Comparative and population genomics approaches reveal the basis of adaptation to deserts in a small rodent

Abstract

Organisms that live in deserts offer the opportunity to investigate how species adapt to environmental conditions that are lethal to most plants and animals. In the hot deserts of North America, high temperatures and lack of water are conspicuous challenges for organisms living there. The cactus mouse (Peromyscus eremicus) displays several adaptations to these conditions, including low metabolic rate, heat tolerance, and the ability to maintain homeostasis under extreme dehydration. To investigate the genomic basis of desert adaptation in cactus mice, we built a chromosome-level genome assembly and resequenced 26 additional cactus mouse genomes from two locations in southern California (USA). Using these data, we integrated comparative, population, and functional genomic approaches. We identified 16 gene families exhibiting significant contractions or expansions in the cactus mouse compared to 17 other Myodontine rodent genomes, and found 232 sites across the genome associated with selective sweeps. Functional annotations of candidate gene families and selective sweeps revealed a pervasive signature of selection at genes involved in the synthesis and degradation of proteins, consistent with the evolution of cellular mechanisms to cope with protein denaturation caused by thermal and hyperosmotic stress. Other strong candidate genes included receptors for bitter taste, suggesting a dietary shift towards chemically defended desert plants and insects, and a growth factor involved in lipid metabolism, potentially involved in prevention of dehydration. Understanding how species adapted to deserts will provide an important foundation for predicting future evolutionary responses to increasing temperatures, droughts and desertification in the cactus mouse and other species.

Keywords: Peromyscus eremicus; bitter taste receptor; hyperosmotic stress; ribosomal protein; selective sweeps; thermal stress.

Figure 1

Diversity and differentiation in the cactus mouse. (a) MDS plot showing relative distance among individuals based on downsampling to a single base at 43.7 million variable sites. Note outlier from Motte on the far left side of the plot. (b) Manhattan plot showing patterns of differentiation based on F_ST between Motte and Deep Canyon Reserves (after outlier removal). (c) Manhattan plot showing patterns of nucleotide diversity π from all samples combined (after outlier removal)

Figure 2

Plot showing mean nucleotide diversity and number of indels as a function of chromosome length (p < .001 in both cases, albeit with opposite trends)

Figure 3

Species tree built in orthofinder2 from protein sequences of 18 species in the Myodonta clade (Order: Rodentia). Beside each species name are the number of gene families that underwent significant (p < .05) expansions (+) or contractions (−). In the yellow box are the functions and number of expanded/contracted gene families in the cactus mouse (Peromyscus eremicus) relative to the closest ancestral node

Figure 4

Scatterplot showing clusters representative of enriched GO terms after semantic reduction in REVIGO for (a) biological process GO terms, and (b) molecular function GO terms. Only the names of GO clusters with a p < 10^–2.5 are shown for visual clarity. The full list of genes and reduced GO terms in REVIGO can be found in Supplementary Tables S3–S5

Figure 5

(a) Density plots comparing distribution of π and (b) Tajima's D across the genome (in red) and across sweepfinder2 candidate regions only (in black). Values are calculated in 1 kb non‐overlapping windows along the genome. Dashed vertical lines show the means across the genome in red and across sweepfinder2 candidate regions only in black. Means across the genome and across sweepfinder2 candidate regions only are significantly different in both cases (p < .001)