View Program - 2022 Population, Evolutionary, and Quantitative Genetics Conference

Selection on new, deleterious mutations affects the evolutionary dynamics of nearby linked loci. While the classic background selection (BGS) model (e.g. Charlesworth et al. 1993, Nordborg et al. 1996) accurately predicts variation under relatively strong selection and weak mutation rates, this model breaks down under weak selection and/or high mutation rates. A number of studies have tried to quantify the impact of BGS using a summary statistic, B, which is the expected reduction in pairwise diversity due to purifying selection (McVicker et al. 2009). B varies along a chromosome as a function of the genetic map and the spatial arrangement of the putative selected sites. B has also been used as the basis for maximum likelihood-based estimators of the strength of selection and deleterious mutation rates. Using forward-time simulations, we show that these commonly-used maximum likelihood-based methods lead to erroneous estimates in the weak selection regime, even under the simplest demographic scenarios. We find weak selection also biases our estimates of levels of diversity in the absence of linked selection, an important quantity used to distinguish the effects of drift from natural selection. Furthermore, we explore alternative, tree-based methods to find regions where BGS models may not be suitable and apply these to inferred tree sequences in humans. Finally, we show that these two selection regimes have different implications for genetic load in a species,

When B is shaped like a U: is weak selection on deleterious alleles important?