New SNP markers reveal largely concordant clinal variation across the hybrid zone between Mytilus spp . in the Baltic Sea

Environmental conditions such as a pronounced salinity gradient and postglacial history make the Baltic Sea a suitable area for studying how selection and gene flow affect genetic differentiation in marine species. A cDNA library was used to identify new single nucleotide polymorphisms (SNPs) in Baltic populations of Mytilus spp. mussels. Sixty polymorphic SNPs were used to genotype 642 individual mussels from the inner Baltic, Danish Straits, northwest Denmark and a population of the northeast Pacific. We characterized 49 novel SNP markers that differentiate the populations of the North and Baltic Sea areas. Concordant narrow clines were observed at the entrance of the Baltic Sea for most of these markers. Considerable variance of hybrid index scores was observed in populations with intermediate allele frequencies within the hybrid zone, e.g. in Gedser and Hjelm. The presented results are in accordance with the existence of strong reproductive isolation, probably caused by a combination of exogenous (e.g. adaptation to brackish waters) and endogenous preand post-zygotic factors (e.g. selection against hybrids). The overwhelming majority of new SNPs markers showed a larger representation of M. trossulus than M. edulis genes in the nuclear DNA of Baltic Mytilus species. Finally, we identified a few markers with an elevated level of introgression of M. edulis alleles in the Baltic Sea M. trossulus populations in comparison to the reference M. trossulus population of the Pacific.


INTRODUCTION
Mytilus taxa are an important component of food chains in marine ecosystems and a dominant benthic filter-feeder in the Baltic.The Baltic Sea was colonized by Mytilus spp.mussels about 7000 yr ago, after the most recent freshwater period and the emergence of this area as a marine (brackish) ecosystem (Zillén et al. 2008).The Baltic Sea has a salinity ranging from almost freshwater in the northern part (Gulf of Bothnia) and brackish (6 to 8 PSU) in the inner Baltic to full salinity (from 20 to 30 PSU) in the northern Danish Straits (Kattegat) rising to 33 PSU in the North Sea.Sharp salinity gradients are built up in the Sound and the Belts.
Baltic populations are usually smaller, less genetically variable and more isolated than those in the Atlantic and may also be subject to genetic bottlenecks (Johannesson & André 2006).However, it has been observed that the inner-Baltic populations of Mytilus spp.mussels are more genetically variable than populations outside the Baltic (North Sea and Kattegat).A similar situation has been observed for Ma coma balthica (Nikula et al. 2008).This is because the Baltic was first inhabited by genetically distinct lineages of these species (Mytilus trossulus, Macoma balthica balthica) which hybridized with Atlantic Mytilus edulis and Macoma balthica rubra, respectively, at the entrance to the Baltic Sea (Väinölä & Hvilsom 1991, Luttikhuizen et al. 2012).Baltic popu-lations of Mytilus spp.have a unique genetic composition of loci derived from M. trossulus and M. edulis genomes due to multiple processes: the history of the taxa involving divergence over more than 3 million yr, a hybrid zone at the entrance of the Baltic, in trinsic post-zygotic and pre-zygotic isolation (Bierne et al. 2011) and local adaptation caused by location-specific salinity regimes (i.e.locus-specific selection evidenced by impeded gene flow).
Differential introgression of loci observed across the contact zone may be caused not only by natural selection (McDonald 1994) but also by a large amount of stochastic variation at neutral loci (Bierne et al. 2003).Hybrid zones are very interesting areas for research because they make it possible to study gene flow and reproductive isolation, and provide an opportunity to examine the functional tests of different combinations of mutations.As mentioned above, some genomic regions in species from hybrid zones are not only influenced by neutral processes.Payseur (2010) tried to identify genomic regions involved in speciation using differential introgression in hybrid zones.F ST outlier analysis can be performed to find the regions (loci) influenced by selection or linked to selected loci (Luttikhuizen et al. 2012).Selection against hybrids reducing their fitness has been observed in hybrid zones (Barton & Hewitt 1989, Burke & Arnold 2001).
According to Beaumont (2005), analysis of estimated F ST can be a first step to identifying genes that might be under selection.However, Bierne et al. (2011Bierne et al. ( , 2013) ) indicate some caveats in interpreting the increasing the number of F ST outliers, independent of population structure, and pointed out that genomewide genetic barriers are often multifactorial.For example, many outlying F ST loci observed in gene scan may be attributable to local adaptation as well as to endogenous genetic incompatibilities (caused by, for example, cryptic hybrid zones involving multiple loci involved in pre-and post-zygotic isolation).This approach to the problem indicates that F ST outliers should be considered as being candidate loci requiring further detailed investigation (Gosset & Bierne 2013).
Initially, North Sea and Baltic Sea mussels of the genus Mytilus were characterized using allozymes (McDonald et al. 1991, Väinölä & Hvilsom 1991).Baltic mussels were classified as being very similar and closely related to Pacific M. trossulus.Study of the genetic transition through the Sound and the other Belt Sea straits between the Baltic and North Sea populations using the Gpi allozyme locus was recently summarized by Väinölä & Strelkov (2011).
The work of Väinölä & Strelkov (2011) corroborates the origin of the Baltic Mytilus spp.from the North Pacific.Earlier work on Baltic populations based on gene-targeted PCR markers (e.g.Bierne et al. 2003, Riginos & Cunningham 2005, Kijewski et al. 2006, Väinölä & Strelkov 2011) has shown the Baltic-North Sea contact (hybrid) zone to have a thoroughly mixed genetic composition of M. edulis and M. trossulus, unimodal in character in the Baltic Sea (introgressed M. trossulus) with bimodality observed in Øresund and the Danish Straits.Recent work (Zbawicka et al. 2012) based on single nucleotide polymorphisms (SNPs) detected 5 markers located in genes of the histone family and p53 differentiating the European populations of mussels.
SNPs are variations that occur at the nucleotide level when a single nucleotide differs among or within individuals of a species.They explain 90% of the genetic differences between individuals and are thus very suitable for genetic research and selective breeding applications (Brookes 2007).SNPs located in coding regions (non-synonymous and synonymous substitutions) can be used to distinguish loci under selective pressure from neutral loci (Morin et al. 2004).Synonymous changes may not be neutral be cause of the efficiency of same codon translation: many organisms preferentially use certain synonymous codons (Carlini & Stephan 2003).Depending on the rate of recombination and the nature of selection, variation at synonymous SNPs can also be in fluenced by selection on neighboring regions (Nielsen et al. 2005).Multilocus scans used to compare different populations for several loci can identify regions in the genome carrying a mutation arising from local adaptation (Schlötterer 2002).SNPs have been widely used for differentiation studies at the individual, population and species level (Pariset et al. 2009, Quintela et al. 2010, Williams et al. 2010), and also for eco logical and conservation studies (Vignal et al. 2002).
New SNPs, discovered based on the expressed sequence tag (EST) sequences of the Baltic mussel, were used as markers for Mytilus taxa in an analysis of populations from the Baltic Sea region.F ST outlier analysis was used to identify candidate loci for which gene flow is impeded across the Danish Straits.The purpose of this study was to apply SNPs to identify markers and genes affected by hybridization in the Danish Straits and to examine the variability between them in the geographic cline (along the hybrid zone).The study also investigated the concordance of multigene clines at the Baltic Sea entrance and the extent of the barrier to gene flow.

Identification and genotyping of SNPs
EST sequences (deposited in GenBank under accession numbers: KJ871031-KJ871077) were used for SNP discovery using the Staden computer programs (Staden et al. 2001).These sequences were aligned with appropriate GenBank EST sequences (mainly from Mytilus galloprovincialis and M. edulis) using the ClustalX program v.1.83(Thompson et al. 1997) with default settings.The possible effect of the SNPs, resulting in change or no change in amino acid se quence (non-synonymous or synonymous changes, respectively), was predicted on the basis of the open reading frames (ORFs) identified (see Table S1 in the Supplement at www. int-res.com/articles/ suppl/ b021 p025 _ supp.pdf).
Genotyping of SNPs was performed using the Sequenom MassARRAY iPLEX platform (Gabriel et al. 2009).PCR and extension primers were designed using the Assay Design 3.1 program (Sequenom); analysis and scoring were performed using Typer 3.4 software (Sequenom).SNPs were classified, based on manual inspection, as 'failed assays' (meaning that the majority of genotypes could not be scored and/or the samples did not cluster well according to genotype), 'monomorphic SNPs', or 'polymorphic SNPs'.Assays were designed for 84 candidate SNPs (77 based on EST sequences and 7 already identified by Zbawicka et al. 2012); 5 of which (SNP1B, SNP2A, SNP2B, SNP3D and SNP6B) were characteristic of the M. trossulus genome and 1 of M. galloprovincialis (SNP1C).Seven SNPs, already identified, were renamed BM201B, BM201C, BM202A, BM202B, BM203C, BM203D and BM206B.

DNA sampling
Mytilus spp.samples, consisting of 630 individuals in total, were collected from 27 localities (Fig. 1, Table S2 in the Supplement).Adult mussels of mixed age and size (15 to 50 mm) were collected.Two samples, Veno Bight (VEM) and Logstor Bredning (LOG), were obtained from the Limfjord (northwest Denmark), which is highly influenced by North Sea water.Sixteen samples were obtained from the Kattegat and northern part of the Belt Sea: Aarhus Bight (ARH), Augstenborg Fjord (AUG), Gilleleje (GIL),  S2).Samples were stored frozen at −20°C until the SNP analyses.Twenty-seven samples consisted of 16 to 29 individuals each; the sample from Canada consisted of 12 individuals.DNA was isolated from mantle tissue using a modified CTAB method according to Hoarau et al. (2002) and suspended in sterile, filtered, distilled water.

Analysis of genotype data
SNP markers and populations were analyzed for genetic diversity, proportion of polymorphic SNP (P o ), observed (H o ) and expected (H e ) heterozygosity, and inbreeding coefficient (F IS ) using Arlequin v.3.5 (Excoffier & Lischer 2010).The statistical significance of F IS (> 0) was tested by 10 000 permutations of alleles between individuals.Departures from Hardy-Weinberg equilibrium (HWE) were tested by exact test, and significance was determined by Markov chain Monte Carlo simulations.Pairwise analysis in Arlequin was used to calculate mean pairwise F ST values defining population differentiation.F ST values at individual SNPs were calculated using the AMOVA function of the same program.Permutation testing with 1000 iterations was used to calculate p-values for mean and locus-by-locus F ST values.Arlequin was also used to detect loci under selection through genetic structure analysis.The neutral distribution of F ST was simulated with 30 000 iterations at a 95% confidence level, and characterized by estimating the 0.05 and 0.95 quantiles of the distribution.In the first step, the neutral expectation was based on the overall mean value of F ST calculated for all SNPs.First, the analysis was conducted for all SNPs, then SNPs with F ST values outside the 0.95 limits corresponding to the null hypothesis were removed and a new analysis was performed with the recalculated value of F ST .SNPs with F ST values above the 0.95 limits after the second analysis were also considered to be outlier loci (Acheré et al. 2005).This procedure reduces bias in the estimation of F ST by removing the highly diverged loci.The F ST distance matrix of sampling sites obtained in Arlequin was used to construct a neighbour-joining tree illustrating the genetic re latedness of all populations, using MEGA v.4 (Tamura et al. 2007).Correction for multiple tests of overall significance values was carried out using the Bonferroni procedure of Rice (1989).
The frequency distribution of the score for a hybrid index, giving the percentage of M. trossulus characteristic alleles, was calculated for all loci with F ST > 0.5 (Table S1).A score of 0 indicates a pure M. edulis, whereas a score of 1 indicates a pure M. trossulus.Only one, the most differentiated SNP coming from the same fragment, was used.
GENEPOP 4.1 (Rousset 2008) was used to test for linkage disequilibrium (LD) between all pairwise combinations of 51 loci (1 SNP per contig or fragment, marked in Table S1) for all populations separately, and for 2 groups of populations: (1) populations from the North Sea and the northern Danish Straits, and (2) populations from the inner Baltic Sea grouped together.Correspondence analysis (CA; Benzécri 1992), implemented in GENETIX (Belkhir et al. 2003), was used for visualizing the genetic substructure at population and individual levels.The result is presented as a scatter plot, with the axes representing the contribution of inertia of the data matrix in a way that can be considered analogous to the total variance in allelic frequency (Benzécri 1992).

SNP validation
In total, 340 putative SNPs were found, distributed across 146 contigs.Eighty-four candidate SNPs were chosen to genotype 642 individual mussels from 28 populations.The assay design failed for 16 SNPS, 6 did not provide an acceptable quality score and 2 were monomorphic in all samples.The remaining 60 SNPs were polymorphic with a high quality score (above 90%) in most (98%) of the sampled individuals.SNP annotation is presented in Table S1 in the Supplement at www. int-res.com/articles/ suppl/ b021 p025 _ supp.pdf.Seven SNPs were identified in contigs that included the histone and hsp70 genes (Table S1), as reported in the study of European populations of the Mytilus spp.mussels (Zbawicka et al. 2012).ORFs were identified in the majority of fragments.ORF identification was not possible in only 6 contigs (6 SNPs).Of the 60 SNPs, 47 (78.3%) were located in coding regions, and the majority (40; 85.1%) were synonymous.Of the SNPs used, 7 (11.6%)were located in noncoding regions (Table S1).
Of the 60 SNPs, only 3 (BM20A, BM21C and BM9C) had more than 2 alleles each (Table S1).The mean (± SD) minor allele frequency (MAF) for SNPs with 2 alleles was 0.14 ± 0.09.SNPs with MAF < 0.2 were highly represented in the study data.The highest average MAF values were observed for samples from the southern Danish Straits (Table 1).
Very little LD between pairs of 51 markers was found.Only 7 pairs of loci (BM202A vs. BM203D, BM11A and BM201B; BM203D vs. BM11A and BM201B; BM11A vs. BM201B; BM12A vs. BM67B) out of a total of 1275 were in highly significant LD (p < 0.0001) in 3 populations from the southern part of the Belt Sea: 1 pair in EGH, 7 in GED and 4 in HJE.One pair of SNPs (BM12A vs. BM67B) was in highly significant LD in the specified 3 samples and in one group of populations of predominantly the Mytilus tros sulus type.For 3 pairs of loci with BM201B, highly significant LD was found in 1 population (GED).
The frequency distribution of the hybrid index (HI) in the study populations is presented in Fig. 2. The percentage of M. edulis characteristic alleles (at loci with F ST > 0.5), showed clinal variation for the Kattegat-Danish Straits-Baltic Sea region.The greatest variance of HI was observed in populations GED and HJE with the highest number of loci with departures from HWE (Table 1).

Genetic and Hardy-Weinberg equilibrium
The percentage of polymorphic SNPs (P o ) ranged from 35.0 to 98.3% among populations, and was highest in populations from the southern Danish Straits.The most homogenous population was the M. edulis population from the North Sea, whereas almost all loci in 2 populations from the southern Danish Straits (GED and HJE) were polymorphic.
Most loci were in HWE in the different populations.The biggest fraction of SNPs that were not in HWE (p < 0.01) was observed in samples from the southern Danish Straits (GED and HJE) (Table 1).Only 5 loci showed significant departures from HWE after a Bonferroni correction.BM17B (ribosomal protein) was not in HWE in 5 populations from the southern Danish Straits and Baltic Sea, while BM33B (cytochrome c oxidase) was not in HWE in 3 populations from the same region because some subunits are encoded in the mitochondrial genome.H o for 60 loci among all populations was in general lower than H e .Heterozygosity among populations was generally lowest in the North Sea and the northern Danish Straits, increased toward the south, and decreased slightly in the inner Baltic (Table 1).The mean within-population fixation index F IS (averaged over all polymorphic loci in each population) was 0.15, showing a significant excess of homozygotes in 9 populations and in all study regions (Table 1).

Genetic differentiation among populations
We investigated the potential of SNP markers for discriminating between populations of Mytilus spp., concentrating on mussels from the Baltic Sea region.The level of genetic differentiation among the studied populations was high (F ST = 0.257).F ST values were sig ni ficantly greater than zero between most of the pairs of samples (reaching values as high as 0.874) (Table S3 in the Supplement at www. int-res.com/articles/ suppl/ b021 p025 _ supp.pdf).Values not significantly different from zero (indicating the absence of differentiation) were observed in comparisons within groups of 10 northern Danish Straits samples as well as the 8 inner Baltic Sea samples.The mean pairwise F ST values were 0.06, 0.191 and 0.344 for the North Sea vs. northern Danish Straits, northern vs. southern Danish Straits and southern Danish Straits vs. inner Baltic, respectively, all of which were highly significant (p < 0.01).AMOVA also revealed significant differences between these 3 groups.The highest dif ferentiation was observed between the North Sea samples and the Baltic Sea with the Canada group (F ST = 0.704).However, comparing the Baltic Sea and Canada samples, F ST values were also high (0.258) and significant.In order to assess the introgression of Baltic populations, F ST values between M. edulis and Baltic M. trossulus were

Population
A CA test was carried out to characterize population structure.CA results for individuals are shown in Fig. 6; for higher resolution, only samples from the Baltic Sea environments are presented.The first 2 axes accounted for 91% of the total variation.Axis 1 shows a separation between the North Sea and Danish Straits popu lations with a predominance of M. edulis (groups G1, G2 and G3) and populations possessing the M. trossulus genes (G4 and G5).The North Sea and northern Danish Straits individuals formed a very tight group, partially overlapping with some southern Danish Straits individuals, while the inner Baltic group had a greater dispersion.Individuals from the 3 most south-eastern Danish Straits populations GED, HJE and ORE either grouped with M. edulis, M. trossulus or were situated between them.Results for all populations are shown on Fig. S1 in the Supplement at www. int-res.com/articles/ suppl/ b021 p025 _ supp.pdf.

SNP features
We described and characterized 60 SNPs as potential markers for Mytilus spp.mussels from the North and Baltic Sea regions.These SNPs were used to genotype 642 mussels collected from 28 localities.
Genotyped data very clearly differentiated populations and allowed an accurate description to be made of the M. edulis × M. trossulus hybrid zone in the Danish Straits.
Based on the analysis of linkage disequilibrium, it can be concluded that almost all studied SNPs were independent characters.Seven pairs of loci, all with highly significant LD, were characteristic of the M. trossulus genome and could thus account for the linkage signature.The highest level of LD (as was to be expected even without physical linkage) was observed in populations from the centre of hybridization (southern part of the Belt Sea), in particular in the GED and HJE samples.The greatest variance of hybrid index score, which is a straightforward multilocus measure of linkage disequilibrium in hybrid zones (Kruuk et al. 1999), was observed in the same populations.Analysis of inter-marker disequilibria in mixed populations often reveals LD between any 2 alleles at different loci with different allele frequencies, even when the loci are unlinked (Gorroochurn et al. 2007).Such a situation was observed in the study of SNP markers in Scottish and Norwegian M. edulis−M.trossulus populations (Zbawicka et. al 2012).

Baltic Mytilus hybrid zone characteristics and SNPs as genetic markers
The variability between markers and genes that could be affected by hybridization in the geographic polymorphisms (SNPs) analysed.Analysis was done for 2 groups: G1−G2 (populations with a predominance of Mytilus edulis genes) and G5−G6 (populations with a predominance of M. trossulus genes).SNPs that had an F ST value above the 0.95 or below the 0.05 quantile (upper and lower solid lines) were considered to be outlier loci.Each SNP is indicated by a diamond cline were examined.When the results from an analysis of allele frequencies (Table S4 in the Supplement at www. int-res.com/articles/ suppl/ b021 p025 _ supp.pdf) and outlier analysis were compared, the 49 SNPs found in 4 DNA fragments and 40 EST contigs were candidates for markers differentiating Baltic Sea region populations and affected by hybridization in the Danish Straits.Grouping of the populations had an impact on the identification of outlier markers.Groups were defined on the basis of the F ST distance matrix and CA analysis results.Twenty six SNPs were significantly involved in the differentiation between mussels from the North Sea and Baltic area (Fig. 4), showing very sharp differences in frequencies (Fig. 5) in a similar way to allozymes (Väinölä & Hvilsom 1991, Väinölä & Strelkov 2011) and 2 of the 4 nuclear DNA markers studied by Stuckas et al. (2009): lysin M7 and male mtDNA.An evenly distributed upward trend of the number of individuals with M. trossulus genes can be observed from the North Sea to the inner Baltic and Canada for another 23 SNPs (Table S4).Although specific alleles at 3 SNPs (BM22A, BM60A and BM3B) were more frequent in the inner Baltic than in the other study region, the differences were not significant.Adding the introgressed populations at the border of the hybrid zone to the groups results in overinflating the heterogeneity in differentiation level and consequent reduction of outliers.
Introgressive hybridization is very strong for maternally inherited mtDNA and a little weaker for the nuclear ITS and SNP5B markers, resulting in complete or almost complete replacement of M. trossulus with M. edulis mtDNA or fragments of nuclear DNA (Quesada et al. 1995, Kijewski et al. 2006, Zbawicka et al. 2012).In the present work, a similar phenomenon was observed, with a high frequency of the alleles characteristic of pure M. edulis occurring in 5% of the SNPs (e.g.BM2G, BM86A and BM98C) (Table S4).In contrast, about 23% of the SNPs that were characteristic of a putatively pure Canadian M. trossulus population were observed in Baltic populations at a very high frequency (90 to 100%) (e.g.BM11A, BM12A and BM203D) (Table S4).A effect had already been ob served for the EF-bis marker and for allozyme data, and was described as a cline (e.g.Gpi locus) (Väinölä & Hvilsom 1991, Kijewski et al. 2006, Stuckas et al. 2009, Väinölä & Strelkov 2011).This may have resulted from selection and differential introgression (Bierne et al. 2003).None of the inner Baltic individuals had only M. trossulus-or M. edulis-specific SNP alleles.Another characteristic of this region was differing levels of introgression observed at some loci over relatively short distances (southern Danish Straits, e.g.VEJ and HJE, ORE and GED).Stuckas et al. (2009) argued for discordance of the shape of multilocus clines across the Baltic contact zone between M. edulis and M. trossulus, based on 3 nuclear and 2 mitochondrial markers with very small sample sizes, while Väinölä & Hvilsom (1991) and Väinölä & Strelkov (2011) both reported concordant allozyme clines with more extensive sampling.The analysis of SNP frequency in this study showed clinal variation for the region comprising the Kattegat, Danish Straits and Baltic Sea.Alleles predominant or common in the North Sea area were nearly absent from the other areas.In this work, most of the studied SNPs showed a concordant and abrupt genetic shift around the Øresund and Danish Belts, which confirms the data presented for some selected SNPs (Fig. 5).
The presence of genome-wide gene tic barriers to gene flow and the appearance of the F ST outliers is partly the result of the history of these mussels and the hy brid zone that, with long periods of allopatric isolation, would have permitted the accumulation of many in trinsic incompatibilities throughout the genome (Bierne et al. 2011).Local adaptation caused by different salinity that leads to non-neutrality, favouring one gene type inside the Baltic and another outside it, should also be taken into account.Thus, the barriers to gene flow can have a multifactorial character, both exogenous and endo genous.The existence of a semi-permeable genetic barrier is confirmed by the trend of evenly distributed SNP frequency across the hybrid zone (BM6C, BM16A, BM33B, BM73B and BM88A).Baltic populations of M. tros sulus are all more introgressed than the Pacific population, shown by comparison of F ST values between M. edu lis and both M. trossulus popu la tions.However, in a locus-specific analysis, 8 SNPs characteristic of a putatively pure Canadian M. trossulus population were observed in inner Baltic populations with 100% frequency (Table S4).
Similar to the mussel Mytilus spp., genomic clines over the Sound and the Danish Belts with selection for certain variants of genes have also been observed in the bivalve Macoma balthica (Luttikhuizen et al. 2012) and some fish species (e.g.André et al. 2010).Based on allozyme loci in M. balthica, it had also been observed that the transitions in different characters do not coincide exactly, and the extent of introgression varies among loci (Nikula et al. 2008).Barriers to Baltic-Atlantic gene flow were also observed in other species including pike, whitefish, stickleback and bladderwrack (Wennerström et al. 2013).

Characteristics of the studied Mytilus spp. populations and interpopulation differentiation
Two populations from the southern Danish Straits (from the eastern islands of Falster and Moen: GED and HJE) had the largest number of polymorphic loci (almost 100%), the highest gene diversity, and considerable departures from HWE (Table 1).These re - sults indicate the location of the centre of the hybrid zone, which is characterized by strongly impeded gene flow.In some populations (KUL, ORE and HJE), approximately 17% of loci had very large (> 0.6) within-population fixation index values (F IS averaged across all loci) (Table S5 in the Supplement at www. int-res.com/articles/ suppl/ b021 p025 _ supp.pdf).The ma jority of these loci were nearly homozygous for 1 allele within a population, with the absence of any heterozygotes or a single heterozygote only, and a few individuals homozygous for the alternative allele.An excess of homozygotes of the different taxa was observed.In a few cases, F IS had values of 1, indicating the existence of 2 alternative homozygotes without any heterozygotes (e.g.BM11A, BM17B, BM54A and BM55A).Pure M. edulis populations were mostly homozygous for this alternative allele, and the M. trossulus population from Canada was predominately homozygous for the second allele.This phenomenon was observed to the greatest extent in the Scottish and Norwegian populations of Mytilus spp., and was probably caused by recent mixing of M. trossulus with M. edulis (Zbawicka et al. 2012).In contrast, in the Danish Straits there is a long history of divergence, involving endogenous and exogenous barriers to gene flow as well as local adaptation (Bierne et al. 2013).Väinölä & Hvilsom (1991) observed a similar situation for some allozymes and explained it as selection against hybrids in later generations.These phenomena can be useful in determining the adaptive differentiation of populations identified from differences in allele frequencies among populations (i.e.F ST ).
Based on the present study, it can be concluded that the boundary separating populations with a predominance of M. edulis genes from those with a predominance of M. trossulus genes is located around the eastern islands of Falster and Moen, where the salinity is 10 to 14 PSU.For most of the examined SNPs, allele frequencies change abruptly, creating concordant narrow clines at the Baltic Sea entrance (Kattegat and the Sound).This study revealed the complexity of the Baltic hybrid zone and clearly showed how populations of the Danish Straits and Øresund are more an admixture of the 2 taxa than a unimodal hybrid swarm in the inner Baltic.In conclusion, the majority of new SNPs showed greater frequency of M. trossulus than M. edulis genes in the nuclear DNA of Baltic Mytilus spp.The results indicate the high potential of the new SNPs in the study of hybridization zones, and in particular as a new tool to study population structure and illustrate admixture level of mussels of the genus Mytilus.

Fig. 1 .
Fig. 1.Geographic locations of 27 Mytilus spp.sampling sites in the Baltic and North Seas.SeeTable 1 for definitions of site abbreviations

Fig. 3 .
Fig. 3. Neighbour-joining tree of Mytilus spp.populations based on the F ST distance matrix

Fig. 4 .
Fig. 4. Detection of outlier loci.Distribution of F ST values as a function of the heterozygosity under a neutral model for the 60 single nucleotidepolymorphisms (SNPs) analysed.Analysis was done for 2 groups: G1−G2 (populations with a predominance of Mytilus edulis genes) and G5−G6 (populations with a predominance of M. trossulus genes).SNPs that had an F ST value above the 0.95 or below the 0.05 quantile (upper and lower solid lines) were considered to be outlier loci.Each SNP is indicated by a diamond

Fig. 5 .
Fig. 5.The single locus clines in the Kattegat through the Sound for the most differentiated 20 loci described by allele frequency as a function of distance from North Sea populations

Fig. 6 .
Fig.6.The first 2 axes of the correspondence analysis (CA) computed from the single nucleotide polymorphism (SNP) data on Mytilus spp.populations from Baltic Sea region.Each dot (point) is an individual.Seven populations of M. edulis−M.trossulus from the southern Danish Straits are indicated with a different symbol (G3 and G4), whereas populations from the North Sea (G1), north Danish Straits (G2) and inner Baltic Sea (G5) are shown with the same symbol for each group