Trans-Atlantic Distribution and Introgression As Inferred from Single Nucleotide Polymorphism: Mussels and Environmental Factors

Large-scale climate changes influence the geographic distribution of biodiversity. Many taxa have been reported to extend or reduce their geographic range, move poleward or displace other species. However, for closely related species that can hybridize in the natural environment, displacement is not the only effect of changes of environmental variables. Another option is subtler, hidden expansion, which can be found using genetic methods only. The marine blue mussels are known to change their geographic distribution despite being sessile animals. In addition to natural dissemination at larval phase-enhanced by intentional or accidental introductions and rafting-they can spread through hybridization and introgression with local congeners, which can create mixed populations sustaining in environmental conditions that are marginal for pure taxa. The species have a wide distribution in coastal regions of the Northern and Southern Hemisphere. In this study, we investigated the inter-regional genetic differentiation of the species complex at 53 locations in the North Atlantic and adjacent Arctic waters and linked this genetic variability to key local environmental drivers. Of seventy-nine candidate single nucleotide polymorphisms (SNPs), all samples were successfully genotyped with a subset of 54 SNPs. There was a clear interregional separation of species. However, all three species hybridized in the contact area and created hybrid zones with mixed populations. Boosted regression trees (BRT) models showed that inter-regional variability was important in many allele models but did not prevail over variability in local environmental factors. Local environmental variables described over 40% of variability in about 30% of the allele frequencies of spp. For the 30% of alleles, variability in their frequencies was only weakly coupled with local environmental conditions. For most studied alleles the linkages between environmental drivers and the genetic variability of spp. were random in respect to "coding" and "non-coding" regions. An analysis of the subset of data involving functional genes only showed that two SNPs at Hsp70 and ATPase genes correlated with environmental variables. Total predictive ability of the highest performing models ( between 0.550 and 0.801) were for alleles that discriminated most effectively from and whereas the best performing allele model (BM101A) did the best at discriminating from and . Among the local environmental variables, salinity, water temperature, ice cover and chlorophyll concentration were by far the greatest predictors, but their predictive performance varied among different allele models. In most cases changes in the allele frequencies along these environmental gradients were abrupt and occurred at a very narrow range of environmental variables. In general, regions of change in allele frequencies for occurred at 8-11 psu, 0-10 C, 60%-70% of ice cover and 0-2 mg m of chlorophyll , M. edulis at 8-11 and 30-35 psu, 10-14 C and 60%-70% of ice cover and for at 30-35 psu, 14-20 C.