Measuring Pre- and Post-Copulatory Sexual Selection and Their Interaction in Socially Monogamous Species with Extra-Pair Paternity

Abstract

When females copulate with multiple males, pre- and post-copulatory sexual selection may interact synergistically or in opposition. Studying this interaction in wild populations is complex and potentially biased, because copulation and fertilization success are often inferred from offspring parentage rather than being directly measured. Here, I simulated 15 species of socially monogamous birds with varying levels of extra-pair paternity, where I could independently cause a male secondary sexual trait to improve copulation success, and a sperm trait to improve fertilization success. By varying the degree of correlation between the male and sperm traits, I show that several common statistical approaches, including univariate selection gradients and paired t-tests comparing extra-pair males to the within-pair males they cuckolded, can give highly biased results for sperm traits. These tests should therefore be avoided for sperm traits in socially monogamous species with extra-pair paternity, unless the sperm trait is known to be uncorrelated with male trait(s) impacting copulation success. In contrast, multivariate selection analysis and a regression of the proportion of extra-pair brood(s) sired on the sperm trait of the extra-pair male (including only broods where the male sired ≥1 extra-pair offspring) were unbiased, and appear likely to be unbiased under a broad range of conditions for this mating system. In addition, I investigated whether the occurrence of pre-copulatory selection impacted the strength of post-copulatory selection, and vice versa. I found no evidence of an interaction under the conditions simulated, where the male trait impacted only copulation success and the sperm trait impacted only fertilization success. Instead, direct selection on each trait was independent of whether the other trait was under selection. Although pre- and post-copulatory selection strength was independent, selection on the two traits was positively correlated across species because selection on both traits increased with the frequency of extra-pair copulations in these socially monogamous species.

Keywords: cryptic female choice; evolution; extra-pair paternity; fertilization; gametes; mating systems; passerines; post-copulatory sexual selection; pre-copulatory sexual selection; reproduction; sperm competition.

Figure 1

Schematic representation of simulation procedure and explanation of aims. Males (or the eggs they fertilize) are represented by unique colors. The male trait is indicated by a circle of varying size and the sperm trait by an arrow of varying length and width. The circle and arrow are dashed when they do not indicate specific individuals. Populations had either only pre-copulatory selection active (Steps B-2 and C-1), only post-copulatory selection active (Steps B-1 and C-2), or both episodes active (Steps B-2 and C-2). For each of these combinations, varying correlations between the male trait and sperm trait were tested. The female was assigned a number of extra-pair copulations (EPCs), E, by drawing from a Poisson distribution with species-specific mean m; E did (B-2) or did not (B-1) depend on the within-pair male’s male trait. From the available extra-pair males (four shown here, for illustration), E actual copulation partners were drawn, randomly (B-1) or depending on the extra-pair male’s male trait (B-2). The probability that each egg was fertilized by extra-pair sperm was calculated, either depending only on E and a species-specific term s (C-1, Equation (1)), or depending on E, s, and the relative sperm quality of the within-pair male and extra-pair copulation partners (C-2, Equation (2)). For extra-pair eggs, the sire was assigned from among the extra-pair copulation partners either randomly (C-1) or depending on their sperm trait (C-2). Illustrations of available sperm for population C-1 are from step B-1, and for C-2 from B-2. Aims are explained, with reference to methods sections and results text, figures and/or tables.

Figure 2

Spurious significant results for male traits (i.e., detected selection in individual populations where the male trait did not affect copulation success) were relatively uncommon for most analytical approaches. Pink colors indicate estimated negative selection and green colors indicate positive selection; intense colors are significant test results, while lighter colors are non-significant. Analytical approaches were (A): unpaired t-tests of males that lost paternity within their own broods (were cuckolded) or did not; (B): unpaired t-tests of males that gained extra-pair young (EPY) in other broods or did not; (C): paired comparisons of extra-pair sires to the within-pair males that they cuckolded; (D): univariate selection analysis regressing offspring sired on the male trait; and (E): multivariate selection analysis regressing offspring sired on both traits (with test results shown for the male trait only). Type 1 error is expected to produce only 5% (75) significant tests per column.

Figure 3

Spurious significant results for sperm traits (i.e., detected selection in individual populations where the sperm trait did not affect fertilization success) were common for most analytical approaches when the sperm trait correlated to a male trait under selection. Pink colors indicate estimated negative selection and green colors indicate positive selection; intense colors are significant test results, while lighter colors are non-significant. Analytical approaches were (A): unpaired t-tests of males that lost paternity within their own broods (were cuckolded) or did not; (B): unpaired t-tests of males that gained extra-pair young (EPY) in other broods or did not; (C): paired comparisons of extra-pair sires to the within-pair males that they cuckolded; (D): univariate selection analysis regressing offspring sired on the sperm trait; (E): multivariate selection analysis regressing offspring sired on both traits (with test results shown for the sperm trait only); (F) regression of the sperm trait on the number of detected extra-pair copulation partners; (G) regression of proportion of extra-pair brood sired on the sperm trait of extra-pair males, including only broods where the male sired at least one extra-pair young; and (H): regression of the proportion of within-pair young (WPY) on the sperm trait of the within-pair male, including only broods where the male lost paternity. Type 1 error is expected to produce only 5% (75) significant tests per column.

Figure 4

The strength of direct selection on male traits (top row (A,B)) or sperm traits (bottom row (C,D)) depending on the species’ value of m (left column; Poisson-distributed mean number of extra-pair copulation partners) and s (right column; variable affecting the likelihood that an extra-pair copulation results in fertilization). Shaded areas represent 95% confidence interval. Direct selection strength was estimated as the mean value from the multivariate selection analysis from populations where both traits were under selection, across all levels of male trait vs. sperm trait correlation. Raw values are presented; statistical tests controlled for phylogenetic relatedness.