4.1 Evidence for reproductive isolation
The present findings confirm the occurrence of two reproductively
isolated populations of brown trout in lakes Bunnersjöarna as previously
reported by Ryman et al. (1979).
We find indications that Deme II is reproductively isolated from Deme I,
i.e. that gene flow from Deme I to Deme II is rare or non-existing. This
conclusion is supported by the following observations i ) a
strikingly lower level of genetic diversity within Deme II as compared
to Deme I for all variability measures used (Tables 1, 2; Figure 6a, b),
including that 53 loci of the 96 SNP array are monomorphic in Deme II
but polymorphic in Deme I (Table 1), and ii ) the highF ST between the two demes which is in the same
order as between populations from reproductively isolated lakes. In
contrast, we cannot rule out a small amount of gene flow from Deme II to
Deme I based on the present genomic and SNP array data. Deme II does not
show any private alleles with the SNP array, and the STRUCTURE analysis
suggests some minor admixture of Deme II into the Deme I cluster (Figure
2c).
The amount of divergence between the demes in Bunnersjöarna is confirmed
to be large through-out the genome and it appears high in comparison to
observations from other cases of sympatry in salmonids
(Jorde et al., 2018). The observedF ST values between the demes from the Pool-seq
data (Nei´s F ST=0.13), as well as from the 96 SNP
array (Nei´s F ST=0.14; Weir & Cockerham´sF ST= 0.24), are considerably higher than those
observed in other cryptic, sympatric salmonid populations including
Arctic charr (Adams, Wilson, & Ferguson,
2008; Wilson et al., 2004), Atlantic
salmon (Aykanat et al., 2015), and brown
trout (Andersson et al., 2017;
Palmé, Laikre, & Ryman, 2013). Rather,
the amount of divergence between the demes appears more in line with
what is found between sympatric salmonids showing substantial phenotypic
divergence such as different morphotypes of Arctic charr
(Conejeros et al., 2014;
Gordeeva, Osinov, Alekseyev, Matveev, &
Samusenok, 2010) or brown trout (Ferguson
& Taggart, 1991).
Our outlier analyses indicate that most differentiation is caused by
genetic drift but a number of genes may be under diversifying selection
in the Lakes Bunnersjöarna demes (Table S4, Figures S5, S6). We found
genes with high F ST involved in growth process.
It is interesting to note that a striking difference with respect to
growth was observed between the two demes with fish from Deme II being
markedly smaller than fish from Deme I
(Ryman et al., 1979). Further, several of
the genes putatively under selection were associated with reproductive
functions. This, along with the strong genetic differentiation observed
between the demes, indicates differences in reproductive characteristics
between the two demes. Spatial separation due to separate spawning
grounds in streams resulting from a strong homing behavior is a typical
population separator in brown trout
(Ferguson, Reed, Cross, McGinnity, &
Prodöhl, 2019). In Bunnersjöarna such a mechanism has been suggested
where Deme II fish have been observed to primarily occur relatively
close to the inlet stream in the southern lake, whereas Deme I fish are
found in the northern lake with the outlet stream towards Lake Ånnsjön
(Ryman et al., 1979).
A large number of putatively selected genes were associated with
metabolic processes including genes associated with “NADH oxidation”.
Interestingly, LDH functions as the catalyst to all these processes
(Parra-Bonilla, Alvarez, Al-Mehdi,
Alexeyev, & Stevens, 2010; also see
https://www.genome.jp/dbget-bin/www_bget?ko:K00016). Putatively
selected genes were also associated with immunological processes, the
roles of which are currently unknown in brown trout. However, adaptive
divergence of immune related genes has been identified in Atlantic
salmon (e.g. Kjærner-Semb et al., 2016)
and it has been suggested that such adaptive genes have largely
contributed to genetic divergence between Atlantic salmon populations in
Norway. Observations of signatures of selection in particularly
duplicated regions of the salmon genome have led researchers to
hypothesize that genome duplications might increase the opportunity for
evolutionary adaptations (Kjærner-Semb et
al., 2016). Genome-wide association studies still remain to be carried
out in brown trout but might clarify this issue further. However, it is
also important to note that we observe no clear genome regions of
selection – rather the signature (elevated differentiation) is diffuse
across the genome with a weak signal with underlying loci being
difficult to identify. This may be due to either many false positives
(no real underlying selection) or that selective differences are highly
polygenic. Further work is needed to resolve this issue.