Recent genetic changes in enhanced populations of sea trout ( Salmo trutta m . trutta ) in the southern Baltic rivers revealed with SNP analysis

The genetic structure of a fish population is usually thought to be stable over time. In the southern Baltic, Salmo trutta m. trutta (sea trout) populations have been characterized by low degree of genetic differentiation. All studied populations had been heavily stocked with mixed material for many years prior to the sampling period, including releases of Pomeranian sea trout to the Vistula River in Poland, Southern Baltic. However, the strategy of stocking became river based a few years before the sampling began. Juveniles from artificial reproduction are released only to their parental river, which reduces the mixing of the gene pool of fish from different populations. Changes in sea trout populations in the southern Baltic over time were studied using single nucleotide polymorphisms (SNPs). Genetic composition of populations of sea trout in the Vistula and Drwȩca river system were found to increasingly resemble the non-admixed hatchery population from Aquamar (Miastko, Poland), whereas the Pomeranian populations were stable. The implementation of a new stocking strategy for the restoration and protection of Vistula sea trout was noted as possible explanation. With the increase of natural breeding, artificial enhancement of sea trout populations should be reduced.


Introduction
The supplementing of diminishing endangered natural populations or restitution of extinct populations with reared specimens has been a common practice applied as a countermeasure to deterioration of living resources and the environment (Araki and Schmid 2010;Laikre et al. 2010).The enhancements have often been reported to have had negative effects on natural populations of fish (Ward 2006).These enhancements have included releases of artificially reared stocks or escapes of large numbers of individuals characterized by lower fitness (Satake and Araki 2011;Baskett et al. 2013;Milot et al. 2013;Naish et al. 2013) and lower genetic diversity have been reported for some species, including Adriatic sturgeon (Boscari and Congiu 2014), Korean starry flounder (An et al. 2013) and black sea bream in Japan (Blanco Gonzalez and Umino 2009).Other effects have been introductions of nonindigenous populations, such as salmon in Poland (Poćwierz-Kotus et al. 2015a;Bernaś et al. 2016); or alien populations of the same species and their subsequent hybridization a Corresponding author: rwenne@iopan.gda.pl with local native populations, as with salmon and brown trout in Bulgaria (Chelenkova et al. 2012), Norway (Thaulow et al. 2012), Hungary (Horváth et al. 2013) and in Spain (Madeira et al. 2005), pikeperch in Finland (Salminen et al. 2012); unintentional escapes from hatcheries as red sea bream in Japan (Blanco Gonzalez et al. 2015), salmon in Norway (Liu et al. 2013) and Canada (Fraser et al. 2008).Accidental or deliberate introductions of exotic and invasive species to new environments, often beneficial economically, causes distortions in the functioning of local ecosystems and can threaten native populations of fish and shellfish, as in the case of brown trout in Japan (Hasegawa and Maekawa 2008), Patagonia (Vigilano et al. 2007;Young et al. 2010) and in North America (Turek et al. 2016), mussel Mytilus in Europe and South Africa (Kijewski et al. 2009;McQuaid et al. 2015 and oyster Crassostrea gigas (Carlton 1979;Meehan et al. 1989;Miller et al. 2012;Lallias et al. 2015).Stocking and introductions to marine farms can also be the means of spreading diseases, as in salmon in Chile (Murray 2013;Marshall et al. 2014) and oyster Crassostrea gigas in Europe (Mineur et al. 2014).The application of genetic research facilitated more effective stocking e.g. as in sea bass in the USA (Darden et al. 2013) and a reduction of their negative effects in white seabass populations (Gruenthal and Drawbridge 2012) and bay scallop (Bert et al. 2011).
In the last decades, a range of new molecular genetic techniques have been used to study fish and shellfish evolution, and changes in wild populations including those associated with exploitation, stocking and research related to aquaculture (e.g.Utter et al. 1987;Wenne 1992;Poteaux and Berrebi 1997;Halvorson et al. 2001;Was et al. 2004;Poćwierz-Kotus et al. 2007;Wenne et al. 2007;Filipowicz et al. 2008;Beaumont et al. 2010;Poćwierz-Kotus and Wenne 2010;Poćwierz-Kotus et al. 2010;Wenne et al. 2011;Kohout et al. 2013;Nakajima et al. 2014).Microsatellites are regularly used in fisheries related applications and research (Was and Wenne 1998;Griffiths et al. 2013;Perrier et al. 2013;Pukk et al. 2013;Bernaś et al. 2014;Koljonen et al. 2014;Thaulow et al. 2014;Valiquette et al. 2014).However, the high potential of more recently discovered and developed single nucleotide polymorphisms (SNPs) to uncover population structures and find genomic regions (islands) under selection has become increasingly recognized (Lamaze et al. 2012;Pustovrh et al. 2012;Ozerov et al. 2013;Milano et al., 2014;Poćwierz-Kotus et al. 2015b;Sušnik Bajec et al. 2015).SNPs are biallelic markers characterized by codominant inheritance, lack of homoplasy and an infinite model of mutation (Morin et al. 2004;Beacham et al. 2010;Seeb et al. 2011) Analyzing the drawbacks of use SNPs the most important are lower discrimination power in comparison with microsatellites, non-neutrality, possibility of random genetic drift and ascertainment bias.Glover et al. (2010) demonstrated that identification of highly informative SNP markers from the screening of larger pools represents a powerful technique to create molecular tools, both to study individual membership of populations and for population differentiation.
Brown trout (Salmo trutta L.) is a palearctic salmonid species, which is naturally distributed in Europe and adjacent waters (Klemetsen et al. 2003).It was introduced to, and subsequently colonized drainages, lakes and coastal waters in North (Canada and USA) and South America (Chile and Argentina), Kerguelen, New Zealand, southern Australia and Japan.The anadromous form of brown trout, sea trout (S. t. trutta) usually spawn in freshwater streams and rivers with juveniles generally staying 1-3 years in the natal rivers before smoltification from the parr stadium and migration to the sea.A decline in natural reproduction resulted in a significant reduction in the number of smolts descending to the sea which, in turn, justified development of massive stocking in Europe and most Baltic countries since many years including Poland (Bartel 1993).On average between the years 2002-2012, 34 mln smolts were released in Baltic countries, including 9.5 mln in Poland (ICES 2013).Many more alevins and fry were released; 7 mln in Poland in 2012 alone.
In Poland there are about 25 rivers, in which sea trout occur (Pedersen et al. 2012).Building dams without appropriate passages prevented sea trout migration and spawning up the rivers.Regular stocking of the Vistula and Pomerania rivers has been conducted since the 1960s.But this was carried out without regard to the origin of the sea trout stocked.Therefore, populations were mixed before the 1990s.Mixed stocking may have started in the nineteenth century but political upheavals over the period make a precise tracking of the stocking process difficult.All populations in rivers are considered as mixed after approximately 120 years of occasional, and more than 50 years of intensive stocking.According to the HELCOM (2011) report, the status of the sea trout populations from the Vistula River was presented as 'reared' (no, or almost no, natural reproduction) with migratory hindrances.Of the populations from Pomeranian rivers: Parsȩta and Słupia (Fig. 1) were qualified as mixed, with natural reproduction and with large-scale fish releases; whereas Rega and Wieprza rivers showed lower (below 10% of smolts) natural reproduction.In the case of the Drwȩca River, a tributary of the Vistula, the level of wild smolts was much lower than in Pomeranian rivers (HELCOM 2011).
In the past, the Vistula sea trout had the largest catchments compared to other river populations in the Baltic Sea drainage (Bartel 1988) and consisted of two stocks: winter and summer.The main part of the population used to be the winter stock ( Żarnecki 1963).Th decline in the sea trout Vistula population, compared to the 1950s and 60s, was caused by the construction of a dam in Włocławek in 1968 (Bartel 1993).The low number of eggs obtained from Vistula specimens resulted in the translocation of stocking material in the 1970s from Pomeranian sea trout, originating from the Parsȩta -Słupia rivers area, to the Vistula River.This translocation caused the decay of original Vistula sea trout characteristics (Borzȩcka 2010).
Several million sea trout at early stages of development (alevins, fry, and parr) and smolts have been released annually into the Vistula River and its tributaries.For smolts released to the Vistula River and its tributaries, homing was very high and ranged from 94% to 99.3% (Dȩbowski and Bartel 1995).However, smolts released into Pomeranian rivers estuaries had a significantly lower percentage of homing.The stocking of the Pomeranian rivers comprises 30-40% of the total of Polish sea trout releases (Pedersen et al. 2012).The 'Stocking of Polish marine areas' programme of The Ministry of Agricultural and Rural Development Board currently regulates both stocking and the duties of particular fishing region users.Juveniles from artificial reproduction are released only to their parental river, which restricts the mixing of the gene pool of fish from different populations as happened in the past, before 1990s.In addition, the passages in Poland have been improved since and this may also have contributed to the enhanced restitution of natural spawning.Studies carried out in the middle of 1990s using allozyme markers, shortly after implementation of the more restrictive stocking strategy indicated that the genetic structure of populations from the Pomeranian rivers Rega and Parsȩta, as well as from the Słupia and Vistula rivers was disturbed (Woźnicki et al. 1999;Łuczyński et al. 2000;Wenne et al. 2000).A low level of differentiation between Vistula River and łupia River populationshas been demonstrated using mtDNA (Włodarczyk and Wenne 2001), and growth hormone gene 2 (GH2) marker analysis (Włodarczyk et al. 1999).A slightly higher level of differentiation was observed by the application of microsatellite DNA (Was and Wenne 2002;2003).However, no molecular markers have been found suitable for application in tracing the origin of spawners entering the Vistula river.Newer studies based on SNP methodology enabled the differentiation of the Vistula and Pomeranian (Słupia) sea trout populations (Drywa et al. 2013).
The aim of this paper was to find changes over 13 years in genetic structure of very weakly spatially differentiated and intensively enhanced sea trout populations in southern Baltic, Poland.

Sample collection and DNA extraction
Two sets of sea trout samples collected from 6 rives in 1996 and 2009 were studied.Approximately 5 mm 2 fin clip samples were collected from 595 individuals.Adult spawners were caught in traps in 5 rivers and electrofished in Parsȩta River in 1996 and 2009, and adult spawners from the Aquamar hatchery in Miastko, Poland in 2005 used for artificial spawning (Fig. 1, Table 1).The hatchery stock in Aquamar is thought to represent a native winter migrating group of sea trout in the Vistula River and is used for production of stocking material for releases in the Vistula only.Fin clips were stored in 95% EtOH solution at 4 • C prior to DNA extraction.Genomic DNA was extracted using Genomic Mini Kit (A&A Biotechnology).DNA quality was assessed by running sample aliquots on 1% agarose gels.Samples were quantified using the Epoch TM Spectrophotometer System (BioTek) and diluted to a final concentration of 20ng/μl in TE buffer.

SNP genotyping and evaluation of genetic differences between populations
Assays were designed for 62 candidate SNPs selected from results of genotyping with the Atlantic salmon custom design Illumina iSelect SNP array containing 15 225 markers of 24 samples of sea trout from Vistula and Slupia rivers (Drywa et al. 2013;2014;Poćwierz-Kotus et al. 2014).SNP genotyping was performed using the Sequenom MassARRAY iPLEX platform (Gabriel et al. 2009) in Centre for Integrative Genetics, Department of Animal and Aquacultural Sciences, Norwegian University of Life Sciences, Ås, Norway.Polymorphic loci in the studied sea trout populations were selected from genotyping results using locus-by-locus AMOVA (Analysis of Molecular Variance) after 10 000 permutations.Population genetic differentiation was analyzed using global weighted average F-statistic over loci and pairwise F ST (Weir and Cockerham 1984;Weir 1996) with 1000 permutations and significance level P < 0.05.The critical probability value for each test was adjusted using the Bonferroni correction (Rice 1989) using FSTAT software 2.9.3.2 (Goudet 1995).Genetic diversity was assessed by number of polymorphic SNPs (P O ) and values of observed (H O ) and expected (H E ) heterozygosity (Nei 1978;1987).All tests, except Bonferroni correction, were carried out by Arlequin 3.5.1.2(Excoffier and Lischer 2010).
Takezaki's POPTREE program was used to create a neighbor joining tree using D A distances (Nei 1978) with 1000 bootstrap replications (Takezaki et al. 2010).The identification of distinct genetic populations and individuals assignment to them was performed with a Bayesian-model-based, clustering analysis using Structure 2.3.4 (Pritchard et al. 2000).Individuals were assigned to predefined K populations (from K =1-16 with 10 iterations for each K), where each K was characterized by a set of allele frequencies for each locus.An admixture model with no a priori information on the origin of the individuals with 10 000 repetitions as burn in, and 100 000 repetitions after burn in, was applied.Individuals were probabilistically assigned to one or more clusters if their genotypes indicated a mixing.The true K was estimated following the method described by Evanno et al. (2005) using STRUC-TURE HARVESTER program (Earl and vonHoldt 2012).The CLUMPP 1.1.2(Jakobsson and Rosenberg 2007) software was used to average the results from the ten replicate cluster analyses.Full Search algorithm and 10 000 random permutations were applied.
Correspondence between populations was assessed using a three-dimensional Factorial Correspondence Analysis (FCA) implemented in GENETIX 4.05.2 software (Belkhir et al. 2000;Benzécri 1992).The ability of the selected markers set to assign individuals to the most likely population, indicated by STRUCTURE 2.3.4 software, was assessed using ONCOR software (Kalinowski et al. 2007).Assignment tests to evaluate the accuracy of identification of individuals to populations by leave-one-out method were performed.Two approaches were applied.First, with a priori baseline populations and second with reporting groups based on clustering results.

Characterization of SNPs, quality control and basic statistics
A total of 595 sea trout specimens from 6 rivers and a hatchery in Poland were genotyped.Candidate SNPs were divided into three groups: failed (lack of genotyping results), monomorphic and polymorphic.Missing data rate, set at a 5% level, reduced the number of specimens to 576 and loci to 47, of which 14 loci were monomorphic and another 4 had minor allele frequency (MAF) values lower than 0.01.The remaining 29 loci were used for further analysis.AMOVA analysis resulted in the selection of 22 polymorphic loci (P O ) with statistically significant results in all populations.Table 2 shows frequencies of alleles for the 13 studied sea trout populations.
The most common polymorphism was A/G appearing in 12 loci (Table 3).The global F ST was 0.03.The highest F ST value for individual loci obtained in AMOVA was 0.095 for locus no.14 and the lowest was 0.007 for locus No. 21.The highest heterozygosity values were observed for 6 loci with values around 0.5.And the lowest heterozygosity values were present in 3 loci with values less than 0.1 (Table 3).Most of the loci were in Hardy-Weinberg equilibrium (HWE).The highest numbers of loci in H-W disequilibrium were found in the hatchery population TM 05 (Table 4).This may have been an unfavorable effect occurring in breeding: founder effect, nonrandom mating and limited gene pool, or it may have been a high number of closely related individuals in the sample analyzed (Was and Bernaś 2016).SNPs were polymorphic (P O ) in different numbers for particular populations, from 18 for TR 96 population to 22 for 4 populations; TV 96, TD 96, TW 96 and TS 09.The percentage of polymorphic SNPs ranged from 82% to 100% in populations.Average gene diversity over loci in populations ranged from 0.170 for TM 05 to 0.270 for TW 09 (Table 4).A significant excess of homozygotes was noticed in 3 populations, mainly in those collected in 1996 (TV 96, TR 96) and in the population TD 09 (with the largest excess).

Genetic differences between populations
Genetic differences between pairs of populations ranged from 0 to 0.16 (Table 5).Forty seven (47.44%) of the F ST values greater than 0 were statistically significant (P < 0.05).The samples collected from rivers in 2009 showed greater differentiation than those from 1996.Vistula samples from 2009 (TV 09 and TD 09), when compared to samples from the Pomeranian rivers from 2009 showed greater differentiation than shown by a similar comparison of the same sampling sites from 1996.After Bonferroni correction 8 (10%) of the F ST values were statistically significant (P < 0.00064).Among the other pairs of populations statistically significant differentiation was found for the Vistula and Wieprza populations from 2009.The hatchery population (TM 05) was the most divergent and differentiated from all other samples, but less divergent from Vistula (TV 09) and Drwȩca (TD 09) samples.Neighbor-joining (NJ) trees constructed on the basis of the D A distances from 22 loci showed genetic relationships between the analyzed populations.In samples from the 1996 differentiation between Vistula River and Pomeranian rivers was not observed.Even more, the TV 96 group was clustered with TS 96 and TP 96 (Fig. 2a).13 years later Vistula sea trout populations (including Drwȩca River) were significantly different from Pomeranian populations.The analysis of 7 populations demonstrated similarity between Vistula (TV 09) and its tributary Drwȩca (TD 09) and a hatchery sample TM 05 (Fig. 2b).
STRUCTURE analysis was carried out for 22 selected loci.The highest value of delta K was estimated for K = 3 and reached 1900.56 without other significant peaks (Figs.3a  and 3b).A Factorial Correspondence Analysis (FCA) was carried out to investigate genetic relationships among the 12 samples from rivers (Fig. 4).The first two axes accounted for 69% of the total variation.The first axis (42.42% variance) showed separation between all Pomeranian samples from 1996 and 2009 and Vistula River samples from 1996, and the Vistula trout samples from 2009: TD 09 and TV 09.In addition, all samples from 1996 formed a very homogenous group while Pomeranian samples from 2009 were more spread.The leaveone-out test, in ONCOR, found 17.7% of fish correctly assigned back to the river from which they were sampled.After the formation of two main reporting groups, by structure analysis results, the proportion of correctly self-assigned individuals increased to 82%.In the first group consisting of all samples apart from hatchery TM 05, self-assignment accounted for 85%.In the second, hatchery TM 05, self-assignment was 78%.

Discussion
The negative effect of dams on populations of freshwater and anadromous fish as manifested by blocking spawning and foraging migrations, population fragmentation or changed water regimes have been reported often (e.g.Jager et al. 2001;Morita and Yamamoto 2002).All of them have genetic consequences, for example reduction in effective population size, shift in migration-selection-drift equilibrium and in local adaptations (Meldgaard et al. 2003;Faulks et al. 2011;Paris et al. 2015).A divergence of brown trout populations isolated by dams established as early as middle ages has been found (Hansen et al. 2014).
Considerable temporal stability of genetic composition of brown trout populations have previously been demonstrated (Hansen et al. 2002;Jensen et al. 2005).Studies, based on microsatellite DNA, in some cases resulted in the observation of temporal changes in the genetic composition, which may be explained by drift in small persisting populations of brown trout (Ostergaard et al. 2003).However, in larger populations of salmon, temporal variation in allele frequencies was due to the effect of hatchery practices (Koljonen et al. 2005).Hansen et al. (2009) have shown admixture of brown trout populations in Denmark with hatchery trout.The admixture may have caused reduced local adaptation.However, in contrast to Danish populations, Polish populations have been very  weakly differentiated due to low homing in Pomeranian populations and long term artificial supplementation (Dȩbowski and Bartel 1995).The Vistula sea trout populations underwent a bottleneck effect in early 70ties due to the construction of a dam in Włocławek in 1968 (Was and Bernaś 2016).
Part of genetic unique genotype composition of the native populations in Vistula was lost before the intensive stocking with the Pomeranian sea trout was carried out and cannot be restituted in living papulations.Temporal changes in genetic composition in the studied populations of sea trout in Poland, southern Baltic as identified using SNPs, though low, were observed and concerned mainly the Vistula populations.Values of pairwise F ST for populations sampled in 2009 were higher suggesting that they are more diverse than those from 1996.There were no significant values in pairwise F ST in the Pomeranian and Vistula river populations (TV 96 and TD 96).This may be explained by intensive enhancement and the practice of mixing the stocking material carried out in the past.Samples from the Vistula River and Drwȩca River collected in 2009 were not significantly differentiated from each other, but were different from each Pomeranian river.
The pairwise comparison between individuals from the same rivers collected at different times (1996 and 2009) revealed that over these 13 years the biggest change was observed for the Vistula River (TV 96 and TV 09) and its tributary the Drwȩca River (TD 96 and TD 09) while Pomeranian populations were more stable over time.Pairwise differences of F ST values between the hatchery population and all the wild populations were statistically significant.The changes in the genetic structure of sea trout populations from Vistula and Drwȩca rivers over the period 1996-2009, observed in this study, may be linked to the new strategy of the 'Stocking of Polish marine areas' programme of The Ministry of Agricultural and Rural Development Board.The reduction of the putative Pomeranian genotypes among Vistula populations may also be a result of this stocking strategy, and may lead to a partial recovery of the native Vistula sea trout population.Despite the release of hatchery offspring of individuals chosen for artificial spawning solely into the parental river, specimen selection is not currently based on genetic tests.This leaves a real possibility that spawning individuals can be erroneously selected from 'contaminated' stock, with a pool of ancestors from the Pomeranian populations.Therefore a further genetic monitoring of the source spawners for production of hatchery offspring needs to be developed and performed.However, implementation of genetic monitoring can reduce but not entirely eliminate, negative effects of releases of fish reared in hatcheries (Kallio-Nyberg et al. 2015).Supportive breeding programmes, in which the gametes used for carrying out breeding are collected from spawner fish obtained from the same river area to which the resulting juvenile fish are subsequently released may help to overcome the problem of potential lack of local adaptation but not necessarily a partial loss of allelic richness (Selly et al. 2014).A further process of re-adaptation of the restituted population in Vistula River will be needed over some generations.

Conclusion
Salmo trutta m. trutta populations in the southern Baltic have been characterized by low degree of genetic differentiation mainly due to admixture prior to the 1990s.The results of the present study prove the occurrence of temporal changes in populations of sea trout in Poland in the period 1996-2009.These changes can be considered to reflect the historic pattern of differentiation between 2 main groups of populations (Vistula and Pomeranian) and, in the studied period, may result from the implementation of a new stocking strategy for the restoration and protection of Vistula sea trout.Clearance of existing and building of new passes for migration of fishes around dams, restitution of river niches appropriate for natural spawning of sea trout could also have influenced the observed changes.In the last 15 years dozens of new fish passes have been constructed on Pomeranian rivers.Also an old fish pass  (Belkhir et al. 2000;Benzécri 1992).on the Włocławek dam was rebuilt in 2015.All these activities have greatly increased the possibility of sea trout migration.With an increase of natural breeding, artificial enhancement of sea trout populations should be reduced.
Bootstraped neighbor-joining tree based on the D A distances from 22 loci using POPTREE2 software (Takezaki et al. 2010) for sea trout populations sampled in 1996 (a), and 2009 and a hatchery population in 2005 (b).
A Factorial Correspondence Analysis (FCA) and NJ tree revealed a strong separation of the Aquamar (Miastko) hatchery population.The nearest to it was the Vistula sample collected in 2009.The heterozygosity values inferred from SNPs in this study were at a similar level, around 0.3, in populations sampled in 1996 and 2009.The hatchery population Aquamar was bred from the winter Vistula sea trout and it did not seem to have the admixture of the Pomeranian sea trout genotype.It accounts for 15-60% of yearly releases to Vistula in 2000s.The stocking material used for releases in Drwȩca River originated from spawners caught in lower river course and sometimes also from specimens electrofished below Włocławek Dam in the Vistula River.Vistula and Drwȩca populations seem to become more similar to the hatchery sample from Aquamar over time, which may indicate influence of releases of hatchery fish.Increased divergence between Vistula sea trout and Pomeranian populations can be considered as step towards partial restoration of native population in the Vistula basin.
(a) ΔK calculated as ΔK = m|L (K)|/s[L(K)] against K values for sea trout populations to identify the optimum number of genetic units.The data generated with STRUCTURE HARVESTER program (Earl and vonHoldt 2012).(b) Proportion of membership of 526 individuals from 12 sea trout populations for K = 3 calculated by STRUCTURE v.2.3.4 (Pritchard et al. 2000) software and averaged by CLUMPP v.1.1.1 (Jakobsson and Rosenberg 2007) software.Plots were generated by DISTRUCT v.1.1 (Rosenberg 2004) software.

Fig. 4 .
Fig. 4. First two axes of the Factorial Correspondence Analysis (FCA) determined from SNP data for studied trout populations in Poland.Each point represents one sample location.The analysis was performed with Genetix software(Belkhir et al. 2000;Benzécri 1992).

Table 1 .
The location and time of sampling.

Table 3 .
Genetic variability of 22 SNPs in 13 sea trout populations.MAF = Minor Allele Frequency; rs = reference SNP ID from dbSNP; H O -observed heterozygosity; H E -expected heterozygosity.Loci marked with an asterisk have a highly significant deviations (P < 0.01) from Hardy-Weinberg equilibrium.

Table 4 .
Genetic variability of 13 sea trout populations among 22 SNPs.H O -observed heterozygosity; H E -expected heterozygosity; P Opolymorphic loci; Loci no. with highly significant HWE departure (P < 0.01) in populations are listed in parentheses.F IS -the inbreeding coefficient (values with P < 0.05 are marked in bold).

Table 5 .
F ST pairwise differences for 22 SNPs.Bolded values are statistically significant.Values with an asterisk are statistically significant after Bonferroni correction.