Genome-wide variation and transcriptional changes in diverse developmental processes underlie the rapid evolution of seasonal adaptation

Abstract

Many organisms enter a dormant state in their life cycle to deal with predictable changes in environments over the course of a year. The timing of dormancy is therefore a key seasonal adaptation, and it evolves rapidly with changing environments. We tested the hypothesis that differences in the timing of seasonal activity are driven by differences in the rate of development during diapause in Rhagoletis pomonella, a fly specialized to feed on fruits of seasonally limited host plants. Transcriptomes from the central nervous system across a time series during diapause show consistent and progressive changes in transcripts participating in diverse developmental processes, despite a lack of gross morphological change. Moreover, population genomic analyses suggested that many genes of small effect enriched in developmental functional categories underlie variation in dormancy timing and overlap with gene sets associated with development rate in Drosophila melanogaster Our transcriptional data also suggested that a recent evolutionary shift from a seasonally late to a seasonally early host plant drove more rapid development during diapause in the early fly population. Moreover, genetic variants that diverged during the evolutionary shift were also enriched in putative cis regulatory regions of genes differentially expressed during diapause development. Overall, our data suggest polygenic variation in the rate of developmental progression during diapause contributes to the evolution of seasonality in R. pomonella We further discuss patterns that suggest hourglass-like developmental divergence early and late in diapause development and an important role for hub genes in the evolution of transcriptional divergence.

Keywords: adaptation; development; diapause; ecological genomics; phenology.

Fig. 1.

Progressive trajectories of transcript expression across time common to diapausing pupae of both the apple and hawthorn host race. Relative expression values are log2 fold changes relative to the 2-mo time point of each host race. (A) Heat map of all 2,578 transcripts significantly differentially expressed across time in both host races; columns A3–A6 and H3–H6 represent values of apple and haw at 3 to 6 mo relative to 2 mo. (B) Trajectories of transcript expression within each of the six modules identified by clustering of coexpression analysis of 2,578 transcripts differentially expressed across time, including number of transcripts (n) associated with each module. Select functional categories identified through enrichment analyses of all transcripts combined are listed. Trajectories represent the approximate 95% confidence intervals ($\pm$ 2 SE) of all genes up-regulated and down-regulated (average expression values above and below zero, respectively) over time. (C) Transcripts associated with the Wnt signaling pathway (subset with a maximum absolute log2 fold change >1) and (D) associated with neuron differentiation (cell fate genes) and pruning (ftz-f1; displayed values are averages of apple and haw flies).

Fig. 2.

CNS gene expression varies between host races. (A) DE was most pronounced (number of transcripts significantly DE) early and late in the overwintering trajectory, with relatively little DE at 3 and 4 mo. (B) Trajectories of transcript expression within each of seven modules identified by clustering analysis of 2,902 transcripts DE between apple and haw host races, including number of transcripts (n) associated with each module, and select functional categories identified through enrichment analyses of transcripts in select modules and of all transcripts combined (full enrichment results in Dataset S2). Expression values are log2 fold changes relative to the 2 mo time point in hawthorn pupae, and shaded trajectories represent the approximate 95% confidence intervals ($\pm$ 2 SE) of all genes up-regulated and down-regulated (expression values above and below zero, respectively) over time. (C) Transcripts involved in the OXPHOS pathway (I, ubiquinol-cytochrome c oxidoreductase and II, F0/F1 ATP synthase families), ribosomal proteins (III), and the core nucleosome proteins (IV) were DE in apple relative to haw flies. (D) Transcripts significantly DE between host races had greater connectivity. Panels I through III illustrate connectivity as network plots where circles are transcripts and connecting lines are edges depicting correlations of time series expression exceeding r = 0.94. Panel IV illustrates that transcripts DE between host races have significantly higher median hub scores; the plot compares empirical cumulative distribution functions (ECDFs).

Fig. 3.

Genome-wide distributions of SNPs. (A) Distribution of all host-associated (FDR < 0.05), and emergence-associated (FDR < 0.05 in apple and haw flies) SNPs within annotation categories. Fisher exact tests comparing counts in the genome vs. associated SNPs were performed within each category to test for higher- or lower-than-expected prevalence of associated SNPs. (B) Distribution of emergence-associated SNPs across chromosomes and LD classes (L = low, M = medium, H = high) in haw flies (Upper) and apple flies (Lower). χ² tests were performed to test whether associated SNPs were nonuniformly distributed across chromosomes and LD classes. Several key functional categories enriched in the set of SNPs associated with emergence in both host races are listed.