Genomic differentiation between temperate and tropical Australian populations of Drosophila melanogaster

Abstract

Determining the genetic basis of environmental adaptation is a central problem of evolutionary biology. This issue has been fruitfully addressed by examining genetic differentiation between populations that are recently separated and/or experience high rates of gene flow. A good example of this approach is the decades-long investigation of selection acting along latitudinal clines in Drosophila melanogaster. Here we use next-generation genome sequencing to reexamine the well-studied Australian D. melanogaster cline. We find evidence for extensive differentiation between temperate and tropical populations, with regulatory regions and unannotated regions showing particularly high levels of differentiation. Although the physical genomic scale of geographic differentiation is small--on the order of gene sized--we observed several larger highly differentiated regions. The region spanned by the cosmopolitan inversion polymorphism In(3R)P shows higher levels of differentiation, consistent with the major difference in allele frequencies of Standard and In(3R)P karyotypes in temperate vs. tropical Australian populations. Our analysis reveals evidence for spatially varying selection on a number of key biological processes, suggesting fundamental biological differences between flies from these two geographic regions.

Figure 1.—

Genome-sequence coverage is equivalent across chromosome arms in normally recombining regions and more variable in low-recombining regions. Mean sequencing coverage is plotted for Queensland (blue) and Tasmania (red) populations. Dark colors indicate regions of normal recombination; lighter colors indicate low-recombining centromeric and telomeric regions. Bars give standard error.

Figure 2.—

Size of differentiated regions is similar in areas of normal and low recombination and larger on chromosome 3R. We calculated mean F_ST in nonoverlapping 1-kb windows across the D. melanogaster genome. Groups of windows in the top 1% tail of the F_ST distribution were grouped together into larger differentiated regions separated from one another by at least five consecutive windows with mean F_ST in the bottom 90% tail (see materials and methods). (a) We plot the size distribution of these differentiated regions for normally recombining (blue) and low-recombining (gray) areas of the genome. Bars indicate standard error. (b) We plot the size distribution of differentiated regions found in normally recombining regions of chromosome 3R (blue) and the size distribution of differentiated regions in normally recombining regions of other chromosome arms (gray). (c) We plot mean F_ST (bottom) and mean polymorphism (π, top) across chromosome 3R. Blue lines indicate average values over 25-kb windows slid every 10 kb; red lines show 200-kb windows slid 50 kb at a time. The gray box indicates the location of the cosmopolitan 3R-Payne inversion.

Figure 3.—

Largest highly differentiated regions occurred at the tip of the X chromosome (a) and in the middle of chromosome 2R (b). Highly differentiated regions are indicated in gray. We plot mean F_ST across each chromosomal region, blue lines indicating 10-kb windows with 1-kb slides and red lines indicating 50-kb windows with 20-kb slides. Annotated genes are drawn across the top of each panel.

Figure 4.—

Regions of high population differentiation localize within the Sfmbt gene on chromosome 2L. We plot individual-position F_ST (blue) and mean F_ST within 1-kb windows (red) across the chromosome. The red dotted line indicates F_ST cutoff for the top 2.5% of 1-kb windows. Individual genes are drawn across the top (black); exons are in blue, 3′-UTRs in light gray, and 5′-UTRs in dark gray.

Figure 5.—

3′-UTRs and unannotated regions are overrepresented in the most-differentiated genomic regions. We calculated the enrichment for each annotation type in the 1% (a), 2.5% (b), and 5% (c) tail of 1-kb F_ST regions, relative to each type's distribution across all 1-kb windows in the normally recombining portion of the genome. Results are shown separately for autosomes and the X chromosome. An enrichment score of 1.0 (indicated by a solid vertical line) indicates no enrichment or depletion; values >1 indicate an overabundance of that type in the F_ST tail, whereas values <1 indicate underabundance.

Figure 6.—

Elevated differentiation between Queensland and Tasmania populations localizes to the 3′-UTR of the Hex-t2 gene. We plot the F_ST of individual genomic positions against the structure of the Hex-t2 gene. Exons are drawn in black, the 5′-UTR is dark gray, and the 3′-UTR is light gray. The bottom panel shows predicted secondary structures of Queensland and Tasmania 3′-UTR regions. Queensland positions indicated by arrows are polymorphic, with the major allele at left; corresponding positions in Tasmania are fixed for what is the minor allele in Queensland.

Figure 7.—

Elevated nonsynonymous F_ST in two melanogaster protein-coding genes. We plot individual-position F_ST along the gene structure. Exons are drawn in black, the 5′-UTR is dark gray, and the 3′-UTR is light gray. Nonsynonymous polymorphisms are shown in red; synonymous and noncoding polymorphisms are shown in blue. (a) A nonsynonymous fixed difference between Queensland and Tasmania is associated with elevated F_ST at the txl gene. (b) Elevated F_ST at a fixed protein-coding change in Irc. (c) Structural homology models of Queensland (orange) and Tasmania (turquoise) Irc; the V317I substitution is potentially involved in direct ligand interaction.

Figure 8.—

Elevated nonsynonymous differentiation in NtR localizes to the major immunogenic region (MIR) of the ligand-binding domain (LBD). (a) We plot positional F_ST across gene structure, with exons drawn in black, 5′-UTR in dark gray, and 3′-UTR in light gray; methyltransferase and ligand-binding domains are indicated by green and red, respectively. Nonsynonymous polymorphisms are shown by red circles. (b) We plot highly differentiated E/D and I/V polymorphisms on the predicted 3D structure of the NtR LBD. In both cases, the major allele in Queensland (E, I) is shown in orange, and the major allele in Tasmania (D, V) is shown in turquoise.

Figure 9.—

Coordinated differentiation in norpA (a) and the 3′-UTR of per (b), a known target of norpA splicing regulation. We plot individual-position F_ST along the gene structure. Exons are drawn in black, the 5′-UTR is dark gray, and the 3′-UTR is light gray.

Figure 10.—

A large region of increased copy number in Queensland occurs on chromosome 3R. We plot the average number of sequence reads for each 1-kb window across this region, both for the Queensland (blue) and for the Tasmania (red) populations. Genes in this region are drawn across the top. The gray box indicates the inferred region of increased copy number in Queensland.