Within‐population variation in female mating preference affects the opportunity for sexual selection and the evolution of male traits, but things are not as simple as expected

Abstract

Females from the same population usually have phenotypic variation in their mating preferences. However, the effects of this within‐population variation on the sexual selection acting on males are still unclear. We used individual‐based models to explore how within‐population variation in female preference (i.e. which male trait value is preferred) and preference strength (i.e. how strong the preference is) affects the opportunity for sexual selection (Is) and the evolution of a sexually selected male trait. We found the highest Is values when females had high variation in preference and an open‐ended preference function. The lowest Is occurred when the magnitude of variation in female preference and male trait value were the same and preference function was closed. Male trait exaggeration was higher when there was high within‐population variation in preference and females had an open‐ended preference function. Also, higher male trait variation was maintained by high variation in preference, but only for a closed preference function. Thus, we found that only within‐population variation in female preference, not in preference strength, influences the opportunity for sexual selection and the evolution of sexually selected male traits. Moreover, we found that the shape of the preference function (i.e. open‐ended or closed) and the magnitude of within‐population variation in female preference compared to male trait variation also influences the Is and consequently the evolution of male traits.

Selection on sperm size in response to promiscuity and variation in female sperm storage organs

“…Our analytical model indicates that the change from stabilizing to disruptive selection should occur when the among-female preference variance is greater than the sum of the among-male variance in sperm traits and the variance parameter in the female preference function (Eq 1, which we do not vary in this simulation). Our simulation results, however, show stabilizing selection under more conditions than expected, also compared to previous models where females sampled a large subset of males (Millan et al ., 2020). Specifically, when the number of copulation partners is low, stabilizing selection can occur even with high among-female variation in preference.…”

“…To calculate absolute fit, we modeled the fit quality as a Gaussian function which attains its maximum when the male’s trait value, y , matches the female’s preference, x . The parameter σ U , akin to standard deviation, controls the strength of the preference (sensu Millan et al ., 2020). For the sake of simplicity, we use a value of σ U =1 in all simulations.…”

“…Females are assigned a preference (i.e., sperm storage organ size) on the same scale, such that the fit between sperm and preference is best when the trait values are equal. We varied population-level SD in female preference (values of 0.5, 1, 1.5, and 2; comparable to the variation explored by Millan et al . 2020), but, for simplicity, the population mean preference was always equal to the sperm preference mean.…”

“…See Millan et al (2020) for relevant work with a different assessment model, and Van Doorn et al, (2001) and van Doorn & Weissing (2002) for models relevant for broad-cast spawners with high mate sampling. Further work is thus needed to understand how variation in female sperm storage organs impacts selection on sperm.…”

“…Simultaneous assessment strategies can give different results from other assessment strategies (Janetos, 1980; Jennions & Petrie, 1997; Muniz & Machado, 2018), and to the best of our knowledge, continuous variation in female preferences has not been modeled with simultaneous assessment with a reasonable number of copulation partners (for an internally fertilizing species). See Millan et al . (2020) for relevant work with a different assessment model, and Van Doorn et al ., (2001) and van Doorn & Weissing (2002) for models relevant for broad-cast spawners with high mate sampling.…”