Examining the potential for resource‐dependent female reproductive fluid‐sperm interactive effects in a livebearing fish

Sexually selected traits can be costly to produce and maintain. The amount of resources available to an individual is therefore expected to influence investment in costly sexual traits. While resource‐dependent expression of sexually selected traits has traditionally been examined in males, resource limitation can also influence how sexual selection operates in females. Female reproductive fluids are thought to be costly to produce and may play an important role in shaping the outcome of postcopulatory sexual selection by influencing sperm performance. However, we know surprisingly little about whether and how female reproductive fluids are influenced by resource limitation. Here, we examine if resource restriction influences female reproductive fluid‐sperm interactive effects in the pygmy halfbeak (Dermogenys collettei), a small internally fertilizing freshwater fish where females store sperm. After experimentally altering female diets (high vs. restricted diets), we compared how female reproductive fluids influence two key metrics of sperm quality: sperm viability and velocity. While female reproductive fluids enhanced sperm viability and velocity, we found no evidence that female diet influenced the interactive effect between female reproductive fluids and sperm viability or velocity. Our findings build on the growing evidence that female reproductive fluids influence sperm performance and call for further attention to be devoted to understanding how resource quantity and quality influence how female reproductive fluids affect sperm performance.

future reproduction (Stearns, 1989;van Noordwijk & de Jong, 1986;Williams 1966). The quantity and quality of resource acquisition and/or allocation among individuals can also influence how animals invest in multiple costly traits (Simmons et al., 2017;Stearns, 1989;Supriya et al., 2019;van Noordwijk & de Jong, 1986). Nutritional status and resource acquisition are therefore expected to influence how animals invest in sexual traits (Rowe & Houle, 1996;Warren et al., 2013). Traditionally, sexual selection research has focused on resource dependent expression of pre-copulatory traits in males, focusing on courtship or competition behaviours and ornaments (Cotton et al., 2004). More recently, research effort has examined resource-dependence of ejaculate traits involved in post-copulatory sexual selection (Fernlund Isaksson et al., 2022;Macartney et al., 2019). However, limited resources can also influence how sexual selection operates in females. Sampling and discriminating among potential mates can be costly and pre-copulatory female mate choice can consequently be influenced by resource limitation (Cotton et al., 2006). Female mate choice can also continue after mating in the post-copulatory process of cryptic female choice, when females' bias sperm use of one male over another (Eberhard, 1996;Thornhill, 1983). Yet, we know surprisingly little about whether cryptic female choice, or the function of putative mechanisms of cryptic female choice (e.g., female reproductive tract fluids), are influenced by resource limitation (Cardozo & Pilastro, 2018).
The effects of female reproductive fluids on sperm performance can be dependent on the duration of female reproductive fluid-sperm interactions. In guppies, for example, sperm viability remained elevated over a 3-h period when incubated in female reproductive fluid, while sperm viability declined sharply in the control solution, magnifying the difference in sperm viability between the treatments over time (Gasparini & Evans, 2013). Moreover, following insemination, male-derived sperm associated proteins are increasingly replaced with female-derived proteins over time during sperm storage in Drosophila melanogaster (Diptera: Drosophilidae;McCullough et al., 2022), suggesting that female effects on sperm performance may be time-dependent.
Female reproductive fluids may be costly to produce (Cardozo & Pilastro, 2018). For example, female reproductive fluids comprise up to one-third of the volume of the total reproductive material of females from a range of salmonid species (Lahnsteiner et al., 1995;Wojtczak et al., 2007). Female reproductive fluids typically contain high concentrations of ions, enzymes, proteins, amino acids and hormones (Kor & Moradi, 2013;Lahnsteiner et al., 1995Lahnsteiner et al., , 1999Siddique et al., 2016). Production of these components of female reproductive fluids may come at a cost. Sperm performance can be influenced by specific combinations of proteins present in the reproductive fluids of some females but not others, suggesting that females may invest differently in their reproductive fluid composition (Johnson et al., 2020). Moreover, immunologically active major histocompatibility proteins present in female reproductive fluid has been implicated as mechanism facilitating cryptic female choice by differentially influencing sperm performance among males in a range of species (e.g., Geßner et al., 2017;Jokiniemi et al., 2020). Since the immune system is costly to produce and maintain, immune function is expected to exhibit condition dependence (Blanco et al., 2001;Cotter et al., 2004;Deerenberg et al., 1997). The effectiveness of female reproductive fluids in influencing sperm performance could consequently vary with resource quality or quantity. However, while investment in a voluminous, biologically active reproductive fluid could be costly, little direct evidence on the costs of female reproductive fluids or their constituent components are available.
The effect of female reproductive fluids on sperm performance can vary among females (e.g., Evans et al., 2012;Fitzpatrick et al., 2020;Jokiniemi et al., 2020;Poli et al., 2019;Simmons et al., 2009;Urbach et al., 2005;Wojtczak et al., 2007). The functional impacts of female reproductive fluids are likely determined by their composition (Zadmajid et al., 2019), which may contribute to the among-female variation in sperm performance (Rosengrave et al., 2009). If female reproductive fluids are costly, then the composition of female reproductive fluids may depend on a female's diet and/or nutritional state, which in turn has implications for their potential function during cryptic female choice (Cardozo & Pilastro, 2018). Yet whether resource availability influences female reproductive fluid composition and their function remains unclear.
Dietary manipulation in sheep influences female reproductive fluid composition, with increased reproductive hormone concentrations detected in follicular fluid when females received nutrient supplementation (Guo et al., 2019;O'Callaghan et al., 2000). Similarly, dietary supplementation increased fatty acid content and reproductive hormone levels in follicular fluid in pigs (Warzych et al., 2011;Zhou et al., 2010). Surprisingly, only one study has examined resource-dependent effect of female reproductive fluids on sperm performance (Cardozo & Pilastro, 2018). In the guppy (Poecilia reticulata; Cyprinodontiformes: Poeciliidae), sperm swam faster when incubated in female reproductive fluids from females fed a high quantity diet compared to sperm incubated in female reproductive fluids from females fed a restricted quantity diet (Cardozo & Pilastro, 2018). Thus, the limited available evidence suggests that female reproductive fluids may indeed be costly and their function in influencing sperm performance can be resource dependent.
Here, we investigate whether female reproductive fluids influence sperm velocity and viability and whether such effects are resource dependent in the pygmy halfbeak (Dermogenys collettei; Beloniformes: Zenarchopteridae). Halfbeaks are a small, internally fertilizing viviparous freshwater fish from Southeast Asia (Greven, 2010;Meisner, 2001). Halfbeaks live in mixed-sex shoals and readily interact in intersexual behaviours of courtship and competition Greven, 2010). Female halfbeaks mate multiply and can store sperm for several months (Greven, 2010; EFI personal observation) suggesting that post-copulatory sexual selection in males is strong Fernlund Isaksson et al., 2022). A recent study revealed resource-dependency of post-copulatory investment in male halfbeaks (Fernlund Isaksson et al., 2022). In particular, male halfbeaks maintained on restricted diets had shorter beaks (elongated lower jaws), invested less in testicular tissue, and performed less courtship towards females, while sperm viability, swimming velocity and morphology were unaffected by resource restriction (Fernlund Isaksson et al., 2022). However, it is not known whether resource restriction influences reproductive traits in female halfbeaks. Therefore, we employed a diet restriction experiment where female halfbeaks were fed either a high or restricted amount of food. We then examined how female reproductive fluid from these females influenced sperm performance (sperm velocity and viability) using a split-ejaculate experimental design.
We predicted that (i) female reproductive fluids will increase sperm performance compared to a control, (ii) the effect of female reproductive fluids on sperm performance will be greater in females fed a high quantity diet and (iii) the effect of female reproductive fluid on sperm performance will be more apparent when sperm are incubated in female reproductive fluid for longer durations.

| Experimental setup
Experimental fish were obtained from F3 and F4 descendants of halfbeaks wild-caught from Tebrau River, Malaysia, that were bred in an aquarium facility at Stockholm University, Sweden. The parents of the experimental fish were kept in 3 mixed-sex groups (~20-30 fish in roughly equal sex ratios) in aquaria containing ~2 cm of gravel, aeration and live and artificial plants, maintained on a 12:12 light: dark cycle at ~26°C. Offspring were collected from parental tanks (note that we did not keep track of the genetic history of individual offspring), housed together in 4 L aquaria (up to five offspring/ aquaria) for up to 2 months, and fed live Artemia salina nauplii (artemia; Anostraca: Artemiidae) twice per day. After ~1-2 months, 20-30 juveniles born in the same calendar month were moved to 72 L tanks until the onset of sexual maturity. During this period of development, juveniles were fed ground flake food and live artemia twice per day, 6 days/week. In addition, previously frozen and thawed D.
melanogaster were fed to fish once per week. At the onset of sexual maturity, fish were separated and reared in sex-specific tanks (55 or 160 L, 15-30 fish per tank) to ensure that all females were unmated prior to the experiment. At 2.5-4 months of age, females were allocated to the experimental diet treatment.
To experimentally test whether female reproductive fluids influence sperm viability and velocity, and whether resource availability influences the function of female reproductive fluids, we performed a resource manipulation experiment on adult female halfbeaks. The experiment was approved by Stockholm Animal Research Ethical Board (permit number 3967-2020).

| Female diet treatment
Adult females (n = 52) aged ~2.5-4 months from female-only tanks were photographed laterally and weighed for body mass, and subsequently moved to individual 4 L tanks (25 × 16 × 12 cm) containing 1 cm gravel, four plastic plants, 2-3 snails, aeration and a circular floating feeding arena. Body length was measured from the lateral pictures (see below). Pairs of size matched females (to ensure no size-difference prior to diet treatments) were randomly assigned to either a high or restricted diet treatment group. Females were fed different amounts of fruit flies (D. melanogaster) throughout the diet treatment (sensu Fernlund Isaksson et al., 2022). Females assigned to the high-diet treatment were fed 36 fruit flies/week (6 flies/day, 6 days/week), while females assigned to the restricted diet treatment were fed 9 fruit flies/week (1 fly/day on 3 days and 2 flies/day on 3 days each week). The number of flies fed in the high diet treatment was based on pilot observations indicating that adult halfbeaks were satiated after eating around six fruit flies. Thus, the amount of food available to females in the restricted diet treatment was ~25% of the amount available to females in the high-diet treatment (note that we did not standardize fruit fly mass between the diet treatments so the difference in food availability between treatments represents an estimate). Females were maintained on their respective diet treatments for 32-33 days.
On the final day of diet treatments, females were photographed and weighed, reproductive fluid was extracted, and sperm assays were performed (see below). The experiment was performed in seven blocks, where 6-8 females were added to the diet treatments in each block and therefore assayed together, leading to a total sample size of 52 females. However, one block was removed from all analysis (6 females) due to technical issues during sampling (one of the sperm viability dyes failed to fluoresce, likely due to overexposure to light), reducing the total sample size to 46 females (23 high diet and 23 restricted diet females).

| Female body mass, length and condition
Female body mass (before and after diet treatment) was measured by placing individual fish in a small bowl with water on a tared balance (Mettler Toledo XS105). Both before and after the diet treatments, body mass was measured twice for each female and the average of the two measurements was used for analyses. Females were photographed laterally before and after the diet treatment using a digital camera (Canon 800D with macro lens EF-S 60 mm). Body length (from the anterior portion of the eye to the caudal peduncle) was measured (before and after treatment) from lateral images using the segmented line tool in ImageJ (v1.52i; Schneider et al., 2012). Using body weight and length values, we calculated a female body condition index using Fulton's K length-weight relationship (K = body mass/(body lenth 3 )). Fulton's K is a commonly used to assess condition in fishes (e.g., Nash et al., 2006), with larger Fulton's K values indicating that an individual is heavier for a given body length, suggesting the individual is in better somatic condition.

| Female reproductive fluids and sperm extraction
The effects of female reproductive fluids and female resource availability on sperm quality were assessed using a split-ejaculate design, where sperm from one male was split across female reproductive fluids from two females (one from the high and one from the restricted diet treatments) and a control treatment. Below we describe a single split-ejaculate replicate, with a total of 23 replicates (for 46 females) performed during the experiment.
For each replicate, two females (one from the high and restricted diet treatments) were paired and assigned a new experimental identity to ensure that sperm assays were performed by a researcher blind to the treatment group of each female (J.L.F.). For each pair, females were euthanized in a benzocaine solution (150 μL benzocaine per 1 mL ethanol), washed in deionized water and placed upside down on a Styrofoam cradle. Using a Drummond micropipette with the tip inserted ~2 mm into the female's gonoduct, 1.5 μL of a saline solution (0.9% NaCl) was injected into the female's reproductive tract, and then ~1-1.5 μL of the diluted reproductive fluid was retrieved (henceforth called female reproductive fluids) and placed in a 0.5 mL Eppendorf tube. This procedure was repeated five times to obtain a sufficient volume of female reproductive fluid to use in sperm assays. This approach has been used to extract female reproductive fluids from other internally fertilizing fish (e.g., Devigili et al., 2018;. Females were processed sequentially. To control for potential order effects during female reproductive fluid extraction, a researcher who was not blind to the female's treatment group (E.F.I) ensured that female reproductive fluid from the high and restricted diet treatments was extracted first in an equal number of trials.
Then, a male randomly chosen from a male-only tank was euthanized in a benzocaine solution (as above), washed in deionized water and placed on a glass slide. Sperm bundles were stripped into a saline solution (henceforth called sperm) by applying gentle pressure on the male's abdomen. A 10 μL volume of sperm was transferred from the slide to a 0.5 mL Eppendorf tube and mixed with 20 μL of Hank's Balanced Salt Solution (HBSS, Sigma-Aldrich). HBSS activates halfbeak sperm and is used when assessing halfbeak sperm quality (e.g., Fernlund Isaksson et al., 2022;Reuland et al., 2021). The sperm bundles were left to disassociate in the HBSS/saline solution for 2 min.
Under our standard laboratory conditions, sperm typically remain motile for several hours when activated in a HBSS/saline solution (personal observation).

| Sperm assays
Sperm viability and velocity were measured from separate subsamples taken from the HBSS activated sperm solution (see Figure S1). From each pair of females in a replicate, a 2 μL (sampling time 1) and 3 μL (sampling time 2) aliquot of female reproductive fluid was distributed to two 0.5 mL Eppendorf tubes, respectively.
Reliably extracting more than 5 μL of diluted female reproductive fluid from each female was challenging. Therefore, the additional volume of solution (i.e., 3 μL instead of 2 μL) used in sample time 2 was done to allow a subsample to be extracted to assess sperm velocity (see below). For the control treatment, a 2 μL (sampling time 1) and 3 μL (sampling time 2) aliquot of the same saline solution (0.9% NaCl) used to extract reproductive fluids from females was added to two 0.5 mL Eppendorf tubes, respectively. After sperm had incubated for 2 min in the HBSS/saline solution, an equal volume of the solution was added to each Eppendorf tube containing either female reproductive fluid or a control solution. In total, sperm was distributed into 6 vials, two per treatment, three of which with a total of 4 μL (sampling time 1) and three with a total of 6 μL (sampling time 2). Sperm viability was assayed after  Our aim was to assess sperm velocity across the different treatment groups for all of the 23 males assessed. However, we were unable to obtain sperm velocity measures in two experimental blocks (n = 8 males) due to excessive sperm sticking to the glass slide and cover slip. After switching to disposable glass slides (described above) sperm swam freely and we were able to assay sperm velocity from a subset of 15 males, representing a total of 30 females each in the high and restricted diet treatment and 15 control samples (total of 45 sperm velocity samples). The threshold for defining static cells was set to 25 μm/s for VCL. After applying this criterium, 42 samples were used in the analysis (15 from high diet, 14 from restricted diet, and 13 from the control treatment).
To assay sperm velocity, a 2 μL subsample was taken from the sampling time 2 aliquot and was placed on a disposable glass slide fit- Rather than focusing only on one of these velocity metrics, and to avoid issues of multi-collinearity in our analyses, we performed a principal component analysis (PCA) to reduce the dimensionality of the CASA generated data (including n = 15 high diet, n = 14 restricted diet, and n = 13 control samples). The PCA returned three principal components (PCs), out of which only the first PC (PC1) had eigenvalue <1, which accounted for 99% of the total variance in the data and was loaded nearly equally by all sperm velocity parameters (Table S1). Therefore, we focused only on sperm velocity PC1 in our analyses.
Because sperm swimming speed typically decreases over time in fishes (Cosson et al., 2008), previous sperm velocity assays performed in halfbeaks have examined sperm soon after sperm activation (typically within 2-5 min, Fernlund Isaksson et al., 2022; Reuland et al., 2021). In this experiment, our aim was to allow female reproductive fluid-sperm interactions more time to influence sperm performance. However, under standard laboratory practices, halfbeak sperm velocity is slower than initial sperm velocities when assessed 1-h after activation (pilot data from n = 5 males assessed at a time of 2 min and 1 h after sperm activation; unpublished data). Therefore, to allow for female reproductive fluid-sperm interactive effects to develop, while acknowledging that sperm velocity may decline over time, we chose to assay sperm swimming speed after 20 min of incubation with female reproductive fluids (or a control)

solution.
However, by assaying sperm 20 min after activation, comparatively few sperm cells were sampled in the velocity assay compared to other experiments on halfbeaks. Specifically, we sampled an average of 59 ± 7 cells (mean ± SE, range: 1-149 cells) per sample, which represents ~40% fewer sperm sampled on average compared with a previous study assessing sperm velocity in halfbeaks (Fernlund Isaksson et al., 2022). Therefore, to account for this reduced sampling and the variance in the number of sperm assessed per sample, we performed two complementary analyses. First, we examined mean sperm velocity values for each male. In addition, we also analysed velocity data in a manner which is independent of number of cells sampled by only assessing the fastest swimming sperm in each sample (using the maximum VCL of each sample, and then including the VSL and VAP values from the same sperm cell) and performing a PCA on this data to generate a maximum velocity PC1 score (see Table S2).

| Statistical analysis
Body mass, length and condition (Fulton's K) were analysed using separate linear mixed models (LMM) with the trait as response variable, and diet treatment, time (before or after treatment) and their interaction as fixed factors. Additionally, female identity was added as a random effect to account for the repeated measures applied to each female, and experimental block was added as a second random factor to account for any underlying differences between the seven blocks.
Sperm viability was analysed using a generalized linear mixed model (GLMM) with a binomial family function. The number of live and dead sperm cells were treated as a bound response variable using the cbind function, with treatment group (high diet, restricted diet and control) and incubation time (10 or 30 min) modelled as fixed factors. Additionally, male ID and experimental block were added as random factors to account for repeated sampling among males at the different sampling points and underlying variation in sperm traits between males and variation among blocks, respectively. Unfortunately, the model structure described above was overdispersed (dispersion ratio = 18.27). Therefore, an observation-level random effect was added to account for overdispersion (Harrison, 2015), which corrected overdispersion in the final model (dispersion ratio = 0.11). While there remains debate surrounding the use of observation-level random effect for addressing overdispersion in binomial models (Harrison, 2015), alternative approaches to reduce dispersion (e.g., using beta-binomial models; Harrison, 2015) did not correct overdispersion in the model. As we prioritized presenting a model without overdispersion, we favoured the GLMM model that included an observationlevel random effect.
Furthermore, sperm velocity was examined using an LMM, with sperm velocity PC1 as the response variable, treatment as fixed factor, and male ID and block as random factors to account for any underlying differences between males (who were repeatedly sampled among treatments) and blocks. We also constructed a linear mixed model with the maximum sperm velocity PC1 employed as the response variable, treatment as fixed factor, male ID and block as random factors.
All models were initially fitted with all possible interaction terms in place. In cases where the interactions were significant, they were retained and presented in the models summarized in the results. When interactions terms were not significant they were retained in the model but not interpreted (note that excluding nonsignificant interaction terms did not qualitatively alter the results).
Linear and generalized linear mixed models were run in the lme4 package (Bates et al., 2015). Significance values were obtained by running an ANOVA from the car package (Fox & Weisberg, 2011).
When applicable, post hoc Tukey tests performed from emmeans package (Lenth, 2019). For GLMMs, overdispersion was assessed using the check_overdispersion function in the performance package (Lüdecke et al., 2021). Principal component analyses were performed using prcomp function in base R. R version 4.0.2 (R Core Team, 2020) was used.

| Effect of diet treatment on body size and condition
Diet treatment influenced female body mass and length, with a significant interaction between diet treatments and time period detected for both body size measures (Table 1). Post hoc tests revealed that body mass (Tukey post hoc pairwise comparison: t 41.3 = 0.31, p = 0.99) and length (t 40.9 = −0.07, p = 1.0) did not differ between the diet treatment groups at the start of the experiment (Figure 1a,b). However, at the end of the diet treatments, females maintained on the high diet were heavier (t 41.3 = 5.69, p < 0.001) and longer (t 40.9 = 3.14, p = 0.02) compared to females maintained on restricted diet (Figure 1, Table 1). Females on the high diet significantly increased their body mass (t 44 = −20.86, p < 0.001) and length (t 44 = −24.75, p < 0.001) during the experiment, while body mass (t 44 = 1.69, p = 0.34) did not change for females maintained on the restricted diet (Figure 1a). Female body length increased over the course of the experiment for females maintained on the restricted diet (t 44 = −9.83, p < 0.001), albeit to a lesser extent than for females maintained on the high diet (Figure 1b). A significant interaction between diet treatment and time period was also detected when assessing female condition (Fulton's K, Table 1C). Although female condition did not differ between diet treatment groups at the start of the experiment (t 81 = 1.55, p = 0.41), female condition differed between diet treatment groups at the end of the experiment (t 81 = 9.36, p < 0.001). Females from the high diet treatment maintained their condition throughout the experiment, with condition being similar before and after the diets were applied (t 44 = 2.02, p = 0.20; Figure 1c). In contrast, the condition of females from the restricted diet treatment significantly reduced over the course of the experiment (t 44 = 10.24, p < 0.001; Figure 1c).

| Sperm viability
Sperm viability was influenced by the presence of female reproductive fluid and incubation time (Table 2, Figure 2). Post hoc tests revealed that sperm viability was higher when sperm were exposed to female reproductive fluid than when maintained in a saline control solution (high diet vs. control: z = −7.03, p < 0.001; restricted diet vs. control incubation: z = −6.64, p < 0.001; Table 2A, Figure 2). However, sperm viability did not differ between female reproductive fluid of high and restricted-diet treatment females (z = 0.37, p = 0.93; Figure 2). However, the main effect of time in the model indicates that sperm viability decreased over time, and the nonsignificant interaction term in the model indicates that the decrease in viability occurred across all treatment groups (Table 2A, Figure 2).

| Sperm velocity
Female reproductive fluid also had a positive effect on sperm swimming velocity. Post-hoc tests revealed that sperm velocity PC1 scores were higher when sperm were swimming in female

| DISCUSS ION
Our results demonstrate that sperm performance is improved when  Note: Sperm viability and velocity were compared when incubated in female reproductive fluid from females maintained on the high and restricted diet treatment and a control (saline) solution -collectively referred to a Treatment in the table. The effect of treatment and incubation time (10-and 30-min incubation) on (A) sperm viability was assessed. Sperm velocity was assessed using (B) sperm velocity PC1, the first-principal component describing sperm swimming parameters when using all sperm velocity data and (C) maximum sperm velocity PC1, the first-principal component describing sperm swimming parameters when only examining the fastest sperm in each sample. Note that sperm velocity was only assessed once and therefore there is no time effect in these models. The total number of males assessed (N, note that males were assessed across each treatment in a split-ejaculate design), test statistics (χ 2 ), and p-value are presented for each predictor.

TA B L E 2
Female reproductive fluid effects on sperm quality.
restricted quantity diet. Therefore, our rare experimental examination offers no evidence that the differences in female diet quantity used in this study determine how female reproductive fluids influence sperm performance in halfbeaks.
Our findings contribute to the growing evidence that female reproductive fluids enhance sperm performance (Zadmajid et al., 2019). Specifically, sperm viability and velocity were enhanced in female reproductive fluids in halfbeaks. This increase in sperm performance could have important fitness implications as sperm viability and velocity are key determinants of fertilization success across a wide range of animals (Simmons & Fitzpatrick, 2012), including fishes (Fitzpatrick, 2020). In particular, in internally fertilizing Poeciliid fishes, sperm viability and velocity predict competitive fertilization success following artificial insemination of sperm from two rival males (Boschetto et al., 2011;Fitzpatrick & Evans, 2014;Gasparini et al., 2010;Smith, 2012). Whether sperm swimming velocity predicts fertilization success in halfbeaks has yet to be investigated, although this represents an interesting target of future investigations. Nevertheless, the stark effect of female reproductive fluid on sperm performance in halfbeaks highlights the merits of examining sperm performance in the medium that they have coevolved to operate in.
Our findings contrast the only other study to experimentally examine how female nutritional state influences the effects of female reproductive fluids on sperm. Cardozo and Pilastro (2018) showed that sperm velocity was enhanced when incubated in female reproductive fluids from females fed a high quantity diet compared to females fed a low quantity diet in guppies. The contrasting results could be explained by several factors. First, female reproductive fluids may simply be more sensitive to nutrient intake in guppies than in halfbeaks. Second, the duration of the interaction between female reproductive fluid and sperm in halfbeaks may have been too brief, as sperm incubated in female reproductive fluids for a maximum of 30 min in our study. In guppies, the positive effect of female reproductive fluids on sperm viability is more prevalent when assayed after a 3-h incubation compared to when assayed shortly after activation (Gasparini & Evans, 2013

F I G U R E 3
The effects of female reproductive fluid from females maintained on the high (orange) or restricted (blue) diet treatment and a control (saline) solution (grey) on sperm velocity. Boxes of the box plots illustrate the median (centrally located line), upper and lower quartile (top and bottom of the box), while the extended lines represent the 95% confidence intervals of the median. The raw data points are superimposed on top of the plots.
weight loss, which could indicate that the restriction in resource quantity we applied was not sufficient (or applied long enough) to uncover resource-dependent effects. Alternatively, females of both diet treatments may have obtained enough macro-nutrients (e.g., protein, carbohydrates and fats) or micro-nutrients (e.g., vitamins and minerals) from the experimental diet applied in this study (i.e., fruit flies) to mask potential resource-dependent effects on female reproductive fluid production. Since specific combinations of proteins in female reproductive fluids can influence sperm performance (Johnson et al., 2020), experimental examinations on a range of diets that vary in quantity and the ratio of macro-and micro-nutrients in a multidimensional nutritional geometry framework (sensu Ng et al., 2018) will help clarify whether and how nutritional state influences female reproductive fluid-sperm interactions.
Whether the positive effects of female reproductive fluids on sperm performance that we detected has broader impact on postcopulatory processes in halfbeaks remains to be seen. In a growing number of species, female reproductive fluids play a key role in facilitating cryptic female choice (Firman et al., 2017;Gasparini et al., 2020).
For example, female reproductive fluids in guppies favour sperm from unrelated males, suggesting that female reproductive fluid can act as a mechanism to reduce inbreeding .  (Devigili et al., 2018) and Atlantic cod (Gadus morhua; Gadiformes: Gadidae; Beirão et al., 2015). This suggests that female reproductive fluids can act as a pre-zygotic barrier to reduce hybridization. In this study, we did not directly test cryptic female choice, inbreeding avoidance or hybridization avoidance, although investigating these topics in halfbeaks would be an interesting extension of the findings of our study -particularly if examined in a multidimensional nutritional geometry framework.
Our study highlights the complexity of post-copulatory sexual selection as well as the importance of replicating studies. This study doubled the available literature on resource-dependent female reproductive fluid-sperm interactions and suggests that further studies are required to appreciate the potential for female nutrient status to influence post-copulatory processes. Despite not finding support for resource-dependent responses, our findings suggests that there is scope for female reproductive fluids to act as a mechanism for cryptic female choice in halfbeaks, since female reproductive fluids from female halfbeaks did influence sperm performance. Future studies should therefore examine other aspects of female reproductive fluids, in addition to investigating the effect of components of female reproductive fluids on sperm performance. Interactions between sperm and female reproductive fluids are crucial for the fertilization process and understanding the mechanisms involved can be useful for a range of applications, including experimental design and fertility biology.

ACK N O WLE D G E M ENTS
This work was funded by a grant from Vetenskapsrådet (2017-04680) to J.L.F. We thank Alessandro Devigili, Mirjam Amcoff and Rebecca McNeil for their input on experimental design and help with laboratory work. We also thank two anonymous reviewers for thorough and constructive feedback.

CO N FLI C T O F I NTE R E S T S TATE M E NT
The authors declare no conflicts of interest.

PE E R R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1111/jeb.14166.

DATA AVA I L A B I L I T Y S TAT E M E N T
The raw data files and R script associated with this study are deposited on Dryad (https://doi.org/10.5061/dryad.n02v6 wx1h).