Genome-wide fixation under viability selection

Abstract

We investigate through simulations the effect of random sweepstakes, recurrent bottlenecks, dominance, and epistasis on selection a diploid population when the phenotype is determined by the genotype at a number of unlinked sites. We assume there are two types at each site, the wild type and a type conferring selective advantage. The focus is on identifying main trends in how the stated elements impact selection when selection acts on at least two unlinked site. We give examples of fixation trajectories under an array of scenarios, and record corresponding numerical estimates of the probability of fixation of the fit type at all the sites in question each time, and the time to fixation conditional on fixation. The results suggest that joint fixation of the fit type at the sites can occur in a number of scenarios, despite starting with the fit type in a single copy at each site. Random sweepstakes generally reduce both the probability and the time to fixation (conditional on fixation). Recurrent strong bottlenecks generally reduce the probability of fixation, but the effect of bottlenecks on the time to fixation (conditional on fixation), depends on the dominance mechanism in question. We identify an example of epistasis where fixation can occur, when in the absence of epistasis fixation does not occur at all or with much lower probability. Our results, coupled with an extensive analysis of genomic data of Atlantic cod(4), is further evidence for pervasive strong positive selection as being a key driving force in the evolution at least of highly fecund populations.