The relation between recombination rate and patterns of molecular evolution and variation in Drosophila melanogaster

Abstract

Genetic recombination associated with sexual reproduction increases the efficiency of natural selection by reducing the strength of Hill-Robertson interference. Such interference can be caused either by selective sweeps of positively selected alleles or by background selection (BGS) against deleterious mutations. Its consequences can be studied by comparing patterns of molecular evolution and variation in genomic regions with different rates of crossing over. We carried out a comprehensive study of the benefits of recombination in Drosophila melanogaster, both by contrasting five independent genomic regions that lack crossing over with the rest of the genome and by comparing regions with different rates of crossing over, using data on DNA sequence polymorphisms from an African population that is geographically close to the putatively ancestral population for the species, and on sequence divergence from a related species. We observed reductions in sequence diversity in noncrossover (NC) regions that are inconsistent with the effects of hard selective sweeps in the absence of recombination. Overall, the observed patterns suggest that the recombination rate experienced by a gene is positively related to an increase in the efficiency of both positive and purifying selection. The results are consistent with a BGS model with interference among selected sites in NC regions, and joint effects of BGS, selective sweeps, and a past population expansion on variability in regions of the genome that experience crossing over. In such crossover regions, the X chromosome exhibits a higher rate of adaptive protein sequence evolution than the autosomes, implying a Faster-X effect.

Keywords: Drosophila melanogaster; Hill–Robertson interference; background selection; crossing over; heterochromatin; recombination; selective sweeps.

Fig. 1.

Ratio of the number of derived nonsynonymous mutations per nonsynonymous site to the number of synonymous mutations per synonymous site, for three categories of frequencies of derived variants. AC, autosomal crossover regions; XC, X chromosome crossover regions; NA, autosomal NC regions; NX, X chromosome NC regions.

Fig. 2.

Correlations between diversity statistics and the numbers of sites in coding sequences in the five NC regions for nonsynonymous diversity π_A, synonymous diversity π_S, and the ratio π_A/π_S. ρ, Spearman’s rank correlation coefficient, with significance denoted by asterisks (***<0.001; *<0.05).

Fig. 3.

B values for the five NC regions (red dots) against the number of coding sequence sites in each region. The blue line shows the effects of HRI on B due to BGS, predicted by Kaiser and Charlesworth (2009). The error bars are the standard errors of B obtained from the diversity statistics for the NC regions as described for table 2.

Fig. 4.

Relations between the effective recombination rate and the means of several variables for genes in the C regions, after grouping genes into bins defined by rates of crossing over. The x axis gives the mean effective recombination rate (cM/Mb) for each bin. Autosomal genes (A) are shown in green and X-linked (X) genes in red. Values for NC regions are indicated by the filled points at the extreme left of each panel but are not included in the correlation or regression analyses (black: the five NR blocks; green: autosomal NC genes; red: X-linked NC genes). ρ: Spearman’s rank correlation coefficient, with significance denoted by asterisks (***<0.001; **<0.01; *<0.05). The lines are least-squares regressions but should be regarded only as indicative, in view of the binning of the data.