Genomic changes associated with adaptation to arid environments in cactophilic Drosophila species

Abstract

Background: Insights into the genetic capacities of species to adapt to future climate change can be gained by using comparative genomic and transcriptomic data to reconstruct the genetic changes associated with such adaptations in the past. Here we investigate the genetic changes associated with adaptation to arid environments, specifically climatic extremes and new cactus hosts, through such an analysis of five repleta group Drosophila species.

Results: We find disproportionately high rates of gene gains in internal branches in the species' phylogeny where cactus use and subsequently cactus specialisation and high heat and desiccation tolerance evolved. The terminal branch leading to the most heat and desiccation resistant species, Drosophila aldrichi, also shows disproportionately high rates of both gene gains and positive selection. Several Gene Ontology terms related to metabolism were enriched in gene gain events in lineages where cactus use was evolving, while some regulatory and developmental genes were strongly selected in the Drosophila aldrichi branch. Transcriptomic analysis of flies subjected to sublethal heat shocks showed many more downregulation responses to the stress in a heat sensitive versus heat resistant species, confirming the existence of widespread regulatory as well as structural changes in the species' differing adaptations. Gene Ontology terms related to metabolism were enriched in the differentially expressed genes in the resistant species while terms related to stress response were over-represented in the sensitive one.

Conclusion: Adaptations to new cactus hosts and hot desiccating environments were associated with periods of accelerated evolutionary change in diverse biochemistries. The hundreds of genes involved suggest adaptations of this sort would be difficult to achieve in the timeframes projected for anthropogenic climate change.

Keywords: Cactophilic Drosophila; Comparative genomics; Heat stress; Host adaptation; Transcriptomics.

Fig. 1

Phylogenetic relationships of the five repleta group species and 19 previously sequenced Drosophila species based on concatenated codon alignments of 1802 orthogroups shared by all species. The divergence time was estimated using the RelTime [102] package in MEGA7 [93]. All bootstrap values for nodes from 1000 iterations were equal to 1 and the ancestral repleta group branch is indicated with a hash

Fig. 2

Consensus phylogenetic tree estimated using IQ-Tree with 1802 genes and 1000 bootstraps for the repleta groups species showing the numbers of orthogoups arising in the internal branches ((blue) and generating inparalogues in the terminal branches (green), plus the total numbers of inparalogues in the terminal branches (brackets, green) and the number of positively selected genes in all branches (red), togther with plots of these four metrics against the lengths of the respective branches

Fig. 3

Summary of significant enrichments (FDR-corrected P < 0.05) of 23 sets of GO terms for biological processes in the 18 evolutionary analyses in Fig. 2 and in the over-represented species in the six key fuzzy-c means clusters from the heat stress transcriptome analyses (Fig. 5 below)

Fig. 4

GO subsets most represented in the 18 evolutionary analyses, where representation is expressed in terms of either the total number (top) or percentage (below) of genes in the subset involved in the respective gene gain or positive selection events. Only subsets containing more than ten genes were analysed and only those where more than ten genes (top) or more than 5% (below) were involved in the events in any one of the analyses are shown. These values are shaded. However for those subsets included on these criteria, any non-zero involvement in any of the other analyses is also shown, but not shaded. See the Methods section for further details of the methods used to generate this figure

Fig. 5

Fuzzy c-means clustering of genes that were differentially expressed (normalized values) in D. hydei and D. buzzatii. The y-axes are standardized expression values with mean = 0 and variance = 1

Fig. 6

GO subsets most represented among the genes in the species in excess that lack orthologues in the same cluster in the other species in the fuzzy-c means clusters in the differential expression analyses. Representation is expressed in terms of either the total number (top) or percentage (below) of genes from the subset in question in the species in excess (again excluding any that also had orthologues in the other species in the same cluster). Only subsets containing more than 10 genes were analysed and only those where more than ten genes (top) or more than 5% (below) were involved in any one of the six clusters analysed are shown. These values are shaded. However for those subsets included on these criteria, any non-zero involvement in any of the other analyses are also shown, but not shaded. See the Methods section for further details of the methods used to generate this figure