Sexual selection and sexual antagonism are important drivers of eco-evolutionary processes. The evolution of traits shaped by these processes depends on their genetic architecture, which remains poorly studied. Here, implementing a quantitative genetics approach using diallel crosses of the bulb mite, Rhizoglyphus robini, we investigated the genetic variance that underlies a sexually-selected weapon that is dimorphic among males and female fecundity. Previous studies indicated that a negative genetic correlation between these two traits likely exists. We found male morph showed considerable additive genetic variance, which is unlikely to be explained solely by mutation-selection balance, indicating the likely presence of large-effect loci. However, a significant magnitude of inbreeding depression also indicates that morph expression is likely to be condition-dependent to some degree and that deleterious recessives can simultaneously contribute to morph expression. Female fecundity also showed a high degree of inbreeding depression, but variance in female fecundity was mostly explained by epistatic effects, with very little contribution from additive effects. We found no significant genetic correlation, nor any evidence for dominance reversal, between male morph and female fecundity. The complex genetic architecture underlying male morph and female fecundity in this system has important implications for our understanding of the evolutionary interplay between purifying selection and sexually antagonistic selection.

Inbred lines of the bulb mite, Rhizoglyphus robini, were established from wild-caught individuals using full sib x sib mating for 14 generations. Males of the species are dimorphic, with fighter males expressing an enlarged and thickened third pair of legs which are used in combat with other males, in contrast, scrambler males have legs with all approximately equal thickness. The morph of each P-generation male was recorded and in subsequent generations of inbreeding, the same male morph was used, thus giving inbred lines fixed for fighter and scrambler alleles.

Using diallel crossing, males and females from each inbred line were crossed with other inbred lines. The offspring produced by each cross were either 1) individually isolated and the fecundity of F1 females determined or 2) reared in small groups and the frequency of each male morph recorded.

Using a Bayesian modeling approach, we partitioned the variance of each trait into additive effects, parental-sex effects, dominance effects and epistatic effects, plus noise. Furthermore, we investigated dominance relationships of these traits by calculating array covariances.

All analysis was performed in R, using packages litterDiallel, MCMCglmm, ggplot2 and lme4 (for their references please see main text).