MICROSATELLITE-BASED GENETIC VARIABILITY AND DIFFERENTIATION OF HATCHERY AND FERAL COMMON CARP CYPRINUS CARPIO L. (CYPRINIDAE, CYPRINIFORMES) POPULATIONS IN CROATIA

Common carp production has an important status in Croatian aquaculture. In addition, the sport fishing of common carp in open waters is very popular, but it is often based on stocking from fish farms. Using fifteen microsatellites, 243 individuals from 5 hatchery and 5 feral populations have been analyzed. A total number of 148 alleles were recorded. However, the mean number of alleles per locus was remarkably low. Pairwise FST values (0.026-0.130) were significant (P<0.01), demonstrating differentiation among populations. The Markov chain method test showed that all the populations deviated from HWE (P <0.05). After sequential Bonferroni correction only the Vrana lake was in HWE in all the loci but MFW20. The factors that may result in genetic divergence and significant reduction of the observed heterozygosity are discussed. AMOVA results for 10 populations indicate that the percentage of the variation among populations was 6.26%, which is lower than the variation within populations (91.04%).


INTRODUCTION
The Common carp (Cyprinus carpio L.) is an economically important species to the aquaculture of Croatia.Common carp culture in Croatia developed considerably at the turn of 19 th to 20 th century.The first carp stocks were introduced from Germany and the Austro-Hungarian Empire (nowadays Czech Republic and Hungary).Decades of efforts in selection resulted in the well-known Našice and Poljana strains, kept in the live gene bank in Szarvas (Treer and Kolak 1994;Gorda et al. 1995).The hybrids of Našice and Israeli Dor-70 strains happen to be the most successful crossbreeds in Israel (Wohlfarth and Moav 1990).However, according to the discussion at the Budapest conference, Flajšhans and Gall (1995) presumed that the Našice strain had disappeared from its original fish farm due to uncontrolled crossbreeding.
Nowadays, fish farmers try to certify their stocks through the Croatian Chamber of Commerce as genuine Croatian product, which is also a quality stamp for the fish (Treer et al. 1996).But this poses potential risks associated with the loss of genetic variation and an increase in inbreeding (Schonhuth et al., 2003).Therefore, the information on genetic diversity of these hatchery stocks is ur-gently required in order to sustain the quality of the broodstock.
Sport fishing of common carp in open waters is often based on stocking from fish farms.Hence, the hybridization of feral populations becomes an increasing problem (Memiş and Kohlmann 2006).
Microsatellites are highly variable genetic markers that are inherited codominantly in a Mendelian pattern.In comparison to other molecular markers, microsatellite markers are sensitive and promising in population genetics studies, especially those populations that are closely related (O'Connel and Wright 1997).As genetic variability of feral and hatchery stocks of common carp has been successfully investigated in Chinese feral and hatchery varieties (Zhou et al. 2004), microsatellites were considered to be a good molecular marker system in our investigation, too.
The aim of our study was to explore genetic variability within and among five hatcheries and five feral populations from different Croatian regions using microsatellite DNA, a hyper-variable molecular marker.

MATERIALS AND METHODS
Fin clip samples of 243 individuals of C. carpio were collected in 2005.Details of sampling localities, sample codes and sample sizes are presented in Fig. 1 and Table 1.Fish were obtained from the following fish farms: Našice, Grudnjak, Končanica, Poljana and Draganići in the Croatian part of the Pannonian Plain.Semi-intensive common carp culture is characteristic for all of these farms.Feral common carps were caught from some of the largest Croatian rivers, the Sava, Danube, Drava and Kupa, and also from Lake Vrana.The attribute "feral" does not necessarily mean "native", moreover, it stands for self-reproducing populations in natural waters (rivers and lakes).
Total DNA was extracted from individual caudalfin samples using DNAeasy Tissue Kit (QIAGEN).Some extractions gave low DNA yields so the final elution was brought down to 100 µl.Microsatellite variation was examined at fifteen microsatellite loci (MFW1, MFW4, MFW7, MFW9, MFW12, MFW31, MFW16, MFW20, MFW23, MFW29, MFW3, MFW13, MFW17, MFW26, MFW28) with primer sequences described in Crooijmans et al. (1997).The temperature profile of the PCR was 95°C for 15 min of an initial denaturing cycle followed by 30 cycles of denaturation at 94°C for 30 s, annealing at 60°C for 90 s, an extension cycle at 72°C for 1 min followed by final extension step at 72°C for 10 min.One primer of each pair was fluorochrome labeled to enable the determination of allele sizes by ABI 3730 Genetic Analyzer and the ABI GeneScan software.
The genetic diversity of each population has been estimated as the mean number of alleles per locus (A), observed (Ho) and expected heterozygosities (He) following Nei (1978) using GENETIX 4.05.2 (Belkhir et al. 2004).Inbreeding coefficients (F IS ) for each population and pairwise F ST values among all the pairs of populations were calculated and tested by permutations using the same software.Allelic richness based on 11 individuals per locus and population was calculated using FSTAT (Goudet 2001).Each population was tested for departure from Hardy-Weinberg equilibrium expectations using the Markov-chain method with 10,000 dememorization steps, 500 batches and 10,000 subsequent iterations.Linkage disequilibrium was tested across all the pairs of loci using GENEPOP 4.0 (Raymond and Rousset 1995).Levels of significance for this test were adjusted by sequential Bonferroni correction (Rice 1989).3.
To estimate and test the significance of genetic differentiation among feral populations and hatcheries, as well as among all the populations, we performed a hierarchical Analysis of Molecular Variance (AMOVA) implemented in ARLEQUIN 3.0 (Excoffier et al. 2005).The estimated components of molecular variances were tested against zero using 10000 permutations.
Genetic distances among populations (Cavalli-Sforza and Edwards 1967) were calculated using the PHYLIP program package, version 3.67 (Felsenstein 1993).An UPGMA dendrogram was reconstructed based on the genetic distance matrix using the same program.The reliability of the dendrogram was estimated by bootstrapping (1000 replicates) and implemented by PHYLIP program package.The dendrogram was visualized in TreeView, version 1.6.6 (Page 1996).

RESULTS
Variation in number of alleles as well as allele size ranges are shown in Table 2.A total of 148 alleles were recorded over all the loci.The number of alleles per locus varied from 6 (MFW12 and MFW17) to 20 (MFW20).
The average number of alleles for all the populations varied from 2.2 (Našice) to 7.4 (Danube).Variability levels across populations estimated as observed heterozygosities ranged from 0.369 to 0.612, and estimated as expected heterozygosities ranged from 0.654 to 0.736 (Table 3).
The Vrana Lake had the highest observed heterozygosity of all the populations while Kupa had the lowest.The Markov chain method test showed that all the populations deviated from HWE (P <0.05).After sequential Bonferroni correction, only Lake Vrana was in HWE in all the loci but MFW20, and only MFW9 and MFW29 loci were in HWE in all the populations (Table 4).
F IS value varied from 0.2146 (Vrana lake) to 0.4014 (Grudnjak).The number of private alleles in all ten populations was relatively low (one in hatchery populations; three in feral populations) (data not shown).
AMOVA results for 10 populations indicate that the percentage of variation among the populations was 5.62%, which is lower than the variation within the populations (61.26%) (Table 5).Genetic differentiation among 10 populations, as well as among the feral and hatchery populations was both significant.
The pairwise F ST values among all the populations are shown in Table 6.The largest F ST genetic distance (0.424) was measured between Draganići and Drava, indicating that these two populations diverged the most, while two geographically very close hatchery populations, Končanica and Poljana, had the smallest genetic distance (0.136).
The UPGMA dendogram generated from a matrix of Nei's genetic distances among the populations (Fig. 2) revealed three cluster groups among the ten populations.In the first cluster the Grudnjak population was separated from the others.Four feral common carp populations were in the second cluster, and hatchery populations in the third cluster, together with the feral Kupa population.Furthermore, the dendrogram showed that genetic divergences among the studied hatchery populations of common carp were relatively small.Consequently, with the exception of the Grudnjak population, all the studied populations might belong to the same group.

DISCUSSION
A precise estimate of population structure and genetic distances from microsatellite data depends on the sample size, number of loci, number and size range of alleles (Ruzzante 1998).Scoring six loci in Atlantic cod (Gadus morhua), Ruzzante (1998) examined the effect of sample size on seven genetic distance measures and two structure metrics.He concluded that 50-100 individuals are needed for a good estimation.The sample size in the present study was obviously lower than the one recommended.Therefore, these results should be taken with caution.
On the other hand, by using fifteen highly polymorphic loci the right resolution could be achieved in order to differentiate the common carp populations studied.Instead, genetic variability might have been underestimated because of the rare alleles and genotypes that were absent from the samples due to the suboptimal sample size.Such missing alleles and genotypes might also be the reason for significant A significant deviation from Hardy-Weinberg equilibrium found in all the loci could also be explained either by a sample bias or by the presence of null alleles in all populations.In the presence of null alleles, the heterozygotes possessing a null allele would be erroneously recorded as homozygotes for the variant allele and this would lead to a deficiency of heterozygosity (Table 4).
In comparison to natural stocks, lower genetic variability in hatchery stocks has been reported for many species, e.g., turbot, Scophthalmus maximus (Coughlan et al. 1998), common carp (Kohlmann et al. 2005), Japanese flounder, Paralichthys olivaceus (Sekino et al. 2002), Atlantic salmon, Salmo salar (Skaala et al. 2004), Kuruma prawn, Marsupenaeus japonicus (Luan et al. 2006).The findings of the present study are similar to the earlier observations.This is mainly due to a small founder population and an ultimately small effective population size (N e ), (Falconer and Mackay 1996).
Hence, we suggest that the reduction of allelic diversity in hatchery stocks might be the result of founder events or occasional bottleneck effects during the breeding process.Among hatchery stocks, the highest heterozygosity is found in Poljana (0.5), which could be the result of refreshing from new fish stocks (Božić, 2009).Our results support the prediction of Kirpitchnikov (1999) about the genetic similarity of stocks originating from the same geographic region.The largest distance between Drava and Draganići is likely the result of the geographical separation.In addition, the large distance between Grudnjak and the other nine populations is the result of the geographical separation of the stocks and the effects of different selective breeding.The Končanica stock is genetically very close to Našice.These two fish farms have been managed by the same organization for many years, so the same spawning material has probably been used in both farms (Treer et al., 2000).The results of this study indicate that in Croatia common carp stocks that are well defined no longer exist.On the other hand, it is possible that uncontrolled mixing has made some stocks disappear, as probably happened with the stock of Končanica.
The introduction of hatchery carp is a big issue for feral populations in open waters (Memiş and Kohlmann 2006).This is evident in Kupa population, which by genetic distance belongs to the hatchery populations.
This study is the first genetic study to deal with Croatian common carp based on microsatellite DNA markers.It has demonstrated that microsatellite markers are a powerful tool in monitoring the genetic condition of different strains of common carp in Croatia.A higher genetic variability of the feral carp populations in comparison to the hatchery stocks has been found.This is particularly important in light of the global threat to feral carp populations.In Croatia, five hatchery stocks of common carp have important status in aquaculture, so our results on the genetic variability within/among them and the relationship among them can provide a new background of knowledge in population conservation and breeding programs.

Fig. 2 .
Fig. 2. UPGMA dendogram based on DA distance for 10 populations of common carp using 15 microsatellite loci.The abbreviation of each population is shown in Table3.

Table 1 .
Details of samples, population abbreviations, population names, locations, population types and sample size of common carp (Cyprinus carpio L.) in Croatia

Table 2 .
Characteristics of microsatellite common carp markers

Table 3 .
Genetic variations at all the loci in ten populations

Table 4 .
Deviations from Hardy-Weinberg proportion for 10 populations in 15 loci.Values in italics indicate the loci that significantly deviate from HW (P < 0.05)

Table 5 .
AMOVA analysis results for 10 populations of common carp based on 15 loci

Table 6 .
FST values among all pairs of the populations of common carp.All the values are significant (p<0.01)