Selection, recombination and demographic history in Drosophila miranda

Abstract

Selection, recombination, and the demographic history of a species can all have profound effects on genomewide patterns of variability. To assess the impact of these forces in the genome of Drosophila miranda, we examine polymorphism and divergence patterns at 62 loci scattered across the genome. In accordance with recent findings in D. melanogaster, we find that noncoding DNA generally evolves more slowly than synonymous sites, that the distribution of polymorphism frequencies in noncoding DNA is significantly skewed toward rare variants relative to synonymous sites, and that long introns evolve significantly slower than short introns or synonymous sites. These observations suggest that most noncoding DNA is functionally constrained and evolving under purifying selection. However, in contrast to findings in the D. melanogaster species group, we find little evidence of adaptive evolution acting on either coding or noncoding sequences in D. miranda. Levels of linkage disequilibrium (LD) in D. miranda are comparable to those observed in D. melanogaster, but vary considerably among chromosomes. These patterns suggest a significantly lower rate of recombination on autosomes, possibly due to the presence of polymorphic autosomal inversions and/or differences in chromosome sizes. All chromosomes show significant departures from the standard neutral model, including too much heterogeneity in synonymous site polymorphism relative to divergence among loci and a general excess of rare synonymous polymorphisms. These departures from neutral equilibrium expectations are discussed in the context of nonequilibrium models of demography and selection.

Figure 1.—

Mean divergences for synonymous sites, small and large introns, and flanking regions between D. miranda and D. pseudoobscura for 106 neo-X-linked genes. Error bars indicate two standard errors. Synonymous site divergence is significantly greater than large (Wilcoxon's two-sample test, P < 0.0001) but not small intron divergences (Wilcoxon's two-sample test, P = 0.31). Large introns evolve significantly slower than short introns (Wilcoxon's two-sample test, P = 0.0038).

Figure 2.—

Frequency distribution of different types of polymorphisms. Low refers to a sample frequency f of , intermediate refers to a sample frequency of –, and high refers to a sample frequency of .

Figure 3.—

Estimates of the fraction of mutations driven to fixation by positive selection. Error bars indicate 90% confidence limits determined by a standard nonparametric bootstrapping procedure.

Figure 4.—

Approximate Bayesian posterior distributions of ρ/θ, for autosomal and neo-X- and X-linked loci. The modes and 95% C.I.'s are listed in Table 2.