Quantifying the Variation in the Effective Population Size Within a Genome

The effective population size (N-e) is one of the most fundamental parameters in population genetics. It is thought to vary across the genome as a consequence of differences in the rate of recombination and the density of selected sites due to the processes of genetic hitchhiking and background selection. Although it is known that there is intragenomic variation in the effective population size in some species, it is not known whether this is widespread or how much variation in the effective population size there is. Here, we test whether the effective population size varies across the genome, between protein-coding genes, in 10 eukaryotic species by considering whether there is significant variation in neutral diversity, taking into account differences in the mutation rate between loci by using the divergence between species. In most species we find significant evidence of variation. We investigate whether the variation in N-e is correlated to recombination rate and the density of selected sites in four species, for which these data are available. We find that N-e is positively correlated to recombination rate in one species, Drosophila melanogaster, and negatively correlated to a measure of the density of selected sites in two others, humans and Arabidopsis thaliana. However, much of the variation remains unexplained. We use a hierarchical Bayesian analysis to quantify the amount of variation in the effective population size and show that it is quite modest in all species-most genes have an N-e that is within a few fold of all other genes. Nonetheless we show that this modest variation in N-e is sufficient to cause significant differences in the efficiency of natural selection across the genome, by demonstrating that the ratio of the number of nonsynonymous to synonymous polymorphisms is significantly correlated to synonymous diversity and estimates of N-e, even taking into account the obvious nonindependence between these measures.