Experimental sexual selection affects the evolution of physiological and life-history traits

Abstract

Sexual selection and sexual conflict are expected to affect all aspects of the phenotype, not only traits that are directly involved in reproduction. Here, we show coordinated evolution of multiple physiological and life-history traits in response to long-term experimental manipulation of the mating system in populations of Drosophila pseudoobscura. Development time was extended under polyandry relative to monogamy in both sexes, potentially due to higher investment in traits linked to sexual selection and sexual conflict. Individuals (especially males) evolving under polyandry had higher metabolic rates and locomotor activity than those evolving under monogamy. Polyandry individuals also invested more in metabolites associated with increased endurance capacity and efficient energy metabolism and regulation, namely lipids and glycogen. Finally, polyandry males were less desiccation- and starvation resistant than monogamy males, suggesting trade-offs between resistance and sexually selected traits. Our results provide experimental evidence that mating systems can impose selection that influences the evolution of non-sexual phenotypes such as development, activity, metabolism and nutrient homeostasis.

Keywords: experimental evolution; life-history evolution; metabolism; physiology; polyandry; sexual selection; trade-offs.

FIGURE 1

Kaplan–Meier plots for (a) juvenile development time, (b) desiccation resistance and (c) starvation resistance. ‘+’ indicates right censored individuals. Monogamy (M) = pink and light blue; elevated polyandry (E) = red and dark blue; males = dashed lines; females = solid lines. Shaded areas show confidence intervals

FIGURE 2

Effects of sex and selection treatment on metabolic rates and mediator variables. a) Posterior estimates of the means for mediator variables (body mass and activity). Note that females are larger than males, and that E females are somewhat larger than M females. b) Posterior estimates of effect sizes of selection treatment (E minus M) on metabolic rate (measured as volume of O2 consumed) controlling for activity, body mass or neither mediator variable. The x‐axes show the posterior estimate of standardized effect size (Cohen's d), that is, a value of 1 would mean that the E treatment has a mean that is larger by 1 standard deviation. The horizontal bars show the 66% and 95% quantiles and the median of the posterior distribution. All response variables have been mean‐centred and divided by the overall standard deviation (such that the dashed line at zero marks the mean across sexes, treatments and cycles). For brevity, only the first cycle is shown for activity and metabolic rate; see Supporting information for all cycles

FIGURE 3

Posterior estimates of the treatment effect size for both sexes, for each of the five metabolites. A positive value indicates that the mean metabolite content is higher in the E treatment than the M treatment, whereas a negative value denotes M > E. A strongly supported treatment effect is implied by most of the posterior lying to one side of zero. The vertical bars show the 66% and 95% quantiles and the median of the posterior distribution. This plot was created using posterior predictions of the means that were adjusted for differences in dry weight between treatments