Patterns of Genome-Wide Diversity and Population Structure in the Drosophila athabasca Species Complex

Abstract

The Drosophila athabasca species complex contains three recently diverged, prezygotically isolated semispecies (Western-Northern, Eastern-A, and Eastern-B) that are distributed across North America and share zones of sympatry. Inferences based on a handful of loci suggest that this complex might be an ideal system for studying the genetics of incipient speciation and the evolution of prezygotic isolating mechanisms, but patterns of differentiation have not been characterized systematically. Here, we assembled a draft genome for D. athabasca and analyze whole-genome re-sequencing data for 28 individuals from across the species range to characterize genome-wide patterns of diversity and population differentiation among semispecies. Patterns of differentiation on the X-chromosome vs. autosomes vary, with the X-chromosome showing better phylogenetic resolution and increased levels of between semispecies divergence. Despite low levels of overall differentiation and a lack of phylogenetic resolution of the autosomes for the most closely related semispecies, individuals do exhibit distinct genetic clustering. Demographic analyses provide some support for a model of isolation with migration within D. athabasca, with divergence times <20 kya. The young divergence times of the semispecies of D. athabasca, together with strong levels of sexual isolation, makes them a promising system for studying the evolution of prezygotic isolation and speciation.

Keywords: Drosophila; population structure; prezygotic isolation; speciation.

Fig. 1

Overview of the D. athabasca semispecies complex and collection locations. (a) Semispecies are morphologically identical, but exhibit semispecies-specific courtship songs most easily quantified by differences in interpulse interval (IPI; the time from the end of a pulse to the start of the next) at High-Repetition-Rate (HRR) bursts. The average IPI along with standard deviations for each semispecies is indicated underneath a typical waveform. Western-Northern and Eastern-B exhibit similar IPIs, however their ranges do not overlap in nature (b). Semispecies ranges are depicted by different colors, Western-Northern (WN) = red, Eastern-A (EA) = blue, Eastern-B (EB) = green. Abbreviations indicate sampling locations (see supplementary table S1, Supplementary Material online for details).

Fig. 2

Population genetic estimates for D. athabasca. (a) Nucleotide diversity within semispecies, and (b) Dxy, and (c) Fst among semispecies for each Muller element, estimated across the genome in 10 kb windows.

Fig. 3

Phylogenetic relationships and population structure among D. athabasca semispecies. Maximum likelihood trees for non-overlapping 50 kb windows along the genome and results from ADMIXTURE analysis for the (a) autosomes and (b) X chromosome. Individuals are color coded by semispecies song type.

Fig. 4

Principal component analysis of X-linked and autosomal SNPs in D. athabasca. (a) Principal component analysis of autosomal SNPs. First principal component shows a clear separation of WN (red) from both Eastern semispecies but no differentiation between EA (blue) and EB (green). PC2 shows separation of EA and EB, while PC3 suggests geographic structure within WN. (b) Principal component analysis of X-linked SNPs shows similar patterns as autosomal SNPs, however PC1, which separates WN from the two Eastern semispecies, explains a much larger amount of the variation.

Fig. 5

Shared and private X-linked and autosomal SNPs in D. athabasca. (a) Venn diagram of autosomal SNPs. Most SNP’s are private to each semispecies, followed by SNP’s shared among all three semispecies. (b) Venn diagram of X-linked SNPs. Most SNPs are private to each semispecies, followed by SNPs shared between the two Eastern semispecies.

Fig. 6

Demographic estimates for the D. athabasca complex. (a) Divergence model for D. athabasca demographic history along with ∂a∂i maximum likelihood estimates of population demographic parameters under an isolation with symmetric migration model inferred from (b) autosomal and (c) X-linked 4-fold synonymous sites. Mean, standard deviation, and 95% confidence intervals are derived from ms simulations. Migration rates are scaled per generation.