Temporal allele frequency changes in large-effect loci reveal potential fishing impacts on salmon life-history diversity

Abstract

Fishing has the potential to influence the life-history traits of exploited populations. However, our understanding of how fisheries can induce evolutionary genetic changes remains incomplete. The discovery of large-effect loci linked with ecologically important life-history traits, such as age at maturity in Atlantic salmon (Salmo salar), provides an opportunity to study the impacts of temporally varying fishing pressures on these traits. A 93-year archive of fish scales from wild Atlantic salmon catches from the northern Baltic Sea region allowed us to monitor variation in adaptive genetic diversity linked with age at maturity of wild Atlantic salmon populations. The dataset consisted of samples from both commercial and recreational fisheries that target salmon on their spawning migration. Using a genotyping-by-sequencing approach (GT-seq), we discovered strong within-season allele frequency changes at the vgll3 locus linked with Atlantic salmon age at maturity: fishing in the early season preferentially targeted the vgll3 variant linked with older maturation. We also found within-season temporal variation in catch proportions of different wild Atlantic salmon subpopulations. Therefore, selective pressures of harvesting may vary depending on the seasonal timing of fishing, which has the potential to cause evolutionary changes in key life-history traits and their diversity. This knowledge can be used to guide fisheries management to reduce the effects of fishing practices on salmon life-history diversity. Thus, this study provides a tangible example of using genomic approaches to infer, monitor and help mitigate human impacts on adaptively important genetic variation in nature.

Keywords: Baltic salmon; SNP; fisheries management; fisheries‐induced evolution; genetic stock identification; temporal genomics.

FIGURE 1

The geographic origin of (a) salmon stocks in the genetic baseline used for genetic stock identification in this study, and (b) sampling sites of wild baseline samples in the Tornio‐Kalix River system. The darker lines depict national borders, including the Tornio/Torne River that flows on the border of Finland and Sweden.

FIGURE 2

Collection of catch samples from (a) the coast of Gulf of Bothnia (years 2019–2020), and (b) from the Tornio and Kalix Rivers (years 1928–2020). Fishing areas on the maps: (a) see Table 1 for names of coastal fishing areas; (b) R1 = Tornio River Downstream area, R2 = Pello‐Lappea, R3 = Kihlanki, H1 = Kiviranta, H2 = Kengisfors, H3 = Kalix River Downstream area, H4 = Rödupp. The darker lines depict national borders, including the Tornio/Torne River that flows on the border of Finland and Sweden. Note that the bifurcation linking Tornio and Kalix (marked with a line) allows fish to access Tornio headwaters via the lower Kalix mainstem and vice versa.

FIGURE 3

Estimated stock proportions of wild salmon catches from (a) the coastal fishing areas and (b) the Tornio River during fishing seasons 2019–2020. The catches were split temporally into quartiles (the four points per stock and area) based on the number of genetic samples. The points represent median quartile days (i.e. median catch date of individuals in each quartile), counted as days since May 1 (see Table S1). The error bars denote 95% credible intervals. For illustrative purposes, estimated proportions of Kymi, Råne, Simo and Vindel River stocks are not shown. TK refers to Tornio‐Kalix.

FIGURE 4

Relationship between mean age at maturity of salmon (i.e. sea age, as number of winters spent at sea before first spawning) and genotypes of the (a) vgll3 and (b) six6 genes, analysed across the entire Baltic salmon catch dataset from 1928 to 2020. E and L represent the alleles associated with early and late maturation, respectively.

FIGURE 5

vgll3*L (associated with older age at maturity) allele frequency in (a) coastal wild salmon catches, and in (b) Tornio River catches during 2019–2020. The lines depict a relationship between vgll3*L and catch date, fitted with a GAM (smoother used: y ~ s(x, bs = “cs”)), whereas the grey area around the lines illustrates the uncertainty of the fitted relationship (95% confidence intervals). The histograms above show the estimated daily catch sizes (number of salmon caught per day) in these areas for the duration of the fishing season. The data points represent the vgll3 genotype of individual samples. The points are jittered on the y‐axis to aid figure interpretation. TK refers to Tornio‐Kalix.

FIGURE 6

Long‐term changes in vgll3*L allele frequency in the Tornio/Torne River salmon catches from 1928 to 2020. The line depicts a relationship between vgll3*L and sampling year, fitted with a GAM (smoother used: y ~ s(x, bs = “cs”)), whereas the grey area around the lines illustrates the uncertainty of the fitted relationship (95% confidence intervals). The data points represent the vgll3 genotype of individual samples. The points are jittered on the y‐axis to aid figure interpretation.