|Home | About | Journals | Submit | Contact Us | Français|
Post-copulatory sexual selection can select for sperm allocation strategies in males [1, 2] but males should also strategically allocate non-sperm components of the ejaculate [3, 4] such as seminal fluid proteins (Sfps). Sfps can influence the extent of post-copulatory sexual selection [5–7] but little is known of the causes or consequences of quantitative variation in Sfp production and transfer. Using Drosophila melanogaster, we demonstrate that Sfps are strategically allocated to females in response to the potential level of sperm competition. We also show that males who can produce and transfer larger quantities of specific Sfps have a significant competitive advantage. When males were exposed to a competitor male, matings were longer and more of two key Sfps, sex peptide  and ovulin , were transferred, indicating strategic allocation of Sfps. Males selected for large accessory glands (AGs, a major site of Sfp synthesis) produced and transferred significantly more sex peptide, but not more ovulin. Large AG males also had significantly increased competitive reproductive success. Our results show that quantitative variation in specific Sfps is likely to play an important role in post-copulatory sexual selection and that investment in Sfp production is essential for male fitness in a competitive environment.
In insects, Sfps produced in the male AGs significantly increase male fitness: for example, by promoting sperm storage, temporarily increasing female egg-laying rate and decreasing female sexual receptivity , and thus increasing progeny production and delaying sperm displacement/competition . However, Sfp production is limited (Sirot et al. in revision & [12–14]), hence males should allocate Sfps prudently. We therefore predicted that major factors likely to influence Sfp transfer would be the quantity of Sfps that a male has available and a male’s ability to adjust Sfp transfer in response to the potential level of post-copulatory competition.
To experimentally manipulate the Sfp investment potential of males we artificially selected for large (L) or small (S) AGs in replicate pairs of populations. Unselected (U), but otherwise identically cultured populations were also maintained (giving 6 lines in total: L1, L2, U1, U2, S1, S2; see Experimental Procedures). Artificial selection produced a consistent and significant divergence in male AG size. The AG size of L males was significantly larger than that of either S or U males, but we detected no correlated responses in terms of body size or testis size (Figures 1 & S1). Unexpectedly, selection for reduced AG size was ineffective, resulting in no decrease in AG size in S lines (Figure 1) and no significant difference in AG size between S and U lines (Figure S1A). Thus, whilst it was possible to select for significantly increased AG size, there was a minimum stable AG size.
To quantify Sfp production and transfer in the selected lines, we used enzyme-linked immunosorbent assays (ELISAs; see Experimental Procedures & Supplemental Data). We measured two Sfps predicted to influence male competitive reproductive success: sex peptide (SP; aka Acp70Aa) and ovulin (aka Acp26Aa). Both SP and ovulin increase female egg-production, though by different mechanisms [8, 15–17], and SP additionally causes dramatically decreased receptivity [18, 19]. As expected, changes in AG size altered the quantity of SP produced, with L male AGs containing significantly more SP than either U or S male AGs, with no significant difference between S and U male AGs (Figure 2A). However, whilst the trend was in the same direction for ovulin production, there was no statistically significant difference between the lines (Figure 2B). Thus, it cannot be assumed that responses to increased AG size are consistent between specific Sfps, in terms of the quantity of protein produced. A potential mechanism underlying the differences between SP and ovulin responses to selection is that these Sfps differ in their sites of synthesis within the AGs. The AG is comprised of about 1000 main cells and 40 secondary cells per gland . SP is synthesized only in the main cells , whereas ovulin is synthesized in both the main and secondary cells . Our selection for overall AG size could have disproportionately affected proteins produced only in the main cells, such as SP. It will be interesting in the future to compare how different selection pressures affect Sfp production. For example, whether Sfps evolve differently when under selection generated by different levels of male-male competition and how any changes compare to those occurring under selection for AG size (as tested here).
Because we detected no difference between U and S males in AG size or Sfp production, we focused on comparing the S and L lines for measures of Sfp transfer. We found a striking difference in the amount of SP transferred by L vs. S males: L males transferred significantly more SP to females than did S males (Figure 3A). There was, however, no significant difference in ovulin transfer among the different lines (Figure 3B). Moreover, the trend for ovulin was in the opposite direction from that for SP (Figures 3A & 3B). The pattern of Sfp transfer associated with differences in AG size was therefore similar to – though more pronounced than – that of Sfp production (Figure 2), and was consistent across matings in which a competitor male was present as well as those in which no competitor male was present (Figure 3A & B; and see ‘Competition affects mating duration and Sfp transfer’ below).
Thus, despite having larger AGs, L males did not transfer a uniformly larger ejaculate: they transferred increased amounts of SP but not of ovulin, resulting in changes to the SP:ovulin ratio received by females. Selection on AG size did not significantly alter mating duration, (Figure 3C; however, mating duration was influenced by the presence of rival males, see ‘Competition affects mating duration and Sfp transfer’, below). Therefore, differences between S and L males in the amount of SP transferred during mating were not explained by divergence in mating duration and were instead likely due to differences in SP production or allocation.
To test whether increased transfer of SP benefitted L males, we measured the competitive reproductive success of selection line males in two ways. Firstly, we tested the L, U and S males in a two-mating sperm displacement ability (SDA) assay . Females were mated first to a competitor male, and then to a selection line male 48hrs later. Females, and the competitor males, were homozygous for the recessive sparklingpoliert (spapol) eye phenotype. Consequently, the paternity of the offspring produced could be assessed visually (see Experimental Procedures). We found no significant differences in SDA between L, U and S males (F2,3 = 1.28, p = 0.40). Thus, under these conditions, there was no evidence that L line males benefitted from increased SP transfer by increasing their ability to displace a prior male’s sperm (although the AG and testis size did account for some variance in SDA: see Supplemental Data).
Secondly, we conducted a multiple-mating competition assay over a 10-day period. Selection line males were housed with spapol females and spapol competitor males. We measured the reproductive success (number of progeny sired) of the males and sampled their mating frequency. We found no overall differences between L, U and S males in the number of matings obtained by selection line males relative to competitor males (χ2 = 4.2, df = 5, p = 0.52; data not shown), showing that there were no detectable differences between treatments in pre-mating competitive ability. However, selection regime had a significant effect on the number of progeny sired. L males sired significantly more offspring than did either U or S males, and there was no significant difference between U and S males in progeny production (Figure 4). Thus, the reproductive success of males paralleled the pattern of differences in AG size, SP production and SP transfer to females (L > U = S). This suggests that males gained significant fitness benefits from the ability to transfer larger quantities of specific Sfps, such as SP. The mechanism for this increased reproductive success is likely to be via the increased transfer of SP and/or other Sfps that elevate egg production, delay the onset of sperm competition and/or function in sperm defence, in the L males. Whilst SP has no direct effect on sperm displacement ability, it benefits males primarily through its ability to decrease female receptivity [18, 19]. This results in a higher ‘per mating’ share of reproduction for males in multiple-mating situations because the inter-mating interval is increased . It is also possible that other Sfps that play roles in sperm defence [23–25] are produced and transferred in higher quantities by the L males. Candidates include Acp36DE which is essential for sperm storage  and consequently for sperm competitive ability , or CG9997, 1652/56, 17575, and Acp29AB which affect sperm retention in storage [28, 29]. The only current antibodies available for these Sfps cross-react with other proteins, thus making their ELISAs difficult to interpret ([30, 31]; L. Sirot, K. Ravi Ram & M. F. Wolfner, unpublished data). However, once highly-specific antibodies are obtained, it will be important to look at quantitative variation in a range of Sfps to explore their effect on post-copulatory sexual selection.
To test whether males can plastically allocate Sfps, we measured the Sfp transfer by AG selection line males that had been exposed to different potential levels of sperm competition. Male social environments were experimentally varied in the 24hrs prior to and including mating. Males experiencing ‘competition’ were housed in pairs, whilst males experiencing ‘no competition’ were housed alone (see Experimental Procedures). This experimental manipulation could vary both sperm competition ‘risk’ and ‘intensity’ [32, 33] and hence we refer to the ‘level’ of sperm competition only.
The presence of a rival competitor male had a strong and consistent effect on both Sfp transfer and mating duration. Significantly more SP and ovulin were transferred to females, and matings were significantly longer, when a competitor male was present prior to and during mating (Figure 3). These results support the hypothesis that males tailor the quantity of Sfps transferred in relation to the potential level of competition. Recent theory has addressed how Sfp allocation could be affected by various factors including sperm precedence (i.e. 1st or 2nd male), the relative influence of sperm versus Sfps on fertilization success , and the exploitation of the Sfps of rival males . Moreover, models of sperm allocation (e.g. [32, 33]) could be applicable to Sfps, wherever increased Sfp quantities directly influence the outcome of sperm competition in the same way as sperm numbers. However, D. melanogaster males benefit from SP transfer, and potentially ovulin transfer, primarily by increasing their paternity prior to female remating and hence before sperm competition occurs . The consequences of this specific effect for ejaculate allocation have not yet been explored directly by formal theory. Nevertheless, our results are consistent with the hypothesis that it is advantageous for a male to increase the transfer of receptivity-inhibiting and short-term fecundity-enhancing Sfps when his mate is likely to encounter subsequent mating attempts by competitor males. Recent findings of Bretman et al.  support this idea: female post-mating responses (increased egg production and decreased receptivity) were significantly stronger when their mates had been housed with competitors prior to mating. Our results suggest that differences in Sfp transfer are likely to be the underlying mechanism, because these post-mating responses are stimulated by Sfps  such as SP [18, 19] and ovulin . Bretman et al.  also found that the increases in female post-mating responses result entirely from prior exposure of males to competitors and not to the presence of competitors at the time of mating. Hence, males who are most successful in pre-mating competition do not induce greater post-mating responses, indicating that such males do not transfer increased levels of Sfps. Our results are therefore consistent with the strategic allocation of Sfps by males, and not higher Sfp transfer by the most successful pre-mating competitors.
It will be important to determine whether, in addition to Sfps, male D. melanogaster strategically allocate sperm. SP is known to bind to sperm in the mated female  but it is not currently known at what stage this occurs (i.e. pre- or post-insemination) and hence whether sperm numbers and SP transfer efficiency are linked. Ovulin, and at least some of the SP, is transferred free from sperm, and short-term SP responses are shown by females mated to males that lack sperm [35, 36]. Thus, Sfp transfer efficiency is unlikely to be obligately linked to sperm number. However, determining the form of any association between Sfp and sperm quantities will be crucial to test ejaculate composition and allocation theory [1, 2].
Although there are clear male reproductive benefits associated with the ability to transfer large quantities of some Sfps, AG size was close to a minimum level in our starting lab population: selection for smaller AGs was unsuccessful. AG size might be subject to truncation selection if a minimal investment is required to avoid too much depletion of Sfps from small AGs over successive matings. Sfp depletion leads to dramatically decreased male fertility and paternity assurance [12, 14], thus there is likely to be strong selection on males to avoid depletion. However, AGs could be costly to develop, maintain or fill with Sfps, in which case AG size could trade off against other life history traits. So far there is no evidence of any such trade-offs in terms of development time or virgin male survival in our AG selection lines (C. Fricke & T. Chapman, unpublished data). However, trade-offs could have been minimised in our selection lines by rearing conditions that reduced competition for resources, with low densities of flies and excess food. This may have permitted the evolution of larger AGs in the L lines without the costs that would usually inhibit such investment under natural or standard lab-cage conditions [37, 38].
The receipt of Sfps such as SP can be costly to females [39, 40] and can potentially mediate sexually antagonistic coevolution . In certain experimental evolution studies, rapid changes in female resistance to male-induced harm have been observed (e.g. [42, 43]). In these studies male-male interaction was eliminated or reduced, through enforced monogamy  or female biased sex-ratios  respectively. Our results suggest that under such conditions males would plastically (i.e. immediately) reduce the level of Sfp transfer, which would reduce mating costs to females. Selection on female resistance would therefore immediately be relaxed even before evolutionary changes in males occurred. Thus, plastic Sfp allocation could potentially select for rapid intersexual coevolution.
Our results show that, in D. melanogaster, Sfp allocation is plastic, can evolve rapidly under selection and is more complex than has hitherto been considered. It will be important to determine whether Sfp allocation is as taxonomically widespread as sperm allocation . Testing this should be possible (Sirot et al. in revision), because the functions of specific Sfps are known in species ranging from arthropods to mammals , antibodies have been developed to Sfps in several species (e.g, fruit flies [9, 30, 31]; carp ; bulls ; humans ), and bioinformatic, proteomic and RNA-interference tools to aid the discovery and characterisation of new Sfps are becoming increasingly available [48–50]. More theory is also needed to generate predictions of how males should invest in, and allocate, the Sfps that play important roles in post-copulatory sexual selection [3, 4], and crucially how females should evolve in response to Sfp allocation. A potential application of our work is the manipulation of males used in biological and genetic insect-pest management. Males released for pest control are often poor in both acquiring mates and inducing the post-mating behavioural changes that are stimulated by Sfps [10, 51]. Our results demonstrate the potential for increasing the reproductive competitiveness of mass-reared males by selecting on AG size or by selecting on the ability of males to induce female post-mating responses.
The Dahomey wild-type stock used in these experiments is as previously described . Competitor males were from a Dahomey wild-type stock into which we had introduced the recessive spapol eye mutation . For all experiments, fly food was supplemented with live yeast granules.
To initiate the selection scheme, 50 Dahomey males for each replicate (2 replicates each for L, S and U lines) were each housed with two virgin females in ASG medium vials (1% agar, 8.5% sucrose, 2% yeast, 6% maize meal and 2.5% Nipagin). After 1 day, males were removed and housed individually for 3–7 days to allow Sfp replenishment. Females were discarded after laying eggs for several days. To generate the L and S lines, male AGs were dissected and AG perimeters measured . Scoring was done blind with respect to line identity and repeated scoring on the same samples gave 96% repeatability. Virgin progeny were collected from 8 families per replicate of males with the largest or smallest AGs for L and S lines respectively to propagate the subsequent generation. 8 vials per U line were chosen at random. For subsequent generations, virgin males and females from the 8 highest and lowest ranked families (for the L and S lines respectively), and the randomly chosen U families, were housed in single-sexed family groups of ~12 and aged for 3–10 days. Virgin females from families ranked 1, 3, 5 and 7 were mated to virgin males from families 2, 4, 6, 8 and vice versa, to ensure that there were no matings between full sibs. Individual males were housed with two females, allowed to mate for one day and then maintained alone to replenish AGs, as above. 25 males per line were dissected each generation (L and S lines) and scored and selected as above (no selection was imposed in generation 20, 21, 23, 26–29 or 31–37). After 40 generations, lines were kept under relaxed-density, unselected conditions at 18°C in bottle culture.
The direct response of AG size to high and low selection was measured as part of the selection process as described above (25 males per line per generation). To test for correlated responses, and changes relative to the U lines, we measured AG size, body size (using wing area as a proxy) and testis size of males that were the offspring of females from generations 16 and 38. Measures of wing or testis size were the averages of the size of the left and right of these organs for a given individual, wherever possible. To measure the quantity of SP and ovulin produced by the males we dissected the AGs and performed ELISAs, as described below.
To test the quantity of Sfps transferred to females during mating, we raised males from the L and S lines, and Dahomey females, at a standardised density of 100 larvae per sugar-yeast (SY) medium vial . Virgin Dahomey females and selection line males were collected on ice and stored 5 per vial in single sex groups. Three days later, males of each line were placed either 1 (‘no competition’ treatment) or 2 (‘competition’ treatment) per vial, using ice anaesthesia, 24 hrs before matings took place. On the day of the matings, 1 female was introduced into each of the male vials and the duration of mating was recorded. 25 minutes after the start of mating, females were aspirated into microcentrifuge tubes, flash-frozen in liquid nitrogen and then stored at −80°C for subsequent ELISAs (see below). Matings that lasted for less than 10 minutes and more than 25 minutes were removed from the dataset. Experiments were conducted in 6 blocks between generations 47 and 55 generations after the establishment of the selection lines (selection was relaxed after generation 40, but the differences in AG size between L and S males were still present: e.g. generation 47, F1,2 = 29.8, p = 0.032; generation 49, F1,2 = 57.6, p = 0.017). Final sample sizes for mating duration analyses were n = 78–88/replicate line/treatment. For Sfp transfer analysis sample sizes were n = 51–61 and n = 29–66/replicate line/treatment, for ovulin and SP respectively.
For both reproductive success assays we competed selection line males (offspring of generation 16 flies) against rival spapol males. SY fly medium was used throughout. Firstly, we tested sperm displacement ability (SDA). Single spapol females were placed with single spapol males, and the males were removed after a single, observed mating (Day 1). On Day 2, single 2-day-old selection line males were introduced to each female and observed until mating occurred. Males were removed after mating. Females were transferred to new food vials on days 3, 4, 6, 8, 10, 13, 15 and 17. The number of progeny produced from eggs laid between day 3 and 17 was counted and scored for eye phenotype. Sperm displacement ability was calculated as a/(b+1) where a is the number of progeny sired by the second male to mate and b is the number of progeny sired by the first male .
Secondly, we tested male reproductive success in a multiple-mating competition assay lasting 10 days. 1 day-old selection line males were housed in groups of 10 in SY vials. Experimental vials were then set up, each containing 2 selection line males, 2 virgin spapol males and 2 spapol virgin females (sample sizes were L1 = 23 L2 = 22, U1 = 22, U2 = 28, S1 = 30, S2 = 27). These groups of flies were maintained for 10 days, and transferred onto fresh SY food on days 1, 4 and 8. Vials were observed for matings, at 20 min. intervals over a 3-hr period on days 2, 4, 7 and 9, and the eye phenotype (wild-type or spapol) of the mating male was recorded. Offspring produced during the 10 day period were counted and paternity was assessed by recording eye colour phenotype. The number of matings obtained and progeny sired in competition provide measures of pre- and post-mating competitive ability.
ELISAs were conducted are described in more detail in the Supplemental Data; see also Sirot et al. in revision. Briefly, flies were dissected and ground in Dulbecco’s Phosphate Buffering Solution with protease inhibitors. The protein samples from male AG or female lower reproductive tracts were then aliquoted into wells on four replicate ELISA plates, two for ovulin and two for SP. The plates were incubated overnight at 4°C with shaking. On the following day, the plates were incubated with block, then primary antibody against ovulin or against SP, and then horseradish-peroxidase (HRP)-conjugated secondary antibody for one hour each at room temperature with shaking. The level of ovulin or SP was detected through a reaction of the HRP with 3,3′,5,5′-tetramethylbenzidine substrate (KPL, Gaithersburg, Md.). The reaction was stopped by the addition of 100 ul 1 M H3PO4 after the wells developed a deep blue colour or after 30 min. Optical density at 450 nm (OD450) was determined using a Molecular Devices kinetic microplate reader. The OD450 value of a blank well was subtracted from the OD450 values o all of the other samples on its plate. Resulting OD450 values for the samples on one plate were regressed against OD450 values of the same samples from the replicate plate. Points with residuals greater than three standard deviations were considered to have low repeatability and were removed. The OD450 values of the two replicate plates were averaged and converted to male AG equivalents using a standard curve of generated from male AGs of Canton S males.
Data analyses were performed using Excel, JMP (ver. 5, SAS Institute Inc.) and R (v2.8.0) software in Mac OS X. The realised heritability for each AG selection line was calculated as the regression of cumulative response on cumulative selection differential measured over the first 13 generations. Analyses of AG sizes, Sfp production, Sfp transfer, mating duration, SDA and reproductive success (i.e. the number of progeny sired) were performed using linear mixed effects models (nlme package in R). For AG sizes, Sfp production, SDA and reproductive success the fixed effects were selection regime (L, U, S), and the random effect was replicate within regime. For Sfp transfer and mating duration, competition was an additional fixed effect, and line within regime was nested within block for random effects. For the additional SDA analysis (see Supplemental data) the fixed effects were AG size, testes size, wing area and selection regime. Extreme outliers were detected using Grubb’s tests and data points with p-values < 0.0001 were excluded from further analysis (2 excluded for SP data, 4 for ovulin and 3 for mating duration). For multiple comparisons in mixed effects models we used Tukey tests in the multcomp package in R. Mating frequency data was analysed using Chi-Squared tests. Progeny counts from the 10-day multiple-mating competition assay were transformed to the power of 1.2 to improve normality. SDA data were normalised by a cube-root transformation. AG, testis and body size data were Box-Cox transformed where necessary. Sfp transfer data were log transformed to improve normality (1 was added to all data points to make them positive). Analyses on untransformed data produced qualitatively identical results.
Funding was provided by the NERC (studentship to JL), the BBSRC (research grant to TC and SW), the Royal Society (University Research Fellowship to TC), the NIH (Ruth L. Kirschstein NRSA fellowship 1F32GM074361 to LKS, research grant 1R01HD38921 to MFW), the Human Frontiers Science Program (Short-Term Fellowship to SW) and the Lloyd’s Tercentenary Foundation (Fellowship to SW). We thank Andy Barnes, Mara Lawniczak, Yasmine Driege, James Boone, Claudia Fricke and Dave Gerrard for assistance with experiments; Bregje Wertheim for invaluable guidance on statistical analysis; Eric Kubli for SP antiserum; Judy Appleton and Lisa Daley for advice on ELISA methods; Andrew Clark for use of his microplate reader; Tom Pizzari for advice on the manuscript and 2 anonymous reviewers for helpful comments.
The authors have declared that no competing interests exist.
Author contributions.TC, JRL, LKS, SW and MFW conceived and designed the experiments; AJB, TC, JRL and SW performed the selection and mating experiments; NB and LKS performed the ELISAs; FCFC, LS and SW analysed the data; TC, LKS, SW and MFW wrote the paper.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.