Abstract
In species with large population sizes such as Drosophila, natural selection may have substantial effects on genetic diversity and divergence. However, the implications of this widespread non-neutrality for standard population genetic assumptions and practices remain poorly resolved. Here, we assess the consequences of recurrent hitchhiking (RHH), in which selective sweeps occur at a given rate randomly across the genome. We use forward simulations to examine two published RHH models for D. melanogaster, reflecting relatively common/weak and rare/strong selection, respectively. We find that unlike the rare/strong RHH model, the common/weak model entails a substantial degree of Hill-Robertson interference, which has implications for the rate of beneficial mutation and for the simulation of RHH models. We also find that the common/weak RHH model is more consistent with our genome-wide estimate of the proportion of substitutions fixed by natural selection between D. melanogaster and D. simulans (19%). Finally, we examine how these models of RHH might bias demographic inference. We find that RHH has relatively minor effects on the inference of recent between-population demographic parameters, while having stronger effects on inference of longer term ancestral population parameters. Thus, even for species with important genome-wide impacts of selective sweeps, neutralist demographic inference can have some utility in understanding the histories of recently-diverged populations.