In contrast to earlier views of reproduction as a cooperative affair between two mating partners, it has become evident that sexual encounters between males and females are commonly accompanied by conflict between sexually interacting partners 1. A basic understanding of this phenomenon can be found in Bateman's principle, which holds that a female's reproductive output is primarily limited by access to resources while male fitness is primarily limited by access to female gametes 23. The most important implication of this is that females are selected to be choosy about whom to mate with, while males can profit from mating with many females. Therefore, with the exception of strict monogamy, the genetic interests and optimal reproductive strategies of mating partners will rarely coincide.

Under these circumstances a trait that is advantageous for one mating partner but harmful to the other can fuel repeated cycles of adaptation and counter-adaptation in the two sexes, i.e. a coevolutionary arms race 4. The last decade has seen many contributions showing that such sexual antagonism between the sexes can severely impact the evolution of reproductive behavior, physiology and morphology in animals 567. For instance, conflict over mating in water striders has led to the evolution of harassment behaviors and grasping structures in males and anti-grasping traits in females 8. Although sexual conflict in plants is less extensively studied, here supporting evidence is also accumulating 910. In addition to antagonistic traits in males and females, competition between males can indirectly cause sexual conflict when traits beneficial in sperm or pollen competition - as a corollary - harm females and lower their fitness 11121314. For example, early induction of stigma receptivity by pollen donors promotes avoidance of pollen competition in Collinsia heterophylla, but is associated with lower seed set of pollen recipients 15. However, investment of limiting resources into expensive pollen or ejaculates 16 that maintain sperm competitiveness has also been demonstrated to come at non-trivial costs to other life-history traits in males 171819. It is therefore not surprising that males modulate ejaculate size according to sperm competition risk, which can even lead to sperm-limited females 20. Hence, the expression of traits beneficial in sperm competition and female manipulation, e.g. male accessory gland proteins, is costly to their bearers, such that males can become limited in the number of fertilizations they can secure.

Ultimately, the evolutionary outcome of sexual conflict for both sexes depends on the relation between costs and benefits of mating interactions, which define the fitness pay-off 4. Benefits are typically expressed in additional eggs, sired offspring or enhanced offspring quality, which can all increase an individual's fitness. Costs can also take several forms. First, costs of production that result from, for instance, sperm and seminal fluid synthesis. Second, costs of manipulation result when receipt of manipulative compounds shift reproductive investment away from the optimum. Third, costs of collateral harm result from loss of resources, e.g. when received compounds (or their metabolites) need to be detoxified. All costs accumulate over an individuals' lifetime and lead to loss of fitness.

In gonochorists (i.e. separate sex species), costs and benefits are normally estimated separately for males and females, and can be inferred by comparing populations or closely related species under different selection pressures 2122. Importantly, direct comparison of costs of mating for both sexes requires a common measure of reproductive investment, which is problematic in separate sex species. In hermaphrodites, however, a basic assumption of sex allocation theory is the existence of a fixed budget for reproduction, thus resources spent on the male function will trade off against female reproductive investment and vice versa. As a consequence, costs of donation and subsequent de novo synthesis of ejaculates can be detected via a decrease in female reproductive investment. Second, the cost of manipulation of reproductive investment, i.e. sex allocation, can be traced by proxies of fitness in both reproductive functions. Again, when predictions of sex allocation theory hold, manipulation by donors could for example, result in an increase in egg production, and a concurrent decrease in male reproductive investment. Third, as collateral harm usually affects both sexes, the costs of harm can be traced through a decrease in fecundity, but since the male function should be affected too, ideally, ejaculate investment should be measured as well.

The flexibility of sex allocation in hermaphrodites is of great importance since it allows them to adapt quickly to local changes in reproductive opportunities. However, plasticity in sex allocation increases vulnerability to exploitation by mating partners. The latter occurs because selection favors sperm donors (hereafter donors) that transfer ejaculates that increase egg production in receiving partners (hereafter recipients), creating potential for sexual conflict over sex allocation 2324, which is predicted to evolve readily 25. Costs of such manipulation fall in the second category mentioned above, and may be detected through decreased investment in the male function or increased egg output and/or enhanced growth (size advantage model 26). Despite the attention for sexual conflict in hermaphrodite mating systems, data on the costs of expression of antagonistic traits is rare (or involved the concurrent elimination of both sexual functions 27), although several authors point out that such data would be essential for determining the evolutionary consequences of antagonistic traits 42829.

The hermaphroditic pond snail Lymmaea stagnalis has been used as a model to study reproductive ecology [e.g. 303132]. De Visser et al. 31 found that snails that were allowed to copulate produced half as many eggs as non-copulating individuals, despite the presence of stored allosperm in all snails. This led them to conclude that in non-copulants male resources were re-allocated to the female function, as sex allocation theory predicts 2333. This study has been considered the most clear-cut demonstration of resource re-allocation between male and female function, and thus a direct trade-off as predicted by sex allocation theory. Quite to the contrary, a study by Van Duivenboden et al. 30 showed that inhibition of egg laying occurs upon copulating as a female, suggesting that compounds in the ejaculate may be responsible. This idea is corroborated by recent work 3435 showing that seminal proteins transferred during copulation induce a delay in egg mass production.

To distinguish between the two hypotheses for the decreased fecundity upon copulation, we performed two experiments using L. stagnalis to determine the costs associated with copulation and the mechanism underlying loss of fecundity. In the first experiment we restricted the roles in which snails could copulate. The costs of mating as a male and a female were expressed in a common measure of investment, namely the decrease in egg production compared to non-mating animals. In the second experiment, we specifically assessed the investment in ejaculates (i.e. male investment) by determining the number of sperm and amount of seminal fluid transferred during a single copulation. Taken together, these experiments provide the first quantification of the costs of ejaculate receipt and donation in a simultaneously hermaphroditic animal.

We found that single sex-acting individuals experienced a similar decrease in fecundity as did reciprocally mating individuals. From the perspective of the female function this suggests receipt of ejaculates to cause reduced fecundity, because reciprocals received an equal number of ejaculates compared to females. In addition, reciprocals mated on average the same number of times as donors did. Without knowledge about the cost of ejaculate donation this may suggest that all of the observed reduction in egg production after mating is due to receipt of an ejaculate, as has been proposed before 30. Surprisingly, however, for animals restricted to the male role, egg production was decreased to the same extent as in females and reciprocals. Since focal males did not receive any ejaculates, any decrease in fecundity can only be attributed to copulations as donors and suggests a re-allocation between male and female function. However, focal males donated the same number of times as reciprocals did, which seems incompatible with the observed costs of mating in reciprocals.

This paradox highlights the importance of also quantifying ejaculate investment to different recipient types. In fact, we found that sperm donation in the first copulation after an isolation period was twice as high towards imposed females compared to hermaphrodites. Provided that these sperm investment patterns persist over consecutive matings, the increased investment in these ejaculates provides a possible mechanism for the observed decrease in fecundity of male acting snails compared to reciprocals. For reciprocally mating snails, on the other hand, ejaculate receipt as well as ejaculate production and delivery costs may be responsible for decreased fecundity in mating individuals. Based on these results we conclude that Van Duivenboden et al. 30 and De Visser et al. 31 were both partly correct.

The results of our second experiment suggest that the main costs of ejaculates arise from enhanced investment in sperm and not from seminal fluid investment. Although this outcome is only based on the costs of a single copulation, it is surprising for three reasons. First, previous work on L. stagnalis showed that inhibition of egg laying is the result of at least one seminal protein and not of the presence of sperm 37. Therefore, one may expect that the amount of seminal fluid would be modulated instead of sperm. However, the dry weight of donor prostates is independent of recipient type. In this respect it is a tentative idea that not the total volume, but rather the relative quantities of transferred proteins within an ejaculate are modulated. Such strategic allocation of seminal peptides has been shown to occur in Drosophila melanogaster 38 indicating that such mechanisms could also operate in other species where post-copulatory sexual selection is important.

Second, although it is known that pond snails are capable of strategic sperm donation 39, the result that more sperm are transferred to imposed females compared to hermaphrodites is unexpected. This is simply because feminized snails are not more fecund than hermaphrodites when kept in isolation 3035. Donors are therefore unlikely to benefit from increased sperm investment towards operated snails. Although speculative, it is in principle possible that surgery interferes with a cue used by donors for partner assessment. This may alter the donor's expectation of egg load and/or risk of sperm competition, but an active role of the recipient in accepting larger ejaculates cannot be ruled out. Nonetheless, sperm allocation was found to be tailored according to recipient type. Subsequent copulations with other recipients of the same type may yield similar sperm investments 40. Conversely, repeated large donations could also quickly deplete sperm stores, resulting in decreasing investments 20. It therefore remains to be tested in pond snails whether sperm donation in subsequent copulations continues to follow this pattern of male investment.

Third, in combining the results from the two experiments it appears that imposed females (in the "Receive" treatment) suffer decreased fecundity from receipt of increased numbers of sperm. This is the first indication that a "sperm effect" may be present in L. stagnalis. The previous absence of a sperm effect 37 is possibly a consequence of the methods used. Namely, in the latter paper, sperm was recovered from excised sperm storage organs and mixed with single purified prostate proteins before experimental insemination. This practice excludes any male compound that may originate from either the sperm duct or the vas deferens and penial complex, which may be present in naturally transferred ejaculates. Especially the sperm duct seems relevant in this respect, since its epithelium contains three types of excretory cells that contain proteinaceous compounds 41. Similarly, although in fruit flies the accessory male gland products have received most attention, their ejaculatory duct and ejaculatory bulb also add molecules to seminal fluid 4243. Clearly, these additional secretions deserve closer investigation with respect to their functions in animal reproductive biology.

In female copulants, net egg investment is decreased, since fecundity losses are not balanced by increased investment per offspring (i.e. dry weight per egg). This implies that the fitness of both mating partners is affected, which is indicative of a harmful strategy. Inflicted harm can be adaptive if female responses to harm benefit the donor. Alternatively, harm is collateral when traits involved in intraspecific competition also have deleterious effects on females as a side-effect 444. Under the experimental conditions tested, we find no indication that pond snails increase the remating interval or egg output which would suggest the harm to be adaptive [but see 45]. However, the fact that pond snails are highly promiscuous 32 and sperm of multiple partners can be stored for three month after insemination 46, harm could arise collaterally from adaptations to sperm competition.

In the hermaphrodite L. stagnalis we find high costs of mating for both mating roles. From the decrease in fecundity of imposed males we infer that donor costs result from investment in sperm and seminal fluid. Female copulants are negatively affected by the receipt of ejaculates. Interestingly, the fecundity reduction correlated with sperm numbers in the first copulation, while differences in seminal fluid quantities were not found. This "sperm effect" may be a novel feature of pond snail reproductive ecology. In addition, we conclude that two earlier hypotheses on the causes of fecundity decrease upon mating are both correct 3031, depending on the mating role an individual snail assumes.

Although the current study lacks measures of lifetime fecundity, we show that the costs of mating in both sex roles are considerable in our model system. Such costs are important determinants of reproductive dynamics in populations, since they affect fitness pay-offs. Future studies could combine mating costs with proxies for fitness, so that fitness pay-offs in units of investment via both reproductive functions in hermaphrodites can be used to evaluate the impact of sexually antagonistsic interactions and harm. The quantification of the costs of receipt and donation of ejaculates, as we present here, is therefore a necessary step, by providing insight into the economics of sexual conflict in a hermaphrodite.

JNAH and JMK have conceived and designed the study and carried it out, performed the statistical analyses and drafted the manuscript. JE was instrumental in conceiving the study, contributed to the statistical analyses and participated in drafting the manuscript and revised it critically for important intellectual content. JNAH coordinated all steps of the process. All authors have read and approved the final manuscript.