Phylogeographic pattern and the origin of turtles
The northern part of the Pannonian Basin represents the edge of the European pond turtle’s range. Peripheral populations may exhibit lower genetic variability (Eckert et al. 2008; Pironon et al. 2017), but our study show that genetic variability of E. orbicularis in northern Pannonia is relatively high in both mtDNA and nuclear markers. The high genetic variability of some studied populations, however, appears to be the result of recent introductions of individuals from other parts of the species’ range. Therefore, the first aim of our study was to confirm the origin of the examined individuals, specifically their affiliation with known phylogeographic mtDNA lineages. Within the entire European pond turtle’s range, ten distinct mtDNA lineages are distinguished, with populations from Sicily belonging to lineage III have recently been described as the cryptic species of E. trinacris (Fritz et al. 2005, 2007; Vamberger et al. 2015). Within each lineage, individual haplotypes are further distinguished. Some have wide distribution, while others, typically in the southern parts of the species’ range, are geographically restricted due to their isolation in Pleistocene refugia, from which they did not spread northward during the Holocene (Lenk et al. 1999; Fritz et al. 2007, 2009; Pedall et al. 2011).
The Pannonian region is inhabited by lineage II, which, however, has a much wider distribution extending from southeastern Balkans through the central Balkans west of the Carpathians to the northern parts of the Pannonian Basin. Populations belonging to this mtDNA lineage are also disjunctly distributed in eastern Germany, western Poland, and central and southwestern France and adjacent parts of Spain (Fritz et al. 2007; Pedall et al. 2011). During the postglacial recolonization of higher latitudes, turtles belonging to lineage II reached as far as Great Britain and central Scandinavia but subsequently disappeared from many regions due to climatic changes in the Holocene (Sommer et al. 2007, 2009). The vast majority of E. orbicularis individuals examined in this study belonged to the expected lineage II and possessed a widely distributed haplotype IIa. In the Marcelová population, we found a haplotype (designated as II-Marcelová), which was not published in previous studies or gene bank datasets. The new haplotype differs by only one mutational step from the haplotype IIa, suggesting that it might have originated relatively recently in situ and might be therefore autochthonous and endemic. However, confirming this hypothesis would require testing a larger number of individuals from different areas covering the distribution of lineage II, including the Pannonian Basin. The presence of individuals carrying the II-Marcelová haplotype makes the population in Marcelová genetically unique and thus worth of conservation (e.g., Fraser and Bernatchez 2001; Coates et al. 2018; Quiroga et al. 2019).
One male captured in Marcelová had haplotype Ib, which according to available data originates from eastern Greece and Bulgaria but was recorded as non-native in various places in Europe (e.g., Germany or Spain; Fritz et al. 2007; Velo-Antón et al. 2011). The origin of haplotype Ib in Marcelová is thus questionable, but it is likely that it could have arrived here by the introduction of a limited number of turtles from the southeastern Balkans. Such a practice was rather recently conducted in the wider area of the studied region, particularly in Czechia (Šebela 2017; Horváth et al. 2020) and it is not excluded that it also took place in Slovakia. An alternative hypothesis assumes that haplotype Ib is original in the northern part of Pannonia and that spread here from the southeastern Balkans after the last glaciation. However, previous phylogeographic studies have shown that lineage I spread from the Balkan refugium northward across the lowlands east but not west of the Carpathians (Lenk et al. 1999; Fritz et al. 2007), and therefore this hypothesis seems less likely. Verification of both hypotheses would require a more intensive genetic survey of Pannonian populations.
The turtles bred in the Šúr Biological Station, whose offspring were released in several locations in southwestern Slovakia, belonged to three haplotypes. In addition to haplotype IIa, which is common and native to the Pannonian region, one female had haplotype Ia, which has a wide distribution from the southeastern Balkans north to Poland and the Baltic states, and in the east to Turkey, Georgia, Russia, and Kazakhstan (Lenk et al. 1999; Fritz et al. 2007; Pedall et al. 2011). Three males had haplotype IIb, only known from northeastern Germany and western Poland (mainly in the Oder basin). Haplotypes Ia and IIb are therefore not typical for the Pannonian region. The origin of individuals with these haplotypes remains unclear, as according to the official records of the State Nature Conservancy of the Slovak Republic, the turtles currently living in the Šúr Biological Station were caught in Hungary as part of a restitution program (Havranová et al. 2017). However, the occurrence of haplotypes that are not specific to Pannonian populations raises the question of whether turtles from other parts of the area have been released at the biological station, or whether non-native individuals have already been imported from Hungary. Since mtDNA is maternally inherited, the successful reproduction of a female with haplotype Ia would result in the occurrence of this haplotype at the biological station and localities where her offspring were released.
In addition to haplotype IIa, haplotypes IIl and IVa were also identified at three Austrian localities. According to gene bank data, haplotype IIl is only known in the Donau Auen NP and is probably endemic to this Austrian population. Nine of the 30 analysed individuals belonged to haplotype IVa. Mitochondrial lineage IV is not native to Pannonia and occurs in the Adriatic part of the Balkan and Apennine Peninsula (Lenk et al. 1999; Fritz et al. 2007; Pedall et al. 2011). Thus, the occurrence of lineage IV in the Donau Auen NP is not autochthonous, as previously documented by Velo-Antón et al. (2011). In addition to haplotype IVa, Velo-Antón et al. (2011) also recorded other allochthonous haplotypes in the Donau Auen NP, demonstrating the human-mediated introduction of turtles from eastern Europe, eastern Balkans, Turkey or Ponto-Caspian region (haplotype Ia), northeastern Germany or the adjacent part of Poland (haplotype IIb), and the western Mediterranean (haplotype Va). According to unpublished cytochrome b haplotype sequences, pond turtles might have been introduced to the Donau Auen NP also from other parts of the Pannonian Basin and the Balkans, and from southern Italy (M. Schindler, unpublished data).
Due to the presence of two diverged mtDNA lineages (II and IV), turtles from the Donau Auen NP form significantly differentiated groups in the haplotype network and show the highest values of nucleotide diversity (π). However, in nuclear microsatellites, all individuals from the Austrian sites form a single cluster, indicating that the introduced Adriatic turtles successfully mate with native turtles and that between mtDNA lineages II (taxonomically classified as subspecies E. orbicularis orbicularis) and IV (taxonomically classified as E. orbicularis hellenica; Fritz et al. 2005) is a high rate of hybridization and extensive gene flow. These results are consistent with the studies focused on zones of the secondary contact, which show substantial introgressive hybridization between E. orbicularis subspecies (Pedall et al. 2011; Vamberger et al. 2015; Raemy et al. 2017).
Population genetic structure
The mitochondrial gene cytb is not sufficiently variable within the Pannonian Basin to allow a detailed analysis of the population-genetic structure of the European pond turtle here. Therefore, microsatellites as highly variable nuclear markers, were used for this purpose. Based on these markers, all examined individuals of E. orbicularis were assigned to the most likely number of clusters K = 6. Membership in specific clusters, however, needs to be carefully assessed at sites where only one individual was captured. Nevertheless, several important findings emerge from the population-genetic structure.
The two analysed hatchlings of the European pond turtle from Devínske Jazero were assigned to the common cluster with the Austrian samples. The similarity of individuals from Devínske Jazero and the Donau Auen NP is also indicated by the low values of the Fst statistic, which expresses the degree of population differentiation. In eastern Austria, the European pond turtle occurs sporadically along the Morava River, but a strong population is found especially in the Donau Auen NP (Gemel 2001; Grillitsch and Cabela 2001; Schindler 2008). The genetic similarity of the Austrian population with the population in Devínske Jazero is obvious. We can therefore rule out the hypothesis that the turtles in Devínske Jazero have an origin in the Šúr Biological Station, where they are bred as part of a restitution program. Based on the current genetic data, however, we cannot determine whether the turtles spread to Devínske Jazero from Austria, or whether they lived here in hiding in the past and were discovered only recently. If they had spread from Austria, we might expect the presence of the mtDNA lineage IV, which is relatively common in the Donau Auen NP, also in Devínske Jazero. However, both individuals from Devínske Jazero belonged to the mtDNA lineage II (haplotype IIa), typical for the Pannonian Basin.
Another significantly different cluster, distinguished from microsatellite genotypes, was composed of individuals from Tajba. The uniqueness of the Tajba population can be explained by the fact that the locality belongs to the Tisza River basin, which is a different river catchment. The turtles primarily use water corridors such as rivers for dispersal. These corridors served as important landscape elements already during the postglacial recolonization of the northern parts of Europe (Joger et al. 2007). Therefore, populations from the Tisza and Danube basins may be genetically different. However, to verify this hypothesis, we would need to analyse more populations belonging to both basins.
Other populations in southwestern Slovakia and Hungary showed less differentiation and their population-genetic structure did not show a significant geographic pattern. However, it is interesting to classify the populations in Marcelová and Veľky Lél into two different Structure clusters. Significant differentiation was also found in the Fst statistic between these populations, which are approximately only 25 km apart. However, differences were also noted in the representation of mtDNA haplotypes. While haplotypes IIa, II-Marcelová and in one individual also Ib were found in Marcelová, only haplotype IIa was detected in Veľky Lél. If there was significant dispersal and gene flow between these two populations, we would expect to find haplotypes specific for Marcelová in Veľky Lél too. However, it should be emphasized that similar haplotype composition between both populations would only be evident in the case of female dispersal, since mtDNA is maternally inherited. Female philopatry is known for European pond turtles (Bona et al. 2012; Fay et al. 2023), so mtDNA may be more structured between populations than nuclear genes (Ciofi et al. 2017). Nevertheless, both types of markers, cytb sequences and microsatellites, suggest that there is no significant dispersal and gene flow between the two populations, probably due to the large distance and the absence of ecological corridors that might facilitate dispersal.
Genetic variability
The northern edge of the Pannonian Basin represents the northern limit of the continuous distribution of the mtDNA lineage II of the European pond turtle. Further north, only populations in eastern Germany and western Poland are found, but they are geographically isolated from the rest of the continuous distribution (Fritz et al. 2007; Pedall et al. 2011; Prusak et al. 2013). However, this isolation is relatively recent, as populations belonging to lineage II colonized the area of the Czech Republic and Germany after the last glaciation but disappeared during the Holocene. (Široký et al. 2004; Sommer et al. 2007, 2009). Populations of plants and animals at the edge of the species range may have reduced genetic variability and higher genetic differentiation for historical (e.g., as a result of postglacial dispersal) and contemporary (e.g., low effective population size, stronger genetic drift, spatial isolation of populations, bottleneck effect) processes (Pironon et al. 2015, 2017; López-Delgado and Meirmans 2022; Singhal et al. 2022).
A higher variability of mtDNA is in southern populations of E. orbicularis, where glacial refugia were located, and decreases towards the north as a result of the postglacial dispersal (Lenk et al. 1999; Fritz et al. 2007; Pereira et al. 2018). This phylogeographic pattern also corresponds to our study. If we do not take into account the probably non-native haplotypes Ia, Ib, IIb and IVa, found at the Šúr Biological Station, in Marcelová and in the Austrian sites, we found three haplotypes (IIa, IIl and II-Marcelová) in the studied Pannonian populations. In the southern parts of the range, the haplotype diversity is significantly higher (Fritz et al. 2009; Vamberger et al. 2015; Ciofi et al. 2017). On the other hand, it is important to draw attention to the local haplotypes, which might have evolved relatively recently in situ, and which occur in the Donau Auen NP (haplotype IIl) and in Marcelová (haplotype II-Marcelová). We still lack more detailed data on the distribution of these haplotypes, but if they occurred only in these studied localities, it would mean that local haplotypes increase genetic diversity in peripheral populations and indicate that even populations from the edge of the range can significantly contribute to the maintenance and generation of overall genetic diversity species (Channell and Lomolino 2000).
Comparison of genetic diversity at microsatellite loci between our and other studies is challenging because different studies use a different set of loci. Despite this limitation, heterozygosity seems to be similar and allelic diversity either similar or only slightly lower in Pannonian populations compared to populations from other parts of the range (Velo-Antón et al. 2008; Molnár et al. 2011; Prusak et al. 2013; Ciofi et al. 2017; Pedall et al. 2017; Pereira et al. 2018; Ilhan et al. 2021). This simple comparison indicates that marginal populations do not necessarily have reduced genetic variation (Eckert et al. 2008; Pironon et al. 2017).
Conservation implications for Slovak populations
In our study, most of the samples came from Slovakia, where E. orbicularis is assessed as a critically endangered species (Kautman et al. 2001). Until the mid-20th century, turtles were relatively common in the southern regions of the country, but due to wetland drainage and habitat loss, many populations have disappeared (Jablonski et al. 2015). Currently, only a few breeding sites are known in Slovakia. Besides the breeding sites, single adult individuals are sporadically observed in southern regions of Slovakia (Jablonski et al. 2015; Havranová et al. 2017; this study and own unpublished data). However, their origin remains largely unresolved as they have not been genetically investigated.
The vast majority of turtles from Slovak localities analysed for cytb can be considered autochthonous to the Pannonian Basin. Exceptions are individuals from the Šúr Biological Station and Marcelová, where individuals carrying non-native haplotypes were found. Juveniles from the breeding facility at the biological station have been released in several locations in western Slovakia (Burešová et al. 2001; Havranová et al. 2017), where the presence of non-native haplotypes can potentially be expected.
Based on microsatellite analysis, the Slovak populations of E. orbicularis can be divided into three groups. The first group consists of the population in Devínske Jazero, which is genetically close to the Austrian population in the Donau Auen NP. The second group consists of Tajba in the Východoslovenská nížina lowland (southeastern Slovakia) and the third, albeit relatively heterogeneous group, consists of the populations in the Danube region (Podunajská nížina lowland, southwestern Slovakia). To preserve the original genetic structure of E. orbicularis in Slovakia, we recommend not to translocate individuals between regions and populations. Translocation for the purpose of increasing genetic variability is not even necessary, as the studied populations show relatively high genetic variability (see e.g., Berger-Tal et al. 2020).
In future studies, it would be appropriate to focus on mapping of E. orbicularis in southern Slovakia and to genetically examine populations that were not analysed in this study. In the westernmost part of Slovakia (Záhorie region adjacent to Austria), we recommend carrying out genetic screening to detect the non-native mtDNA lineage IV, which occurs in the Donau Auen NP as a result of the human-mediated introduction. If the turtles from the Austrian localities were to disperse as far as Slovakia, then this lineage could also be expected here. Furthermore, we recommend a genetic analysis of populations established from the individuals bred in the Šúr Biological Station. This could reveal the presence of non-indigenous haplotypes and genetic variability that can potentially be reduced due to the founder effect.