4.1 Population structure and evolutionary history of Harbour porpoises in the North Atlantic and adjacent waters
By using whole-genome resequencing data, as previous authors suggested (Fontaine et al., 2007; 2014; Ben Chehida et al., 2021), we found evidence of three harbour porpoise subspecies and at least three populations within the North Atlantic subspecies (Wiemann et al., 2010; Lah et al., 2016). The three subspecies are P.p. relicta in the Black Sea, P.p. meridionalis in Iberia and P.p. phocoenain the rest of the Atlantic. Black Sea porpoises seem to be a relict population from past glaciations based on the suggested lack of genetic exchange with the other subspecies studied (Figure 2A,B). Our sample set included two samples from Iberian waters, but only one of them was genetically distinct and hence suggestive of belonging to a putative Iberian/Mauritanian subspecies (Fontaine et al., 2007; 2014). PCA results (Figure 2A,C) show that Iberian sample No-2 does not cluster with the Black Sea nor with the Atlantic subspecies. Admixture results grouped this sample with the Atlantic cluster, possibly sinceNGSAdmix does not create a new cluster for only one sample. Nevertheless, while no other sample had any Black Sea ancestry, this sample had a 10% membership to the Black Sea cluster, further suggesting its distinct evolutionary trajectory. More samples from the Iberian/Mauritanian waters should be resequenced to categorically identify these porpoises as an own separate subspecies.
Regarding the population structure within the North Atlantic subspecies,P.p. phocoena , we did not find evidence of separation between West and East North Atlantic porpoises, but we confirmed the two proposed distinct porpoise lineages in the peripheral waters of the Baltic Sea (Lah et al., 2016; NAMMCO, 2019). On one hand, PCA, admixture analysis (Figure 2) and Fst levels (Figure 6) show that CA, ICE, BAS and NOS porpoises belong to the same population, the so-called Atlantic population. On the other hand, BES and PBS porpoises clustered separately. Between NOS and BES, there is some gene flow (as indicated by partial assignments to the two respective clusters for some specimens; blue and red in Figure 2). Three very distinct PBS individuals stand out in the genetic structure analysis (Figure 2, 6, S4), which we assigned to the PBS population. This third Baltic cluster was interpreted as the Proper Baltic Sea population for a series of reasons: first, this cluster only comprised porpoises of the PBS region (Figure 2, 6, S14); second, the three porpoises assigned to the PBS population were bycaught during the breeding season, when a separation between putative BES and PBS porpoises occurs (Carlén et al., 2018); third, these three porpoises were bycaught in the easternmost locations (16-18.58W) and in areas known to be important for the PBS population (Carlén & Evans, 2020), i.e., one in the Gdansk Bay and the other two in the waters surrounding the Swedish island of Öland (Figure S4). These waters present the greatest densities of harbour porpoises during the breeding season in the PBS region (Amundin et al., 2022) and have been reported as potentially important breeding grounds for the PBS population.
The result indicating lack of genetic structure over long distances in the open Atlantic is at odds with the fine-scale population structure we observe in the Baltic region. In an area separated by less than 1,000 kilometers we identified three distinctive lineages (Figure S4C): Atlantic population in the Skagerrak strait, Belt Sea population in the Danish Belts, the Sound and Arkona basin, and the Proper Baltic population in the Baltic proper. We did not find an IBD pattern (Figure S14C), highlighting that geographical distance does not have a major impact on the genomic variation and is unlikely to be a driver of the genetic differences found in North Atlantic harbour porpoises. It should be mentioned, though, that previous authors have found weak patterns of IBD in North Atlantic harbour porpoises (Fontaine et al., 2007; Ben Chehida et al., 2021) when including more sampling locations along the harbour porpoise distribution range. Resequencing more specimens from the open Atlantic may be needed to assess the potential role of geographical distance on North Atlantic harbour porpoise population differentiation.
The demographic history analysis shows that North Atlantic harbour porpoises have been strongly influenced by Pleistocene glaciations, especially since the Last Glacial Period (LGP), when around 25 rapid climate fluctuations occurred until the end of the Last Glacial Maximum (LGM), ~ 19,000yBP (Kindler et al., 2014). TheNe of the three harbour porpoise subspecies were highly correlated until the onset of the LGP (~110,000yBP) when the Iberian and Black Sea subspecies curves split from the North Atlantic subspecies. Our divergence estimate differs from that of previous authors using a portion of the mitogenome (Fontaine et al., 2014) that dated the most common recent ancestor of North Atlantic subspecies during the LGM. Similar differences in divergence estimates have been reported before in the finless porpoise, when authors using mitochondrial control region data (Wang et al., 2008) inferred a much younger divergence than authors using whole-genome resequencing data (Zhou et al., 2018). The Iberian and Black Sea subspecies originated from the ancestral North Atlantic population when a small group of individuals may have colonized the Mediterranean and Black Sea after the LGP, which left an imprint in the genome as a founder effect. Both Iberian and Black Sea subspecies had lowNe (Figure 3B), which increased sensitivity to genetic drift which in turn lead to loss of genetic variation, high inbreeding levels (Figure 4A, S5) and a positive Tajima’s D (Figure 4C).Ne curves for populations of the North Atlantic subspecies started to diverge around ~75,000yBP, which roughly coincides with two rapid climate fluctuations during that same period (Kindler et al., 2014). This may suggest that BES and PBS lineages have diverged from the Atlantic population during an interglacial period before the formation of the Baltic Sea at the end of the LGM. However, our demographic analysis also indicate that after the LGM, the Ne of the Baltic populations increased. This would be compatible with a scenario in which - as the ice sheets retreated from northern regions at the end of LGM - a group of porpoises colonized the newly formed Baltic Sea, originating the modern BES and PBS populations. Nevertheless, other processes like gene flow and linked selection could be cofounding factors in the SMC++ demographic inferences (Mazet et al., 2016; Schrider et al.,2016), thus these results must be interpreted with caution and time/Ne estimates should not be taken literally.
The maximum likelihood tree inferred with Treemix is also compatible with the existence of three subspecies in the North Atlantic and adjacent waters. Previous phylogenies reconstructed with mitogenome data (Ben Chehida et al. 2020) found that BLS porpoises are basal in the North Atlantic, thus we were confident rooting our tree with the BLS subspecies. The first split in our graphical representation of historical relationships among North Atlantic harbour porpoises was the IBE subspecies from the North Atlantic subspecies. Among the North Atlantic subspecies, Treemix analysis placed present-day NOS harbour porpoises as basal, compatible with a northward post-glacial expansion from a southern refugia. As the ice sheets retreated, the ancestral North Atlantic subspecies may have started to colonize novel environments in the Baltic, Iceland, Barents Sea, and Canadian waters.
4.2 Local adaptation of Baltic porpoises to low salinity levels
Species and populations inhabiting highly divergent environments are expected to be under different selective pressures, which could cause each local population to evolve traits that provide an advantage under its local environmental conditions (Kawecki & Ebert, 2004). However, the role of ecological specialization on population differentiation and speciation remains poorly understood (Savolainen, Lascoux & Merilä, 2013). This is particularly true for cetacean species, with only a few recent studies attempting to address the genetic basis of local adaptation (Barceló et al., 2022; Pratt et al., 2022; de Greef et al., 2022; Louis et al., 2021; Zhou et al., 2018). Ecologically and geographical marginal environments often host populations at the edge of the species distribution and under extreme selection regimes (Johannesson & André, 2006). Examples of such populations are the Belt Sea and Proper Baltic Sea harbour porpoise populations that occur in the peripheral waters of the Baltic Sea, separated from the North Sea by a pronounced salinity gradient.
The GEA results show that including BLS samples in the RDA had a major impact on the outcome of the analysis, especially on the number of putative SNPs inferred to be under selection. The genomic variation of BLS porpoises was highly associated with high temperature (Figure S12), with ~6,000 SNPs correlated with SST and only a few associated with the other variables (Table S4). As water temperatures in the Black Sea are significantly higher than in the rest of the locations (Figure S6) and BLS porpoises are highly divergent, we could not discern whether these ~6,600 SNPs were indeed associated with temperature or were rather very distinct for different evolutionary pressures or pronounced genetic drift in BLS porpoises. Thus, to identify candidate genes associated with environmental variables we focused on the dataset without BLS, where the population divergence was not as strong.
Our seascape genomics analysis provides statistical support for an influence of salinity on population differentiation in the Baltic Sea (Figure 5, S12). SSS was highly significant in both RDA and pRDA, while SST was significant only in the RDA. The salinity gradient could have contributed to the origin of a soft barrier between the Atlantic and the Baltic, leading to adaptive divergence in BES and PBS porpoises, as seen in other cetaceans as the finless porpoise (Zhou et al., 2018). From 272 inferred candidate SNPs, 107 were annotated with genes potentially associated with the salinity gradient in the Baltic (Table S5). We identified three solute carrier group (SLC) genes, a group of membrane transport proteins. Particularly interesting was the SLC10A1 gene, a sodium ion transport with a critical role in the osmoregulation of bile acids in the liver (Kubitz & Häusinger, 2007). We also obtained hits in osmoregulatory genes (AQP9, DYNC2H1) previously inferred to be under selection in cetaceans (Xu et al., 2013, São Pedro et al., 2015, Zhou et al., 2018). Further steps to evaluate the candidate SNPs and genes should include the annotation of Gene Ontology (GO) terms as well as genes within 20kb of the candidate SNP (Pratt et al., 2022).