Parallel genetic adaptation amid a background of changing effective population sizes in divergent yellow perch (Perca flavescens) populations

Abstract

Aquatic ecosystems are highly dynamic environments vulnerable to natural and anthropogenic disturbances. High-economic-value fisheries are one of many ecosystem services affected by these disturbances, and it is critical to accurately characterize the genetic diversity and effective population sizes of valuable fish stocks through time. We used genome-wide data to reconstruct the demographic histories of economically important yellow perch (Perca flavescens) populations. In two isolated and genetically divergent populations, we provide independent evidence for simultaneous increases in effective population sizes over both historic and contemporary time scales including negative genome-wide estimates of Tajima's D, 3.1 times more single nucleotide polymorphisms than adjacent populations, and contemporary effective population sizes that have increased 10- and 47-fold from their minimum, respectively. The excess of segregating sites and negative Tajima's D values probably arose from mutations accompanying historic population expansions with insufficient time for purifying selection, whereas linkage disequilibrium-based estimates of Ne also suggest contemporary increases that may have been driven by reduced fishing pressure or environmental remediation. We also identified parallel, genetic adaptation to reduced visual clarity in the same two habitats. These results suggest that the synchrony of key ecological and evolutionary processes can drive parallel demographic and evolutionary trajectories across independent populations.

Keywords: demographic history; effective population size; fisheries; genetic adaptation; genetic diversity; parallel evolution.

Figure 1.

Population structure of Lake Michigan yellow perch populations. (a) Yellow perch collected from seven Lake Michigan sites, including the comparatively isolated Muskegon Lake, which is connected to the main basin by a narrow channel (see inset). (b) A phylogenetic tree illustrates three distinct groups: Green Bay, Muskegon Lake and the five main basin sample sites. Twenty-five main basin individuals were randomly sampled for visual clarity; see electronic supplementary material, figure S1 for the full phylogenetic tree with all 210 individuals. (c,d) Both principal coordinate analysis of 210 individuals (c) and admixture analyses (d) further support the delineation of samples into three discrete populations. Note that one individual sampled in Muskegon Lake probably swam in from the main basin (electronic supplementary material, figure S2; see ‘Genetic divergence among yellow perch sample sites’ in electronic supplementary material).

Figure 2.

Numbers of SNPs, MAF and Tajima’s D. (a) Green Bay and Muskegon Lake populations have 2.8–3.4 times more SNPs than the five main basin sample sites, despite all sample sites having equal sample sizes, sequencing quality, read depth and missing data. (b) Both Green Bay and Muskegon Lake also have a lower median MAF compared with the five main basin populations, here illustrated with a density plot where the median MAF is depicted with a solid vertical line. (c) The negative Tajima’s D for both Green Bay and Muskegon Lake populations suggests recent population expansion. Collectively, these results suggest that the three populations (Green Bay, Muskegon Lake and the main basin sample sites) have different demographic histories.

Figure 3.

Demographic history of Lake Michigan yellow perch populations. Green Bay and Muskegon Lake sample sites have substantially lower effective population sizes, which average 23 and 90 estimated by NeEstimator v2, respectively, in comparison with main basin populations according to (a) estimates based on whole-genome sequencing data obtained in this study and (b) estimates based on RAD-Seq data obtained from Schraidt et al. [59] (note that y-axis is on a base 10 logarithmic scale). (c) Green Bay and Muskegon Lake have lower genetic diversity than main basin populations, measured here as observed heterozygosity averaged across overlapping windows of 5 MB using a step size of 2.5 MB across each chromosome. Notice that the main basin heterozygosity estimates are highly correlated, confirming higher gene flow among main basin sample sites. (d,e) The demographic history as reconstructed by GONE demonstrates a rapid increase in effective population size in (d) Green Bay and (e) Muskegon Lake since 10–30 generations ago. SGB, South Green Bay; MUS, Muskegon Lake; GRH, Grand Haven; MIC, Michigan City; MIL, Milwaukee; NUB, Naubinway; SUT, Sutton’s Bay; MEN, Menominee; BDN, Big Bay de Noc; LBD, Little Bay de Noc.

Figure 4.

Candidate genes, biological processes and phenotypes involved in parallel genetic adaptation. (a) A total of 13 outlier windows were detected in comparisons between both Green Bay and the main basin sample sites and Muskegon Lake and the main basin sample sites. The outlier window on chromosome 4 contains visual pigment encoding genes, such as opn1lw2. The coding regions, as indicated by horizontal grey bars, of these outlier genes, contain SNPs that are highly differentiated. (b) Seven major categories of biological processes are potentially involved in adaptive differentiation in Green Bay and Muskegon Lake in comparison with main basin sample sites, among which visual perception-related processes (highlighted with red asterisks) further support parallel adaptive differences in vision in Green Bay and Muskegon Lake. (c,d) The genetic differentiation at opn1lw2 is probably associated with water clarity, as evidenced by significant differences in (c) Secchi depths and (d) eye sizes between Green Bay or Muskegon Lake and main basin populations (i.e. Lake Michigan and Grand Haven). The Secchi depth in Green Bay and Muskegon Lake is significantly smaller than that in the southern basin of Lake Michigan (p < 0.001; c), and least-squares means of yellow perch eye diameter in Green Bay and Muskegon Lake is significantly larger than that in Grand Haven (p < 0.001; d). We accounted for differences in total length when comparing eye diameter of yellow perch collected from Green Bay, Muskegon Lake and Grand Haven (electronic supplementary material, figure S11; see Methods).