Genetic Diversity and Structure of Common Carp (Cyprinus carpio L.) in the Centre of Carpathian Basin: Implications for Conservation

Hungary is one of the largest common carp-production countries in Europe and now, there is a large number of local breeds and strains in the country. For proper maintenance of the animal genetic resources, information on their genetic diversity and structure is essential. At present, few data are available on the genetic purity and variability of the Hungarian common carp. In this study, we genetically analyzed 13 strains in Hungary and, in addition, the Amur wild carp, using 12 microsatellite markers. A total of 117 unique alleles were detected in 630 individuals. Low levels of genetic differentiation (Fst and Cavalli–Sforza and Edwards distance) were estimated among strains. The AMOVA showed the low but significant level of genetic differentiation among strains (3.79%). Bayesian clustering analysis using STRUCTURE classified the strains into 14 different clusters. The assignment test showed that 93.64% of the individuals could be assigned correctly into their original strain. Overall, our findings can be contributed to complementing scientific knowledge for conservation and management of threatened strains of common carp.


Introduction
Nowadays many fish species and subspecies are endangered mainly due to habitat loss, overexploitation, pollution, and climate change [1,2]. As a result of highly effective selection programs and robust environmental changes, the risk of losing valuable genetic materials such as old varieties of a certain species, e.g., common carp strains, has increased [3]. Genetic diversity within populations is the basis of their ability to adapt to the changing environment. Lowered genetic diversity might decrease the adaptability and fitness of populations and can result in the increased extinction risk of the populations [4]. The more frequent use of modern strains in new

Sample Collection and DNA Extraction
A total of 630 samples were collected from 13 Hungarian common carp strains, namely Biharugra scaly (BiS), Biharugra mirror (BiM), Böszörmény mirror (BoM), Hortobágy scaly (HoS), Hortobágy mirror (HoM), Hortobágy wild (HoW), Szarvas-15 (SZ1), Szarvas P3 (SzP), Szeged scaly (SzS), Szeged mirror (SzM), Hajdúszoboszló scaly (HaS), Hajdúszoboszló mirror (HaM), and Tata scaly (TaS) in five hatcheries (Biharugra, Debrecen, Hajdúszoboszló, Hortobágy, and Tiszaszentimre) and the National Agricultural Research and Innovation Centre (NARIC), Research Institute for Fisheries and Aquaculture NAIK HAKI gene bank (National Agricultural Research and Innovation Centre, Szarvas, Hungary). We also collected samples of Amur wild carp (AmW) as outgroup for statistical analysis. Details on the number of samples of each strain, their abbreviations, and coordination of sample collection places for common carp are given in Table 1. Tissue samples were obtained from fin clips. Approximately 0.5-3 cm of caudal fin clips were cut and preserved in 96% ethanol and stored at −20 • C until further laboratory analysis. The study was performed on the basis of the permit issued by the Hajdú-Bihar Megyei Kormányhivatal (reference number: 15/2019/DE MÁB). Sampling of fin clips was carried out after anaesthetizing the fish in clove oil solution (1 drop of concentrated oil to 1 L of water). Total genomic DNA was isolated from the fin clips using the EZNA Tissue DNA kit (Omega Bio-Tek, Norcross, GA, USA) following the manufacturer's instructions. Concentration of the isolated gDNA was measured using a Nanodrop 1000 Spectrophotometer. Samples were diluted to 100 ng/µL for further use.

Population Structure
Pairwise genetic differentiation between strains were calculated using the Wrights fixation index (F st ) and Cavalli-Sforza and Edwards' genetic distance (D) [51] using the INA correction method described in FreeNA [44]. The dendrogram was constructed using the genetic distance based on neighbor-joining method [52] with 1000 bootstrap replicates in POPULATION v. 1.2.28 (https://bioinformatics.org/groups/?group_id=84) to explore relationships among the strains [14,53]. Assignment tests were performed using GENECLASS v.2.0 [54] to assign individuals to their strain of origin and translocation events among the strains based on the likelihood of multilocus genotypes using Rannala and Mountain [55] the Monte Carlo resampling of Paetkau et al. [56] with 10,000 simulated individuals and a significance level of 0.05.
The patterns of genetic structure between strains based on allele frequency in microsatellite data were assessed using the analysis of molecular variance (AMOVA) with Arlequin 3.5.2.2 [57]. One thousand random permutations were performed to assess the significance of each pairwise comparison. Strains were grouped for AMOVA in two ways. First, all strains were grouped together, and all loci were considered. Second, the strains were grouped according to their geographical locations (i.e., hatcheries) into six groups ( Figure 1). The pattern of genetic structures among strains was assessed using the Bayesian approach in STRUCTURE 2.3.4 [58], with admixture model and correlated allele frequency [59], a burn-in of 100,000, with 1,000,000 Markov chain-Monte Carlo (MCMC) repetitions and 10 independent runs for each K value (number of clusters, K = 1−15) to check for consistency across runs and identify genetic clusters. The maximum number of clusters was calculated by adding one to the number of strains to allow detection of substructure [60]. The best K genetic cluster was evaluated by calculating ΔK following the Evanno method [61] in STRUCTURE HARVESTER [62]. Then, CLUMPP 1.1.2 [63] and DISTRUCT [64] were used for estimating the highest similarity coefficient over all runs for different values of K and plotting the clustering results.

Microsatellite Markers
All 14 strains of common carp, in a total of 630 individuals, were successfully screened for 12 microsatellite loci, which were polymorphic in all strains. The analysis by MICRO-CHECKER [43] revealed no indications of stuttering or allele dropout. The frequency of null alleles was generally low ranging from 0.000 to 0.317. Among the 12 loci, evidence of null alleles above the potentially problematic threshold was detected in MFW6, MFW11, MFW13, and MFW26 in some strains (Table  3). The pattern of genetic structures among strains was assessed using the Bayesian approach in STRUCTURE 2.3.4 [58], with admixture model and correlated allele frequency [59], a burn-in of 100,000, with 1,000,000 Markov chain-Monte Carlo (MCMC) repetitions and 10 independent runs for each K value (number of clusters, K = 1−15) to check for consistency across runs and identify genetic clusters. The maximum number of clusters was calculated by adding one to the number of strains to allow detection of substructure [60]. The best K genetic cluster was evaluated by calculating ∆K following the Evanno method [61] in STRUCTURE HARVESTER [62]. Then, CLUMPP 1.1.2 [63] and DISTRUCT [64] were used for estimating the highest similarity coefficient over all runs for different values of K and plotting the clustering results.

Microsatellite Markers
All 14 strains of common carp, in a total of 630 individuals, were successfully screened for 12 microsatellite loci, which were polymorphic in all strains. The analysis by MICRO-CHECKER [43] revealed no indications of stuttering or allele dropout. The frequency of null alleles was generally low ranging from 0.000 to 0.317. Among the 12 loci, evidence of null alleles above the potentially problematic threshold was detected in MFW6, MFW11, MFW13, and MFW26 in some strains (Table 3). Table 3. Estimated null allele frequencies in 14 strains of the common carp using 12 microsatellite loci.

Genetic Diversity
Data for all parameters of genetic diversity for the strains are shown in Table 4. All strains showed deviation from HWE at least at one locus (p < 0.01). The mean number of alleles per each strain was 11. The lowest and highest mean number of alleles was observed for Hajdúszoboszló scaly (6.500) and Biharugra mirror (17.580), respectively. A total of 117 private alleles were detected for the 12 polymorphic loci in 630 individuals. The lowest and highest number of private alleles were observed in Szeged mirror and Hortobágy wild, respectively. The mean observed heterozygosity value (H o ) in the 14 strains was 0.840 with a range of 0.720 in Szarvas-15 to 0.990 in Böszörmény mirror. Additionally, the mean expected heterozygosity value (H e ) in all the strains was 0.800 with a range of 0.710 in Szarvas-15 to 0.890 in Hortobágy wild ( Table 4).
The inbreeding coefficient value varied from −0.250 for Böszörmény mirror to 0.083 for Amur wild carp.

Population Structure
The measure of genetic differentiation (F st ) ranged from 0.028 (between Biharugra mirror and Hortobágy mirror) to 0.231 (between Szarvas-15 and Hajdúszoboszló scaly; Table 5). The Cavalli-Sforza and Edwards' genetic distance (D) ranged from 0.347 (between Biharugra mirror and Hortobágy mirror) to 0.723 (between Hajdúszoboszló scaly and Tata scaly; Table 5). The neighbor-joining phylogenetic tree was constructed based on Cavalli-Sforza and Edwards (1967) chord D c genetic distance (Figure 2). The strains studied formed four clusters. Biharugra mirror, Hortobágy mirror, and Biharugra scaly formed one cluster (Cluster 1) connected with Szarvas P3, Tata scaly, Böszörmény mirror, Hajdúszoboszló mirror, Hortobágy scaly, and Hortobágy wild (Cluster 2). These two clusters were connected to cluster 3 (Hajdúszoboszló scaly, Szeged scaly, Szarvas-15, and Szeged mirror). The observed clustering pattern was not consistent with the geographical area of origin. Finally, all samples of Amur wild formed a distinct group connected to the other three previous clusters. Using the population assignment test, 93.64% of individuals were correctly assigned to their strain of origin (Table 4). Table 5. Pairwise F st values (above diagonal) and D c distance (below diagonal) between 14 common carp strains in Hungary based on 12 microsatellite loci.

Strain
HaS Hortobágy mirror) to 0.723 (between Hajdúszoboszló scaly and Tata scaly; Table 5). The neighborjoining phylogenetic tree was constructed based on Cavalli-Sforza and Edwards (1967) chord Dc genetic distance (Figure 2). The strains studied formed four clusters. Biharugra mirror, Hortobágy mirror, and Biharugra scaly formed one cluster (Cluster 1) connected with Szarvas P3, Tata scaly, Böszörmény mirror, Hajdúszoboszló mirror, Hortobágy scaly, and Hortobágy wild (Cluster 2). These two clusters were connected to cluster 3 (Hajdúszoboszló scaly, Szeged scaly, Szarvas-15, and Szeged mirror). The observed clustering pattern was not consistent with the geographical area of origin. Finally, all samples of Amur wild formed a distinct group connected to the other three previous clusters. Using the population assignment test, 93.64% of individuals were correctly assigned to their strain of origin (Table 4).  AMOVA results (Table 6) revealed that 3.79% of the observed variance occurred among strains, whereas 96.03% was explained by differences within strains. When the analysis was performed after reorganizing the strains in the six geographical groups, the percentage of variation associated to their differentiation was low and non-significant (0.42%, p = 0.270), while the difference among strains AMOVA results (Table 6) revealed that 3.79% of the observed variance occurred among strains, whereas 96.03% was explained by differences within strains. When the analysis was performed after reorganizing the strains in the six geographical groups, the percentage of variation associated to their differentiation was low and non-significant (0.42%, p = 0.270), while the difference among strains within geographical groups was higher and significant (3.62%, p < 0.001). These results support the conclusion that a high level of genetic variation within strains is biologically typical for domestic common carp in the Carpathian Basin. The logarithm probabilities Ln p (X/K), using the preliminary STRUCTURE run, related with different numbers of genetic clusters K, calculated from structure analysis of 630 individuals showed the highest value at K = 14, which was followed by K = 2 ( Figure 3). Based on the value of K = 14, individuals from various strains were significantly different from each other. Eight of the analyzed strains (HaS, SzS, HaM, BiS, Sz1, SzP, AmW, and BoM) were characterized by very high membership coefficients, and each appeared to represent different gene pools. In contrast, the other strains showed some level of admixture rates.
Genes 2020, 11, x FOR PEER REVIEW 10 of 16 within geographical groups was higher and significant (3.62%, p < 0.001). These results support the conclusion that a high level of genetic variation within strains is biologically typical for domestic common carp in the Carpathian Basin. The logarithm probabilities Ln p (X/K), using the preliminary STRUCTURE run, related with different numbers of genetic clusters K, calculated from structure analysis of 630 individuals showed the highest value at K = 14, which was followed by K = 2 ( Figure 3). Based on the value of K = 14, individuals from various strains were significantly different from each other. Eight of the analyzed strains (HaS, SzS, HaM, BiS, Sz1, SzP, AmW, and BoM) were characterized by very high membership coefficients, and each appeared to represent different gene pools. In contrast, the other strains showed some level of admixture rates.

Genetic Diversity
It is emphasized that the number of used loci and their traits can be strongly affect estimates of genetic parameters [67]. In this study, we evaluated the genetic variation of 12 microsatellite loci within 13 common carp strains in Hungary and Amur wild carp. Based on the findings, all strains differed in at least one locus in Hardy-Weinberg equilibrium, which is also supported by previous studies (e.g., [14]). Several studies [14,35,36] suggested that the deviation from the HWE in the common carp strains may be caused by homozygote excess. However, the deviation from the HWE reported in some strains is caused by heterozygote excess [14]. Overall, European common carp populations are generally characterized by homozygote excess [22,36], therefore this finding may result from different practices of individual fish hatcheries [14].
We detected 117 alleles using 12 microsatellite markers in the common carp strains from Hungary. Tomljanovic et al. [38] reported 148 alleles in Croatian common carp strains using 15 microsatellite loci. Napora-Rutkowski et al. [14] detected 45 alleles within Polish common carp strains based on 15 microsatellites, whereas Lehoczky et al. [22] identified 80 alleles within Hungarian strains based on four loci. However, different results in the number of alleles in different studies may suggest that some loci are more polymorphic.

Genetic Diversity
It is emphasized that the number of used loci and their traits can be strongly affect estimates of genetic parameters [67]. In this study, we evaluated the genetic variation of 12 microsatellite loci within 13 common carp strains in Hungary and Amur wild carp. Based on the findings, all strains differed in at least one locus in Hardy-Weinberg equilibrium, which is also supported by previous studies (e.g., [14]). Several studies [14,35,36] suggested that the deviation from the HWE in the common carp strains may be caused by homozygote excess. However, the deviation from the HWE reported in some strains is caused by heterozygote excess [14]. Overall, European common carp populations are generally characterized by homozygote excess [22,36], therefore this finding may result from different practices of individual fish hatcheries [14].
We detected 117 alleles using 12 microsatellite markers in the common carp strains from Hungary. Tomljanovic et al. [38] reported 148 alleles in Croatian common carp strains using 15 microsatellite loci. Napora-Rutkowski et al. [14] detected 45 alleles within Polish common carp strains based on 15 microsatellites, whereas Lehoczky et al. [22] identified 80 alleles within Hungarian strains based on four loci. However, different results in the number of alleles in different studies may suggest that some loci are more polymorphic.
In accordance with Lehoczky et al. [22] and Hulak et al. [36], we estimated relatively high values of expected heterozygosity within common carp strains (in our study, the values ranged from 0.710 to 0.890). In contrast, Napora-Rutkowski et al. [14] calculated slightly lower expected heterozygosity values ranging from 0.418 to 0.781. In agreement with Napora-Rutkowski et al. [14], the observed heterozygosity values calculated were slightly higher than the expected heterozygosity values.
Based on the inbreeding coefficient, in accordance with Napora-Rutkowski et al. [14], many strains are characterized by a heterozygote excess, thus it can be stated that the possibility of inbreeding deterioration is relatively low. The large number of heterozygotes obtained in our study can also be explained by the high number of alleles, which may also be influenced by the relatively low number of individuals. It has previously been suggested that 50-100 individuals per population are needed for proper estimations of genetic distance measurements and genetic structural indicators [68,69]. However, Hale et al. [70] have recommended that a sample size of about 25-30 individuals per population is required to accurately estimate the microsatellite-based statistics of genetic diversity. Many studies (e.g., [71][72][73]) have demonstrated that increasing the number of individuals in such studies has little benefit in terms of allele frequency and expected heterozygosity. After all, what is important is to give an accurate estimate of allele frequency and diversity, and not to determine all alleles because, for example, some rare alleles are not informative for assessing the genetic diversity or the genetic structure of a population [71][72][73].

Population Structure
In accordance with our results, previous studies reported low levels of genetic differentiation among several strains of common carp in the Czech Republic ( [36]; mean F st = 0.183) and France ( [35]; mean F st = 0.250). The low F st values can be explained by the inadequate number of elements of the strains, the presence of null alleles, or genotyping error, but this indicator may affect the characteristic features of the strains sampled [14]. The Cavalli-Sforza and Edwards distance supported low levels of genetic differences among the Hungarian strains. The Cavalli-Sforza and Edwards distance have a relatively high sensitivity in detecting genetically similar populations [74]. We found the smallest genetic distance between Biharugra mirror and Hortobágy mirror, whereas the largest genetic difference was observed between Hajdúszoboszló scaly and Tata scaly. This difference can be explained by the geographical distance between these strains (BiM-HoM: geographical distance ≈110 km; HaS-TaS: geographical distance: ≈300 km). While the NJ phylogenetic tree ( Figure 2) did not well support the geographical grouping of the common carps, the strains from the Tiszántúl Region (Eastern Hungary) show differences, either phenotypically (mirror, scaly, wild) or geographically. The strains grouped in the first cluster come from the gene bank of the Research Institute for Fisheries and Aquaculture NAIK HAKI (Szarvas, Hungary) and the Debrecen hatchery. This result shows that the strains in the Debrecen hatchery (Szeged scaly and Hajdúszoboszló scaly) can be traced back to the founding individuals from the gene bank (Szarvas-15 and Szarvas P3). Individuals from strains originating from the other four regions show mixing, with the exception of strains from the Biharugra hatchery, which are grouped into a separate cluster. Within the cluster, in the case of the Hortobágy wild we obtained the highest individual private allele (Apr = 24) as well as the highest allele richness (1.890) among these strains. The result of the unique allele richness and the location of the cluster allow us to conclude that Hortobágy wild can be considered as a separate strain, which may be the result of artificial selection. The clustering of strains from geographically close hatcheries suggest occasional exchanges of breeding animals between hatcheries. The strains from the gene bank were grouped into two clusters, but they showed a mixture by strains and phenotype (cluster 1: Szarvas-15 and Szeged mirror, and cluster 2: Tata scaly and Szarvas P3). This result may indicate that the gene bank strains are closer to each other, the genetic distance between them is smaller compared to the strains derived from the hatchery, as similarly found by Lehoczky et al. [22]. Another reason may be that the Szarvas strains were developed from the common carp strains collected from all over Hungary, and that is the reason for the closer relationship of the Szarvas strains to some of the other strains. Furthermore, one possible reason for the Tata scaly and Szarvas P3 located in cluster 2, the reason why the Szarvas P33 scaly carp has been isolated from the Tata homozygote scaly (SSnn) common carp strain with individual selection (oral communication). Thereafter, to consolidate the genetic structure and external characteristics of the new line, inbreeding was used during four consecutive generations, combined with very strict phenotypic selection. The inbred line P33 became the maternal line of Szarvas P31 and P34 hybrids. P3 is the founder population of P33, which is produced from P3 by inbreeding. The Hajdúszoboszló mirror and Böszörmény mirror strains appear in one cluster, which may be due to the fact that in the case of Hajdúszoboszló mirror the Böszörmény mirror strains was used for variety improvement purposes. The Hortobágy mirror and the Hortobágy scaly strains are grouped in two separate clusters, which allows us to conclude that they can be two genetically distinct strains. The first national common carp breeding program led by HAKI back in the 1950s was focused on Hortobágy common carp strains. Hortobágy is the largest fish hatchery and farm in Hungary, with continuous selection and genetic improvement, where strains are maintained under controlled conditions.
In accordance with other studies (e.g., [14,22,67]), over 90% common carps were correctly assigned to their strain of origin. The results of the AMOVA test showed high levels of variance within strains, which suggests high diversity at the level of individuals, but does not support the traditional distinction of strains.
STRUCTURE-based analyses ( Figure 3) indicated that the highest ∆K value was obtained when K = 14, where the individuals from various sites differed significantly. This result can be used as evidence for a relatively high genetic diversification of common carp strains in Hungary. Based on the results, we can propose that after their creation, the different strains are genetically pure and kept clean in the case of eight strains (HaS, SzS, HaM, BiS, Sz1, SzP, AmW, and BoM). It is known that a stable polymorphism can be maintained if a heterozygous advantage (overweight) effect exists. In the case of the other six strains (HoS, HoM, HoL, BiM, SzM, TaS), we can already observe genetic evidence for mixing. Well-designed breeding programs can improve this picture, but require careful attention, as uncontrolled mixing can even result in the loss of strains, as has probably already happened, for example, with the Croatian Končanica stock [75].

Conservation Implications
Common carp is the main fish species in pond aquaculture production in Hungary [76], where 33 strains of the species have been identified [18,77]. In addition, almost all Hungarian fish hatcheries distinguish further strains that do not have variety recognition. With few exceptions, the phenotypic difference is negligible between strains [22]. Intensification of breeding and selection programs with high levels of stocks can degrade the genetic basis of the species and lead to the extinction of varieties and the uniformization of populations [36,78]. It is extremely important to genetically describe and preserve our strains, not only for proper maintenance of animal genetic resources, but also in order to be able to choose suitable breeding and selection strategies. Our results on their genetic variability and the relationships between them can be provide a new background for population conservation [79] and breeding programs.
Overall, the microsatellite loci examined in our study proved to be quite effective in characterizing different genetic variabilities within and between the common carp strains. Our findings indicate relatively high values of expected heterozygosity for loci within common carp strains from Hungary. It is known that a stable polymorphism can be maintained if the effect of heterozygous advantage exists. It is also suggested that the applied fish farming practice is able to preserve and possibly improve a certain level of genetic diversity for generations [80][81][82]. In some strains (e.g., Hortobágy scaly, Biharugra scaly, Szeged mirror, and Amur wild carp), based on the inbreeding coefficient, lower heterozygosity values can be seen. These strains require more attention from breeders, as in these cases the risk of inbreeding deterioration can already threaten.
Author Contributions: B.T.: sample collection, laboratory work, sample procession, article writing; R.K.: data analysis, article review; M.R.A.: data analysis, article writing; Z.B.: sample collection, laboratory work, article review; M.F.: article review, funding acquisition; P.B.: article review; G.K.: article review; S.K.: funding acquisition, supervision, sample collection, article writing. All authors have read and agreed to the published version of the manuscript.

Funding:
The work was supported by the European Regional and Development Fund and the Government of Hungary within the project GINOP-2.3.2-15-2016-00025. The work/publication is supported by the EFOP-3.6.3-VEKOP-16-2017-00008 project. The project is co-financed by the European Union and the European Regional Development Fund.