Female × Male Interactions in Drosophila Sperm Competition

See allHide authors and affiliations

Science  08 Jan 1999:
Vol. 283, Issue 5399, pp. 217-220
DOI: 10.1126/science.283.5399.217


In several organisms, the success of a male's sperm in multiply inseminated females depends on the male's genotype. InDrosophila, the female also plays a role in determining which sperm are successful. Pairwise tests among six isogenic lines ofDrosophila melanogaster were performed to determine whether there is a genotype-specific interaction in the success of sperm. The success of a particular male's sperm was found to depend on the genotype of the female with which he mates, providing evidence for an interaction with profound evolutionary consequences.

Males and females face very different problems in trying to assure that their gametes are used maximally. Males that can co-opt females into using their sperm are at an advantage over other males. Ignoring the complexities of mating behavior, there is an enormous opportunity for variation in the success of different males' sperm in multiply mated females. Genes that give such an advantage to one male's sperm are expected to increase in the population, even if this increase causes a decline in the viability or fertility of the mother or offspring (1). Females cannot afford to let males be the sole determinant of which gametes are used, especially if evolution in males results in deleterious consequences for the female (2–4).

In the case of Drosophila, females mate before stored sperm are exhausted, resulting in sperm from more than one male being present in the storage organs of females (5). Extensive variation among male genotypes in the competitive success of sperm in multiply mated females has been documented for Drosophila melanogaster(6, 7). Similarly, females differ widely among genotypes in their tendency to exhibit last-male precedence (8). A preclusion of the evolutionary response by females results in the increased competitiveness of sperm, which comes at a cost to the female's life-span (3). This result suggests that not only is there substantial genetic variation among males in sperm competitiveness, but also that females are normally evolving in response to male variation in these traits. These results beg the question of whether there are specific interactions between the genotypes of males and females that determine the outcome of sperm competition. We designed a study to quantify interactions in gametic use and found that these interactions are quite strong.

Sperm displacement by a pair of male genotypes (9) was tested in both mating orders, with the order (bwD , wild) considered as a test of the “offense” ability of the wild male's sperm to displace the resident tester (bwD ) male's sperm (10) (Table 1). The other order (wild,bwD ) was considered as a test of “defense” and scored how well the wild male's sperm resisted being displaced by the tester male (11). Additional tests were done to verify that multiple mating within any trial time was rare (12). If the different genotypes varied in viability, this might appear as a spurious sperm displacement effect (13). The segregation ratios were used to score viabilities in four separate crosses, including +i/bwD × +j/+j, +j/+j ×bwD /+i,bwD /+i × +j/+j, and +j/+j × +i/bwD , where the female genotype is given first, the allele of maternal origin is given before the slash, and i and j represent all six lines. Segregation ratios were scored in more than 1500 vials (>45,000 flies) and were found to be heterogeneous among the 36 crosses by an analysis of variation (ANOVA) (not shown), but the magnitude of the differences was not great (Fig. 1). If the fractions of wild and bw offspring were k and (1 – k), then dividing the counts of wild and bwoffspring by 2k and 2(1 – k), respectively, will compensate for the segregation effects.

Figure 1

(left). Segregation ratios for the 36 pairwise combinations of six lines. Each point represents the average across replications of the four segregation tests, and each has an average standard error of 0.014. The six lines represent the six female genotypes (solid circle, 1; solid diamond, 2; open circle, 3; open square, 4; cross, 5; and open triangle, 6). The means were found to be significantly heterogeneous, so departures from Mendelian segregation of offspring counts were compensated in estimates of the sperm displacement parameters P1 and P2.

Table 1

Experimental design.

View this table:

Variability among the 36 crosses in the degree of sperm displacement is extensive (Fig. 2). An inspection of the line means suggested that both male and female effects on P1 and P2 are evident, which is consistent with earlier studies. (P1 is the fraction of offspring sired by the homozygous wild-type male when he is the first male to mate and is generally well under ½; P2 is the fraction of offspring for each cross that were fathered by the second male, when the second male to mate is wild type.) A three-way ANOVA was used to test the null hypothesis that the degree of sperm precedence was the same in all crosses (Table 2). The tests were performed both with and without compensation for the segregation differences. Test location (California, Pennsylvania, or Texas) was treated as a cross classification (along with the line), because environmental conditions were different in the three locations and line × location interactions are equivalent to biologically interesting gene × environment interactions. On average, 19.3 replicate tests were done for each of the 36 crosses at each location, and 124,881 offspring from 6246 vials were scored.

Figure 2

(right). (A) The defense component of sperm displacement, P1, plotted for all 36 combinations of male and female genotypes. The interaction between male and female genotypes is evident from the crossing of the lines connecting female genotypes. Symbols are as in Fig. 1. (B) The offense component, P2, is plotted as in Fig. 2A.

Table 2

Analysis of variance of sperm displacement parameters P1 and P2. df, degrees of freedom; MS, mean square;F comp, F statistic for segregation-compensated data; F uncomp,F statistic for uncompensated data; dash, not applicable.

View this table:

The location × female × male interaction was significant in both tests, meaning that the rank order of female × male interactions in P1 and P2 differed at the different sites. A conservative test of the female × male interaction was to use a composite MS, including the three-way interaction MS as the denominator, and this test yielded a significant female × male interaction for P1 only. When the ANOVA was performed for each location separately, the tail probabilities for the female × male interaction in the defense tests were 0.2787, 0.0001, and 0.0518 for California, Pennsylvania, and Texas, respectively; the offense tests had respective tail probabilities of 0.2085, 0.0001, and 0.0102. The likely cause for the lower significance of the California samples was that the California sample size was 20% that of the Pennsylvania sample and 30% that of the Texas sample.

We also constructed the null distribution of the interaction MS by randomly permuting the observed data, shuffling the observed measure (P1 or P2 in this case) in relation to the classification variables that were being tested. For each permuted data set, the MS was calculated again. With sufficient replication (n = 10,000), the permuted samples provided the best picture of the null distribution that was compatible with the sample structure, although ideally the test would conserve the marginal male and female effects (14). The MS of female × male interaction was found to fall in the null distribution with 0 out of 10,000 replicates having a greater P1 and 12 out of 10,000 having a greater P2. The case for female × male interaction in both components of sperm competition is quite strong.

The relative success of the sperm from the same or different strains is compared in Fig. 2. The rank of competitive ability of the line 1 sperm in offense tests in females of line 1 was 3 (out of 6). The ranks for the other five lines were 3, 4, 5, 6, and 6 (out of 6). A permutation test showed that the probability of getting ranks as high or higher than these by chance was <0.002, so it appears that females are avoiding sperm from their own line, which is consistent with an inbreeding avoidance. There was no such association in the defense tests [ranks were 1, 1, 5, 4, 3, and 5 (out of 6)]. Offense and defense components appear to be physiologically distinct, because, consistent with previous findings (11), offense and defense components of sperm displacement were uncorrelated [Spearman rank correlation of the line means of P1 versus P2 was 0.155 (P < 0.365)].

Physiological mechanisms suggest that female × male interactions in the manifestation of sperm precedence ought to be expected. The uptake of seminal proteins by females varies widely across species, and often, radiolabel incorporation is seen in somatic tissue (15). There is abundant evidence that male accessory gland proteins, which are transmitted to the female in seminal fluid, mediate egg laying, sperm use, remating, and other behavioral changes in the female (16). There also is apparently strong conspecific recognition of sperm, which may be important in the evolution of isolating mechanisms (17). Seminal proteins are highly variable and undergo rapid molecular evolution, which is consistent with a male-female coevolutionary arms race (18).

In order to extend models of sperm competition (19) to the situation wherein sperm success depends on both the female and male genotypes, a matrix of sperm displacement parameters must be constructed for all mating combinations. Computer simulations of the simplest two-allele model with female × male interactions demonstrate that not only can polymorphisms be maintained by this model, but this model can exhibit complex cycling dynamics. The mode of selection suggests that rapid allele turnover would occur and that polymorphism among alleles with strong differences in sperm competitive ability is not unlikely.

When laboratory populations were allowed to evolve in such a way that female-specific responses could not occur, remating was faster and female fitness was reduced (3, 4). This implies that evolution results in male traits that are in conflict with female fitness. Females normally respond to this evolution to keep the deleterious effects of male evolution in check, and evolutionary changes in females, in turn, change the males' mating and sperm competitive ability. The male-female signaling that is evidently involved in decisions about remating and sperm use is not simple, but it may have profound fitness consequences. The resulting antagonistic coevolution may be responsible for the rapid rate of molecular evolution of genes encoding seminal proteins.

  • * To whom correspondence should be addressed. E-mail: c92{at}


View Abstract

Navigate This Article