Female reproductive fluid concentrations affect sperm performance of alternative male phenotypes in an external fertilizer

There is growing evidence that the female reproductive fluid (FRF) plays an important role in cryptic female choice through its differential effect on the performance of sperm from different males. In a natural spawning event, the male(s) may release ejaculate closer or further away from the spawning female. If the relative spatial proximity of competing males reflects the female pre‐mating preference towards those males, then favoured males will encounter higher concentrations of FRF than unpreferred males. Despite this being a common situation in many external fertilizers, whether different concentrations of FRF can differentially influence the sperm performance of distinct male phenotypes (favoured and unfavoured by the female) remains to be elucidated. Here, we tested this hypothesis using the grass goby (Zosterisessor ophiocephalus), a fish with distinct territorial‐sneaker reproductive tactics and female pre‐mating preference towards territorial males, that consequently mate in an advantaged position and whose sperm experience higher concentrations of FRF. Our findings revealed a differential concentration‐dependent effect of FRF over sneaker and territorial sperm motility only at low concentrations (i.e. at the distance where sneakers typically ejaculate), with increasing FRF concentrations (i.e. close to the eggs) similarly boosting the sperm performance of both sneaker and territorial males. The ability to release sperm close to the eggs is a prerogative of territorials, but FRF can likewise advantage the sperm of those sneakers that are able to get closer, allowing flexibility in the direction of female post‐mating choice.

| 1199 PINZONI et al. et al., 2017). The latter occurs whenever polyandrous females can bias sperm utilization during and/or after mating. Unlike sperm competition, that has been extensively studied and recognized as a powerful force shaping male behaviour, morphology and physiology (Birkhead & Møller, 1998;Birkhead & Montgomerie, 2020;Fitzpatrick, 2020;Simmons, 2001;Simmons & Wedell, 2020), cryptic female choice has remained for a long time the least investigated and understood mechanism of sexual selection (Firman et al., 2017). This is mostly due to the experimental challenges associated with the study of these cryptic processes, and to the difficulties in disentangling their effects from those driven by sperm competition (Firman et al., 2017;Thornhill, 1983).
Although multiple routes of cryptic female choice, acting at different stages of the reproductive process (from right after mating to the moment of fertilization), have been identified (Eberhard, 1996), the mechanistic bases of these processes often remain obscure, particularly in externally fertilizing organisms (Kekalainen & Evans, 2018).
However, since in externally fertilizing species females often have limited control over the male competitors participating in the reproductive event, sexual selection is expected to be particularly intense in its post-mating component, favouring the evolution of processes of cryptic female choice (Evans & Sherman, 2013).
Recently, female reproductive fluid (FRF) has been suggested to be a possible key mediator of cryptic female choice in both internal and external fertilizers (Firman et al., 2017). This fluid surrounds the eggs before and during fertilization and is kept inside the female reproductive tract in internal fertilizers or released along with the eggs in external fertilizers (Zadmajid et al., 2019). Here, we adopt the term FRF (as in (Gasparini et al., 2020)) to avoid specific taxa-related terminology since this fluid can have different origins (ovarian, oviductal, follicular and/or coelomic), but is in any case the medium with which sperm interact during fertilization (Gasparini et al., 2020;Zadmajid et al., 2019). FRF has been shown to positively affect various sperm performance traits, typically associated with sperm competition success, across a variety of different species, including a number of external fertilizers. In fish for example, FRF can mediate sperm attraction, enhance sperm viability, increase sperm velocity and sperm motility, prolong the duration of sperm motility, and modulate sperm trajectory and propulsion efficiency (Alonzo et al., 2016;Butts et al., 2012;Elofsson et al., 2006;Gasparini & Evans, 2013;Lehnert et al., 2016;Lymbery et al., 2017;Myers et al., 2020;Poli et al., 2019;Rosengrave et al., 2009;Urbach et al., 2005). Moreover, sperm performance in presence of FRF can also depend on specific male-female interactions, due to a differential influence of FRF on the ejaculates of different males (Alonzo et al., 2016;Gasparini & Pilastro, 2011;Lehnert et al., 2016Lehnert et al., , 2017Lymbery et al., 2017;Myers et al., 2020;Poli et al., 2019;Rosengrave et al., 2016;Stoltz & Neff, 2006). Such differential effects on sperm performance represent a crucial mechanism through which females can bias fertilization, thereby offering significant potential for cryptic female choice (Firman et al., 2017).
In externally fertilizing fish, for instance, FRF has been shown to favour ejaculates of unrelated males during fertilization to avoid inbreeding (Lehnert et al., 2017) and to bias the fertilization success towards the preferred male phenotypes in a fish species characterized by alternative male reproductive tactics (further details can be found later in relation to our research question) (Alonzo et al., 2016).
Yet, many factors of the external fertilization process's ecology have been neglected in this context, first and foremost the proximity of the male competitors to the spawning female.
It is a widely held assumption that how close a male is to the female while she releases her eggs is an important predictor of his paternity success, with the probability of fertilization increasing with the degree of male proximity to the female (Blanchfield et al., 2003;Egeland et al., 2015;Hutchings et al., 1999;Poli et al., 2021;Stoltz & Neff, 2006). Growing evidence indicates that paternity success in externally fertilizing species is indeed affected by the relative position of the competing males and the female, and that females, when possible, select a spawning site close to their preferred mate (Li et al., 2015;Poli et al., 2021).
However, so far, no one has considered the implications of this for FRF-mediated post-mating sexual selection, specifically, whether different concentrations of FRF (experienced by different males) might change or amplify the paternity outcomes discussed above.
During fertilization, sperm encounter a gradient of increasing FRF concentrations in their journey towards the eggs and this is crucial to localize the eggs. Not all the sperm encounter the same gradient, and, in particular, sperm from different males may experience different FRF concentrations due to their relative proximity to the female during spawning.
The occurrence of such different concentrations of FRF may provide a mechanistic way for females to favour the preferred male at post-mating level via the FRF effects on sperm, if those effects are concentration dependent.
Among external fertilizers, species showing male alternative reproductive tactics are excellent study systems to assess the possible role of FRF concentrations in mediating directional cryptic female choice, because of the pronounced pre-mating female preference for one of the two male phenotypes (usually the territorial/nesting male) that, as a consequence, gains closer vicinity to the female during spawning. The territorial phenotype is strongly preferred by females because it secures them with direct and/or indirect benefits, but females cannot prevent opportunistic males (sneaker/satellite) from participating in the reproductive events (Oliveira et al., 2008).
Furthermore, opportunistic males often produce more sperm than territorials, to boost their post-mating success. Therefore, females might exhibit mechanisms of reinforcing their pre-mating preferences in the post-mating competition (Alonzo et al., 2016).
Since the territorial male usually ejaculates closer to the egglaying female than the sneaker does (Oliveira et al., 2008), sperm of males adopting different tactics usually experience different concentrations of FRF, depending on the male proximity to the female.
Females might, therefore, exert a choice on the ejaculates of males adopting different reproductive tactics through the effect of different concentrations of FRF on their sperm performances.
Here, we test this hypothesis using the grass goby Zosterisessor ophiocephalus (Pallas), an externally fertilizing fish in which territorial/ sneaker (or bourgeois/parasitic, according to (Taborsky, 1997)) male tactics have been widely investigated (Locatello et al., 2007(Locatello et al., , 2013Mazzoldi et al., 2000;Scaggiante et al., 1999). In this species, territorial males during the breeding season dig and defend a nest (a burrow under the seagrass rhizomes), court females (that show a strong premating preference for this phenotype) and perform parental care to the eggs. Territorial males release sperm in the form of mucous trails on the nest ceiling, where the females release eggs one at the time. In contrast, sneaker males try to steal some fertilization by entering the nest and releasing ejaculates during the spawning event (Mazzoldi et al., 2000;Scaggiante et al., 1999). The level of sperm competition is intense in this species (Locatello et al., 2007(Locatello et al., , 2013Mazzoldi et al., 2000;Scaggiante et al., 1999), and sneaker males invest disproportionately more in sperm production than territorial ones.
However, the two male phenotypes do not differ in sperm velocity and viability (measured both in water and in their own seminal fluid (Locatello et al., 2007)). Under natural spawning conditions, sperm of sneaker males experience a lower concentration of FRF than those of territorials, as they are usually released in an unfavourable spatial position, further from the spawning female ( Figure 1). In the hypothesis of a directional post-mating choice (i.e. in favour of territorial males), we expect territorial sperm to be systematically favoured over sneaker sperm at all concentrations. To test this prediction, we assessed whether the different concentrations of FRF experienced during spawning elicit differences in the sperm performances of sneaker and territorial males. We measured a range of sperm motility traits in territorial and sneaker males in presence of different concentrations of FRF (0%, 10%, and 20%) that mimicked those encountered by the two male phenotypes during the spawning event. We used the FRF of the same female with sperm from two males with different tactics, in order to disentangle tactic-specific effects of FRF from the ones driven by female's identity. Moreover, in several species the sperm performance in presence of FRF has been shown to depend on specific male-female interactions, that might be influenced by female status/quality (Cotton et al., 2006). To explore the role of female identity, we performed a second experiment (with a full factorial block design) using the higher FRF concentration (i.e. simulating the optimal proximity to the eggs during spawning) and testing whether FRF of different females could differentially affect the sperm motility of males of the two alternative tactics. In doing this we also checked if the FRF effect was eventually influenced by some female condition parameters and/or by the time of the reproductive season.

| Animal sampling and handling
Grass goby females, territorial and sneaker males were sampled, in nests full of eggs, in the Venetian Lagoon during their breeding season (March-May 2021) and transported to the Hydrobiological Station "Umberto D'Ancona" in Chioggia (Venice, Italy). In the field, we performed the initial attribution of male tactics, taking advantage of previous information on the occurrence and size distribution of males performing either as territorial or sneaker (Mazzoldi et al., 2000). In particular, to avoid determination mistakes, we categorized male tactics using the extreme values of size distribution, discarding individuals of intermediate size (11-18 cm) (Mazzoldi et al., 2000). Thus, sneakers were considered as such only if a male larger than 18 cm was present in the nest where they were sampled and if their size was lower than 11 cm. Overall, the total length range of individuals used in our experiments was 6-11 cm for sneakers versus 18-23 cm for territorials.
All individuals were kept for a maximum of 5 days in separate tanks with continuous water exchange (20°C ± 1°C) under a 14:10 artificial light-dark cycle and fed daily with fresh mussel. Prior to gametes' collection all individuals were anaesthetized in a water solution of MS 222 (tricaine sulphate; Sandoz), measured (SL = distance between the snout and the base of the tail fin) and weighed. Males' field categorization as territorial or sneaker was then validated according to the characteristics of their ejaculate trails (fluid and white in sneaker males, because of the high sperm content, and dense and opaque in territorial males, because of the lower sperm count and the higher mucin content) (Scaggiante et al., 1999) and to their sperm production (sperm number of sneakers = 1 443 975 ± 128 167, territorials = 512 030 ± 31 662). After gametes' collection, all individuals were released, unharmed, at the site of collection.

| Exp. A-Effect of different concentrations of FRF on male tactics
We tested the effect of the presence and concentration of FRF on sperm performance of territorial and sneaker males. Sperm of F I G U R E 1 Schematic representation of a territorial male's nest, illustrating the (simplified) positions of the competing males and the female during spawning, together with the pattern of FRF concentration inside the nest. F, female; S, sneaker male; T, territorial male. Created with Biore nder.com. each male were exposed to three different treatments: 0% FRF (seawater only), 10% of FRF and 20% of FRF in seawater. The real amount of FRF surrounding the eggs is difficult to estimate, as it continuously changes with time since egg deposition lasts for several hours, in nests containing a water volume of around 17.8 l (Mazzoldi et al., 2000). However, a FRF 20% dilution is likely a plausible average concentration experienced by territorial male sperm, that are released from sperm trails laid in close proximity to eggs.
20% FRF dilution is also the concentration used in similar studies in other externally fertilizing fish species (Butts et al., 2012;Lehnert et al., 2016;Poli et al., 2019). In contrast, sneaker males entering the nest are usually kept at a distance from the egg laying females by territorial males (Scaggiante et al., 2005). Considering the nest volume and the amount of fluid released by the females we assume that, commonly, the sperm of sneakers might experience a range of FRF dilution from 0% to 10%. Thus, we measured sperm motility traits in 0%, 10% and 20% concentrations of FRF in order to assess the sperm performance of the males in conditions mimicking those experienced by the sneaker and territorial males.
We collected FRF from 15 ready-to-spawn females and ejaculates from 15 territorial and 15 sneaker males (see gametes and FRF collection below) and performed computer-assisted sperm analysis (CASA) in the three different treatments (0%, 10% and 20%) in randomized order for both male tactics.

| Exp B.-Effect of male reproductive tactic, female identity and their interaction
We tested whether different females favour the performance of sperm of a specific tactic (interaction female identity x male reproductive tactic), while controlling for some female condition and environmental parameters (female size, total volume of FRF produced, and time of the reproductive season). We applied a full-factorial block design in which the performance of sperm of 2 males of each tactic (2 territorials, 2 sneakers) were tested in presence of the FRF of 2 different females (see Figure 2). The sperm of each male, either territorial or sneaker, were tested with the FRF of two different females, and the FRF of the same female was used for 4 males, 2 sneakers, and 2 territorials. The use of a different female for each pair of a territorial and a sneaker male, and of two different territorial or sneaker males (2 pairs) allowed to reveal the potential effect of the interaction between female identity and male tactic while controlling for individual interactions (female identity x male identity).
The repeatability of sperm traits in FRF (10% and 20%) was checked in a separate trial using 10 males (5 sneaker and 5 territorials) and 10 females, with two replicates for each male and for each FRF concentration. The repeatability of sperm motility traits in water was already well documented in previous studies (Locatello et al., 2007).

| Gametes and FRF collection
Eggs were obtained from anaesthetized ready-to-spawn females through a gentle pressure on their swollen abdomen (previously dried to prevent accidental activation of the eggs by water) and collected on standard microscope slides. The FRF surrounding the eggs was collected with a Gilson P2 micropipette (mean fluid volume per female: 23.6 μL ± 2.4) and preserved at −80°C until 1 hour before analysis, when it was thawed and diluted with filtered seawater at 20°C ± 1°C to the concentration of interest (10% or 20% according to the experiment). FRF was maintained in ice until analysis. To obtain the 10% and the 20% solution, 2 or 4 μl of FRF was diluted with 18 or 16 μl of solution (seawater + sperm) respectively.
Ejaculates were obtained through a gentle pressure on the abdomen (previously dried to avoid accidental activation of sperm by water) of anaesthetized males and collected with a Gilson pipette.
Ejaculate samples were centrifuged at 13.300 g for 3 min at 4°C to separate sperm from the supernatant seminal fluid (mean fluid volume: territorial: 93.6 μL + 9.8; sneaker: 16.4 μL + 1.8) and sperm cells were re-suspended in an extender inactivating medium (3.5 g l −1 NaCl, 0.11 g l −1 KCl, 0.39 g l −1 CaCl2, 1.23gl −1 MgCl2, 1.68gl −1 NaHCO3, 0.08gl −1 glucose, pH 7.7) (Fauvel et al., 1999). FRF and sperm samples were maintained on ice until analysis (within 1 h from collection). As sperm concentration varies among males and is significantly higher in sneakers than territorials, the volume of inactivating solution was individually adjusted to reach a standardized final concentra-

| Sperm analysis
For each assay, 6 μL of inactivated sample was activated by adding 1 μL of male's own seminal fluid and 13 μL of (i) 0% FRF or (ii) 10% FRF or (iii) 20% FRF, in random order.
Activated sperm samples were incubated for 2 min, a time sufficient to ensure the homogeneous activation of sperm while avoiding exhaustion, since in this species sperm remain active for more than 30 min (Scaggiante et al., 1999). Three μl of sample was then placed

| Statistical analysis
All analyses were conducted using R v 3.6.3 (R Core Team, 2020).
Repeatability was tested using the "rptR" package with Gaussian distribution for the variables VCL, LIN, and BCF, and Proportion distribution for sperm motility, and based on 1000 permutations. All means are shown with associated standard error.
In both experiments, VCL, LIN and BCF were analysed using linear mixed effect models ("lmer" function of the "lme4" package), while sperm motility was analysed with a generalized linear mixed effect model ("glmer" function of the "lme4" package) assuming a binomial error distribution and logit link function. In experiment A, we tested the effect of the treatment (0% FRF, 10% FRF, 20% FRF), of male tactic, and of their interaction by including in the mixed effect models the treatment and the male mating tactic as fixed factors and the male identity and the female identity as random factors with fixed intercept. In experiment B, the mixed effect models included male mating tactic as a fixed factor and male identity, female identity and the interaction between male mating tactic and female identity as random factors with fixed intercept. When a significant interaction male tactic * female identity was detected, female size, volume of FRF produced and time of the reproductive season were added in the model as covariates, to test for a potential influence of the female condition.
The associated p-values of the fixed factors were obtained from the "anova" function of the "lmerTest" package using Satterthwaite's approximation to calculate the denominator degrees of freedom. p values for the random factors were obtained by likelihood ratio tests of the full model against the model with the specific random factor removed. Post hoc pairwise comparisons were performed with the function lsmeans (package "lsmeans"), applying a Holm correction for multiple comparisons of interest. Assumptions of the linear models were checked by inspection of residuals' distribution, using the package "DHARMa". Overdispersion in the generalized linear models was checked with the function "check_overdispersion" (library "performance"). We detected overdispersion for the first glm (experiment A) but not for the second (experiment B -dispersion ratio = 0.864, Pearson's Chi-Squared =64.778, p-value =0.794). To account for overdispersion in the first glm, we included an observation level random effect (test results after random effect inclusion: dispersion ratio = 0.195, Pearson's Chi-Squared = 14.656, p-value =1).

| Repeatability
Repeatability of sperm performances in FRF was high at both concentrations and for all sperm traits considered (p < 0.001 for all traits at both concentrations, for details, see Table S1).

| Exp. A-Effect of different concentrations of FRF on male tactics
Sperm curvilinear velocity (VCL) was affected by treatment (F 2,3.9 = 8.21, p = 0.04), but not by the male reproductive tactic (F 1,22.6 = 0.58, p = 0.45), or by the interaction between treatment and tactic (F 2,47.2 = 0.77, p = 0.47). In both tactics sperm velocity increased from 0% (seawater) to 10% and 20%, with the increase between the 10% and the 20% FRF treatments seemingly more pronounced in territorial males ( Figure 3a, Table 1), but with no significant difference between tactics.
In the presence of both concentrations of FRF, sperm linearity (LIN) decreased compared to 0% (seawater), with no difference between 10% and 20% FRF, for both male tactics ( Figure 3b and Table 1), even though the effect of the treatment was found to be marginally non-significant (F 2,2 = 14.16, p = 0.07). There was no effect of male reproductive tactic (F 1,24.6 = 0.09, p = 0.76), or of the interaction between treatment and tactic (F 2,47 = 0.41, p = 0.67).
Interestingly, sperm motility (MOT, i.e., the proportion of motile sperm) was significantly affected by the treatment (χ 2 2 = 39.43, p < 0.001), and by the interaction between treatment and tactic (χ 2 2 = 7.09, p = 0.03). There was no effect of the male reproductive tactic (χ 2 1 = 0.20, p = 0.66). These results indicate how increasing concentrations of FRF mediated an improvement in sperm motility for both male tactics, but with a different trend in the two tactics ( Figure 3d and Table 1). It is specifically in the comparison between 0% and 10% FRF that the effect significantly differs between tactics (interaction treatment*tactic: χ 2 1 = 6.78, p = 0.009). Post hoc pairwise comparisons showed, indeed, that FRF significantly affect sperm motility of territorial males already at a low concentration (10%), whereas sperm motility of sneaker males was not affected by the lower concentration of FRF (10%) ( Table 1). Then, when the concentration of FRF reaches 20% (mimicking close proximity to the eggs), also sneaker sperm motility benefits of FRF similarly to territorial sperm. Sneaker sperm indeed showed significantly higher motility at 20% compared to 10% FRF (Table 1) For none of the sperm traits analysed we evidenced significant differences between the two tactics within each treatment (all Pr(>|t|) > 0.1, for details, see Table S2 of the supplementary materials).

| Exp B.-Effect of male reproductive tactic, female identity and their interaction
From the full-factorial experiment (Figure 2), we found no effect of male reproductive tactic for any of the sperm traits considered F I G U R E 3 Effect of different concentrations of FRF compared with seawater (0 fluid) on different sperm traits of sneaker (solid line) and territorial males (dashed line) (means ± SE): (a) sperm curvilinear velocity VCL (μm/s), (b) sperm linearity LIN, (c) sperm beat-cross frequency BCF, (d) percentage of motile sperm (sperm motility MOT). Asterisks indicate significant differences among treatments levels for both male tactics, as evidenced by pairwise comparisons (Table 1).

Sperm trait Contrast
Sneaker Territorial ( Table 2), as expected from the previous experiment. Sperm velocity, linearity and beat cross frequency were not affected neither by female identity, nor by the interaction between female identity and male reproductive tactic (Table 2). Sperm motility, however, was significantly affected by female identity, and we found a significant effect also of the interaction between female identity and male reproductive tactic (Table 2). This suggests that the fluid of different females differentially influenced the proportion of motile sperm of a specific male reproductive tactic. We, therefore, assessed whether the female condition (body size and volume of FRF produced) or the time of the reproductive season (date of the experiment) were driving the observed effects on sperm motility mediated by the female identity, but none of these covariates was found to be significant (all Pr(>|z|) > 0.3).

t/z-ratio p t/z-ratio p
Finally, the male identity had a significant effect on all these traits, except for sperm beat-cross frequency (

| DISCUSS ION
Our results showed that FRF has positive effects on sperm performance, corroborating a growing number of studies on the topic (see (Gasparini et al., 2020;Zadmajid et al., 2019)). However, only few studies have investigated whether the FRF has a differential effect on alternative reproductive tactics (Alonzo et al., 2016;Lehnert et al., 2017), and none have explored the possible role of In nature, sneakers of the grass goby usually release their sperm at the nest entrance, thus their sperm encounter low concentrations of FRF, or no FRF at all (Figure 1). In getting closer to the female (simulated by a 10% FRF concentration) sneakers can only partially benefit of FRF with an increase in sperm velocity but not in sperm motility. However, territorial males experience higher FRF concentrations, as they lay sperm trails very close to the eggs (in our design simulated with 20% FRF). Territorial males benefit from increased proximity to females in terms of both sperm velocity and motility, with a more pronounced disparity in sperm motility between the alternative tactics when the difference in their distance from the female is greater. In such a common scenario, the direction of female post-mating choice seems, therefore, congruent with pre-mating preferences: the differential effect of FRF might influence the outcome of sperm competition by decreasing the numerical advantage of the sneakers, who release more sperm but further away from the female compared to territorial males (Mazzoldi et al., 2000). Nonetheless, should a sneaker male be able to get close enough to the female to benefit from the same FRF concentration as the territorial male, then both tactics will likely have the same chances at fertilization, or the fertilization game might be even skewed towards sneakers, since, in addition, they release a higher number of sperm (Mazzoldi et al., 2000).
Furthermore, it is already known that sneaker males have the ability to exploit the seminal fluid of competitor territorial males, whose sperm performance is in turn penalized by the presence of sneaker seminal fluid (Locatello et al., 2013). Such evolutionary dynamics of ejaculate exploitation and impairment in sperm competition mediated by male seminal fluid, together with the differential effect of FRF here described, makes the grass goby an ideal study system to investigate the combined selective pressures of both sperm competition and cryptic female choice.
This concentration-based effect of FRF is of particular interest also because the two distinct territorial-sneaker mating tactics in the grass goby are expressed as an ontogenetic gradient (Scaggiante et al., 2004). In this species, younger males can indeed switch from sneaking tactic to territorial one in favourable social conditions, such as nest and ripe females' availability, with the opportunity to spawn and perform parental care (Scaggiante et al., 2004).

TA B L E 2
Estimates and significance levels for the fixed factor "male tactic" and the random factors of linear mixed effect models: Female ID, interaction between Female ID and Male tactic and Male ID for all the sperm traits considered. In bold p-values < 0.05. Females might, therefore, gain indirect genetic benefits not only by having their eggs fertilized by the territorial male but also by a successful sneaker male (the one able to get as close as the territorial to the spawning female). Whether the position a sneaker male is able to obtain during fertilization is associated with its quality, and hence indirect genetic benefits for the female, would require specific follow up experiments.
In natural conditions, therefore, the ideal post-mating strategy for females may be to use the proxy of male spatial positioning instead of discriminating among alternative mating tactics per se.
Grass goby females, overall, appear to be flexible in their pre-and post-mating choices. At the pre-mating level they strongly favour territorial males, most likely because of the direct benefits associated with such preference, but at the post-mating level the direction of the choice is not that fixed. If sneaker males are successful enough to benefit from a high concentration of FRF they are provided with the same chances of fertilization as the preferred territorial male.
This possibly suggests a shift in the relative importance of direct and indirect (genetic) benefits from the pre-to the post-mating events, with the former driving females in their pre-mating choice, and genetic benefits gaining relative more importance at the post-mating level, when the mate quality transcends the mating tactics adopted by males.
Our findings also highlight a significant effect of female identity and an interaction between female identity and male reproductive tactics on sperm motility. This indicates that different females affected differentially, via their FRF, the sperm motility of males of the two alternative reproductive tactics. This, in light of the results of the first experiment, confirms that, in the grass goby, in which sperm of territorial and sneaker males have similar quality (Locatello et al., 2007), the FRF did not consistently favour the sperm of a specific male phenotype. Our observations concur with the evidence of a widespread variability in female preferences across the animal kingdom (Cotton et al., 2006;Rodríguez, 2020;Rosenthal & Ryan, 2022). Preferences can be opposite in different females of the same population, and even in the same female over the course of life, suggesting adaptive flexibility in preferences (Rosenthal & Ryan, 2022), with females often getting more permissive as mates become scarce, risks increase or time is short (Hedrick & Dill, 1993;Pennebaker et al., 1979;Rand et al., 1997).
Variation in mating decisions might often result from conditiondependent differences in female quality, and/or from contextdependent environmental factors, since mate choice is a costly behaviour in many species (Cotton et al., 2006;Griggio & Hoi, 2010;Hingle et al., 2001). Here, we controlled for some female parameters (female size, total volume of FRF produced and the date of the experiment) to understand whether the female condition or the time of the reproductive season were somehow driving the observed effects mediated by the female identity, but none of these covariates was found to be significant. Future studies are needed to shed light on the role of female physiological status, such as stress level and hormonal condition, in the influence exerted by the FRF on female post-mating preference, as well as on the potential proximate mechanisms, on both male and female side, implicated in this interaction.
Emerging evidence suggests that females may evaluate different aspects of mate quality in separate moments of the mating event: male phenotype and overall quality before mating, sperm quality and competitiveness afterwards (Evans & Garcia-Gonzalez, 2016;Firman et al., 2017;Rosengrave et al., 2008). The mechanism mediated by FRF described here provides a potential example of this shift in male quality assessment from pre-to post-mating arenas. This mechanism of differential influence, based on the specific concentration of FRF encountered by competing ejaculates, could indeed translate into a directional cryptic choice under specific circumstances (i.e. favouring only territorial males when they are the sole proximate males to the female), while also depending on the competitiveness of the partners involved (i.e. favouring also those sneakers that are able to get closer and benefit of higher FRF concentration).
Overall, this study adds a new piece to the puzzle of how FRF can mediate post-mating sexual selection in external fertilizers, unravelling the importance of experimentally simulating, as much as possible, the natural spawning conditions to better decipher the multiple factors that play in the complexity of the real world.

ACK N O WLE D G E M ENTS
We are grateful to A. Sambo, C. Breggion, M. Panin and F. Poli for the endless field assistance. Financial support was provided by Italian Ministry of University to MBR (DOR-Unipd 2020-2021). 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://www. webof scien ce.com/api/gatew ay/wos/peer-revie w/10.1111/jeb.14199.

DATA AVA I L A B I L I T Y S TAT E M E N T
The datasets and R scripts supporting this article are available on the