Single nucleotide polymorphisms unravel hierarchical divergence and signatures of selection among Alaskan sockeye salmon (Oncorhynchus nerka) populations

Abstract

Background: Disentangling the roles of geography and ecology driving population divergence and distinguishing adaptive from neutral evolution at the molecular level have been common goals among evolutionary and conservation biologists. Using single nucleotide polymorphism (SNP) multilocus genotypes for 31 sockeye salmon (Oncorhynchus nerka) populations from the Kvichak River, Alaska, we assessed the relative roles of geography (discrete boundaries or continuous distance) and ecology (spawning habitat and timing) driving genetic divergence in this species at varying spatial scales within the drainage. We also evaluated two outlier detection methods to characterize candidate SNPs responding to environmental selection, emphasizing which mechanism(s) may maintain the genetic variation of outlier loci.

Results: For the entire drainage, Mantel tests suggested a greater role of geographic distance on population divergence than differences in spawn timing when each variable was correlated with pairwise genetic distances. Clustering and hierarchical analyses of molecular variance indicated that the largest genetic differentiation occurred between populations from distinct lakes or subdrainages. Within one population-rich lake, however, Mantel tests suggested a greater role of spawn timing than geographic distance on population divergence when each variable was correlated with pairwise genetic distances. Variable spawn timing among populations was linked to specific spawning habitats as revealed by principal coordinate analyses. We additionally identified two outlier SNPs located in the major histocompatibility complex (MHC) class II that appeared robust to violations of demographic assumptions from an initial pool of eight candidates for selection.

Conclusions: First, our results suggest that geography and ecology have influenced genetic divergence between Alaskan sockeye salmon populations in a hierarchical manner depending on the spatial scale. Second, we found consistent evidence for diversifying selection in two loci located in the MHC class II by means of outlier detection methods; yet, alternative scenarios for the evolution of these loci were also evaluated. Both conclusions argue that historical contingency and contemporary adaptation have likely driven differentiation between Kvichak River sockeye salmon populations, as revealed by a suite of SNPs. Our findings highlight the need for conservation of complex population structure, because it provides resilience in the face of environmental change, both natural and anthropogenic.

Figure 1

Populations of sockeye salmon from the Kvichak River drainage. Location in southwest Alaska appears in rectangle of upper left insert. Numbers correspond to populations within Alagnak (1 - 9) and Kvichak (10 - 31) subdrainages. Population legends--tributaries: squares; mainland beaches: triangles; and island beaches: circles. Grey crosses indicate two waterfalls that may represent velocity barriers for migrating adults.

Figure 2

Detection of outlier SNPs using BAYESCAN. Populations included in the analysis: (a) the entire Kvichak River drainage, (b) after excluding Lake Clark, and (c) Iliamna Lake only. F_ST: locus-specific genetic divergence among populations; log₁₀Bayes Factor: decision factor in logarithmic scale (base 10) to determine selection; a vertical line indicates "decisive" evidence for selection. Filled circles represent candidates for selection; empty circles represent putatively neutral loci. Marker labels have been simplified (the prefix "One_" is missing).

Figure 3

Frequency plots of the major allele for four outlier SNPs in sockeye salmon. SNPs correspond to (a) One_HpaI-99, (b) One_GPH-414, (c) One_MHC2-190, and (d) One_MHC2-251 genotyped across 31 Alaskan populations (see Table 1 for population codes). Squares represent lakes from the Alagnak subdrainage (Battle: black; Kukaklek: grey; Nonvianuk: empty), whereas circles represent lakes from the Kvichak subdrainage (Clark: black; Sixmile: grey; Iliamna: empty).

Figure 4

Detection of outlier SNPs using Arlequin 3.5. Assuming a hierarchical model of migration and excluding Lake Clark populations. F_ST: locus-specific genetic divergence among populations; h₀/(1 - F_ST): a modified measure of heterozygosity per locus. Dashed lines indicate lower and upper 95% confidence intervals for variation in neutral F_ST as a function of h₀/(1 - F_ST). Filled circles represent candidates for selection; empty circles represent putatively neutral loci. Marker labels have been simplified (the prefix "One_" is missing).

Figure 5

Unrooted neighbour-joining tree between sockeye salmon populations. Based on Cavalli-Sforza & Edwards [51] chord distances for 38 nuclear ('No Outliers') SNPs. Segmented lines indicate the Alagnak (blue) and Kvichak (red) subdrainages; squares represent lakes from Alagnak (Battle: black; Kukaklek: grey; Nonvianuk: empty squares), whereas circles represent lakes from Kvichak (Clark: black; Sixmile: grey; Iliamna: empty). Numbers on branches represent percentages (only >50% are shown) of bootstrap support after generating 1000 resampled trees. R² depicts how well branch length and the variation of the distance matrix are correlated.

Figure 6

Principal coordinate analysis (PCO) between sockeye salmon populations from Iliamna Lake. Based on (a) 38 SNPs ('No Outliers'), (b) 39 SNPs including One_MHC2-190 ('Outlier 1'), or (c) 39 SNPs including One_MHC2-251 ('Outlier 2'). Population numbers can be found in Table 1. Population legends--tributary spawners: empty squares; island-beach spawners: black circles; mainland-beach spawners: grey triangles; Iliamna River (#23, a late-spawning population, Table 1): black square.

Figure 7

Scatterplots of genetic (F_ST) vs. geographic distances between sockeye salmon populations from Iliamna Lake. Scatterplots used (a) 38 SNPs ('No Outliers'), (b) 39 SNPs including One_MHC2-190 ('Outlier 1'), or (c) 39 SNPs including One_MHC2-251 ('Outlier 2'). Legend for population comparisons--tributary-tributary: yellow squares; tributary-island beach: purple circles; tributary-mainland beach: red circles. Regression lines and coefficients of determination (R²) were only reported for tributary-island beach (purple) and tributary-tributary comparisons (black) as no significant isolation-by-distance patterns were found for tributary-mainland beach comparisons.