variable across species), but not strictly seminal fluid	[111]	n/a

(a). Male roles

Taxa with alternative reproductive tactics or otherwise predictable male ‘roles’ (sensu [112]) in sperm competition often exhibit role-specific seminal fluid expression (table 2a). In these cases, individuals can potentially use information on their own status to optimize their seminal fluid (and sperm) production to suit the expected fitness returns on investment for that tactic or role, rather than the population average. Such differential expression appears to be widespread, with examples from mammals, insects and especially fishes, but the extent to which this represents plasticity versus fixed genetic differences between genetically-determined male morphs or tactics remains to be established.

(b). Cues of sperm competition risk and intensity

Environmental cues such as population density or the number of rival males encountered may enable individuals to fine-tune ejaculate production to the level of sperm competition they are likely to experience. While this is well established for sperm production (e.g. [92,106,113–115]), several recent studies strongly support that seminal fluid production is also plastically adjusted in this manner (table 2b). This includes plasticity in both the size of the accessory reproductive glands (e.g. [94,97]) as well as the expression of seminal fluid genes and proteins (e.g. [93,95,99,105,110]). Importantly, though, not all seminal fluid genes or proteins vary equally in expression according to sperm competition cues, implying differential seminal fluid composition. Although it seems plausible that different sperm competition environments might well favour different seminal fluid compositions, in only a few cases has the adaptive significance of overall or protein-specific plasticity begun to be addressed (e.g. finding that differential expression of seminal fluid impacts sperm viability in crickets [99] and sperm competition outcomes in flatworms [103]). To date, most studies of seminal fluid plasticity have also investigated seminal fluid expression in stable social environments both prior to and after sexual maturity rather than in response to a change in the environment (but see [95,98]), so obvious outstanding questions here concern whether individuals are equally sensitive to sperm competition cues at different life stages as well as the timescales over which they can respond. Given the high rates of protein turnover found in accessory reproductive glands, rapid adjustments to seminal fluid production might actually be more possible than to spermatogenesis [116].

A further open question is the extent to which seminal fluid reaction norms to sperm competition cues are individual- or genotype-specific. There is certainly growing evidence that seminal fluid traits harbour substantial genetic variation [55,70,104,117,118], but, to date, I am aware of only one system in which the combined effects of genotype and social environmental cues of sperm competition on seminal fluid expression have been systematically investigated. In the flatworm Macrostomum lignano, there was genetic variation in multiple axes of seminal fluid expression variation and widespread evidence for genotype–environment interaction effects on seminal fluid gene expression [103,104], some of which may be functionally relevant to sperm competition outcomes via the manipulative effects of M. lignano seminal fluid on post-mating partner behaviour and thus on the fate of received ejaculates [28,103]. If such genotype–environment interactions are widespread, this could help explain the maintenance of genetic variation in seminal fluid expression and sperm competitive ability [119,120]. Seminal fluid may be an especially amenable trait to study links between genotypes, sexually selected phenotypes and environments (e.g. [121]), because of the relatively simple genetic basis of the underlying traits. Extracting functionally relevant axes of genetic variation or their underlying gene regulatory networks [104,122–124] will further help to reduce dimensionality and define the most evolutionarily relevant seminal fluid variation.

(c). Male condition and age

Another form of plasticity that may be highly relevant to sperm competition outcomes is the extent to which seminal fluid production is condition-dependent (e.g. [125,126]). A recent meta-analysis concluded that nutritional limitation tends to affect the seminal fluid component of the ejaculate more so than aspects of sperm quality [127]. Similarly, male investment in seminal fluid may also be strongly age-dependent: there is evidence for both senescence in key seminal fluid proteins as well as age-related changes in seminal fluid expression explaining variation in male reproductive success (e.g. [128–130], reviewed in [131]). A key future challenge will therefore be to ascertain the relative importance and combinatorial significance of these male role-, sperm competition cue-, condition- and age-mediated aspects of plasticity in the seminal fluid proteome.

5. Seminal fluid allocation

The final biological level of investment variation I wish to explore is the one we still know least about, namely the ability of males to strategically allocate seminal fluid from their accumulated reserves to a specific mating. This could encompass both changes in the overall amount of seminal fluid transferred, as well as finer scale adjustments to its composition. At least in some taxa, the former is relatively easy to measure (e.g. [43]), but the latter is of course much more challenging, and there are few examples of such studies to date.

Once again, Drosophila research has led the way. Two key seminal fluid proteins, ovulin and sex peptide, are differentially allocated according to both sperm competition cues from potential rivals [132] and female mating status [133]. These adjustments appear adaptive, because males transfer more of both proteins under a heightened risk of sperm competition [132], but conserve investment in the fecundity-stimulating ovulin when the female they encounter has already mated [133]. Moreover, artificial selection for larger accessory glands led to greater sex peptide expression that indeed translated into enhanced sperm competitive ability [132]. Recently, expanding such studies to a whole-proteome scale revealed the complex compositional changes that occur in seminal fluid along a sperm competition continuum (with different clusters of proteins responding in different ways); that sperm and seminal fluid exhibit discordant allocation responses (with sperm investment peaking at lower sperm competition levels than, on average, seminal fluid protein abundance); and that, although probably adaptive in the current situation, such responses may incur costs to future reproductive success [134].

These insights have recently been augmented by studies in two vertebrate model systems. Using a social challenge experimental paradigm to induce changes in social status and thus optimal ejaculate quality, Bartlett et al. [135] demonstrated that rapid, adaptive adjustments to sperm velocity in Chinook salmon (Oncorhynchus tswatchysha) are mediated by seminal fluid. These responses occurred after only 48 h, but whether this represents purely plastic allocation, or could also reflect early production plasticity effects, remains to be determined. In red junglefowl (Gallus gallus), seminal fluid proteomes vary across successive ejaculates transferred to the same female followed by an ejaculate transferred to a novel female, encompassing proteins exhibiting depletion, enrichment and differential allocation to first or second females [136], but the overall amount of both sperm and seminal fluid transferred showed clear evidence of a Coolidge effect, i.e. a progressive decline with the same female followed by increased investment to a novel partner [136]. In line with the differing roles of dominant and subordinate males in sperm competition, these responses also partly depended on male social status [136]. This builds on earlier work suggesting seminal fluid may be differentially allocated according to both social status and female attractiveness [137].

6. Integrated ejaculates in the post-genomic era

Genomics-enabled approaches are transforming the types of seminal fluid traits that can be studied and the techniques available to study them. First, gene knockout, knockdown and editing techniques have been used successfully to reveal the functions of seminal fluid in model organisms such as flies (e.g. [27,138,139]) and mice (e.g. [140–142]), but also increasingly non-model organisms such as crickets [99,128], beetles [143] and flatworms [28,144,145]. Second, and partly explaining the first trend, it is increasingly possible to measure the expression of many seminal fluid genes (e.g. [104,122]) and proteins (e.g. [93,134]) simultaneously, overcoming some of the challenges associated with its complexity. Third, the routine application of transcriptomics and genomics approaches even to non-model organisms facilitates a much deeper comparative understanding of seminal fluid evolution (e.g. [146]). Integrating these experimental, expression and comparative approaches enables seminal fluid to be studied from multiple perspectives within the same animal group, unconstrained by the prior existence of genetic and genomic resources and thus enabling research programmes on new models for seminal fluid function and evolution selected specifically for this purpose. As such programmes accumulate, we can both test the generality of insights gained to date from model organisms and gain a better understanding of how diversity in sperm competition levels and other associated features of reproductive ecology impacts on, and is shaped by, male investment in seminal fluid.

Alongside this scope for greater depth and taxonomic breadth in future seminal fluid research, there are several specific gaps in knowledge in need of closer attention. Despite some pioneering studies (e.g. [29,52,147]), one obvious such gap is our limited understanding of female responses to seminal fluid receipt. Similarly, we still have only a rudimentary understanding of the functional and regulatory integration of the seminal fluid proteome, and our grasp of the adaptive significance of plasticity in seminal fluid production and allocation remains patchy. Ideally the fitness effects of such plasticity would be measured directly, but a recurring theme emerging from this review has also been the need for a detailed functional understanding of seminal fluid traits as a prerequisite to even begin predicting adaptive production and allocation patterns. Further, I have already highlighted the need to understand how different environmental factors combine to affect seminal fluid expression, as well as to understand individual and ultimately genetic variation in these reaction norms. Finally, although I have treated them separately here, it will eventually be important to establish links between the different levels of seminal fluid investment variation: do species with, on average, higher levels of sperm competition, for example, exhibit more or less plasticity in seminal fluid traits, or is the degree of plasticity simply reflective of the degree of variability in sperm competition levels? and does greater plasticity in seminal fluid production imply more or less plasticity in terms of seminal fluid allocation?

7. Conclusion

My aim has been to emphasize the near inevitability of seminal fluid as a key adaptation to sperm competition during animal evolution. But the relationship between sperm and seminal fluid investment is complex; with so many ways in which additional ejaculate components transferred together with sperm might influence the reproductive physiology and behaviour of mating partners to increase own reproductive success—both in concert with as well as independently of sperm—this complexity is likewise inevitable. Nevertheless, a survey of male investment strategies reveals some clear, albeit provisional, trends: (i) overall sperm and seminal fluid investment is often positively correlated across taxa, but evolutionary trade-offs sometimes also occur; (ii) seminal fluid expression is highly sensitive to environmental cues of sperm competition level; and (iii) males of several taxa are also capable of strategically allocating both sperm and seminal fluid reserves to a particular mating. What is needed now is a greater cross-talk between theoretical and empirical studies of seminal fluid; more, and more detailed, quantifications of sperm and seminal fluid investment patterns; as well as more integrative studies to link function, investment patterns and their fitness consequences across a broader range of taxa. As I just outlined, this task is certainly made easier by the advent of omics-scale methods to study seminal fluid as a multivariate trait even in non-model organisms, and particularly the increasingly widespread ability to manipulate the individual components contained within it to assess their functional and adaptive significance. As these data accumulate, so too will our ability to make sense of inter- and intra-specific variation in seminal fluid production and allocation strategies and integrate these into our broader understanding of ejaculate evolution under sperm competition.

Data accessibility

Additional data is included in the electronic supplementary material, table S1.

Competing interests

I declare I have no competing interests.

Funding

My recent research in this area has been funded by the Deutsche Forschungsgemeinschaft (grant no. RA-2468/1-1).