https://orcid.org/0000-0002-0713-3944
Abstract
Precise mechanisms underlying sperm storage and utilization are largely unknown, and data directly linking stored sperm to paternity remain scarce. We used competitive microsatellite PCR to study the effects of female morphology, copula duration and oviposition on the proportion of stored sperm provided by the second of two copulating males (S2) in Scathophaga stercoraria (Diptera: Scathophagidae), the classic model for sperm competition studies. We genotyped all offspring from potentially mixed-paternity clutches to establish the relationship between a second male’s stored sperm (S2) and paternity success (P2). We found consistent skew in sperm storage across the three female spermathecae, with relatively more second-male sperm stored in the singlet spermatheca than in the doublet spermathecae. S2 generally decreased with increasing spermathecal size, consistent with either heightened first-male storage in larger spermathecae, or less efficient sperm displacement in them. Additionally, copula duration and several two-way interactions influenced S2, highlighting the complexity of postcopulatory processes and sperm storage. Importantly, S2 and P2 were strongly correlated. Manipulation of the timing of oviposition strongly influenced observed sperm-storage patterns, with higher S2 when females laid no eggs before being sacrificed than when they oviposited between copulations, an observation consistent with adaptive plasticity in insemination. Our results identified multiple factors influencing sperm storage, nevertheless suggesting that the proportion of stored sperm is strongly linked to paternity (i.e., a fair raffle). Even more detailed data in this vein are needed to evaluate the general importance of sperm competition relative to cryptic female choice in postcopulatory sexual selection.
INTRODUCTION
When females sequentially mate with multiple males, one of the strongest predictors of relative paternity shares between competitors tends to be the species-specific pattern of sperm utilization (i.e., sperm precedence; Simmons 2001, 2014). Yet, irrespective of whether sperm precedence on average favors the first or the last male in any given species, competitive fertilization is usually also mediated by various male and female effects. These include differential copulation duration (Parker and Simmons 1994; Arnqvist and Danielsson 1999; Andrés and Cordero Rivera 2000), sperm number (Martin et al. 1974; Parker and Simmons 2000; Gage and Morrow 2003; Boschetto et al. 2011) or sperm morphology differences between males (Garcia-Gonzalez and Simmons 2007; Lüpold et al. 2012; Bennison et al. 2015), the female’s reproductive tract morphology (Miller and Pitnick 2002; Fedina and Lewis 2004; Lüpold et al. 2016), remating interval (Cochran 1979; Colegrave et al. 1995; Evans and Magurran 2001; Xu and Wang 2010; Dubey et al. 2018), sperm ejection behavior (Pizzari and Birkhead 2000; Lüpold et al. 2013; Manier, Belote, et al. 2013), or the time between mating and oviposition (Ueno and Ito 1992; Barbosa 2009). Unfortunately, assessing the relative importance and evolutionary implications of such factors has been impaired because processes controlling fertilization success are typically inferred from patterns of paternity without exact knowledge of how sperm are transferred, stored and used. Such information would be essential for interpreting causes and consequences of broad patterns in paternity after multiple mating (Lessells and Birkhead 1990; Parker et al. 1990; Simmons and Siva-Jothy 1998; Simmons 2001). As has been also the case historically for precopulatory sexual selection (Clutton-Brock 1988; Andersson 1994), assessing the relative importance of male (including sperm) competition and female (including cryptic) choice, as well as any interactions between the two, is difficult if not impossible in the case of postcopulatory selection without more detailed knowledge of the mechanisms mediating competitive fertilization success (Birkhead and Pizzari 2002; Snook 2005; Lüpold and Pitnick 2018).
Sexual selection via sperm competition is indisputably an important evolutionary force (Parker 1970c). Sperm competition has driven the evolution of many male behavioral, physiological, and morphological traits involved in the avoidance of or engagement in competition for fertilization of a given set of ova (Birkhead and Møller 1998; Simmons 2001; Birkhead et al. 2009). In contrast, the female role in determining fertilization outcomes has received comparatively little attention. This imbalance is noteworthy, because females provide the selective environment in which postcopulatory sexual selection occurs (Parker 1970c; Lloyd 1979; Arnqvist 2014; Firman et al. 2017; Simmons et al. 2020). Thornhill (1983, 1984) coined the term “cryptic female choice” to describe female processes occurring during and/or after copulation that bias paternity or offspring production toward a certain male. Eberhard (1996) described over 20 mechanisms, grouped into five categories, that potentially enable females to exert such cryptic choice: female influences on remating, sperm transfer, sperm storage, sperm utilization at the time of fertilization (i.e., sperm selection), and investment in offspring (Eberhard 1996, 1997; Simmons 2001; Firman et al. 2017; Lüpold and Pitnick 2018). Although female influences during any one of the postcopulatory stages can strongly impact the fertilization success of a particular male, such impacts are poorly documented (Birkhead 1998; Simmons 2001; Firman et al. 2017). This paucity of evidence is in part due to the methodological challenges of empirically examining processes occurring within the female reproductive tract (Bussière et al. 2010; Hall et al. 2010; Manier et al. 2010; Lüpold et al. 2013). Certain techniques that have been used to study these phenomena suffer from practical limitations (cf. Birkhead 2000). For example, assigning sperm in storage by way of phenotypic markers such as sperm length can be ambiguous (Hellriegel and Bernasconi 2000) and often requires bidirectional selection to enhance the difference between sperm phenotypes (Pattarini et al. 2006; Bennison et al. 2015). Molecular methods such as competitive microsatellite PCR (Bussière et al. 2010; Hall et al. 2010) have overcome some of these limitations by permitting direct inference about differential sperm storage patterns relative to paternity outcomes in natural populations. Even better resolution of the spatio-temporal patterns and dynamics of postocpulatory processes has been achieved by the use of red or green fluorescently labeled sperm (e.g., Manier et al. 2010) that permit direct observation and quantification of competing sperm across female sperm-storage organs. Studying competing sperm within the female reproductive tract has greatly advanced our understanding of the mechanisms mediating postcopulatory sexual selection (Lüpold et al. 2012, 2013, 2020; Marie-Orleach et al. 2014; Droge-Young et al. 2016).
The yellow dung fly Scathophaga stercoraria is the classic model system for sexual selection studies that has been the focus of a rich array of approaches to date (including field and laboratory behavioral studies, life history assessments, experimental evolution, etc.; Blanckenhorn 2009; Simmons et al. 2020). Male interactions seem to drive precopulatory sexual selection (Parker 1970a, 1970b; Jann et al. 2000), though females retain some postcopulatory influence on the paternity of their offspring by biasing sperm storage and use (Ward 2000, 2007). As in many Diptera (Puniamoorthy et al. 2010), female yellow dung flies have multiple sperm storage organs (spermathecae) into which males cannot directly transfer sperm (Hosken 1999; Hosken et al. 1999; Simmons et al. 1999; Hosken and Ward 2000; Schäfer et al. 2013). Instead, males ejaculate into the bursa copulatrix (Hosken 1999; Simmons et al. 1999), with the phallosome (endophallus) almost directly abutting the spermathecal duct openings (Hosken et al. 1999). Female yellow dung flies have typically three (and rarely four) spermathecae (one called the singlet on one side of the body, plus a pair collectively called the doublet on the opposite side: Ward 2000; Schäfer et al. 2013), each with its own narrow duct (Hosken et al. 1999; Figure 1). Several lines of evidence suggest a possible role for these organs in sperm storage and choice. Theoretical work has shown that separate storage compartments in principle allow differential storage (i.e., transport to the spermathecae) and use of sperm from different males (Hellriegel and Ward 1998; shown by Manier, Lüpold, et al. 2013, in Drosophila simulans). In addition, direct paternity assessments of S. stercoraria sperm have demonstrated that sperm contents differ amongst spermathecae following double matings in the laboratory (Bussière et al. 2010) as well as across spermathecae of wild-caught females (Demont et al. 2011, 2012). Nevertheless, whether any of these biases reflect adaptive cryptic sperm choice of stored sperm by individual females has been debated and remains unclear (Birkhead 1998, 2000; Simmons 2001; Ward 2007).
Dissected female reproductive tract of a yellow dung fly (A), with an inset close-up showing detached singlet and doublet spermathecae (B).
Data directly linking the proportion of stored sperm to paternity are missing for most taxa including yellow dung flies (but see Martin et al. 1974; Fedina and Lewis 2004; Manier et al. 2010; Manier, Belote, et al. 2013; Droge-Young et al. 2016). Relating sperm storage to sperm use is empirically difficult, especially in vertebrates, because unless performed in vivo as in the transparent flatworm Macrostomum lignano (Marie-Orleach et al. 2014), quantifying sperm in storage usually requires sacrificing the female (Bussière et al. 2010; Hall et al. 2010; Manier et al. 2010; Demont et al. 2011, 2012). Further, one cannot be certain that the fraction of sperm remaining in storage after oviposition is similar to the fraction used for fertilizing eggs (although the use of isogenic lines with repeatable ejaculate, sperm storage and use patterns can surmount these obstacles: Lüpold et al. 2012, 2013, 2020). Here, we attempted to overcome these difficulties by manipulating the timing of oviposition with three experimental treatments: 1) females sacrificed without oviposition to observe the typical S2 (the proportion of stored sperm assigned to the second of two copulating males); 2) females allowed to oviposit between the two matings, informing about sperm storage and use in noncompetitive situations; and 3) females allowed to oviposit after the second mating, permitting comparisons of S2 and P2 (the proportion of paternity assigned to the second of two copulating males) within these females, as well as S2 comparisons with females from the first treatment to detect changes in storage after oviposition occurred. These contrasts provide information about the relationship between sperm storage and paternity patterns, and also permit detecting influences of potential (adaptive) sperm allocation by males in relation to female size or egg number, which has been documented in yellow dung flies (Parker et al. 1999) and several other taxa (reviewed in Wedell et al. 2002; Kelly and Jennions 2011). Moreover, these contrasts can reveal differential patterns of sperm utilization by the females (Ward 2000; Manier, Belote, et al. 2013; Manier, Lüpold, et al. 2013).
We specifically tested the following three hypotheses: 1) Oviposition between the two matings (i.e., use of sperm from only the first male) should increase S2. Alternatively, if males (strategically) allocate sperm according to whether the female has just oviposited or not (Parker et al. 1999), prior oviposition should decrease S2. 2) If sperm are largely used at random during fertilization (i.e., in proportion they are present within the various sperm stores; cf. Parker 1990), S2 should be similar for flies without oviposition and flies that immediately oviposited after the second copulation. 3) Finally, a fair raffle in sperm utilization (i.e., no female sperm selection among stored sperm) would be reflected in a strong correlation between the stored sperm (S2) and paternity shares (P2). As fertilization is known to be mediated by male and female behavior and morphology (introduced above), several such traits were considered as potential influencing factors in addition to our oviposition treatment.
MATERIALS AND METHODS
Animal husbandry and laboratory matings
All experimental flies were first-generation descendants of adults collected in the field from dung pats in Fehraltorf, Switzerland, that were kept in the laboratory at 22 °C, 60% relative humidity and 12 h photoperiod with water, sugar, and Drosophila melanogaster supplied ad libitum as prey. We randomly assigned our experimental females (n = 105) to one of the three treatments (n = 35 females per treatment): 1) two copulations without oviposition; 2) oviposition between mating one and mating two; and 3) oviposition after the second mating. We mated all females with two unrelated males on two consecutive days, so that the mating interval (known to affect sperm-storage patterns in yellow dung flies: Bussière et al. 2010) was always one day across all treatments. We chose mating partners at random without prior screening of microsatellite genotypes. For each assigned pair, we transferred a single male from its housing vial to a clean vial (28 mm diameter × 95 mm long), introduced a single virgin female, and then observed the pair to record the copulation duration. After copulation, we immediately separated females and males. We repeated this procedure on the following day for the females’ second matings. In few cases (14 out of 210 copulations), very long copulations were interrupted after 75 min. For females assigned to one of the two treatments with oviposition, we provided a smear of dung immediately after their focal mating to initiate oviposition. Hence, oviposition either took place on the first day after the first copulation, or it occurred on the second day after the second copulation. All females and males used in this series of double matings were frozen at −80 °C late in the evening of the second day.
Unlike the eggs laid after the first mating, which were sired exclusively by the first male, those laid after the second mating had two potential sires. To determine their paternity, we transferred all egg clutches (typically 30–90 eggs) to 100 mL plastic containers with an excess of previously homogenized and frozen cow dung (>2 g/larva; Amano 1983). We reared all offspring in a climate chamber at constant 22 °C, 60% relative humidity, and 12 h photoperiod. We checked the containers daily for emerged adults until emergence stopped. All emerged flies were immediately frozen at −80 °C for subsequent paternity analyses.
Dissections
We isolated stored ejaculates from the previously frozen females that had been dehydrated in 70% ethanol for 24 h before dissection (Tripet et al. 2001; Bussière et al. 2010; Demont et al. 2011, 2012). We carefully removed the posterior portion of the female reproductive tract (including the common oviduct, spermathecal ducts, spermathecae, accessory glands, and bursa copulatrix; Figure 1A) from the rest of the female by grasping the genital valves with forceps and tearing them from the abdomen. We separated the spermathecae together with their ducts from the rest of the reproductive tissue under a binocular microscope (Leica MZ-12, Leica Microsystems GmbH, Wetzlar, Germany) and photographed them (Figure 1B). We then separated, dehydrated, and thus solidified sperm “pellets” from female spermathecal tissue using very finely sharpened dissecting tweezers viewed under a stereo microscope. We took great care to extract the entire ejaculate, and although some sperm may have been lost, this quantity relative to the extracted sperm mass was likely minor. Each sperm pellet was transferred separately to a buffer solution (ATL buffer from the QIAamp® DNA Micro Kit, Qiagen; see below). The three sperm pellets from each female, each originating from a different spermatheca, were amplified and analyzed separately to study the skew in sperm storage across spermathecae. In our analyses, we distinguished the singlet spermatheca (regardless of the side of the body on which it is found) from the middle and outer doublet spermathecae (Hosken et al. 1999). Distinguishing the middle and outer doublet spermathecae was possible by allowing the organs to float during dissection; the middle doublet spermatheca consistently lay between the other two in a horizontal plane (Figure 1B).
Finally, for each female we measured hind tibia length (before storage in ethanol) as an index of body size, as well as the spermathecal area and duct length (from dehydrated tissue), using ImageJ software (Schneider et al. 2012).
Extraction, amplification, and analysis of DNA
We isolated DNA from all flies using Chelex extraction (Demont et al. 2011, 2012). In brief, we removed the heads of flies and transferred them to 96-well PCR plates on ice, and added 100 µL of a 6% Chelex suspension (Chelex 100®, Na+-form, particle size 50–100 mesh, Fluka) into each well using wide-ended pipette tips. Next, we covered each plate with a plastic mat, carefully shook it, and spun down the heads to ensure that the sample was covered in liquid. Using a thermocycler, we incubated plates for 60 min at 55°C, boiled for 9 min at 100 °C, and then cooled down to 20 °C. After removing samples from the thermocycler, we again shook them carefully, spun them down, stored the plates at 4 °C for 10–20 h, and later froze them at −20 °C for at least 24 h until processing the DNA extractions further.
Since we expected the number of DNA copies extracted from the spermathecae to be very low, we used a kit designed for forensic amounts of DNA (QIAamp® DNA Micro Kit, Qiagen AG, Switzerland) to isolate the DNA of these sperm pellets. We followed the recommended protocols, adding carrier RNA to the buffer AL (1 µL dissolved carrier RNA in 200 µl buffer AL) and the minimum recommended amount of elution buffer AE (20 μL) when extracting DNA from sperm pellets in order to retain the highest possible concentration of DNA. We then used the QIAGEN® Multiplex PCR Kit to simultaneously amplify four microsatellite loci: SsCa17, SsCa24, SsCa26 (Garner et al. 2000), and SsCa30 (Garner et al. 2000; Demont et al. 2008).
The total PCR reaction volume for the heads was 6 µL: 1 µL DNA template, 3 µL QIAGEN Multiplex PCR Master Mix, 1.4 µL distilled water, and 0.6 µL microsatellite primer mix (100 µM). The total PCR reaction volume for the sperm was 30 µL (Demont et al. 2011, 2012), using the same mixing ratios as for the heads (note that the total PCR reaction volume for the sperm was only 24 µL in Bussière et al. 2010). Cycling conditions for the heads were as follows: 95 °C for 15 min, then 27 cycles of 94 °C for 30 s, 60 °C for 3 min and 72 °C for 45 s, and finally 60 °C for 30 min. We used the same cycling conditions for the sperm DNA, but for 30 instead of 27 cycles, to account for the lower initial template concentration (Demont et al. 2011, 2012). These conditions do not usually produce large stutter bands (Bussière et al. 2010; Demont et al. 2011, 2012). We separated PCR products on a capillary sequencer (Applied Biosystems 3730 DNA Analyzer) and analyzed the output using the Applied Biosystems GeneMapper® software. Of the four loci, one (SsCa17) was not sufficiently polymorphic for our procedure described below, so for the remainder of the paper we focus only on the other three loci (cf. Bussière et al. 2010).
Statistical analyses
To investigate how morphological traits were associated across females, we calculated Pearson correlation coefficients (r) between female hind tibia length, spermathecal duct length, and square-root spermatheca area, as well as between the different dimensions of the female reproductive tract.
The response variable in all our statistical analyses was S2 (the proportion of stored sperm assigned to the second of two copulating males), estimated by the relative signal intensity of a second male’s alleles in the amplified sample of DNA extracted from the sperm pellet (Bussière et al. 2010; Hall et al. 2010). Since many factors besides the initial concentrations of alleles (e.g., allele length) may contribute to the observed signal strength of a particular allele after PCR (Haberl and Tautz 1999), we corrected measures of relative peak intensity for each of our microsatellite loci to obtain S2 values. A detailed description of this procedure, along with the exact regression coefficients used in the current study, is given in Bussière et al. (2010).
Although we were able to derive estimates of the proportion of stored sperm belonging to each of the males using the procedure described above, this proportion was not based on discrete counts of sperm; instead it was a simple ratio of sequencing band intensity, and both numerators and denominators for band intensity varied widely across sequencing runs based on the quantity of DNA extracted. Therefore, our response variable represents a proportion for which the actual contributing amount of sperm is unknown. As a consequence, we used logit-transforms to reflect the theoretically binomial error in this value, but fitted our models as straightforward linear models, albeit with a logit-transformed response.
We fitted within-female skew in sperm storage across her spermathecae and the effect of oviposition (i.e., the three treatments) using linear mixed-effects models in the lme4 package (Bates et al. 2015) implemented in R 3.6.2 (R Development Core Team 2019). The most complex model we considered included as explanatory variables treatment (a factor with three levels), spermathecal duct length, square-root spermathecal area, both copulation durations (all continuous covariates), all two-way interactions, and all quadratic terms for the continuous covariates. Continuous predictors were mean-centered and standardized prior to fitting models to facilitate parameter estimation and comparisons of effect sizes (Schielzeth 2010). For this model, we included (intercepts only) random effects for the microsatellite locus as well as spermathecal identity nested within female. We chose to fit locus as a random effect despite the fact that it has only three levels (and therefore could pose challenges for estimation) because we wanted to control for multiple measures within a spermatheca, and we were not interested in assessing fixed effects for this term. The variance accounted for by locus was also substantially non-zero, and models in which locus was included instead as a fixed effect did not differ qualitatively from those shown in the Results. We assessed model adherence to assumptions by visual inspections of diagnostic plots.
For the subsample of data for which we had estimates of P2, that is, the treatment in which oviposition followed the second copulation, we additionally investigated the relationship between P2 and S2 using a linear mixed-effects model. We used the S2 (and not P2) values as the response because flies first laid their eggs and were subsequently frozen to have their sperm extracted. Consequently, S2 estimates in the present study describe what is left in the spermathecae after oviposition (rather than assessing how sperm in storage might affect paternity). The most complex model included as explanatory variables spermathecal identity, logit-transformed P2, spermathecal duct length, square-root spermatheca area, both copula durations, and all possible two-way interactions. We also included female identity as the lone random intercept effect (models including locus did not converge, perhaps because of the lower sample size; consequently, we aggregated data for each spermatheca by taking the (logit-transformed) average S2 across all informative loci as the response variable).
For each of the two above-described models, we conducted a series of simplifications to remove interaction terms that did not significantly improve model fit, using parametric bootstrap comparisons (in the PBmodcomp function in the pbkrtest library: Halekoh and Højsgaard 2014) of models with and without the term in question, but always retaining all main effects to allow comparisons between them. Hypothesis tests for coefficients likewise utilized parametric bootstraps, comparing models with and without the focal term that were otherwise identical. We did not remove main effects from models containing significant interactions involving those effects, so we did not derive hypothesis tests for any main effects involved in interactions for the model in question.
RESULTS
Final sample sizes
We provided each of 105 females two mating opportunities (35 for each mating-oviposition treatment), but multiple reasons (including females escaping, dying, failing to mate or oviposit, having four spermathecae with four ducts, or having three empty spermathecae) reduced our final sample size to n = 76 (treatment without oviposition: n = 28; oviposition between the two matings: n = 21 (10 females did not lay eggs); oviposition after the second copulation: n = 27). Of these 76 females, a subset had missing values for one or more measured variable: 4 females had one empty spermatheca (conservatively assuming sperm transfer did not work), one single spermatheca was lost during dissection, and three spermathecae provided ambiguous results after PCR. We constrained data to records with complete observations for the predictors to permit model contrasts.
Reproductive tract dimensions and correlations among them
Mean (± SE) spermathecal duct lengths were 662.32 ± 12.61 µm (n = 66 females), 675.20 ± 13.28 µm (n = 70), and 680.25 ± 11.52 µm (n = 67) for the singlet spermatheca, the middle doublet and outer doublet spermatheca, respectively. Mean (± SE) spermatheca sizes (i.e., square-root spermathecal areas) were 113.53 ± 1.74 µm (n = 66), 114.61 ± 1.03 µm (n = 71), and 115.45 ± 1.22 µm (n = 68) for the singlet spermatheca, the middle doublet and the outer doublet spermatheca, respectively.
The duct lengths and areas of the three spermathecae were all positively correlated, but the spermathecal areas and ducts varied independently of one another (Supplementary Table S1). While the areas of the doublet spermathecae were significantly associated with female hind tibia length, the area for the singlet spermatheca area was not. Interestingly, spermathecal duct lengths did not covary positively with female hind tibia length.
Does oviposition treatment affect patterns of sperm storage across spermathecae?
Our overall analysis of factors affecting patterns of sperm storage skew revealed several significant predictors (Table 1). A single female had an unusually high length measurement for the duct associated with the middle doublet spermatheca, but the model was not sensitive to its exclusion, so we report findings including this case (observable as the three right-most observations in the middle panel of Figure 2). Spermathecal identity was important. In general, the two doublet spermathecae had less second male sperm in storage (middle doublet deviation B = −0.90 logit units; outer doublet deviation B = −2.14 logit units), though this effect depended on the length of the spermathecal duct: both singlet and outer doublet spermathecae showed higher proportions of second male sperm in storage when the relevant duct was long, while the middle doublet was relatively insensitive to duct length (interaction between spermathecal identity and spermathecal duct length: parametric bootstrap Ppb = 0.038; Figure 2). As expected, the copula durations of both males affected the proportion of sperm in storage, with coefficients in opposite directions: longer second male and shorter first male copulations led to higher proportions of second male sperm (first male copula duration Ppb = 0.007; second male copula duration Ppb = 0.010; Figure 3). There was no evidence that these effects depended on one another, as the interaction term between the copula durations did not significantly improve model fit and thus was removed during simplification (Ppb = 0.158). There were also two marginally nonsignificant main effects: both the oviposition treatment 2 with oviposition after the first mating (Hypothesis 1) and, especially, the treatment 3 with oviposition after both matings tended to feature less second male sperm in storage than the treatment without oviposition (Figure 4A; Ppb = 0.062), indicating nonrandom sperm use during fertilization (Hypothesis 2) as well as female and/or (second) male influences on sperm storage across spermathecae. In addition, larger spermathecae showed slightly lower second-male sperm in storage (Figure 4B; Ppb = 0.056).
Summary of the linear mixed effects model for logit-transformed proportion of second-male sperm in storage (S2) as a function of oviposition treatment (the reference level is no oviposition), spermathecal identity (reference is the singlet), and further morphological and behavioral covariates. The model included the locus that provided a given estimate as well as the spermatheca nested within the female as random effects. Total number of observations = 401 (all data); significant terms (P < 0.05) are highlighted in bold, barely nonsignificant terms (P < 0.1) in bold italic. Because we used model contrasts to assess significance of effects, we cannot assess main effects involved in interactions
| Random effects | ||||
|---|---|---|---|---|
| Variance | Number of groups | |||
| Spermathecal identity nested within Female | 1.414 | 165 | ||
| Female | 7.888 | 59 | ||
| Locus | 0.184 | 3 | ||
| Residual | 3.977 | |||
| Fixed effects | ||||
| Term | Estimate | SE | t | Ppb |
| (Intercept) | 4.098 | 0.745 | 5.503 | |
| Treatment (between) | −0.901 | 1.042 | −0.865 | |
| Treatment (after) | −2.141 | 0.949 | −2.256 | 0.062 |
| Spermatheca (middle doublet) | −0.361 | 0.357 | −1.012 | |
| Spermatheca (outer doublet) | −0.666 | 0.375 | −1.776 | |
| Spermathecal duct length | 0.350 | 0.392 | 0.893 | |
| Spermatheca Area | −0.505 | 0.273 | −1.850 | 0.056 |
| Male 1 copula duration | −1.256 | 0.410 | −3.064 | 0.007 |
| Male 2 copula duration | 1.280 | 0.435 | 2.943 | 0.010 |
| Spermatheca (middle): duct length | −0.269 | 0.355 | −0.756 | |
| Spermatheca (outer): duct length | 0.721 | 0.407 | 1.771 | 0.036 |
| Random effects | ||||
|---|---|---|---|---|
| Variance | Number of groups | |||
| Spermathecal identity nested within Female | 1.414 | 165 | ||
| Female | 7.888 | 59 | ||
| Locus | 0.184 | 3 | ||
| Residual | 3.977 | |||
| Fixed effects | ||||
| Term | Estimate | SE | t | Ppb |
| (Intercept) | 4.098 | 0.745 | 5.503 | |
| Treatment (between) | −0.901 | 1.042 | −0.865 | |
| Treatment (after) | −2.141 | 0.949 | −2.256 | 0.062 |
| Spermatheca (middle doublet) | −0.361 | 0.357 | −1.012 | |
| Spermatheca (outer doublet) | −0.666 | 0.375 | −1.776 | |
| Spermathecal duct length | 0.350 | 0.392 | 0.893 | |
| Spermatheca Area | −0.505 | 0.273 | −1.850 | 0.056 |
| Male 1 copula duration | −1.256 | 0.410 | −3.064 | 0.007 |
| Male 2 copula duration | 1.280 | 0.435 | 2.943 | 0.010 |
| Spermatheca (middle): duct length | −0.269 | 0.355 | −0.756 | |
| Spermatheca (outer): duct length | 0.721 | 0.407 | 1.771 | 0.036 |
Area: square-root transformed area; Ppb: parametric bootstrap P-value.
Summary of the linear mixed effects model for logit-transformed proportion of second-male sperm in storage (S2) as a function of oviposition treatment (the reference level is no oviposition), spermathecal identity (reference is the singlet), and further morphological and behavioral covariates. The model included the locus that provided a given estimate as well as the spermatheca nested within the female as random effects. Total number of observations = 401 (all data); significant terms (P < 0.05) are highlighted in bold, barely nonsignificant terms (P < 0.1) in bold italic. Because we used model contrasts to assess significance of effects, we cannot assess main effects involved in interactions
| Random effects | ||||
|---|---|---|---|---|
| Variance | Number of groups | |||
| Spermathecal identity nested within Female | 1.414 | 165 | ||
| Female | 7.888 | 59 | ||
| Locus | 0.184 | 3 | ||
| Residual | 3.977 | |||
| Fixed effects | ||||
| Term | Estimate | SE | t | Ppb |
| (Intercept) | 4.098 | 0.745 | 5.503 | |
| Treatment (between) | −0.901 | 1.042 | −0.865 | |
| Treatment (after) | −2.141 | 0.949 | −2.256 | 0.062 |
| Spermatheca (middle doublet) | −0.361 | 0.357 | −1.012 | |
| Spermatheca (outer doublet) | −0.666 | 0.375 | −1.776 | |
| Spermathecal duct length | 0.350 | 0.392 | 0.893 | |
| Spermatheca Area | −0.505 | 0.273 | −1.850 | 0.056 |
| Male 1 copula duration | −1.256 | 0.410 | −3.064 | 0.007 |
| Male 2 copula duration | 1.280 | 0.435 | 2.943 | 0.010 |
| Spermatheca (middle): duct length | −0.269 | 0.355 | −0.756 | |
| Spermatheca (outer): duct length | 0.721 | 0.407 | 1.771 | 0.036 |
| Random effects | ||||
|---|---|---|---|---|
| Variance | Number of groups | |||
| Spermathecal identity nested within Female | 1.414 | 165 | ||
| Female | 7.888 | 59 | ||
| Locus | 0.184 | 3 | ||
| Residual | 3.977 | |||
| Fixed effects | ||||
| Term | Estimate | SE | t | Ppb |
| (Intercept) | 4.098 | 0.745 | 5.503 | |
| Treatment (between) | −0.901 | 1.042 | −0.865 | |
| Treatment (after) | −2.141 | 0.949 | −2.256 | 0.062 |
| Spermatheca (middle doublet) | −0.361 | 0.357 | −1.012 | |
| Spermatheca (outer doublet) | −0.666 | 0.375 | −1.776 | |
| Spermathecal duct length | 0.350 | 0.392 | 0.893 | |
| Spermatheca Area | −0.505 | 0.273 | −1.850 | 0.056 |
| Male 1 copula duration | −1.256 | 0.410 | −3.064 | 0.007 |
| Male 2 copula duration | 1.280 | 0.435 | 2.943 | 0.010 |
| Spermatheca (middle): duct length | −0.269 | 0.355 | −0.756 | |
| Spermatheca (outer): duct length | 0.721 | 0.407 | 1.771 | 0.036 |
Area: square-root transformed area; Ppb: parametric bootstrap P-value.
Fitted proportion (solid line) of second-male sperm in storage (S2) in each of the three spermathecae as a function of spermatheca duct length. All other partial effects have been controlled to the mean (for continuous characters) and the “no oviposition” treatment. Each point represents a measure of second male stored sperm from a single locus. Model parameters are provided in Table 1.
Fitted proportion (solid line) of second-male sperm in storage (S2) as a function of the mean-centred and standardized copula durations for both males. Observations have been grouped into approximate 33% quantiles of the first male’s copula duration, and the solid lines illustrate partial effects for a first male with a copulation that is 1 SE smaller than average on the left (“short”), at the “mean” in the centre, and 1 SE larger than average in the right facet (“long”). All other partial effects have been controlled to the mean (for continuous characters), and are visualized for the singlet spermatheca and the “no oviposition” treatment. Model parameters are provided in Table 1.
Fitted proportion of second-male sperm in storage (S2) as a function of the oviposition treatment (red points and 95% confidence interval whiskers, panel A) and the size of the spermathecae (black line, panel B). All other partial effects have been controlled to the mean (for continuous characters), and are visualized for the singlet spermatheca. Both of these effects are marginally nonsignificant according to parametric bootstrapping (see model parameters and hypothesis tests in Table 1).
How paternity covaries with sperm in storage for females that oviposited after both matings
Analysing only the treatment 3 with potentially mixed paternity, there was a large effect of paternity (P2) consistent with the prediction that sperm storage and sperm use should generally covary (Hypothesis 3; Table 2). However, this effect was moderated by spermathecal size, such that the strongest association was visible in large spermathecae: for smaller spermathecae, the second male tended to dominate storage at all but the lowest levels of paternity (interaction Ppb = 0.044 in Table 2; Figure 5). Spermathecal area also statistically interacted with first male copula duration: long first male copulations resulted in less second male sperm in storage, but this effect decreased with spermatheca size (Ppb = 0.016; Figure 6). For the largest spermathecae, storage was uniformly high for the second male regardless of the first male’s copula duration. Spermathecal identity was also important, with the singlet storing the highest proportion of last male sperm (middle doublet deviation B = −0.50 logit units; outer doublet deviation B = −0.75 logit units), but this effect again depended on spermathecal duct length. In the singlet, there was an increase in stored second male sperm when its spermathecal duct was long, whereas the middle and outer doublets seemed insensitive to variation in duct length (Ppb = 0.042; Figure 7). The second male’s copula duration did not significantly affect sperm storage within this treatment (Ppb = 0.650).
Summary of the linear mixed effects model for logit-transformed proportion of second-male sperm in storage (S2) as a function of spermathecal identity (reference is the singlet), second male paternity (P2), and further morphological and behavioral covariates. The analysis was limited to the females with potential mixed paternity that oviposited after two copulations (treatment 3) and included female identity as random effect. Total number of observations = 73; significant terms (P < 0.05) are highlighted in bold. Because we used model contrasts to assess significance of effects, we cannot assess main effects involved in interactions (although P2 and duct length (in bold italics) can be judged significant based on their small SE and large t)
| Random effects | ||||
|---|---|---|---|---|
| Variance | Number of groups | |||
| Female | 0.576 | 27 | ||
| Residual | 2.501 | |||
| Fixed effects | ||||
| Term | Estimate | SE | t | Ppb |
| (Intercept) | 1.657 | 0.369 | 4.487 | |
| Spermatheca (middle doublet) | −0.502 | 0.452 | −1.111 | |
| Spermatheca (outer doublet) | −0.752 | 0.474 | −1.586 | |
| Second male paternity (P2) | 0.844 | 0.098 | 8.582 | |
| Spermathecal duct length | 0.819 | 0.367 | 2.231 | |
| Spermatheca area | −0.320 | 0.273 | −1.171 | |
| Male 1 copula duration | −0.368 | 0.283 | −1.303 | |
| Male 2 copula duration | 0.133 | 0.282 | 0.471 | 0.650 |
| Spermatheca (middle): duct length | −1.149 | 0.472 | −2.432 | 0.042 |
| Spermatheca (outer): duct length | −0.903 | 0.506 | −1.784 | |
| P2: area | 0.299 | 0.131 | 2.277 | 0.044 |
| Male 1 copula duration: area | 0.621 | 0.259 | 2.400 | 0.016 |
| Random effects | ||||
|---|---|---|---|---|
| Variance | Number of groups | |||
| Female | 0.576 | 27 | ||
| Residual | 2.501 | |||
| Fixed effects | ||||
| Term | Estimate | SE | t | Ppb |
| (Intercept) | 1.657 | 0.369 | 4.487 | |
| Spermatheca (middle doublet) | −0.502 | 0.452 | −1.111 | |
| Spermatheca (outer doublet) | −0.752 | 0.474 | −1.586 | |
| Second male paternity (P2) | 0.844 | 0.098 | 8.582 | |
| Spermathecal duct length | 0.819 | 0.367 | 2.231 | |
| Spermatheca area | −0.320 | 0.273 | −1.171 | |
| Male 1 copula duration | −0.368 | 0.283 | −1.303 | |
| Male 2 copula duration | 0.133 | 0.282 | 0.471 | 0.650 |
| Spermatheca (middle): duct length | −1.149 | 0.472 | −2.432 | 0.042 |
| Spermatheca (outer): duct length | −0.903 | 0.506 | −1.784 | |
| P2: area | 0.299 | 0.131 | 2.277 | 0.044 |
| Male 1 copula duration: area | 0.621 | 0.259 | 2.400 | 0.016 |
Area: square-root transformed area; Ppb: parametric bootstrap P-value
Summary of the linear mixed effects model for logit-transformed proportion of second-male sperm in storage (S2) as a function of spermathecal identity (reference is the singlet), second male paternity (P2), and further morphological and behavioral covariates. The analysis was limited to the females with potential mixed paternity that oviposited after two copulations (treatment 3) and included female identity as random effect. Total number of observations = 73; significant terms (P < 0.05) are highlighted in bold. Because we used model contrasts to assess significance of effects, we cannot assess main effects involved in interactions (although P2 and duct length (in bold italics) can be judged significant based on their small SE and large t)
| Random effects | ||||
|---|---|---|---|---|
| Variance | Number of groups | |||
| Female | 0.576 | 27 | ||
| Residual | 2.501 | |||
| Fixed effects | ||||
| Term | Estimate | SE | t | Ppb |
| (Intercept) | 1.657 | 0.369 | 4.487 | |
| Spermatheca (middle doublet) | −0.502 | 0.452 | −1.111 | |
| Spermatheca (outer doublet) | −0.752 | 0.474 | −1.586 | |
| Second male paternity (P2) | 0.844 | 0.098 | 8.582 | |
| Spermathecal duct length | 0.819 | 0.367 | 2.231 | |
| Spermatheca area | −0.320 | 0.273 | −1.171 | |
| Male 1 copula duration | −0.368 | 0.283 | −1.303 | |
| Male 2 copula duration | 0.133 | 0.282 | 0.471 | 0.650 |
| Spermatheca (middle): duct length | −1.149 | 0.472 | −2.432 | 0.042 |
| Spermatheca (outer): duct length | −0.903 | 0.506 | −1.784 | |
| P2: area | 0.299 | 0.131 | 2.277 | 0.044 |
| Male 1 copula duration: area | 0.621 | 0.259 | 2.400 | 0.016 |
| Random effects | ||||
|---|---|---|---|---|
| Variance | Number of groups | |||
| Female | 0.576 | 27 | ||
| Residual | 2.501 | |||
| Fixed effects | ||||
| Term | Estimate | SE | t | Ppb |
| (Intercept) | 1.657 | 0.369 | 4.487 | |
| Spermatheca (middle doublet) | −0.502 | 0.452 | −1.111 | |
| Spermatheca (outer doublet) | −0.752 | 0.474 | −1.586 | |
| Second male paternity (P2) | 0.844 | 0.098 | 8.582 | |
| Spermathecal duct length | 0.819 | 0.367 | 2.231 | |
| Spermatheca area | −0.320 | 0.273 | −1.171 | |
| Male 1 copula duration | −0.368 | 0.283 | −1.303 | |
| Male 2 copula duration | 0.133 | 0.282 | 0.471 | 0.650 |
| Spermatheca (middle): duct length | −1.149 | 0.472 | −2.432 | 0.042 |
| Spermatheca (outer): duct length | −0.903 | 0.506 | −1.784 | |
| P2: area | 0.299 | 0.131 | 2.277 | 0.044 |
| Male 1 copula duration: area | 0.621 | 0.259 | 2.400 | 0.016 |
Area: square-root transformed area; Ppb: parametric bootstrap P-value
Fitted proportion (solid line) of second-male sperm in storage (S2) for females that laid eggs after two matings (treatment 3) as a function of second male paternity (P2) and spermatheca size. Observations have been grouped into approximate 33% quantiles for spermathecal size, and the solid lines illustrate partial effects for a spermatheca that is 1 SE smaller than average on the left, at the mean in the centre, and 1 SE larger than average on the right. All other partial effects have been controlled to the mean (for continuous characters), and are visualized for the singlet spermatheca. Model parameters are provided in Table 2.
Fitted proportion (solid line) of second-male sperm in storage (S2) for females that laid eggs after two matings (treatment 3) as a function of spermatheca size and the first male’s (M1) copula duration. Observations have been grouped into approximate 33% quantiles for spermathecal size, and the solid lines illustrate partial effects for a spermatheca that is one SE smaller than average on the left, at the mean in the centre, and 1 SE larger than average on the right. All other partial effects have been controlled to the mean (for continuous characters), and are visualized for the singlet spermatheca. Model parameters are provided in Table 2.
Fitted proportion (solid line) of second-male sperm in storage (S2) for females that laid eggs after two matings (treatment 3) as a function of spermathecal identity and duct length. All other partial effects have been controlled to the mean (for continuous characters). Model parameters are provided in Table 2.
Discussion
Clarifying the mechanics of sperm transfer and use within the female body is crucial for assessing relative contributions of males and females to differential fertilization success (Eberhard 1996; Simmons 2001; Fedina and Lewis 2004; Luck et al. 2007; Pai and Bernasconi 2008; Bussière et al. 2010; Manier et al. 2010; Lüpold and Pitnick 2018). Owing to methodological challenges in investigating internal sperm transfer, cryptic female choice has historically received less attention than sperm competition, despite recognition of its potentially strong influence (Eberhard 1996 1997; Simmons 2001). However, recent advances in direct visualization (Hosken et al. 1999; Arthur et al. 2008; Sbilordo et al. 2009; Manier et al. 2010) or molecular techniques (Bussière et al. 2010; Hall et al. 2010) have helped illuminate the events mediating sperm transfer and use. Using competitive microsatellite PCR to explore sperm storage in yellow dung flies, we here obtained four main results. First, we found suggestive but nonsignificant support for an effect of oviposition on sperm storage, indicating that male allocation and female sperm usage may both affect storage patterns (Hypotheses 1 and 2). Second, we confirmed a previously reported consistent skew in sperm storage across spermathecae (Bussière et al. 2010), with more second-male sperm stored in the singlet than in either doublet spermatheca. Third, morphological (e.g., spermatheca size, duct length) and behavioral (e.g., copula duration) covariates significantly influenced sperm-storage patterns (i.e., S2), demonstrating a complex interplay between female and male influences on sperm storage. Finally, the strong association between S2 and P2 among females that laid mixed-paternity clutches implies that paternity is largely assigned in proportion to the relative numbers of sperm in storage (Hypothesis 3), although our data do not exclude subtle female influences during sperm utilization at fertilization (i.e., sperm selection). Below we discuss all these results and their implications in turn.
Sperm storage across all oviposition treatments and associated postcopulatory processes
Although the effect was marginally nonsignificant, there was a trend for the timing of oviposition relative to mating to influence sperm-storage patterns in the yellow dung fly, highlighting potential effects of female sperm transport (during oviposition) as well as possible male sperm allocation adjustments in response to the oviposition status of the female (Wedell et al. 2002). For flies that did not lay eggs, the mean proportion of last-male sperm in storage S2 was highest, whereas S2 was nonsignificantly depressed in females that oviposited, especially after her second copulation (Figure 4a). It is possible that disproportionally more sperm from the second male were used to fertilize the clutch of eggs, or that oviposition itself interrupted sperm transfer and displacement, for example, by descending eggs pushing sperm out of the bursa copulatrix before they could enter the spermathecae. These alternatives predict different paternity outcomes: in the former case, second-male paternity should be higher than mean S2 (i.e., 59.8%), whereas it need not be different in the latter case. Mean (± SE) second male paternity (P2) was 58.7 ± 7.3%, clearly favoring the latter explanation.
Interestingly, flies that did not oviposit (treatment 1) also showed higher S2 values than those ovipositing after their first copulation (treatment 2), thus using the first male’s sperm. If the only difference between these treatments were the number of sperm used to fertilize the first clutch, the opposite pattern (i.e., higher S2 in treatment 2 with oviposition) would be expected. However, male dung flies are known to adjust sperm investment in copula in response to female characteristics, including the number of eggs that females carry (Parker et al. 1999). Females that carry no mature clutch are a much less valuable resource to males because such females almost invariably will mate again immediately before oviposition. The associated first-male sperm displacement dramatically reduces the net fitness benefits, thus explaining the lower investment in such a copulation. In accordance with this expectation and observations of previous studies (Parker 1970c, 1970d; Parker et al. 1999), we found second-male copulations to be shorter in the treatment 2 featuring oviposition between matings than those in other treatments (linear model with second male copulation duration as response: F2,71 = 4.027, P = 0.02). The trend toward a difference in sperm-storage pattern can therefore potentially be accounted for by this shift in male sperm allocation (Parker 1970c, 1970d; Parker et al. 1999; Wedell et al. 2002). In any case, it is possible that any difference in storage arising as a result of first-male sperm depletion is likely to be swamped by male plasticity in copula duration (Parker 1970c, 1970d; Parker and Simmons 1994).
Our study documents a consistent skew in sperm storage across spermathecae, with more last-male sperm stored in the singlet than in either doublet spermatheca (Figure 1). This pattern was present regardless of treatment, and matches the pattern identified previously by experimentally manipulating mating interval to study its influence on sperm storage (Bussière et al. 2010). A consistently higher level of sperm displacement associated with the singlet spermatheca could explain why field-caught yellow dung fly females store sperm from fewer males in their singlet than in either doublet spermatheca (Demont et al. 2011). Whether this pattern derives from male or female effects cannot be disentangled here, but we think that the requirement for active female musculature for effective sperm transfer makes female influence more likely, as males do not penetrate the spermathecal ducts with their intromittent organs. An active female role in sperm migration into the sperm-storage structures has been implicated in other insects (reviewed by Pascini and Martins 2017) and in birds (Mendonca et al. 2019). Yet, we cannot exclude the possibility of a male effect, for example, by sperm of different males varying in their likelihood of entering one or the other spermatheca. For example, when Lüpold et al. (2012) mated male D. melanogaster of different isogenic lines to females of a consistent genetic background, they documented repeatable genotype-specific patterns in sperm segregation between different female sperm-storage structures (in this case the seminal receptacle and paired spermathecae), thus suggesting that these patterns must be mediated by males directly, or by triggering differential female sperm uptake into these structures.
Another explanation for unequal sperm proportions among spermathecae could be sequential filling of the spermathecae in a consistent order (e.g., by always filling the singlet spermatheca first). However, this seems unlikely because such a scenario would predict an interaction between the second male’s copula duration and the degree to which sperm proportions vary across organs. Both the present data and an earlier study (Bussière et al. 2010) do not support this. Although we cannot determine from our results whether the skew across spermathecae is adaptive, such a pattern is a prerequisite for adaptive sperm selection (Hellriegel and Ward 1998; Hellriegel and Bernasconi 2000; Ward 2000; Bussière et al. 2010). This potential has been demonstrated in Drosophila simulans females, which also show distinct S2 patterns between their seminal receptacle and spermathecae, and which can exploit this to preferentially use sperm from one or the other sperm-storage structure to maximize the fertilization success of their favored male (Manier, Lüpold, et al. 2013).
Copulation durations of both males significantly influenced the fraction of second-male sperm stored also in our experiment. Since the link between copula duration and the number of sperm transferred or paternity success has been documented repeatedly in the yellow dung fly (Parker and Simmons 1991, 1994, 2000; Simmons et al. 1999; Bussière et al. 2010) as well as across diverse taxa (reviewed in Weggelaar et al. 2019), we here forego a detailed discussion of this finding.
In addition to these behavioral covariates, however, female reproductive tract dimensions also influenced S2, which nonsignificantly decreased with increasing spermatheca size, suggesting that sperm displacement is weaker in larger spermathecae. Spermathecal volume correlates with female body size (this study Supplementary Table S1; Parker et al. 1999), and males appear to compensate for the lower sperm displacement rate in large females by copulating longer with them (Parker et al. 1999). S2 was also influenced by spermathecal duct length, but this effect varied across the three spermathecae. In the singlet and outer doublet, S2 increased with duct length, suggesting that long singlet ducts facilitate sperm displacement (e.g., through increased muscular force and thus higher sperm transfer to the spermatheca). By contrast, this pattern was absent in the middle doublet spermatheca. A compelling explanation for this difference across spermathecae remains elusive.
Relationship between S2 and P2 and associated postcopulatory processes
Only females that oviposited after the second copulation produced clutches with potentially mixed paternity, so our investigation of how patterns of sperm storage (S2) and paternity (P2) covary is therefore restricted to this treatment. Overall, S2 and P2 values were strongly correlated for any spermatheca, but the strongest correlation occurred within the singlet spermatheca (r = 0.902). These findings are consistent with the singlet being preferred for storage and fertilization, as has been reported also for the fly Dryomyza anilis (Otronen 1997), but which is not consistently found in yellow dung flies (Ward 1993, 1998; Otronen et al. 1997; Hellriegel and Bernasconi 2000). These correlations exceed values reported by Manier et al. (2010) for the relationship between P2 and S2 in D. melanogaster: in their study, R2 values for the relationship between S2 and P2 were 0.259 and 0.004 for the seminal receptacle and spermathecae, respectively. This suggests that once sperm are stored in yellow dung fly spermathecae, they are largely used according to their numerical representation (i.e., a “fair raffle”: Parker 1990). This does not necessarily imply that females exert no control over sperm use for fertilization, as some variance in the relationship between S2 and P2 remains unexplained (see also Simmons and Siva-Jothy 1998), although some fraction of this variance might be attributable to the fact that it was methodologically impossible to sample spermathecal contents before females completed egg laying. Yet, even if female influences were to be minimal during the fertilization process, differential sperm uptake and storage could easily suffice to bias paternity. For example, the mere fact that females possess more than two spermathecae, the development and maintenance of which likely incur substantial costs (e.g., lines selected to express more spermathecae have decreased fecundity: Ward et al. 2008; Schäfer et al. 2013), suggests that these investments may be adaptively advantageous. Combined with the observed biases in sperm storage across spermathecae (Bussière et al. 2010; Demont et al. 2011, 2012; this study) and varying paternity biases across different female conditions (Ward 2000), at least some female contribution to reproductive outcomes seems likely.
The significant interaction between P2 and spermatheca size indicates that morphology can additionally influence postcopulatory processes. S2 and spermatheca size were inversely related, indicating that spermathecal volume probably plays a role in sperm displacement. Decreased displacement rates for large females have been reported previously, and males appear to compensate for this by copulating longer with larger females (Parker et al. 1999). Nevertheless, Parker et al. (1999) were unable to detect an influence of spermathecal volume on copula duration in their study. Our additional finding that the effect of spermatheca size depends on the duration of the first male’s copulation (but not the second) is intriguing: long copulations by the first male may permit better “sperm defense” when spermathecae are small by physically filling them, making it hard for the second male to quickly and efficiently supplant prior ejaculates.
CONCLUSION
With the help of molecular tools, our work has clarified several exciting new aspects relating sperm storage to sperm utilization in the classic model system for studies of sperm competition, the yellow dung fly. Yet, the complexities of the mechanics involved in sperm storage and use preclude simple explanations. The outcome of competition for fertilization rather seems to be determined by subtle and complex interactions between a female, her multiple mates, the circumstances surrounding the copulations, the intermating interval, the time to oviposition, and prevailing environmental conditions at oviposition. While we continue to steadily advance our knowledge concerning mechanisms mediating postcopulatory sexual selection, convincing evidence for some of these mechanisms (e.g., sperm selection by females) remains elusive. We are convinced that combining direct anatomical observation of processes occurring within females (Manier et al. 2010) with modern molecular methods for quantifying sperm in storage (Bussière et al. 2010; Hall et al. 2010) and direct estimates of paternity success will continue to illuminate mechanisms underlying nonrandom paternity in this and many other species.
SUPPLEMENTARY MATERIAL
Supplementary data are available at Behavioral Ecology online.
Supplementary Table S1 Pearson’s product moment correlations between various morphological measurements in female yellow dung flies. N = 62–71 (some structures were damaged during dissection, leading to slight differences in sample size depending on the pair of traits). Traits showing significant correlations are in bold; there has been no adjustment for multiple tests.
This work is part of the PhD of Marco Demont, who has since left science to lead a private life. We thank Claudia Buser for help with mating trials and dissections.
Funding
This work was funded by the Zoological Museum, University of Zurich, and by a grant from the Claraz-Stiftung to M.D., O.M. (PZ00P3_121777/1, PZ00P3_137514/1, and 31003A_125144/1), W.B. (3100A0-111775), and S.L. (PP00P3_170669) were funded by the Swiss National Science Foundation. L.B. was supported by the University of Stirling.
Author Contributions
Together with P.I.W. and W.U.B., L.F.B. acted as M.D.’s main PhD advisor at the University of Zurich. All these advisors strongly contributed to the conception of the work (performed mainly by M.D.), data analysis and original manuscript preparation. L.F.B. reanalyzed the data with modern statistical approaches, with the help of S.L. and O.Y.M. The latter two contributed substantially to the reincarnation of this work in terms of intellectual concepts, statistical analyses, literature contributions and writing. We are particularly grateful to P.I.W. for his significant initial support, who unfortunately died too early during the course of this work. We hope that the presentation of the data fully reflects his scientific view.
Data availability:
Analyses reported in this article can be reproduced using the data and code provided by Demont et al (2021).
