Figures:
Figure 1: The sampling location for each species of fish. Threespine locations are represented by green triangles, ninespines by blue squares and tubesnouts by orange circles. Labels for each population are used consistently throughout this paper; the first half of the label denotes the species (Ts = thresspine stickleback, Tu = tubesnout and Ns = ninespine stickleback). The second half of the label denotes the state or province where the population was collected (AK = Alaska, USA; BC = British Columbia, Canada; OR = Oregon, USA; AB = Alberta, Canada; NU = Nunavut, Canada). The two Albertan ninespine populations are combined into a single point (NsABm & NsABk) for visual clarity. The base map is projected in Azimutahl equal distances (datum = WGS84) orientated to centre on Canada (latitude = 90 & longitude = -98.4). Ocean water is coloured by the annual range in sea surface temperature (°C) taken from the Bio-ORACLE database (https://www.bio-oracle.org/). Threespine and tubesnout photos were taken by Hazel Cameron-Inglis and used with permission, the ninespine photo was taken by Piet Spanns and used under an open license. The final plot was compiled in R using the sf, ggplot, raster and grid packages.
Figure 2: Genome wide patterns of genetic diversity within the threespine stickleback, ninespine stickleback and tubesnout. A ) F­ST per SNP and B )\({\overset{\overline{}}{H}}_{E}\) per 50Kb window for each species, excluding windows in intergenic regions. Ninespine scores were mapped onto their position on the threespine genome. Threespine and tubesnout FST was downsized by sampling every 100th SNP along the genome, and approximately 70 windows were filtered out of the \({\overset{\overline{}}{H}}_{E}\)plots for visual clarity. The red-dashed lines show the 999th FST and 99th\({\overset{\overline{}}{H}}_{E}\) quantiles. This plot was generated in R using the ggplot and gridExtra packages.
Figure 3: Comparison of genomic patterns among species.A) shows the relationship in average genetic diversity (\({\overset{\overline{}}{H}}_{E}\)) among genes for each species pair. Each point is a gene which is orthologous among the species. The dashed lines represent the 95th and 5thquantile of \({\overset{\overline{}}{H}}_{E}\) in each species. Any points on the bottom left or top right segments of a panel are genes with extreme \({\overset{\overline{}}{H}}_{E}\) that are shared among species. B) is a matrix of \({\overset{\overline{}}{H}}_{E}\)Spearman’s correlations among all population pairs, where the colour represents Spearman’s ρ and the text shows the significance level of a correlation test (* P < 0.05; ** P < 0.01; *** P < 0.001). C) shows the relationship and between the average FST per gene for each species. Colored points are signatures of local adaptation for each species; red for threespine sticklebacks and blue for tubesnouts. Gray points are genes not associated with local adaptation; they are partially transparant to show overlapping genes. No signatures of selection overlaped among species.
Figure 4: Detecting genes with elevated divergence and testing for signatures of convergent evolution. A-B) Show the top-candidate approach where each point is a separate gene. The total number of SNPs is compared to the number of SNP outliers in each gene, with top candidates identified as those genes that exceed the number of outliers expected under a binomial distribution, represented by the jagged red line. C-D) Null-W test results between C)tubesnout orthologs of threespine top candidates and D)threespine orthologs of tubesnout top candidates. The grey curve is the null-distribution of Z-scores from all orthologs of candidate genes in the focal species (i.e. tubesnout orthologs in C and threespine orthologs in D ). The blue points are top-candidate-orthologs, whose values on the y-axis have been jittered for visual clarity. The red dashed line is the 95th quantile of Z-scores. FDR corrections are not shown.