Evolutionary trajectories are determined by selection and the underlying genetic architecture of additive genetic variances and covariances for the traits. To better understand divergence in multivariate phenotypes among natural populations, we need population-specific estimates of selection and genetic architecture of the traits involved; such extensive data are challenging to acquire and few examples exist. Here we do this using males from six populations of the mushroom-feeding fly Drosophila subquinaria. We take advantage of a naturally occurring pattern of reproductive character displacement involving male chemical display traits (cuticular hydrocarbons/CHCs) between populations that are sympatric vs. allopatric with the closely related Drosophila recens. Genetic (co)variance structures for male CHCs from three populations from each region were previously quantified; here we provide novel estimates of sexual selection on these traits within each population via binomial mate choice trials, and we test whether this selection differs between regions. Then, using all these data we implement a comparative approach investigating how differences in sexual selection and genetic architecture can explain the observed pattern of CHC divergence, extending the method beyond past applications to incorporate uncertainty in estimates of selection and genetic (co)variances. We find modest differences in sexual selection between regions, in contrast to past studies, and that divergence is not well explained by this alone. However, explanatory ability improves substantially when genetic (co)variances among the traits are accounted for. We also find that predicted evolutionary responses to sexual selection in sympatry are likely constrained as the genetic ‘line of least resistance’ is not well aligned with selection.

We collected two data sets quantifying cuticular hydrocarbons in males from six populations of D. subquinaria, three sympatric with D. recens in nature and three allopatric to D. recens. CHCs were extracted by separately placing individuals in 100 μl of hexane for 3 min, followed by 1 min of vortexing before the fly was removed and discarded. CHCs were subsequently quantified via gas chromatography with flame ionization detection on an Agilent 6890N dual-channel “fast” (220 V oven) gas chromatograph (Agilent Technologies, Wilmington, DE). The area under each of ten peaks (i.e., unique CHC compounds that could be reliably quantified) was then integrated to quantify their abundances. Relative abundance (i.e., proportional concentrations) of each peak were taken and subsequently logcontrast transformed by dividing the proportional area of each peak by the proportional area of an arbitrarily selected common divisor (peak 5 in this case) and taking the log₁₀ of this quotient. Additional details can be found in the associated publication: Jarvis & Rundle (2026).

Breeding designs. A half-sibling breeding design was conducted separately in each population and CHCs were extracted from three male offspring per half-sib family. These data were previously used in a separate publication to characterized the phenotypic divergence in CHCs as well as the additive genetic (co)variance structure (G-matrix) for this set of traits: see Jarvis et al. (2025), https://doi.org/10.1093/evolut/qpaf009; data: https://doi.org/10.5061/dryad.tdz08kq8d  

Mating data. Binomial mate choice trails were conducted separately for each of the six populations. Each trial had a single virgin female from one of the populations and two virgin males from a standard, reference population. Vials were observed until intromission occurred with one of the two males, at which point all flies were anesthetized with light CO₂. CHCs were extracted from either the successful or unsuccessful male from a given trial (randomly predetermined, 50:50 ratio) and the other male, and the female, were discarded. If mating did not occur within 30 min., the trial was terminated and the flies were discarded.