Figure legends
Figure 1 Map showing the location of Lakes Bunnersjöarna in the mountainous area of central Sweden. The lakes are found in the uppermost part of the water system (elevation indicated, m =meters). Our samples were collected in the northern Bunnersjön (N. Bunnersjön;n = 68) and in the southern Bunnersjön (S. Bunnersjön;n =72). For comparison, we also used material from Lake Ånnsjön (shown here) as well as from several other lakes in other areas (see Figure S1a-d).
Figure 2 Membership coefficient (Q ) plots showing the assignment probability of individual fish (total n =140) to clusters using 96 SNPs. Membership coefficients (Q ) were obtained from STRUCTURE and the most likely number of clusters (K ) for the data was estimated using three different ad hoc methods: log probability of data (lnPrXK ), ΔK , and parsimony index (PI ). Panel a) shows K =2, panel b) K =3, and panel c) K =11 as the most likely numbers of K suggested by ΔK , PI, and lnPrXK , respectively. Each fish is represented by a vertical bar and the order of individuals is the same in all panels, starting with 68 individuals classified to Deme I, followed by 72 individuals classified to Deme II based on genotype in the allozyme locus LDH-1 .
Figure 3 Individual-based neighbor-joining tree illustrating the genetic relationship among brown trout from Lakes Bunnersjöarna using 70 SNP loci. Individuals marked I vs. II (purple vs. blue color) are classified to belong to Deme I vs. Deme II based on their allozymeLDH-I genotype. The tree was constructed based on Nei’s DA distance, and has been compressed to include branches with bootstrap values of at least 70%. Numbers along the branches indicate bootstrap values in percentages. The black dots mark the four individuals that were randomly selected for individual whole-genome sequencing.
Figure 4 Pairwise F ST values between Deme I and II of Lakes Bunnersjöarna estimated from whole-genome Pool-seq data using 5 kb windows across 40 brown trout chromosomes. NA=F ST values from scaffolds not possible to assign to a chromosome. The horizontal black dashed line shows the genome-wide mean F ST=0.13 while the red dashed line marks the 97.5% limiting F ST=0.35.
Figure 5: Distribution of observed (range: 0 to 1) and simulated (“expected”; range: 0 to 0.864) F STvalues based on the POWSIM method (Ryman & Palm, 2006). A total of 12,177,462 SNPs derived from Pool-seq data were analyzed and the mean F ST for these was 0.083 (Table S2). The largest simulated F ST was 0.864 and we had 194 observed F ST values above this value. The framed subfigure is a close-up to show the distribution of the extreme values in the right-hand tail.
Figure 6 a) Estimates of total lengths of runs of homozygosity (LnROH ) of three categories and b) F ROH(fraction of runs of homozygosity, ROH, expanded over the genome) using individual whole-genome sequencing data from each of two individuals per deme from Lakes Bunnersjöarna.
Figure 7 : The genetic relationships among brown trout populations from the two demes of Lakes Bunnersjöarna and from other lakes in the same geographic region and TreeMix analysis assuming potential migration between lakes in the same geographic area where creek waterways between lakes exist (cf. Figure 1, Figure S1, Appendix S1). The scale indicates the proportion of genetic divergence per unit length of the branch (indicated by the scale length).