Sexy sons from re-mating do not recoup the direct costs of harmful male interactions in the Drosophila melanogaster laboratory model system


William R. Rice, Department of Ecology, Evolution and Marine Biology, University of California, Santa Barbara, CA 93106-9610, USA.
Tel.: +1-805-8935793, fax: +1-805-8934724;


The empirical foundation for sexual conflict theory is the data from many different taxa demonstrating that females are harmed while interacting with males. However, the interpretation of this keystone evidence has been challenged because females may more than counterbalance the direct costs of interacting with males by the indirect benefits of obtaining higher quality genes for their offspring. A quantification of this trade-off is critical to resolve the controversy and is presented here. A multi-generation fitness assay in the Drosophila melanogaster laboratory model system was used to quantify both the direct costs to females due to interactions with males and indirect benefits via sexy sons. We specifically focus on the interactions that occur between males and nonvirgin females. In the laboratory environment of our base population, females mate soon after eclosion and store sufficient sperm for their entire lifetime, yet males persistently court these nonvirgin females and frequently succeed in re-mating them. Females may benefit from these interactions despite direct costs to their lifetime fecundity if re-mating allows them to trade-up to mates of higher genetic quality and thereby secure indirect benefits for their offspring. We found that direct costs of interactions between males and nonvirgin females substantially exceeded indirect benefits through sexy sons. These data, in combination with past studies of the good genes route of indirect benefits, demonstrate that inter-sexual interactions drive sexually antagonistic co-evolution in this model system.


Males have been shown to harm females in a broad diversity of taxa (Dean, 1981; McKinney et al., 1983; Kasule, 1986; Arnqvist, 1989; Fowler & Partridge, 1989; Burpee & Sakaluk, 1993; Crudgington & Siva-Jothy, 2000; Moore et al., 2001). The most comprehensive data, however, comes from the Drosophila melanogaster laboratory model system, where it has been shown that interactions with males can reduce both the survival (Cohet & David, 1976; Partridge et al., 1987; Fowler & Partridge, 1989; Partridge & Fowler, 1990; Chapman et al., 1993,1995; Rice, 1996; Civetta & Clark, 2000; Sawby & Hughes, 2001; Pitnick & Garcia-Gonzalez, 2002) and fecundity (Pyle & Gromko, 1978; Chapman et al., 1995; Holland & Rice, 1999; Prout & Clark, 2000; Pitnick & Garcia-Gonzalez, 2002; Friberg & Arnqvist, 2003) of females. Females are expected to evolve resistance to harm from males and if this counter-evolution reduces the efficacy of male reproductive adaptations, then theory predicts antagonistic co-evolution between the sexes (Trivers, 1972; Parker, 1979; Rowe et al., 1994; Rice, 1996; Rice & Holland, 1997).

Empirical support for the hypothesis of antagonistic co-evolution between the sexes can be criticized for three reasons. First, most studies have only measured the effect of specific male phenotypes on components of female fitness, rather than lifetime fitness of females and therefore net selection on females could not be determined. Second, most previous laboratory studies have been carried out in environments to which the population had not previously adapted and as a consequence, the observed sexual conflict may be an artifactual short-term transient rather than co-evolved interactions between the sexes. Third and most importantly, all studies to date have measured female fitness in a single generation and therefore potentially offsetting indirect benefits, which increase the genetic quality of offspring (Parker, 1979; Arnqvist, 1992; Andrés & Morrow, 2003; Cordero & Eberhard, 2003), have not been addressed adequately. Females could gain a net benefit from apparently harmful interactions with males if these exchanges lead to matings which produced offspring with sufficiently increased genetic quality, which in turn produced greater numbers of grand-offspring (indirect benefits). In this case a female expressing a new mutation that made her resistant to costly male fertilization attempts (i.e. persistent courtship, mating attempts and sperm displacement) would lose the indirect benefit of increased genetic quality of her offspring. When direct costs of harmful interactions with males result in a net benefit to females, in the currency of grand-offspring, then a mutation causing females to be resistant to costly male fertilization attempts would not accumulate and sexually antagonistic co-evolution would not ensue.

Here we address all of these criticisms by quantifying both the directs costs and indirect benefits to females, in the currency of lifetime production of offspring, that result from their interactions with males in the environment to which the sexes have co-evolved for over 320 generations. If direct costs outweigh indirect benefits, then on balance, sexual conflict is operating. We focused on one of the major components of potential male–female conflict, i.e. persistent male courtship of nonvirgin females. In our laboratory population females mate soon after eclosion and need only mate once to secure ample sperm to last their lifetime, yet males persistently court nonvirgin females. We specifically considered selection on a hypothetical resistance mutation that protected nonvirgin females from all direct costs due to interactions with males, but also prevented females from obtaining the indirect benefits that accrue from re-mating in response to these interactions. If the percentage direct cost in offspring production is larger than the gain in grand-offspring production, then a mutation conferring resistance to male fertilization attempts would have a net selective advantage and would be expected to spread and lead to sexually antagonistic co-evolution.

As an example of such a resistance allele, consider a new mutation that causes a male pheromone (that normally is expressed exclusively in males) to be expressed in a female after she has mated. Expression of the male pheromone in a mated female interferes with her being sensed as a female by males and thereby prevents males from courting her. Expression of the female resistance gene has two effects: (i) it makes females monoecious because males no longer recognize nonvirgins as potential mates and (ii) it protects females from harm associated with persistent male fertilizations attempts. If a new mutation creating such a resistance allele increases in the population, then this would demonstrate male harm (direct costs) outweigh indirect benefits obtained due to persistent interactions between males and nonvirgin females. In our experiment we set out to measure selection acting on such a new female resistance allele by quantifying the (i) cost to nonvirgin females of interacting with males and (ii) the indirect benefits obtained by nonvirgin females when they re-mate in response to this interaction.

Male–female interactions can reduce female lifetime fecundity due to both harmful male behaviour and harm from the act of copulation and the constituents of the ejaculate. This harm can be offset when re-mating leads to higher offspring quality. Direct benefits through materials resources transferred to females at mating are absent in D. melanogaster (Chapman et al., 1994; Pitnick et al., 1997) but indirect benefits can increase genetic quality of offspring through two routes: (i) in the context of sexual selection and sperm competition when males with above average fertilization success produce sons that are also above average in fertilization success (sexy sons) and (ii) in the context of natural selection on both sexes when males with above average fertilization success produce sons and daughters with above average fitness outside the context of sexual selection (good genes). The good genes route to indirect benefits has been tested extensively in our LHM base population in two previous studies, carried out independently in two different laboratories and found to be absent (Holland, 2002; Brown et al., 2004), so we focused here exclusively on the second alternative of sexy sons.

Our study concluded that any potential indirect benefits via sexy sons that were obtained as a result of secondary mating did not counterbalance the direct cost of nonvirgin females interacting with males. These data confirm that inter-sexual interactions in the D. melanogaster laboratory model system produce the selection needed to drive sexually antagonistic co-evolution.

Materials and methods

Direct costs to nonvirgin females

To quantify the total direct cost to nonvirgin females from all aspects of their interactions with males (Cohet & David, 1976; Partridge et al., 1987; Fowler & Partridge, 1989; Partridge & Fowler, 1990), we compared the lifetime fecundity of females when they experienced minimal exposure to males (just sufficient to insure the fertilization of their eggs) to that when they were continuously exposed to males (which is the condition to which the females have adapted for over 300 generations of culture in the laboratory) as depicted in Fig. 1. To insure that the interactions between the sexes were not measured in a new environment, all of our assays were carried out under culturing conditions (timing of events, age and density of flies, culture medium and other environmental factors) that closely matched those of the base population from which they were taken (LHM, see Supplementary material). Briefly, in the base population the flies were transferred to three consecutive vials each generation: the juvenile competition vials (days 1–12.25), the adult competition vials (days 12.25–14.25) and the oviposition vials (last 18 h of the 2-week generation cycle). Larval development and competition, pupation and early adult stages took place in the juvenile competition vials. Nearly all females were inseminated by the time that they were transferred to the adult competition vials, in which most females mate again (re-mating, see Supplementary material). After interacting for 2 days in the adult competition vials, flies were placed onto fresh food during the last 18 h of their 14-day lifecycle (oviposition vials) and only eggs collected at this time were used to begin the next generation. Accordingly, only eggs laid during this 18-hour period were used to assess lifetime fecundity of females in these experiments.

Figure 1.

Measuring total direct harm (reduction in lifetime fecundity) to females due to their persistent interactions with males. Red-eyed virgin females were first mated to red-eyed males and then they competed with brown-eyed females for a limiting resource (live yeast) in environments with (continuous male exposure treatment) or without (minimal male exposure treatment) brown-eyed males. Lifetime fecundity and re-mating were then scored by culturing individual red-eyed females during the last 18 h of their 2-week life cycle and counting the number and eye colour of their offspring. Subscripts: Rd: red-eyed and bw: brown-eyed.

The assay was carried out in two experimental blocks of pairs of vials each. To produce the flies used to begin each block of the experiment, males from the replicate LHM base population were crossed to females from a replica of this population (LHM-bw, reared with the same protocol as LHM and recurrently backcrossed to this population) that is homozygous for the autosomal brown eye (bw) recessive eye colour marker. The cross was composed of 28 vials of 16 males (bw+/bw+) and 16 females (bw/bw) and the density of fertilized eggs (bw+/bw) was trimmed to 150–200 per vial. These conditions match those of the normal culture of the LHM base population. Virgin females derived from this cross were red-eyed because they were heterozygous (bw+/bw) for the recessive brown-eye colour marker. The virgins were collected on day-9 post-egg deposition and aged for 3 days before being used to begin the experiment (Fig. 1). All other flies, which did not need to be virgins, were collected from a pool of flies that emerged on days 9 and 10 post-egg deposition.

On day 12 of the 14-day generation cycle, two vials containing five red-eyed virgin females (bw+/bw) were combined with red-eyed LHM males (bw+/bw+) for 90 min in a ratio of three males to one female to ensure that all females were mated (Fig. 1, left). Past experiments in our laboratory demonstrated that virtually all virgin females are mated with this protocol. The flies were next immobilized with brief exposure (<30 s) to CO2 and males were removed. The two vials containing the remaining five red-eyed, mated females were then randomly assigned to one or the other of two treatments (Fig. 1, middle). In the ‘minimal male exposure’ treatment, the five red-eyed females were combined with 10 brown-eyed females (bw/bw) with which they competed for a limited amount (10 mg) of live yeast for 2 days. The brown-eyed competitor females were present so that we could measure direct costs of males to females when they competed for a limiting resource (live yeast), as occurs during their normal culture in the base population. In the ‘continuous male exposure’ treatment, the five red-eyed females were combined with 10 brown-eyed females (bw/bw) and 15 LHM−bw males (bw/bw) for 2 days. In this treatment the red-eyed females also competed with the brown-eyed females for a limiting resource (live yeast) but in the presence of courting males with which they could potentially mate (re-mate). At the end of the 48-hour period of adult competition, females were individually placed into food vials and allowed to lay eggs for 18 h (Fig. 1, right). This procedure mimics the duration and point in time (last 18 h of day 14 of the 2-week generation cycle) when females lay the eggs that begin the next generation in the LHM base population, so progeny production at this time represents a female's lifetime fecundity. After 12 days of incubation the family size of each female was recorded (Fig. 1, right). Re-mating by females could be detected because half of their offspring from the second sire (most offspring [85%] are derived from the second sire due to strong sperm precedence in D. melanogaster) express the recessive brown eye marker.

In our experimental design, we measured the influence of male interactions with females in the adult competition phase of their life cycle (days 12.25–14.25, see Supplementary material), but males were absent in both the minimal and continuous male exposure treatments when eggs were laid during the last 18 h of their 2-week generation cycle (the fecundity assay in Fig. 1, which corresponds to the oviposition stage of the life cycle). By not including potential male-harm to females during the oviposition stage of the lifecycle, our estimate of male-induced harm to nonvirgin females represented a conservative lower bound. As a check for any qualitative difference in the effect of male–female interaction during the oviposition period, a smaller experiment was conducted in which five vials of flies were reared under the normal protocol of the LHM base population up until the oviposition stage. At this time half of the females were placed individually in vials with a single male present and half with males absent. Fecundity was then measured over the 18-hour egg-laying period to estimate the effect of male–female interactions in the oviposition vials on the lifetime fecundity of females.

The experimental treatment measured harm to females due to their interactions with males by reducing male density to zero or retaining the same male density as the ancestral population, while keeping female density constant (0 vs. 15 males per 15 females in the adult competition vials, Fig. 1). The effect of reduced male density could be manifest directly, through the elimination of male–female interactions (which is what we set out to measure), or in a nondirect manner by lowering the influence of males on a resource limiting female reproduction. We have previously shown that female fecundity is limited by the relatively small supply (10 mg) of live yeast applied to the surface of the killed-yeast culture medium (see Supplementary material). The male density of zero in the minimal male exposure treatment might augment female fecundity in a nondirect way because males could not eat or foul the small amount of available live yeast. However, a control experiment, which will be published as part of another study, indicated that males have no measurable nondirect influence on female fecundity (see Supplementary material). As a consequence, any observed reduction in female fecundity would be due to females resisting (kicking, wing-flicking and running away from males) the persistent courtship and copulation attempts of males, as well as the harmful effects of additional seminal fluid components received when female mate more than once (Chapman et al., 1995).

Indirect benefits to nonvirgin females

To assess the potential indirect benefits through sexy sons that nonvirgin females gain from persistent fertilization attempts by males, we analysed the fitness of offspring from polyandrous females. Persistent male fertilization attempts potentially enable females to use these interactions to obtain superior mates, i.e. those that that produce sons that are more attractive to females in the context of mating and/or that are better in sperm competition. In our base population, females mate soon after eclosion (within 24 h) and obtain sufficient sperm from a single mating to last their entire lifetime in our discrete-generation, base population (Lefevre & Johnson, 1962). Yet males persistently court previously mated females and most females are polyandrous in our base population (see Supplementary material). Female interactions with males reduce their lifetime fecundity (see below), however, females could potentially recoup this cost by re-mating and producing higher quality sons. Indirect benefits via sexy sons is possible whenever there is standing genetic variation for male adult reproductive success, which is known to occur in out LHM base population (Chippindale et al., 2001; Rice & Chippindale, 2001; Gibson et al., 2002). The degree of this potential benefit depends on the heritability of male fertilization success and the degree to which females increase their bias in re-mating toward males with high heritable genetic quality, compared to their primary mating.

To measure the magnitude of these potential indirect benefits through sexy sons, we gave females an opportunity to mate multiply: a first mating as virgins followed by the opportunity to re-mate during a 2-day period commencing immediately after their first insemination, a time when most females re-mate during the standard culture of the LHM base population (see Supplementary material). If there are indirect benefits from nonvirgin females interacting with males, then these benefits should be manifest in higher genetic quality of sons sired by secondary males, compared to sons from the first mate. Put another way, if indirect benefits offset the direct, harmful effects of persistent male courtship and re-mating, then the counterbalance must be manifest by increased genetic quality of sons (sexy sons), since the good genes route has been determined to be absent (Holland, 2002; Brown et al., 2004) and because male D. melanogaster provide no material parental investment to their offspring (Chapman et al., 1994; Pitnick et al., 1997).

We used a genetic eye colour marker to identify sons from first and secondary sires of multiply mated females (Fig. 2, also see description below). We then measured the lifetime reproductive success of these sons with a protocol that closely matched the environmental conditions of the base population (Fig. 2).

Figure 2.

Measuring total indirect benefits of re-mating through sexy sons. Brown-eyed virgin females were permitted to doubly mate with brown- and red-eyed males, with the order of mating reversed in the primary-sires and secondary-sires treatments (two leftmost panels; during the primary mating, the number of males was reduced to 32 in the third replicate). Next brown-eyed offspring were collected only from multiply mated females (these families are denoted by clear vials in middle panel), which were sired by the first male in the primary-sires treatment and by the secondary males in the secondary-sires treatment. Fitness of sons from primary and secondary sires was next compared, in the context of sexual selection, by competing them with red-eyed males (panel second from right) and then measuring their offspring production (these are the grand-offspring of their mothers through sons) by culturing individual females and scoring the number and paternity of their offspring (right-most panel). Subscripts as in Fig. 1.

Flies used in the primary-sires and secondary-sires treatments were generated by creating duplicates of both the LHM base population and the replica of this population that is homozygous for the recessive brown eye (bw) marker (LHM-bw). These duplicate populations were cultured in parallel with the originals by using the same culturing protocols. From the two duplicate populations, brown-eyed (bw/bw) virgin females were collected from the duplicate of the LHM-bw population on day 9 of its 14-day generation cycle and brown-eyed males (bw/bw) and red-eyed males (bw+/bw+), used to mate the females, were collected from their respective duplicate populations (of LHM-bw and LHM) from a pool of flies that enclosed between days 9 and 11.

For each of the three replicates, approximately 20 vials of 16 brown females each were constructed for each of the two treatments. For the primary mating of these females, they were transferred without anesthesia to new vials containing an excess of males (48 for the first two replicates and 32 for the third replicate; Fig. 2, left). After 90 min for mating (past experiments in out laboratory have shown that virtually all females mate during this period while re-mating rarely occurs) (Rice, 1996; Holland & Rice, 1999), the primary males were removed with a brief exposure (<30 s) to CO2. Females were then combined with secondary males for 48 h, during which time re-mating could occur (Fig. 2, second from left).

In the primary-sires treatment the first males to mate females were brown-eyed and the secondary males were red-eyed. In the secondary-sires treatment the order of males was reversed. Because the females were homozygous for the recessive brown-eye marker (bw/bw), offspring sired by the brown-eyed males were brown-eyed (bw/bw) and those from red males were red-eyed (bw/bw+). At the end of the 2-day period for secondary mating, the flies were anesthetized with brief (<30 s) exposure to CO2, the males were discarded and 14 of the 16 females were transferred to individual vials where eggs were laid for a period of 18 h (Fig. 2, middle). On day 11, sons were collected only from those females that had re-mated (i.e. those females that re-mated during the secondary mating bout; Fig. 2, middle). This was accomplished by discarding all families of offspring derived from females that only mated to the first sires. Those sons resulting from secondary mating in the secondary-sires treatment (determined by eye colour) were ‘sexy sons’. Eight sons from each type (secondary-sires or primary-sires) were randomly collected from the pool of all sons derived from the doubly mated females from each of the original treatment vials in a replicate.

The lifetime fitness of the two types of sons (i.e. from either primary-sires or secondary-sires) was assayed by placing 16 virgin brown-eyed females (3 days post-eclosion; taken from the duplicate LHM-bw population) with eight brown-eyed males (secondary-sires or primary-sires sons derived from the multiply mated females of the previous generation) and eight red-eyed males (competitors, taken from the duplicate of the LHM base population and matched for age with the brown-eyed males) as depicted in Fig. 2 (second from right). These males and virgin females were mixed, without anesthesia, on day 12 of their 2-week generation cycle, the same day that flies are transferred to the adult competition vials during the normal culture of the LHM base population (see Supplementary material). Males competed to fertilize the females for 48 h (the same amount of time that the males and females interact in the adult competition vials within the base population) and then the males were discarded and the females (14 of the 16 were randomly sampled) were transferred to individual egg laying vials for the last 18 h of their 2-week generation cycle (Fig. 2, right). Progeny derived from eggs laid during this time were counted from the families of each individual female and scored for eye colour so to ascertain which male type (competitors or sons from primary-sires/secondary sires) produced each offspring (Fig. 2, right). The laboratory personnel counting the progeny flies were not aware of the experimental treatment from which the flies were taken. Adult lifetime reproductive success of sons from primary and secondary sires was recorded as the average number of brown-eyed offspring sired per female, with averages taken over all of the 14 females cultured for egg deposition from each of the original 20 vials per replicate. The experiment was replicated three times between July 2003 and November 2003.


Direct costs

Females experiencing continuous exposure to males had an average lifetime fecundity that was 15.6% lower than that of females experiencing only minimal exposure to males (Fig. 3, F1,190 = 18.63, P < 0.0001, 2-way anova, factors = treatment [minimal vs. continuous exposure to males] and experimental block [1 vs. 2]). We estimate the 95% lower-bound (based on 10 000 bootstraps of the percentage ratio: 100*[{average lifetime fecundity of females with continuous exposure to males divided by average lifetime fecundity of females with minimal exposure to males}−1]) of this percentage to be −10.0%, so we can be highly confident that persistent male courtship and re-mating reduced a female's fitness by at least one tenth of her lifetime reproductive output. A significant reduction in female lifetime fecundity due to interactions with males remained when we analysed only females that did not re-mate (11.7%, F1,136 = 6.89, P < 0.001, 2-way anova, factors = treatment [minimal vs. continuous exposure to males] and experimental block [1 vs. 2]), indicating that male behaviour alone harmed females. Last, the smaller assay of the direct effect of males on female fecundity during the oviposition stage of the life cycle (which was not included in our measure of male harm to females in adult competition vials) indicated that the presence of males at this time further reduced female fecundity (fecundity of females with males present was 8.84% lower than when males were absent, paired-test, t4 = 3.948, P = 0.02). As a result, our estimate of male harm to nonvirgin females, which was measured in the adult competition vials, is a conservative metric of total male harm.

Figure 3.

Mean lifetime fecundity of females with minimal vs. continuous exposure to males.

Indirect benefits

The point estimates of the fitness of sons from primary and secondary males were not statistically different (Fig. 4, F1,116 = 0.66, NS, 2-way anova, factors = treatment [sons from primary vs. secondary sires] and experimental block [1, 2 or 3]) with the average number of grand-offspring (±SE) from sons derived through primary and secondary sires equaling 15.99 ± 0.636 and 15.22 ± 0.655, respectively. There was, however, a significant noncrossing interaction (P = 0.031) between blocks (1, 2 or 3) and treatment (sons from first vs. secondary sires), so we also analysed the data on a replicate by replicate basis and found that in none of the three replicates was there a statistically significant difference in the fitness of sons from first and secondary sires.

Figure 4.

Mean grand-offspring production through sons produced from primary vs. secondary males that had mated their mothers.

Our lack of a finding of higher reproductive success of sons obtained from secondary sires may reflect lack of statistical power. Past studies of the LHM base population have shown there to be additive genetic variance for the lifetime reproductive success of adult males (Chippindale et al., 2001), so some degree of indirect benefits via the sexy sons route would seem feasible despite the fact that our point estimate is less than zero (mean offspring from secondary-sire's sons – means offspring from primary-sire's sons = 15.22–15.99 = −0.77 ± 0.913). To address this issue we constructed a 95% upper bound for the percentage advantage in fitness of sons from secondary compared to primary sires (based on 10 000 bootstraps of the percentage ratio: 100*[{average grand-offspring through secondary sires divided by average grand-offspring through first sires}−1]). This upper limit of the indirect benefits obtained by females through sexy sons was a 6.13% increase in grand-offspring production. Our estimate of indirect benefits to females is necessarily higher than the true value experienced on a per capita basis because our assay measured indirect benefits obtained by re-mated females and all females do not re-mate. Put another way, all females suffer the behavioural costs of interacting with males but only re-mated females obtain the potentially offsetting advantage of indirect benefits via sexy sons.


Our analysis of the direct cost to nonvirgin females due to their interactions with males indicates that we can be 95% confident that females lose at least 10% of their lifetime fecundity and our analysis of indirect benefits through sexy sons indicates that we can be 95% confident that females gain at most 6.13% more grand-offspring through their sons. The upper bound of 6.13% for an indirect benefit through sons must be discounted by one-half (i.e. reduced to 3.07%) because females do not gain a sexy sons advantage through their daughters. So conservatively using lower and upper bounds for costs and benefits respectively, our multi-generation analysis indicates that cost exceed benefits by at least 3-fold, and therefore that conflict between the sexes is not counterbalanced by indirect benefits through sexy sons. Previous work on our LHM base population indicates that indirect benefits through the good genes route also cannot counterbalance the observed direct costs of females interacting with males (Holland, 2002; Brown et al., 2004). Overall, we have documented a substantial direct cost to females due to interactions with males but we estimate no compensating indirect benefit through sexy sons and we can be 95% confident that any underestimation of indirect benefits due to sampling error cannot account for this disparity between costs and benefits.

Our empirical estimate of the magnitude of indirect benefits from interactions with males has relevance outside the context of sexual conflict. Traditional models of sexual selection via adaptive female preference generally assume that there is some cost associated with female choice in the currency of energy, time, or other resources spent searching for superior males (Andersson, 1994). Our study, when combined with previous studies (Holland, 2002; Brown et al., 2004), indicates that despite substantial standing genetic variance for fitness among males in our LHM base population (Chippindale et al., 2001) and elaborate sexual signalling between the sexes (Speith, 1952), there is no evidence for any substantial benefit to females via indirect benefits, at least in the context of non-virgin females trading-up by re-mating with superior males. Because indirect benefits via sexy sons are at most a 3% fitness gain when averaged across sons and daughters, the cost of female choice, in the context of re-mating by nonvirgin females, must be even smaller than this value for the sexy sons processes to be an effectual mode of sexual selection. Our experiments do not test the general operation of the sexy sons model of sexual selection because they only address the potential advantage of indirect benefits in the context of female re-mating.

Because prior work on the LHM base population found substantial heritable variation for lifetime fitness (Chippindale et al., 2001), it might seem surprising that indirect benefits associated with re-mating have been estimated to be no different from zero in this study (and also in two previous studies, Holland, 2002; Brown et al., 2004). Our data, however, do not demonstrate that there is no indirect benefit associated with re-mating, only that the benefit is small relative to the cost of interacting with males. It is well established that traits closely associated with net fitness typically have low heritabilities due to the presence of substantial nonheritable phenotypic variation (Houle, 1992). Low heritability of fitness is expected to cause adaptive female choice via indirect benefits to be less effective because of a reduced correlation between phenotype (male fitness) and genotype (heritable fitness variation).

At any given time the direct cost to females of interacting with males will depend upon the degree to which the evolution of female resistance lags behind male adaptations that harm the females. Because females generally provide more parental investment than males (Trivers, 1972), males are predicted to evolve to exploit this asymmetry, after which females respond by evolving resistance (Trivers, 1972; Parker, 1979; Rowe et al., 1994; Rice, 1996; Rice & Holland, 1997). As a consequence, females will be perpetually ‘catching up’ in the inter-sexual arms race, generating a chronic lag-load (Maynard Smith, 1976) in females. Here we demonstrated the existence of such a lag load by comparing the fitness of females in male-protected and male-unprotected environments. Our large point estimate of the magnitude of the cost to females of interacting with males (a 16% reduction in lifetime fecundity) indicates that conflict between the sexes is a powerful selective force that is capable of driving rapid antagonistic co-evolution between the sexes.

Because our study compared the total direct costs of males to the total indirect benefits from mating with additional males, it remains possible that indirect benefits may compensate for some components of direct male harm. Nonetheless, in this case a limited form of sexually antagonistic co-evolution could still ensue when adaptations by females to lessen direct costs, while retaining indirect benefits, also reduced the efficacy of male adaptations mediating male–male competition for fertilizations. In this case females receive a net benefit from the interaction with males, but selection would still occurs to reduce the costs, and this can drive counter evolution by males. Collectively the data presented here provide strong evidence that, on balance, inter-sexual conflict occurs in the D. melanogaster laboratory model system that is not offset by indirect benefits, and that this conflict is expected to lead to antagonistic co-evolution between the sexes.


We thank A. Stewart, E. Morrow, E. Cunningham and A. Kaplan for technical help and U. Friberg, B. Holland, D. Promislow, E. Morrow and A. Stewart for helpful comments on early drafts of the manuscript. We also thank E. Gaines and A. Chippindale for help in collecting the data for Fig. S1. This work was supported by grants US National Science Foundation (DEB-0128780 and DEB-0410112) to W.R.R.

Supplementary material

A Word file with more information on this subject is available to download from the following website: