https://orcid.org/0000-0002-1853-7469
Abstract
Bateman’s principles, originally a test of Darwin’s theoretical ideas, have since become fundamental to sexual selection theory and vital to contextualizing the role of anisogamy in sex differences of precopulatory sexual selection. Despite this, Bateman’s principles have received substantial criticism, and researchers have highlighted both statistical and methodological errors, suggesting that Bateman’s original experiment contains too much sampling bias for there to be any evidence of sexual selection. This study uses Bateman’s original method as a template, accounting for two fundamental flaws in his original experiments, (a) viability effects and (b) a lack of mating behavior observation. Experimental populations of Drosophila melanogaster consisted of wild-type focal individuals and nonfocal individuals established by backcrossing the brown eye (bw-) eye-color marker—thereby avoiding viability effects. Mating assays included direct observation of mating behavior and total number of offspring, to obtain measures of mating success, reproductive success, and standardized variance measures based on Bateman’s principles. The results provide observational support for Bateman’s principles, particularly that (a) males had significantly more variation in number of mates compared with females and (b) males had significantly more individual variation in total number of offspring. We also find a significantly steeper Bateman gradient for males compared to females, suggesting that sexual selection is operating more intensely in males. However, female remating was limited, providing the opportunity for future study to further explore female reproductive success in correlation with higher levels of remating.
Introduction
Darwin’s The Descent of man (1871) explored the mechanisms of sexual selection by which “secondary” sexual characters that seemed to provide little advantage for survival could have evolved. Darwin separated sexual selection from natural selection based on its operation through reproductive competition, describing how males are generally more competitive for access to mating partners compared with females, while females are more discriminating of mating partners (Clutton-Brock, 2017). The modern distinction from natural selection commonly describes sexual selection operating via differential success in mating and fertilization (Andersson, 1994; Henshaw et al., 2016; Jones, 2009). Darwin’s ideas were further expanded some decades later by notable biologists Fisher (1930) and Huxley (1938). Although Huxley disagreed with much of Darwin’s theoretical ideas, he aided in synthesizing evidence of sex differences in complex displays and aggressive behavior, while also being the first to suggest that variance in reproductive success should be greater in males than in females, at least in polygamous species (Clutton-Brock, 2017; Huxley, 1938; Otte, 1979). While the work of both Huxley and Fisher readdressed sexual selection to consider the evolution of sex differences in morphology and behavior, a fundamental question still remained. Why should it be that females were typically more selective of mating partners compared with males, whereas males were primarily both more competitive in mating and highly ornamented?
Modern thinking surrounding this key question in sexual selection research is strongly influenced by Bateman (1948). Intrasexual selection, involving competition between individuals of one sex for access to mates, had been overwhelmingly assumed to be, what Bateman termed “intra-masculine,” due to general observation that males were more “eager” to mate, whereas females although “passive” in mating, exerted choice (Bateman, 1948; Dewsbury, 2005; Tang-Martínez, 2010). However, Bateman was more interested in finding an ultimate cause of intrasexual selection, irrespective of the mating system, and therefore used the model organism Drosophila melanogaster in his experiments, at equal sex ratios (Bateman, 1948). Bateman set up 64 experimental populations, which were divided into six “series.” Each series varied in group size (either three females and three males, or six females and six males), the age of flies (1, 3, or 6 days old), the number in which mating and oviposition could occur (3–4 days), and in one series, the population was transferred to a new vial every day for the duration of the experiment. In total, 215 adult females and 215 adult males were used (Bateman, 1948; Snyder & Gowaty, 2007). Moreover, Bateman’s experiment was also the first crucial study of genetic parentage in experimental populations. Flies were taken from several inbred strains that each carried a different heterozygous dominant phenotypic mutation at a single locus. Thus, each male and female within an experimental population carried a different dominant marker mutation, and parentage could be assigned to offspring based on these visible genetic markers. These data were used to infer the number of mates for each male and female in experimental populations: (a) for each male, how many females within the population had offspring that carried his marker mutation and (b) for each female, how many different marker mutations were present in her offspring, and the number of offspring per male and female (Bateman, 1948; Snyder & Gowaty, 2007).
Bateman’s results have since become known as Bateman’s principles. These are as follows: (a) males have higher individual variation in number of offspring (i.e., reproductive success) compared with females, (b) males have higher variance in number of mates (i.e., mating success) compared with females, and (c) the slope of the relationship between mating success and reproductive success (termed the Bateman gradient) is steeper in males than in females (Arnold, 1994; Bateman, 1948; Collet et al., 2014). Bateman’s results provided a theoretical explanation for Darwinian sex roles implying that males typically compete more intensely for mates, whereas females are more choosy with respect to partners and provide more parental care, reasoning that the sexual dimorphism observed in many species could be explained by differences in the intensity of sexual selection (Bateman, 1948). He considered his first two principles, the sex difference in variance of number of mates and number of offspring, to be an indication of intramasculine selection, whereas the third principle, a stronger correlation in males between number of mates and number of offspring, was reasoned as the cause for sexual selection on males.
Importantly, Bateman not only stressed the significance of sex differences in sexual selection for the evolution of sexual dimorphism but also speculated on the ultimate cause for Darwinian sex roles to predominate the animal tree of life. Specifically, he argued that sex differences in sexual selection are ultimately rooted in anisogamy by considering a steeper correlation between mating success and reproductive success for males, compared with females, to be a result of the sex differences in the size and rate of production of gametes, which has since become fundamental to evolutionary theory (Lehtonen, 2022; Schärer et al., 2012; Vries & Lehtonen, 2023). Anisogamy (large, relatively immobile female gametes and small, mobile male gametes) leads to the allocation of resources to mating and offspring being asymmetrical between the sexes. Considering egg production is resource limited, while being energetically costly, females are expected to be typically more choosy of mating partners regarding their genetic quality or with respect to resources provided by the mate. Males, however, can produce an abundance of relatively less expensive sperm and can therefore afford to mate with and fertilize a larger number of mates. Anisogamy results in sexual selection of males for the allocation of resources to costly traits that increase reproductive success. As a consequence, sexual selection should be stronger on males than on females, due to reproductive competition, and the greater potential payoff males obtain from multiple mating (Janicke et al., 2016; Kokko & Jennions, 2008; Lehtonen et al., 2016; Parker, 2014).
Since Triver’s (1972) review readdressed Bateman’s principles, it has been integral to studies of sexual selection and sex differences (Kokko et al., 2012; Parker & Birkhead, 2013). Arnold and Duvall (1994) formalized Bateman’s third principle into a measure of sexual selection, concluding that Bateman gradients were crucial to understanding both the direction and intensity of sexual selection. Thus, the sex that displays a steeper partial regression slope of fecundity as a function of number of mates will be under stronger sexual selection (Arnold & Duvall, 1994). More recent research has often focused on testing Bateman’s principles in sex-role-reversed species, those species in which males care for young, whereas females compete for mates and do not provide parental care (e.g., Eens & Pinxten, 2000; Emlen & Wrege, 2004; Jones & Avise, 2001; Jones et al., 2005). As expected, females were found to have steeper Bateman gradients, and thus are under strong sexual selection, being competitive rather than choosy, whereas sexual selection on males is relaxed, and they are the choosier sex (Andrade & Kasumovic, 2005).
Bjork and Pitnick (2006) tested Bateman’s principles considering the effects of anisogamy, repeating Bateman’s mating experiment with several species of Drosophila that differed in the size ratio of sperm to egg. In D. melanogaster, which have very different sizes of sperm and egg, the Bateman gradient was steeper for males compared with females, whereas in D. bifurca, which are relatively isogamous, the difference in slopes for males and females was not significant, consistent with Bateman’s conclusions (Bjork & Pitnick, 2006). Moreover, Janicke et al. (2016) synthesized 72 studies on 66 animal species, which provided estimates of (a) variance in reproductive success (the opportunity for selection), (b) variance in mating success (the opportunity for sexual selection), and (c) Bateman gradients, to determine whether Bateman’s principles were consistent across the animal kingdom. These results indicated that males showed both a higher opportunity for selection and steeper Bateman gradients, compared with females, with parental care and sexual dimorphism explaining a significant proportion of the observed variation. In addition, based on an extended data set comprising 95 effect sizes, Janicke and Fromonteil (2021) found meta-analytic evidence for a higher opportunity for sexual selection in males and that the sex difference in the opportunity for sexual selection predicts sexual size dimorphism in animals. Thus, the results provided further support for Bateman’s conclusions that anisogamy leads to stronger sexual selection on males and, additionally, that sexual selection is evolutionarily linked to both sexual dimorphism and sex-biased parental care (Janicke et al., 2016).
Despite abundant evidence, Bateman’s principles have not eluded criticism. From a conceptual perspective, Bateman’s principles have been questioned to be universally valid across animals because many studies have provided results that are inconsistent with Bateman’s principles, particularly those in which both sexes experience comparable levels of sexual selection (Hafernik & Garrison, 1986; Jensen et al., 2004; Scott, 1988). Moreover, it is generally recognized that females can benefit from multiple mating (Arnqvist & Nilsson, 2000; Evans & Magurran, 2000; Fromonteil et al., 2023; Stockley et al., 1997), males may discriminate between mating partners (Arnaud & Haubruge, 1999; Sæther et al., 2001), sperm production can be costly (Dewsbury, 1982), and that sperm depletion can result in strategic sperm allocation, a form of mate choice (Wedell et al., 2002). Given these opposing findings, some researchers have suggested that sexual selection theory is fundamentally flawed, suggesting sex roles are instead driven by ecological, demographic, and social conditions, rather than theories set out within the Darwin–Bateman paradigm (Gowaty & Hubbell, 2009; Roughgarden, 2015).
From a more empirical perspective, Bateman’s original study has been questioned based on limitations regarding the experimental design and the statistical analysis (Hoquet et al., 2020; Snyder & Gowaty, 2007). Moreover, Gowaty et al. (2012) produced a complete replication of Bateman’s original experiment. Following Bateman’s methodology, they used the same mutant lines (with one exception), cultured in the same way, with experimental populations kept at the same combination of sex, age, genetic markers, and mating assay duration. Parentage was also assigned using Bateman’s exact method, with only offspring that carried a mutation from each parent (double mutants) included in mating success counts, and the sum of single and double-mutant offspring used to estimate reproductive success. Their repetition uncovered fundamental flaws in Bateman’s method. First, the mutations Bateman used had viability effects, as evidenced by (a) the frequency of double-mutants was lower than expected under Mendel’s law, resulting in miscounting mating success, and (b) considerably fewer single-mutant offspring survived when carrying the mother’s wild-type allele rather than the father’s. Furthermore, genetic paternity could be assigned much more than genetic maternity, resulting in biased estimates of reproductive success. Finally, some individuals were marked as having not mated, despite their phenotypes being present in offspring, but were excluded due to only double-mutant offspring being considered in mating success estimates. They conclude, based on their results, that Bateman’s original experiment contains too much sampling bias for there to be any evidence of sexual selection (Gowaty et al., 2012, 2013).
Combined, these criticisms and the flaws within Bateman’s original experiment have resulted in a concept that, although widely cited and considered to be fundamental to sexual selection theory, remains controversial. Here, we adapt Bateman’s original method, to explore whether replicating Bateman’s experiment, while rectifying the problems raised by Gowaty et al. (2012), finds support for Bateman’s principles in Drosophila melanogaster. We used Bateman’s original method as a template, specifically to (a) measure mating success for both males and females, based on number of mates per individual, (b) measure reproductive success for both males and females, based on total number of offspring per individual, and (c) explore the relationship between mating success and reproductive success for both males and females. These data allow us to calculate Bateman’s classic sexual selection metrics for both sexes: I, the standardized variance in reproductive success or “opportunity for selection”; IS, the standardized variance in mating success or “opportunity for sexual selection”; and ßSS, the Bateman gradient, the slope of a least-squares regression of reproductive success against mating success.
To eliminate the potential for viability effects, experimental populations consisted of focal individuals collected from a wild-type population (LHM), and nonfocal individuals collected from a population that is homozygous for the recessive brown eye (bw-) eye-color marker. This population was established by backcrossing the mutation into a replicate LHM population; thus, all other loci are representative of the wild-type population, effectively eliminating viability effects. While the bw eye-color marker has no effect on female fecundity, it does have an effect on male fitness, with wild-type males showing a 27.2% advantage over bw- competitors (see Appendix D in Stewart et al., 2005). This competitive advantage may result in underestimates of the opportunity for selection and the opportunity for sexual selection (resulting in more conservative comparisons between sexes), but possibly leads to an overestimation of the male Bateman gradient because focal individuals may obtain more offspring per additional partner when experiencing less competition. Furthermore, mating behavior was observed directly to obtain accurate measures of mating success, as opposed to Bateman’s original method of using genetic markers. As advocated by Morimoto (2020), we use Bateman’s three principles as a framework to make three testable predictions: (a) males will have higher individual variation in reproductive success compared with females, (b) males will have higher variance in mating success compared with females, and (c) the slope of the relationship between mating success and reproductive success will be steeper in males than in females.
Materials and Methods
Fly Stocks and Rearing Conditions
Male and female D. melanogaster used in the mating assay were taken from two previously established laboratory populations at the University of Sussex, LHM and LHM-bw. LHM is a large outbred wild-type population that has adapted to laboratory conditions for over 500 nonoverlapping generations (Rice et al., 2006). It was established from a sample of 400 inseminated females collected near Modesto, CA (USA) in 1991 (Rice et al., 2005). LHM-bw is homozygous recessive for the autosomal brown eye-color marker (bw-) and was established by multiple and recurrent backcrossing the mutation into a replicate LHM population (LHM females crossed with LHM-bw males, then F1 offspring crossed with each other to give backcrossed LHM-bw individuals; last completed 9 rounds in 2012).
Both populations were maintained in 25-mm vials containing a cornflour–agar–molasses medium, in incubators set to 25 °C and 65% relative humidity, on a 12:12 hr light:dark cycle. Experimental populations were cultured under the same conditions as the base populations in which they were taken (for a full description, see Rice et al., 2006). The experiments were carried out in October 2017. Briefly, to begin each new generation, eggs laid in juvenile-competition vials are culled to a density of 150–200 eggs per vial. Larval development, pupation, and early adult stages occur in these vials from days 1 to 12. Adult flies are then mixed among adult competition vials and placed into groups at a density of 16 males and 16 females per vial from days 12 to 14. During these 2 days, females compete for a fecundity-dependent limiting resource (6 mg of live yeast on the surface of the food medium), while males compete for fertilizations. Finally, flies are transferred to oviposition vials without live yeast on day 14 for 18 hr and then discarded. Eggs from the oviposition vials are used to start the next generation (Morrow et al., 2005; Rice et al., 2006).
Virgin females were collected from juvenile-competition vials on days 9–11 of the 2-week generation cycle. All flies were discarded between 8 am and 9 am, and subsequently newly eclosed adult females were collected every 3 hr, to ensure they would not remain within the vial (that may have contained newly eclosed males) after they had reached sexual maturity. Females were distributed into same-sex vials at the density required for the mating assay, either 16 LHM-bw, 15 LHM-bw, or 10 LHM. Females ranged from 4 to 7 days old at the time of the mating assay. Males were collected from the same populations subsequent to virgin female collection and again distributed into same-sex vials at the density required for the mating assay, either 16 LHM-bw, 15 LHM-bw, or 10 LHM. Males ranged from 3 to 6 days old at the time of the mating assay.
General Apparatus
Individual flies were moved between vials under CO2 anesthesia using a small synthetic paintbrush, during the process of preparing adult competition vials, sexing flies, and distributing into mating assay vials. During the mating assay, flies were observed in 25-mm vials and medium supplemented with 6 mg of live yeast. An LED microscope ring light was used during mating behavior observation, to differentiate between LHM and LHM-bw individuals using the eye-color marker. Subsequent to the mating assay, females were isolated into individual 10-mm test tubes containing the cornflour–agar–molasses medium.
Experimental Design
In total, ~1,300 flies were used in this experiment. Mating assay vials were separated into two groups: (a) female observational vials (n = 40) and (b) male observational vials (n = 35). Each female observational vial consisted of 1 focal LHM female, 15 LHM-bw females, and 16 LHM-bw males. Each male observational vial consisted of 1 focal LHM male, 15 LHM-bw males, and 16 LHM-bw females (Figure 1). This design has two major advantages over Bateman’s original design. First, since each vial contained one focal individual (LHM) for which mating behavior could be observed, individual-level data for both sexes could be collected. For females, using LHM in a population of LHM-bw allowed for accurate identification during the mating assay based on eye-color marker (LHM exhibit wild-type red eye compared with brown [bw] eye-color marker). For males, this allowed for accurate identification during the mating assay, as well as the ability to determine the number of offspring (reproductive success). As all other females within the vial were LHM-bw, any offspring sired by the focal LHM male (which was homozygous for the dominant wild-type red eye marker) would exhibit red eye color. Second, experimental populations consisted of 16 pairs, which matched the population density and sex ratio that the base population had evolved in. A small subset of the experimental vials did not match this density (nine male vials and five female vials), lacking one to three individuals due to mortality.
Experimental design. In both focal male and focal female observational vials mating behavior was observed for 4 hr, then males were discarded and females remained in the same observational vials for 1–2 days. Subsequently, females were isolated into individual 10-mm test tubes for 18 hr, then discarded. For focal male vials, all females were isolated, for focal female vials only the focal LHM female was isolated. After 11 days, number of offspring and eye color were scored.
Mating assays occurred over 3 days, with roughly equal numbers of male and female mating assays carried out per day. Mating behavior was observed for 4 hr, during which time observational vials were scanned and mating recorded continuously. Only mating pairs observed for longer than 10 min were considered successful matings, based on studies that indicate the mean copulation time for the LHM population is 19.5 min (n = 204; SD = 3.87 min) and minimum copulation time was 11.6 min (Kuijper & Morrow, 2009). The time at which copulation ended was noted to distinguish between number of matings.
After 4 hr, males were discarded and females remained in the same observational vials for 1–2 days, to replicate the culturing conditions of the LHM population as much as possible. Subsequently, females were isolated into individual test tubes for 18 hr to lay eggs. For female observational vials, this was only focal LHM females and bw females were discarded; for male observational vials, all bw females were isolated (Figure 1). After 11 days, reproductive success was scored. For focal females, this was the total number of offspring. For focal males, this was the proportion of each bw female’s total number of offspring which displayed the wild-type red eye. These figures were then added together for an overall number per male observational vial. A small proportion of the number of mate values (n = 4) was altered due to calculations of reproductive success highlighting inaccurate observations of mating success. Genetic parentage analysis (based on the eye-color marker) indicated an additional mating based on the number of females within the population that produced offspring with the focal male’s genotype.
Statistical Analyses
Statistical analyses included two steps. First, we estimated sexual selection metrics and tested for sex differences therein. This was done on relativized data of mating success and reproductive success (i.e., observed values divided by the average obtained for the given sex) to allow for comparison between sexes and with other studies (Anthes et al., 2017; Jones, 2009). In addition to Bateman’s classic sexual selection metrics (i.e., I, IS, ßSS; see above), we also estimated the maximum standardized sexual selection differential (s’max), defined as the product of ßSS and the square root of IS, which provides an upper limit of the strength of precopulatory sexual selection and has been found to be particularly powerful to quantify the actual strength of sexual selection (Henshaw et al., 2016). We applied bootstrapping with 10,000 bootstrap samples to estimate all sexual selection metrics with their 95% confidence intervals (CIs) using the package “boot” (Canty & Ripley, 2017) in R (R Development Core Team, 2021). Sex comparison was done using permutation tests with 10,000 iterations in which P-values are estimated as the proportion of permutations with a difference as large or larger than the observed difference. We arbitrarily defined the sex difference between sexual selection metrics as the estimate obtained from males minus the estimate obtained from females so that positive values indicate a male bias (ΔSex = xMales − xFemale). For completeness, we also report Bateman gradients as fits of linear regressions obtained from the lm function in R (R Development Core Team, 2021).
Second, we decomposed the variance of male reproductive success (i.e. the opportunity for selection) to compare the potential of pre- and postcopulatory episodes of sexual selection. Following an approach detailed elsewhere (Arnold & Wade, 1984; Pélissié et al., 2014; Webster et al., 1995), we used the delta method to decompose the opportunity for selection as the sum of variances of fitness components and of their covariances (see also Collet et al., 2012; Janicke et al., 2015; Pischedda & Rice, 2012). Specifically, we modeled male reproductive success (mRS) as the product of copulatory mating success (cMS; i.e., the number of mating partners), paternity share (PS; i.e., the proportion of offspring sired by the focal male out of the total number of offspring produced by its female mates), and the mate’s fecundity (mFec; i.e., the average number of offspring produced by all mating partners of the focal individual). This allowed us to partition the variance in relative mRS (also called I; see above) into its fitness components as
With Cov defined as
Variance in cMS refers to the fraction of variance in mRS that can be attributed to precopulatory processes and is equal to the male opportunity for sexual selection (Is). By contrast, variance in PS refers to the fraction of variance in mRS caused by precopulatory and postcopulatory episodes of sexual selection (Evans & Garcia-Gonzalez, 2016). In particular, variance in PS includes variance arising from individual differences in insemination success (i.e., the proportion of mating partners that were successfully inseminated) and variance in fertilization success (i.e., the proportion of offspring sired relative to the number of sperm inseminated, which is the outcome of sperm competition and cryptic female choice). Hence, comparison of variance of cMS and PS allows to contrast the potential of precopulatory versus postcopulatory sexual selection.
Values are given as means ± SE unless otherwise stated. All data and R script for analysis, figures, and table are deposited with Zenodo (https://zenodo.org/record/7729031#.ZA9Z5uzMJQK).
Results
The vast majority of the focal individuals copulated during the mating trials. Only four males and one female did not mate. As a consequence, most focal individuals reproduced, with only six males and two females producing no offspring. Importantly, 26 (74.3%) males copulated with more than one female partner, whereas only three (7.5%) females remated with another male.
All estimated sexual selection metrics showed a significant male bias. In particular, we observed an almost threefold higher opportunity for selection (I) and a fourfold higher opportunity for sexual selection (Is) in males compared with females (Table 1; Figure 2A and B). Moreover, mating success had a significant positive effect on reproductive success in males (mean slope ± SE: 1.03 ± 0.12, F1,33 = 68.41, p < .001, R2 = .67), but not in females (mean slope ± SE: 0.37 ± 0.23, F1,38 = 2.58, p = .117, R2 = .06), which translated into a steeper male Bateman gradient (Table 1; Figure 2C). We obtained similar results when excluding individuals that did not copulate during mating trials in terms of a positive male Bateman gradient (mean slope ± SE: 1.05 ± 0.17, F1,29 = 39.18, p < .001, R2 = .57) and a statistically nonsignificant female Bateman gradient (mean slope ± SE: 0.11 ± 0.27, F1,37 = 0.18, p = .675, R2 < .01) both resulting in a significant sex differences (ΔSex, 95% CIs: 0.93, 0.40 to 1.46; p = .025). Finally, according to the findings on the opportunity for sexual selection and the Bateman gradient, we detected a significant sex difference in the Jones’ index, which was six times higher in males (Table 1).
Sexual selection metrics (I, Is, ßss, s’max) and sex differences (ΔSex). Estimates of males (N = 35) and females (N = 40) are shown with 95% confidence intervals obtained from bootstrapping. p-Values indicate the significance of the observed sex difference based on permutation tests. All estimates were calculated on relativized ]data.
| Sexual selection metric | Males (95% CIs) | Females (95% CIs) | ΔSex (95% CIs) | p-value |
|---|---|---|---|---|
| Opportunity for selection (I) | 0.56 (0.25, 0.91) | 0.19 (0.11, 0.28) | 0.35 (0.05, 0.73) | .036 |
| Opportunity for sexual selection (Is) | 0.36 (0.19, 0.53) | 0.09 (0.02, 0.18) | 0.26 (0.08, 0.46) | .007 |
| Bateman gradient (ßss) | 1.03 (0.71, 1.36) | 0.37 (-0.08, 0.97) | 0.67 (0.04, 1.23) | .045 |
| Jones’ index (s’max) | 0.60 (0.38, 0.87) | 0.10 (-0.02, 0.29) | 0.50 (0.21, 0.80) | <.001 |
| Sexual selection metric | Males (95% CIs) | Females (95% CIs) | ΔSex (95% CIs) | p-value |
|---|---|---|---|---|
| Opportunity for selection (I) | 0.56 (0.25, 0.91) | 0.19 (0.11, 0.28) | 0.35 (0.05, 0.73) | .036 |
| Opportunity for sexual selection (Is) | 0.36 (0.19, 0.53) | 0.09 (0.02, 0.18) | 0.26 (0.08, 0.46) | .007 |
| Bateman gradient (ßss) | 1.03 (0.71, 1.36) | 0.37 (-0.08, 0.97) | 0.67 (0.04, 1.23) | .045 |
| Jones’ index (s’max) | 0.60 (0.38, 0.87) | 0.10 (-0.02, 0.29) | 0.50 (0.21, 0.80) | <.001 |
Sexual selection metrics (I, Is, ßss, s’max) and sex differences (ΔSex). Estimates of males (N = 35) and females (N = 40) are shown with 95% confidence intervals obtained from bootstrapping. p-Values indicate the significance of the observed sex difference based on permutation tests. All estimates were calculated on relativized ]data.
| Sexual selection metric | Males (95% CIs) | Females (95% CIs) | ΔSex (95% CIs) | p-value |
|---|---|---|---|---|
| Opportunity for selection (I) | 0.56 (0.25, 0.91) | 0.19 (0.11, 0.28) | 0.35 (0.05, 0.73) | .036 |
| Opportunity for sexual selection (Is) | 0.36 (0.19, 0.53) | 0.09 (0.02, 0.18) | 0.26 (0.08, 0.46) | .007 |
| Bateman gradient (ßss) | 1.03 (0.71, 1.36) | 0.37 (-0.08, 0.97) | 0.67 (0.04, 1.23) | .045 |
| Jones’ index (s’max) | 0.60 (0.38, 0.87) | 0.10 (-0.02, 0.29) | 0.50 (0.21, 0.80) | <.001 |
| Sexual selection metric | Males (95% CIs) | Females (95% CIs) | ΔSex (95% CIs) | p-value |
|---|---|---|---|---|
| Opportunity for selection (I) | 0.56 (0.25, 0.91) | 0.19 (0.11, 0.28) | 0.35 (0.05, 0.73) | .036 |
| Opportunity for sexual selection (Is) | 0.36 (0.19, 0.53) | 0.09 (0.02, 0.18) | 0.26 (0.08, 0.46) | .007 |
| Bateman gradient (ßss) | 1.03 (0.71, 1.36) | 0.37 (-0.08, 0.97) | 0.67 (0.04, 1.23) | .045 |
| Jones’ index (s’max) | 0.60 (0.38, 0.87) | 0.10 (-0.02, 0.29) | 0.50 (0.21, 0.80) | <.001 |
Sex comparison of Bateman’s sexual selection metrics in Drosophila melanogaster. Bar plots show means of bootstrap estimates of (A) the opportunity for selection (I) and (B) the opportunity for sexual selection (Is) with 95% confidence intervals. Bateman gradients (C) are shown as slopes of linear regression of relativized reproductive success on relativized mating success (dashed line) with 95% confidence intervals (shaded area). Data for males in blue and for females in red.
Variance decomposition of male performance (see Figure 3) revealed that the largest fraction of variance in reproductive success can be attributed to individual differences in precopulatory processes with copulatory mating success accounting for 67.4% of the total opportunity for selection. By contrast, paternity share contributed only 10.3% to the variance in reproductive success. Permutations tests confirmed that variance in copulatory mating success was significantly larger compared with variance in paternity share (mean difference, 95% confidence limits: ΔcMS-PS = 0.57, 0.24 to 0.92, p = .013). Similarly, variance in relative copulatory mating success was larger than variance in mate’s fecundity (ΔcMS-mFec = 0.56, 0.26 to 0.92, p = .012), but we did not detect a difference of variances in paternity share and mate’s fecundity (ΔPS-mFec = 0.01, −0.08 to 0.10, p = .971). Covariance between copulatory mating success and paternity share was highly negative, but confidence limits overlapped slightly with zero (mean and 95% confidence limits: covcMS-PS = −0.06, −0.13 to 0.0001, p = .012).
Variance decomposition of male reproductive success. Bars show means of bootstrap estimates of the proportion of variance in male reproductive success (mRS) that can be attributed to variance in copulatory mating success (cMS), paternity share (PS), and the partners’ fecundity (mFec) and the covariances. Error bars indicate 95% confidence intervals.
Discussion
This study finds direct observational support for Bateman’s principles in D. melanogaster, where variation in both mating success and reproductive success was significantly higher in males than in females. Variation in male mating success ranged from zero mates to five mates within the 4-hr observation period. Conversely, the majority of females mated once, and only three individuals from a sample size of 40 mated twice. Variation in reproductive success was also positively correlated with mating success in males (the Bateman gradient), a relationship that was weaker and statistically nonsignificant in females. Furthermore, precopulatory rather than postcopulatory processes accounted for the majority of the variance in male reproductive success, indicating mating success was the most important component of reproductive fitness in males.
These results clearly show that sexual selection is operating more intensely in males; if mating success positively correlates with reproductive success, and therefore increases fitness, then any trait associated with mating success will have a positive selection differential (Jones, 2009). Furthermore, if mating success can be considered a trait, then Bateman gradients are a measure of selection on this trait. That is, mating success, in addition to any trait associated with mating success, will experience persistent, directional selection in males (Jones, 2009). In contrast, the sex with a shallow Bateman gradient, in this case females, will not be under strong sexual selection for any trait associated with mating success, as increased mating success does not positively correlate with increased reproductive success, and thus increased fitness (Arnold & Duvall, 1994; Jones, 2009).
This is further supported by standardized variance measures (Table 1). These figures indicate that males are experiencing greater opportunity for selection (I), greater opportunity for sexual selection (Is), and the maximum standardized selection differential (s’max, termed the “Jones index”) is much greater for males. The Jones index is particularly useful in examining the upper limit on the strength of precopulatory sexual selection because it combines information on the strength and direction of selection with an estimate of the variance of a sexual trait, which is the material on which selection can act. Therefore, s’max can be considered a more complete index with regards to the sexual selection process (Jones, 2009). These results provide strong support for Bateman’s conclusions that (a) variance in number of offspring is higher for males compared to females, (b) variance in number of mates is higher for males compared to females, and (c) the slope of the relationship between mating success and reproductive success is steeper in males.
These results suggest that the Darwin–Bateman paradigm holds for D. melanogaster, in which neither sex provides parental care for eggs or young (Clutton-Brock, 1991), meaning that parental investment includes solely the resources invested in the production of gametes. Thus, as Bateman concluded in his original paper (Bateman, 1948), gamete size is likely responsible for sex differences in reproductive competition and more intense sexual selection on males. In species that mate multiply, such as Drosophila, this creates sexual conflict between males and females, as males compete vigorously for mating opportunities and have evolved mechanisms for increasing the likelihood their sperm is used during fertilization (Bubis et al., 1998), while for females, extra copulations increase offspring genetic diversity (Arnqvist & Nilsson, 2000; Jennions & Petrie, 2000), and they are choosier of mates, having evolved mechanisms for controlling which male’s sperm are used (Bubis et al., 1998; Eberhard, 1996). However, this reduces the number of offspring an individual male sires, thus creating conflict in the optimal remating rate for females (Chapman et al., 2003; Morrow et al., 2005). In D. melanogaster, polyandry has been found to have no positive effects on female fecundity, hatching rate, or larval viability (Brown et al., 2004), which is in line with our finding of a nonsignificant female Bateman gradient. However, male ejaculate contains toxic components; therefore, remating shortens the female’s life span (Chapman et al., 1995). Males also employ chemical mate guarding strategies to prevent additional matings, marking females with unattractive pheromones postmating, to reduce their chances of remating (Ejima et al., 2007). Experimental evidence suggests that proteins in seminal fluid decrease the tendency of female remating and elevate sperm displacement (Chapman, 2001; Mane et al., 1983; Zawistowski & Richmond, 1986).
However, D. melanogaster females do typically mate multiply, both in experimental and natural settings, and wild-caught females have been found to store sperm from many males and produce offspring from up to five males (Imhof et al., 1998; Kuijper & Morrow, 2009; Ochando et al., 1996). Based on this evidence, the number of remating by females in this study is clearly limited, and as a result, does not provide strong evidence for Bateman’s conclusion that female remating does not increase reproductive success. Although variation in female mating success and reproductive success was much more limited compared to males, and males had a significantly positive Bateman gradient, therefore providing overall support for Bateman’s principles, a much greater sample size of additional female matings would need to be obtained to unequivocally support the conclusion that females do not benefit from additional matings.
There are many possible reasons why female remating rate was low in our experiment. Mating behavior was observed over 4-hr time periods, and studies of wild populations of Drosophila suggest that females remate once their sperm load is depleted (Gromko & Markow, 1993). Females store sperm in the paired spherical spermathecae and an elongated tubular seminal receptacle (Pitnick et al., 1999), which is then utilized to fertilize eggs. Subsequent to mating females undergo physiological and behavioral changes, such as increased ovulation and oviposition rates and decreased receptivity to further mating, that either have short-term effects (“copulation effect”) or long-term effects (“sperm effect”) (Fuyama, 1995; Singh & Singh, 2004; Wolfner, 1997). Evidence suggests that females become significantly less attractive to courting males twenty minutes after mating and that female receptivity to males returns much later than attractiveness to males returns (Gromko & Markow, 1993). This would in turn alter the operational sex ratio, such that in a given population the ratio of sexually competing males to females would likely be male-biased (Kvarnemo & Ahnesjö, 1996). Moreover, a longer delay in remating has been suggested to be influenced by the amount of sperm transferred during a first mating, with females that receive large amounts of sperm being less likely to remate sooner (Bubis et al., 1998; Letsinger & Gromko, 1985; Newport & Gromko, 1984).
Studies that implement the method of 2-hr periodic confinement, in which behavior is observed for 2 hr, females are isolated and then re-introduced the next day, experience higher levels of remating than observed in this study. This method of shorter mating assay periods implemented over multiple days probably allows female receptivity and attractiveness to return to “virgin-like” levels, therefore increasing the chances of remating (McRobert et al., 1997; Singh & Singh, 2004). Furthermore, continuous confinement combined with a high density of males can affect remating frequency. Whether males behave aggressively is dependent on the number of males within a group, and male-male aggression is particularly high in the presence of nonvirgin females and food (Hoffmann, 1987; Hoffmann & Cacoyianni, 1990; Wang et al., 2008). Gromko and Gerhart (1984) observed a reduction in remating frequency as population density increased to five pairs per vial under continuous confinement. Considering that experimental populations were held at 16 pairs per vial, it is expected that this pattern would be exacerbated. Interference among males during courtship attempts was observed frequently, much like Gromko and Gerhart (1984), suggesting that high densities of males reduce the ability of males to direct sustained courtship. It is therefore suggested that future studies testing Bateman’s principles in D. melanogaster should not only maintain experimental populations at the density by which they have evolved, to ensure behavior is representative of the population, but also implement confinement periods and observe behavior over a number of days, to achieve a more reasonable number of female remating that more closely matches that of wild populations.
An important aspect of this study that allowed for improved validation of results compared to Bateman’s original experiment was the direct observation of mating behavior. Although this allowed for a more accurate assessment of mating success alongside genetic parentage analysis, it was not completely lacking error. A small proportion of the male data had an inaccurate recording of mating success based on observation of mating behavior, and genetic parentage analysis indicated an additional mating based on the number of females within the population that produced offspring displaying the male phenotype. This is assumed to be a slight error that can be rectified by comparing against genetic parentage analysis, but it highlights the difficulty in assessing accurate measures of mating success based on behavior observation alone. This study goes some way to rectifying methodological issues highlighted in Bateman’s original experiment (Gowaty et al., 2012, 2013), but for large observational populations or model species that are more difficult to observe behaviorally, such as D. melanogaster, genetic parentage analysis is a useful tool to consolidate mating observation.
In conclusion, this study finds support for Bateman’s principles and the theories outlined by Darwin (1871), with the opportunity for future study, particularly in regard to female remating and its influence on female Bateman gradients. Although Bateman’s method was flawed due to viability effects and a lack of behavioral observation to integrate into his analysis, it is evident from this study, together with previous studies that find support for Bateman’s principles, that his results and conclusions were not. It is clear that exceptions to this rule exist, and also that the factors that influence sex differences in reproductive competition and subsequently precopulatory sexual selection can be more complex than being purely driven by anisogamy. However, species for which there is sex-role reversal do not contradict Bateman’s principles, but support the Darwin–Bateman–Trivers’ paradigm, which has been evidenced to be observed overwhelming across the animal kingdom (Janicke et al., 2016). This study lends itself to that support, but future study that explores female reproductive success at higher levels of remating, and perhaps explores the concept of density-dependent remating, will further clarify the growing evidence that female D. melanogaster do not experience increased reproductive success as a result of increased mating success, and thus that males experience more intense sexual selection.
Data availability
Data and R code for analysis and plots is available at https://zenodo.org/record/7729031#.ZA9Z5uzMJQK.
Author contributions
E.H.M. and T.J. conceived the study. N.D. performed the experiments and collected the data. N.D. and T.J. performed the analysis. N.D., T.J., and E.H.M. drafted the paper, with all authors commenting on revisions.
Funding
Funding was provided by the Swedish Research Council (grant number: 2019–03567) and by the Royal Society to E.H.M. as a University Research Fellowship. T.J. was funded by the Centre National de la Recherche Scientifique (CNRS) and the German Research Foundation (DFG grant number: JA 2653/2-1).
Conflict of interest
The authors declare no conflict of interest.
Acknowledgment
We thank Miguel Gómez for comments on a draft manuscript.
