Revealing the selection history of adaptive loci using genome-wide scans for selection: an example from domestic sheep

Abstract

Background: One of the approaches to detect genetics variants affecting fitness traits is to identify their surrounding genomic signatures of past selection. With established methods for detecting selection signatures and the current and future availability of large datasets, such studies should have the power to not only detect these signatures but also to infer their selective histories. Domesticated animals offer a powerful model for these approaches as they adapted rapidly to environmental and human-mediated constraints in a relatively short time. We investigated this question by studying a large dataset of 542 individuals from 27 domestic sheep populations raised in France, genotyped for more than 500,000 SNPs.

Results: Population structure analysis revealed that this set of populations harbour a large part of European sheep diversity in a small geographical area, offering a powerful model for the study of adaptation. Identification of extreme SNP and haplotype frequency differences between populations listed 126 genomic regions likely affected by selection. These signatures revealed selection at loci commonly identified as selection targets in many species ("selection hotspots") including ABCG2, LCORL/NCAPG, MSTN, and coat colour genes such as ASIP, MC1R, MITF, and TYRP1. For one of these regions (ABCG2, LCORL/NCAPG), we could propose a historical scenario leading to the introgression of an adaptive allele into a new genetic background. Among selection signatures, we found clear evidence for parallel selection events in different genetic backgrounds, most likely for different mutations. We confirmed this allelic heterogeneity in one case by resequencing the MC1R gene in three black-faced breeds.

Conclusions: Our study illustrates how dense genetic data in multiple populations allows the deciphering of evolutionary history of populations and of their adaptive mutations.

Keywords: Adaptation; Allelic heterogeneity; Introgression; Selective sweeps.

Fig. 1

Population structure of French sheep. a Average individual ancestry coefficients within breed for four ancestral populations. Breeds are mapped to their geographical origin. Individual ancestry coefficients for (b) four ancestral populations and c 16 ancestral populations. d Maximum likelihood population tree with four migration events

Fig. 2

Maximum likelihood population tree of European sheep breeds estimated using 40 K SNPs. Breeds from this study dataset are in black. Sheep HapMap breeds are from Northern Europe (in grey), Britain (in green), Switzerland (in blue) and West and Middle Mediterranean (in gold)

Fig. 3

Posterior probabilities of the number of independent SNPs in selection signatures. Posterior mean numbers of independent variants are indicated in red. Selection signatures included in this analysis were detected using FLK and all French breeds

Fig. 4

Chromosome 14:14069040 to 14530851. a Local tree and b haplotype cluster frequency plots for Noire du Velay (NVE), Romanov (ROM), Suffolk (SUF) and Texel (TEX) breeds for chromosome 14:14,069,040 to 14,530,851 (with the position of MC1R marked with the black line) and c mutation location and their frequencies in Noire du Velay (NVE), Romanov (ROM), Suffolk (SUF) and Texel (TEX) breeds

Fig. 5

Manhattan plots of hapFLK and FLK results in northern (a) and southern (b) populations. In each panel, the positive values are the -log10(p-value) of the hapFLK test and the negative values are the log10(p-value) of the FLK test. Significant SNPs (FDR < 0.05 for hapFLK and < 0.01 for FLK) are shown in darker colours

Fig. 6

Selection signature around genes ABCG2, NCAPG and LCORL on chromosome 6. a Population tree (left) constructed in the region under selection with haplotype cluster plots shown for each breed (right). hapFLK and FLK p-values for the selection signature in the northern (b) and southern (c) populations