A genome-wide SNP genotyping array reveals patterns of global and repeated species-pair divergence in sticklebacks

Abstract

Genes underlying repeated adaptive evolution in natural populations are still largely unknown. Stickleback fish (Gasterosteus aculeatus) have undergone a recent dramatic evolutionary radiation, generating numerous examples of marine-freshwater species pairs and a small number of benthic-limnetic species pairs found within single lakes [1]. We have developed a new genome-wide SNP genotyping array to study patterns of genetic variation in sticklebacks over a wide geographic range, and to scan the genome for regions that contribute to repeated evolution of marine-freshwater or benthic-limnetic species pairs. Surveying 34 global populations with 1,159 informative markers revealed substantial genetic variation, with predominant patterns reflecting demographic history and geographic structure. After correcting for geographic structure and filtering for neutral markers, we detected large repeated shifts in allele frequency at some loci, identifying both known and novel loci likely contributing to marine-freshwater and benthic-limnetic divergence. Several novel loci fall close to genes implicated in epithelial barrier or immune functions, which have likely changed as sticklebacks adapt to contrasting environments. Specific alleles differentiating sympatric benthic-limnetic species pairs are shared in nearby solitary populations, suggesting an allopatric origin for adaptive variants and selection pressures unrelated to sympatry in the initial formation of these classic vertebrate species pairs.

Figure 1. Levels of heterozygosity and patterns of genetic variation among global stickleback populations

(A) Marine (red) and freshwater (blue) sticklebacks were sampled across the northern hemisphere. Benthic (green) and limnetic (yellow) species-pairs are restricted to a few lakes in British Columbia. See Table S1 for population information. (B) Mean heterozygosity levels (2pq +/− SEM) within freshwater populations are significantly lower than marine populations. (C) Principal component analysis of global populations reveals the first major axis of variation separating Pacific (left) from Atlantic populations (right cluster), and a second axis separating Northern Pacific freshwater populations from Southern Pacific freshwater and Pacific marine populations. SNP ascertainment vectors are shown as grey arrows (see Experimental Procedures). See also Figure S1, and Tables S1 & S2.

Figure 2. Bayesian scan for genomic regions consistently differentiated between marine and freshwater environments

(A) Bayes factor scores for all markers in the genome, with the top scoring regions (ATPase, EDA, and Mucin gene cluster regions on chromosomes I, II, and IV, respectively) indicated in red. (All SNPs with log₁₀ Bayes factors >=1.5 are listed in Table S3) (B) Bayes factors scores found when Pacific and Atlantic Ocean sticklebacks are analyzed as separate regional groups. SNPs with Bayes factors >2.5 in either population are labeled with corresponding gene region in red. (C–K) Allele frequencies in each of the 34 global populations for top ranking SNPs highlighted in panels A and B (freshwater populations: blue points; marine populations: red points; mean freshwater and marine allele frequency: blue and red dashed lines, respectively; Pacific and Atlantic populations grouped on the left and right side of the vertical solid grey line).

Figure 3. F_ST outlier analyses in benthic–limnetic populations

(A) Allele frequency of F_ST outlier SNPs are shown in pie charts for three species-pair lakes. Asterisks and dark outlines around pies indicate lakes in which SNP was found to be an F_ST outlier. Four SNPs appeared to be under selection in all three lakes (red lines, corresponding gene regions indicated) and a further 11 SNPs in two of three lakes (blue lines). (B) Genome graph of allele frequency differences between pooled ecotypes indicating chromosomal position of F_ST outliers. See Table S4 and Figure S2 for more details.

Figure 4

Principal component analyses of benthic (green) and limnetic (yellow) individuals using (A) all 925 polymorphic SNPs, (B) 879 putatively neutral SNPs or (C) 46 putatively selected SNPs. Ellipses represent 95% onfidence intervals for each ecotype within a lake. Global populations projected onto benthic–limnetic variation space in B and C are shown as red (marine) and blue (freshwater) triangles, with the benthic-like or limnetic-like populations labeled with a number code. See Table S1 for population information, Table S4 for information on SNPs linked to adaptive loci in each of the three lakes, and Figure S3 for the global distribution of alleles at SNPs linked to putatively adaptive loci.