Genomic footprints of bottleneck in landlocked salmon population

Abstract

At the end of the last ice age, several Atlantic salmon populations got caught up in the lakes and ponds of the Northern Hemisphere. Occasionally, the populations also got locked when the flow of rivers terminated from reaching the sea due to land upheaval. Therefore, the pattern of evolution shaping the landlocked salmon populations is different from the other anadromous salmons, which migrate between the sea and rivers. According to the theories of population genetics, the effect of genetic drift is expected to be more pronounced in the former compared to the latter. Here we examined this using the whole genome data of landlocked and anadromous salmon populations of Norway. Our results showed a 50-80% reduction in the genomic heterozygosity in the landlocked compared to anadromous salmon populations. The number and total size of the runs of homozygosity (RoH) segments of landlocked salmons were two to eightfold higher than those of their anadromous counterparts. We found the former had a higher ratio of nonsynonymous-to-synonymous diversities than the latter. The investigation also revealed a significant elevation of homozygous deleterious Single Nucleotide Variants (SNVs) in the landlocked salmon compared to the anadromous populations. All these results point to a significant reduction in the population size of the landlocked salmons. This process of reduction might have started recently as the phylogeny revealed a recent separation of the landlocked from the anadromous population. Previous studies on terrestrial vertebrates observed similar signatures of a bottleneck when the populations from Island and the mainland were compared. Since landlocked waterbody such as ponds and lakes are geographically analogous to Islands for fish populations, the findings of this study suggest the similarity in the patterns of evolution between the two.

Figure 1

Phylogenetic relationship among the salmon populations from six locations in Norway. The brown trout was used as an outgroup to root the tree. The NJ tree was constructed using the whole genome data and the bootstrap confidence values were based on 500 replicates. The asterisk (*) denotes 100% bootstrap support. However, the nodes without an asterisk have > 80% but < 100% bootstrap support.

Figure 2

Genomic heterozygosities estimated for five anadromous and one landlocked population of Norway are shown. Error bars denote the standard error of the mean. The genomic diversity of the landlocked salmon population was significantly smaller than those of anadromous populations (at least, P < 0.0001).

Figure 3

Box plot showing (A) the number and (B) the total size of runs of homozygous (RoH) segments in landlocked and anadromous salmon populations. The centre line denotes the median, the boundaries of the box represent the first and third quartiles, and the whiskers show the maximum and minimum values. The mean number of RoH of the landlocked salmon population was significantly higher than those of anadromous populations (at least, P = 0.0003). Similarly, the average size of RoH of the landlocked salmon population was significantly higher than those of anadromous populations (at least, P = 0.002).

Figure 4

(A) The correlation between effective population size and the ratio of nonsynonymous and synonymous diversities (dN/dS). The relationship was highly significant (r = 0.79, P < 0.000001). The best-fitting regression line is shown. (B) The average dN/dS ratio was estimated for landlocked and anadromous salmon populations. Error bars show the standard error of the mean. The ratio observed for landlocked salmons was significantly higher than those of anadromous ones (at least, P = 0.0018).

Figure 5

The stacked bar shows the proportions of homozygous (solid) and heterozygous (stripped) SNVs estimated for landlocked and anadromous salmon populations. (A) Nonsynonymous SNVs (B) Deleterious SNVs (C) Loss of function (LoF) SNVs. The number of homozygous SNVs in landlocked salmons was significantly higher than those estimated for the anadromous salmon, and this is true for the comparisons involving nonsynonymous, deleterious, and LoF SNVs (at least P < 0.0001).