A test of the background selection hypothesis based on nucleotide data fromDrosophila ananassae

We estimated DNA sequence variation within and between four populations of Drosophila ananassae at Om(1D) and vermilion (v) by using single-strand conformation polymorphism analysis and direct DNA sequencing. Om(1D) is located on the X chromosome in a region with a normal recombination rate; v is in a region of low recombination. In each population, levels of nucleotide diversity at v are reduced 10- to 25-fold relative to those at Om(1D). Divergence between D. ananassae and its sibling species D. pallidosa, however, is comparable for both loci. This lack of correlation between levels of polymorphism and divergence led to the rejection of a constant-rate, neutral model. To distinguish among alternative models, we propose a test of the background selection hypothesis based on the observed pattern of differentiation between populations. Although the degree of differentiation (measured by FST) among all pairs of subpopulations is similar at Om(1D), we found substantial differences at v. The two northern populations from Burma and Nepal are very homogeneous, whereas comparisons between northern and southern populations (e.g., between Nepal and middle India) produced large FST values. A coalescent-based simulation of the background selection model (in a geographically structured species with a finite number of demes) showed that the observed homogeneity among the northern populations is inconsistent with the background selection hypothesis. Instead, it may have been caused by a recent hitchhiking event that was limited to the northern species range.