Parallel Selection in Domesticated Atlantic Salmon from Divergent Founders Including on Whole-Genome Duplication-derived Homeologous Regions

Abstract

Domestication and artificial selection for desirable traits have driven significant phenotypic changes and left detectable genomic footprints in farmed animals. Since the 1960s, intensive breeding has led to the rapid domestication of Atlantic salmon (Salmo salar), with multiple independent events that make it a valuable model for studying early domestication stages and the parallel evolution of populations of different origins subjected to similar selection pressures. Some aquatic species, including Atlantic salmon, have undergone whole-genome duplication (WGD), raising the possibility that genetic redundancy resulting from WGD has contributed to adaptation in captive environments, as seen in plants. Here, we examined the genomic responses to domestication in Atlantic salmon, focusing on potential signatures of parallel selection, including those associated with WGD. Candidate genomic regions under selection were identified by comparing whole-genome sequences from aquaculture and wild populations across 2 independently domesticated lineages (Western Norway and North America) using a genome-wide scan that combined 3 statistical methods: allele frequencies (FST), site frequency (Tajima's D), and haplotype differentiation (XP-EHH). These analyses revealed shared selective sweeps on identical SNPs in major histocompatibility complex (MHC) genes across aquaculture populations. This suggests that a combination of long-term balancing selection and recent human-induced selection has shaped MHC gene evolution in domesticated salmon. Additionally, we observed selective sweeps on a small number of gene pairs in homeologous regions originating from WGD, offering insights into how historical genome duplication events may intersect with recent selection pressures in aquaculture species.

Keywords: MHC; aquaculture; artificial selection; domestication; ohnolog.

Fig. 1.

Sample distribution and genetic structure of Atlantic salmon lines from western Norway and North America. a) Geographic distribution of Atlantic salmon population in North America. Aquaculture samples were used from Gaspe (Gaspe, n = 16), St John River (n = 53), and Penobscot River (n = 11). Wild samples came from 8 rivers: Malbaie Charlevoix (n = 10), Bonaventure (n = 10), Petite riviere Cascapedia (n = 10), Laval (n = 10), De la Chaloupe (n = 10), Jupiter (n = 10), Du Vieux Fort (n = 10), and Saint-Paul (n = 10). b) Geographic distribution of Atlantic salmon population from Western Norway. Aquaculture samples come from MOWI Norwegian breeding line (n = 112). Wild counterparts were collected among 17 rivers: Vorma (Vorm, n = 2), Suldalslågen (Suld, n = 10), Vikedalselva i Vindafjord (Vike, n = 10), Etneelva (Etne, n = 2), Eidfjordvassdraget (Eidf, n = 2), Oselva i Os (Osel, n = 2), Loneelva i Osterøy (Lone, n = 2), Flåmselva (Flam, n = 10), Lærdalselva (Laer, n = 2), Årøy (Aroy, n = 10), Daleelva (Dale, n = 2), Flekkeelva (Flek, n = 2), Nausta (Naus, n = 10), Jølstra (Jols, n = 10), Oselvvassdraget (Osen, n = 10), Ryggelva (Rygg, n = 10), and Gloppenelva (Glop, n = 2). c) Scatterplot of PCA. PC1 and PC2 were calculated for all Atlantic salmon samples. d) Population genetic structure analysis. All samples were clustered into 6 and 9 genetic components (K = 6 and K = 9) according to the lowest cross-validation error.

Fig. 2.

Cross-population test on extended haplotype homozygosity (XP-EHH). Each dot is SNPs, while the x axis displays the position across the genome. The y axis represents the −log10(P-value the XP-EHH (cross population extended haplotype homozygosity) values. Diamonds indicate SNPs with significantly high XP-EHH values that overlap genes in aquaculture populations from Western Norway and North America, respectively. We adopted −log10(P-value) = 4 as the cutoff (dashed line) for significant XP-EHH values (Materials and Methods). Gray and black dots are other tested SNPs. Only SNPs with a -log10(P-value) >1 were plotted to ensure the computational efficiency. Labeled genes are genes with exact same SNPs under parallel selection (yellow) or selection on the homeologous regions, with matching colors in both populations showing the WGD-derived gene pairs. The suffix denotes the chromosomal location of the copy. The entire SNP list with significant XP-EHH is available in supplementary table S1, Supplementary Material online.

Fig. 3.

The MHC region under parallel selective sweeps in both continents. a) Zoomed-in view of XP-EHH results on the MHC loci on chromosomes 14 and 27, exhibiting parallel selective sweeps. The x axis shows the genomic position, and the y axis represents the −log10(P-value) of the XP-EHH (cross population extended haplotype homozygosity) values. Candidate SNPs with significantly high XP-EHH values (above the threshold −log10(P) = 4, indicated by the dashed line) are shown, with matching SNPs across populations indicating parallel selection signals. Genes under parallel selection or in homeologous regions with shared selective signals are highlighted accordingly, and genes with signals specific to one population are also labeled. b) Haplotype pattern in the MHC region (126 SNPs from chr14: 64326415-64339381), with each column representing a SNP and each row a phased haplotype from the four population groups. The group is labeled accordingly. c) Distribution of Tajima's D values for each population, shown as histograms. The vertical dashed line indicates the observed Tajima's D value at the MHC locus. Neutral expectations are derived from 500 randomly sampled size-matched regions per population.

Fig. 4.

Expression of genes under parallel selection or parallel selection on the homeologous regions in early development stages and mature tissues. Gene expression is represented as log-transformed TPM. The sweep panel indicates the gene that has undergone selective sweeps in North America or Western Norway. The left heatmap shows the gene expression levels during early development, from late gastrulation to the eyed egg stage. The right heatmap shows the gene expression levels across different tissues in the mature Atlantic salmon. We used publicly available RNA sequencing data from early development stages (AQUA-FAANG project No. PRJEB51855) and from various tissues of healthy fish (NCBI Project No. SRP011583) including brain, eye, gill, gut, head kidney, heart, kidney, liver, muscle, nose, ovary, caecum, skin, spleen, and testis of Atlantic salmon in Europe.