Molecular Phylogenetics and Evolution 92 (2015) 118–123

Contents lists available at ScienceDirect

Molecular Phylogenetics and Evolution

journal homepage: www.elsevier.com/locate/ympev

Short Communication Molecular data suggest a multispecies complex of Phoxinus () in the Western Balkan Peninsula q ⇑ Anja Palandacˇic´ a, , Jernej Bravnicˇar b, Primozˇ Zupancˇicˇ c, Radek Šanda d, Aleš Snoj b a First Zoological Department, Natural History Museum Vienna, Burgring 7, 1010 Vienna, Austria b Department of Science, Biotechnical Faculty, University of Ljubljana, Groblje 3, 1230 Domzˇale, Slovenia c AZV Agency, Dolsko 14, 1262 Dol pri Ljubljani, Slovenia d National Museum, Václavské námestí 68, 115 79 Prague 1, Czech Republic article info abstract

Article history: This study aimed to evaluate the phylogeographic distribution of Phoxinus phoxinus sensu lato populations Received 12 May 2015 in the Balkan Peninsula using molecular methods. For the western Balkan Peninsula, two Accepted 28 May 2015 (Danubian P. phoxinus and Adriatic P. lumaireul) have been suggested, but not confirmed by subsequent Available online 2 July 2015 morphological studies or by genetic data. For the present study, more than 300 specimens were collected from the western Balkans. Keywords: A partial sequence of the mt cyt b gene analysed using Bayesian and ML methods revealed several Western Balkan Peninsula well-supported clades, with distances ranging from 4% to 11%. The clades were corroborated by RAG1 Phylogeography sequence analysis. Thus, the molecular analysis points to a multispecies complex of Phoxinus sp. with Molecular methods Cyprinidae ranges of the clades extending north-west to south-east along the western Balkan Peninsula. In addition, Phoxinus the dispersion of fish through subterranean water connections in is indicated. Ó 2015 Elsevier Inc. All rights reserved.

1. Introduction respectively. Nevertheless, molecular studies of Phoxinus in Europe are scarce, with the only included in wider phyloge- The European is a common name for various members netic studies (for example Zardoya and Doadrio, 1999). of the genus Phoxinus Rafinesque 1820, ubiquitous small cyprinids According to Kottelat and Freyhof (2007), the western Balkan that inhabit streams throughout northern Eurasia, and which have Peninsula is populated by two species of European minnow, with been in Europe regarded as a single species Phoxinus phoxinus P. phoxinus in the Danubian part and P. lumaireul in the Adriatic (Linnaeus, 1758). Such a broad distribution raises the question of part of the peninsula. However, Bianco (2014), studying the mor- whether P. phoxinus represents a single species or a group of spe- phology of 86 individuals from his personal collection, could not cies. Studies of other taxa with similarly large ranges, e.g., Gobio confirm their findings. Also, according to IUCN taxonomic notes gobio or Cobitis taenia, have revealed new, or re-established old, (http://www.iucnredlist.org/details/135679/0), the identity of the taxa (Perdices et al., 2003; Mendel et al., 2008), suggesting the Phoxinus populations from Drin and Skadar drainages requires fur- same might be true for P. phoxinus. Morphology-based studies have ther investigation. The drainages of lakes Skadar and Ohrid are resurrected P. colchicus Berg 1910 for the eastern Black Sea basin known for their high levels of endemicity (e.g. Talevski et al., (Bogutskaya and Naseka, 2004), and re-established P. lumaireul 2009), as is the Balkan Peninsula in general (e.g. Bogutskaya Schinz 1840 for the Adriatic basin, and P. strandjae Drensky 1926, et al., 2012), hinting at an even higher level of diversity of from the Veleka (Bulgaria) and Resowska (Turkey) drainages Phoxinus in the Balkans than already proposed (Kottelat and (Kottelat, 2007). In addition, Kottelat (2007) proposed the exis- Freyhof, 2007). tence of three new species, P. bigerri, P. septimaniae and P. strymon- From the above it is clear that high levels of diversity, complex icus, with restricted distributions in France, Spain and Greece, phylogenies and distributional patterns of many cyprinids that co-exist with the European minnow in the Balkans, combined with a dynamic geological history (Karamata, 2006), suggest that the q This paper was edited by the Associate Editor G. Orti. ‘Balkan minnow’ could comprise a heterogeneous assemblage of ⇑ Corresponding author. divergent lineages (referred to in the present paper as P. phoxinus E-mail addresses: [email protected] (A. Palandacˇic´), sensu lato). To determine the status of the Balkan minnow, both [email protected] (J. Bravnicˇar), [email protected] (P. Zupancˇicˇ), [email protected] (R. Šanda), [email protected] (A. Snoj). phylogenetically and taxonomically, genetic analysis of http://dx.doi.org/10.1016/j.ympev.2015.05.024 1055-7903/Ó 2015 Elsevier Inc. All rights reserved. A. Palandacˇic´ et al. / Molecular Phylogenetics and Evolution 92 (2015) 118–123 119 geographically distinct populations across the entire range is Table 1 (continued) needed. Here we performed the first step in such a comprehensive Abbr. Sampling site No. samp/ Drainage Country Clade task by investigating the phylogeographic patterns of Balkan min- haps now as revealed by cytochrome b (cyt b) and recombination acti- SCA Šcˇavnica 6/2 D SLO 1d vating gene 1 (RAG1) sequences of a large number of individuals VAL Kolubara 12/9 D SRB 1d collected in the Western Balkan Peninsula. JAD Jadar 4/2 D SRB 1d SVR Svrljiški Timok 8/2 D SRB 1d LEP Lepenica 4/4 D BH 2 2. Materials and methods KAR Karanovac 6/3 D BH 2 UGR by Vitovlje 4/2 D BH 2 MIL Miljatska by Sarajevo 2/2 D BH 2 In total, 310 specimens of P. phoxinus sensu lato were collected RAM Lake Ramsko 1/1 A BH 2 from an area extending from Italy to FYR Macedonia (Table 1, BUN 5/4 A BH 2 Fig. 1). Following DNA extraction, parts of the mitochondrial (mt) BRE by Stolac 3/2 A CRO 2 cyt b gene and the nuclear RAG1 gene were sequenced. Cyt b and ZAL 3/2 A BH 2 MUS Mušnica 2/2 A BH 2 RAG1 sequences from voucher species Phoxinus lumaireul PLA Lake Plavsko 4/4 D MNE 3 ( drainage, Bosnia-Herzegovina, Genebank Accession Nos. DRI by Focˇa 3/2 D BH/SRB 4 HM560123 and HM560420) were included in the analyses. BRR Brestovacˇka River 4/2 D SRB 5 KRU Krupacˇko Barje 5/2 D SRB 5 MOA Moravica 3/1 D SRB 5 SKR Skrapezˇ 5/5 D SRB 5 Table 1 LEC Lepenac 4/3 D KOS 5 Phoxinus phoxinus sampling sites from Slovenia (SLO), (CRO), Bosnia- KOM Komanska River 1/1 D BUL 5 Herzegovina (BH), Serbia (SRB), FYR Macedonia (FYROM), Montenegro (MNE), BRO Brodicˇ by Orlic´ 7/3 A CRO 6 Bulgaria (BUL), Kosovo (KOS) and Italy (IT). Abbr.: Abbreviations of sampling MOR Moracˇa 4/4 A MNE 7 locations, correspond to Figs. 1 and 2. Sampling site: location of fish sampling. No. CRN Lake Skadar, Rijeka 2/2 A MNE 7 samp/haps: number of samples and number of haplotypes at each location. Drainage: Crnojevic´a A, Adriatic; D, Danubian; A/D, water connection to both. ⁄, literature unavailable, but VIZ Lake Skadar, Virpazar 7/4 A MNE 7 in the area of A or D. Clade: 1–8, where 1a–d are subdivisions of Clade 1; sampling ZET Zeta 4/4 A MNE 7 sites with samples clustering into different clades (CER, ZRM, ZAL, MUS) are recorded SUS Sušica 2/2 D MNE/ 7 twice, for each corresponding clade. BH ZAL Zalomka 4/2 A BH 7 Abbr. Sampling site No. samp/ Drainage Country Clade MUS Mušnica 6/2 A BH 7 haps TRE Trebišnjica 4/3 A BH 7 BEL Belski potok 10/3 A SLO 1a OHR Lake Ohrid, Sateska 6/3 A FYROM 8 CER Cerknišcˇica 6/5 D/A SLO 1a River IDR Idrijca 1/1 A SLO 1a KRN Lake Krn 4/3 A⁄ SLO 1a MAL Koromacˇi 4/1 A⁄ SLO 1a NAD Nadizˇa 3/3 A SLO 1a NAN Nanošcˇica 6/2 A/D SLO 1a Sequences were edited by eye and aligned with MEGA 5.0 OSP Osp 4/2 A SLO 1a POD Podgora 6/2 D SLO 1a (Tamura et al., 2011). Phylogenetic trees were constructed from PRE Predjama 6/1 A SLO 1a cyt b alignment using Bayesian inference (BI) with BEAST 1.8.0 NRE Notranjska Reka 4/1 A SLO 1a (Drummond et al., 2012) and Maximum-Likelihood using PhyML SOC Socˇa 2/2 A SLO 1a (Guindon et al., 2010). Distances between groups were calculated VIP Vipava 4/2 A SLO 1a in MEGA 5.0. An unrooted minimum-spanning network was con- MIR 2/2 A CRO 1a RIZ Rizˇana 6/3 A SLO 1a structed using the median joining algorithm (Bandelt et al., 1999) FON Fontana Vera, Ticino 6/5 A IT 1a implemented in Network 4.6 (www.fluxus-engineering.com) with RRI Risirg ripa, Ticino 2/2 A IT 1a default settings. TIC Ticino 6/3 A IT 1a For RAG1 sequences, the gametic phase of heterozygous indi- ZRM 8/5 A CRO 1a KRK Krk 3/1 A CRO 1a viduals was resolved using Phase 2.1 (Stephens et al., 2001; RAS Raša 3/1 A CRO 1a Stephens and Scheet, 2005), which is a reliable alternative to clon- KOL Kolpa 8/5 D SLO 1b ing (Harrigan et al., 2008), and, as for cyt b, an unrooted PET Petrinjcˇica 3/2 D CRO 1b minimum-spanning network was calculated. RIJ Rijecˇina 5/3 A CRO 1b For details on DNA extraction, polymerase chain reaction (PCR) SUN Sunja 1/1 D CRO 1b MRE Mrezˇnica 4/4 D CRO 1b conditions, sequencing, trees and network construction see Una by Kostanjevica 1/1 D CRO 1b Supplementary material. LIC Licˇka Jesenica 3/2 D CRO 1b LOV Lovinac 3/2 A⁄ CRO 1b ZRM Zrmanja 3/2 A CRO 1b UNB Una by Bihac´ 5/4 D BH 1b 3. Results RSS Rašaška River 1/1 D CRO 1b SAN Sana by Sredice 3/3 D BH 1b Sequencing resulted in a 1089 bp cyt b alignment that con- ˇ BLO Blošcica 7/2 D SLO 1c tained 395 polymorphic sites, no gaps or missing data, and which BOH Lake Bohinjsko 4/2 D SLO 1c CER Cerknišcˇica 4/1 A/D SLO 1c was collapsed to 152 unique haplotypes. Sequencing of the 841 bp JEV Jevnica 1/1 D SLO 1c RAG1 fragment proved challenging, and only 72 samples were suc- LJU Ljubljanica 4/3 D SLO 1c cessfully determined, of which 27 were heterozygous. Additionally, LOZ Lozˇnica 4/2 D SLO 1c six samples were omitted from further analysis due to low statis- RAD Radulja 6/4 D SLO 1c PTU Ptuj 5/4 D SLO 1d tical support (<0.8) for inferred gametic phase. RAG1 alignment RAT Ratkovski potok by 4/2 D SLO 1d contained 41 polymorphic sites. All sequences are deposited in Prosenjakovci GenBank under accession numbers KT166456 to KT166834. 120 A. Palandacˇic´ et al. / Molecular Phylogenetics and Evolution 92 (2015) 118–123

Fig. 1. A map of Phoxinus phoxinus sensu lato sampling locations on the western Balkan Peninsula and in Northern Italy with three letter abbreviations for the sampling sites (corresponding to abbreviations of sampling sites in Table 1) and colour codes that denote clades based on mtDNA cytochrome b gene phylogeny.

The phylogenetic trees constructed with Bayesian inference and restricted to specimens from Lake Plavsko (MNE) and Drina River Maximum-Likelihood from cyt b aligned sequences revealed eight (BH), respectively. Clade 5 (red) included sampling sites in clades (Clade 1 to Clade 8; posterior probability (PP) 1; bootstrap Serbia, Bulgaria and FYR Macedonia, of which some belonged to values 100%; Fig. 2a), extending along the sampling area from the watershed, some are sinking streams and some flow the north-west to the south-east (Fig. 1). Distances between the into the Aegean Sea. Interestingly, sampling location SVR belonged clades ranged from 4% to 11% (Table 2). In the densely sampled to Clade 1d, even though it is part of the same watershed as BRR area of Clade 1, an additional four well-supported (PP > 0.9) (Timok river system) in Clade 5. Clade 6 comprised samples from sub-clades 1a to 1d were identified, with distances ranging from a tributary to the River (CRO). In Clade 7 (blue), samples from 2% to 4% (Table 2). The deeper nodes on the tree had weak support, the southern Adriatic river system clustered together, including with both low PP and bootstrap values, especially the relationship rivers flowing into Lake Skadar and sinking streams flowing between Clades 1 and 2, and among 3, 4 and 5. A haplotype net- directly or indirectly into the . Samples from locations work constructed from the cyt b sequences (Supplementary mate- in Mušnica and Zalomka (BH) clustered into two different clades: 2 rial) corroborated the structure of eight clades revealed by the and 7. Clade 8 contained samples from Lake Ohrid. trees. More on the water connections of each sampling location is The distribution of the clades detected in the trees crossed the reported in Supplementary material. watershed boundaries and did not correspond to the two main The network constructed from 134 phased RAG1 haplotypes watersheds of the sampling area—the Adriatic and the Danubian. (67 sequenced specimens, including GenBank sequence Many of the clades found in the (Danube) tributaries were HM560420) roughly corroborated seven clades detected in the detected also along the Adriatic coast, even in the Aegean river sys- cyt b phylogenetic analysis (Fig. 2b). However, the structure pro- tem (e.g. LEP). One exception was Clade 1d, which was found in a vided by RAG1 may also fit other separation schemes, as the few localities along the Danubian watershed in Slovenia and Serbia clades defined with cyt b have multiple connections among (dark green in Fig. 1). Clade 1 also included Italian, Slovenian and themselves (Clades 1 and 2, and 5 and 7). Among samples from Croatian samples from the North Adriatic area (Clade 1a, light the ‘mixed’ localities in the rivers Mušnica and Zalomka, which green) and samples from tributaries of the upper Sava (Clade 1c, contain mitochondrial haplotypes from two different clades, only localities from Slovenia) and middle Sava (Clade 1b, SLO and hybrids with nuclear variants from both clades were detected CRO samples). Samples from locations in CER (SLO) and ZRM (marked in yellow, Fig. 2b). (CRO), respectively, clustered into two different clades: 1a and 1c, and 1a and 1b. from river RIJ (CRO), which flows directly 4. Discussion into the Adriatic Sea, exhibited haplotypes clustered to Clade 1b, rather than as expected to 1a. Mitochondrial cyt b gene analysis of more than 300 specimens Clade 2 (orange) comprised samples from tributaries of the collected in the Western Balkan Peninsula revealed eight clades lower Sava and rivers flowing directly into the Adriatic Sea, either with a range of genetic distance of 4–11%. Some of these clades aboveground or underground. Clades 3 and 4 (violet) were were also detected with partial nuclear RAG1 gene analysis. A. Palandacˇic´ et al. / Molecular Phylogenetics and Evolution 92 (2015) 118–123 121

Fig. 2. Molecular analyses based on cytochrome b and recombination activating gene 1. (a) Phylogenetic tree was constructed from cyt b collapsed alignment and includes 152 unique haplotypes originating from this study and cyt b sequence from voucher species Phoxinus lumarieul (Genebank HM560123). The tree was created using Bayesian inference (BEAST 1.8.0, Drummond et al., 2012). Branches carry posterior probabilities and bootstraps from the Maximum-Likelihood analysis (PhyML; Guindon et al., 2010). Weakly supported nodes are grey. Three letter abbreviations denote sampling sites (Fig. 1, Table 1). Outgroups chosen on BLAST similarity search and literature are Alburnoides bipunctatus ohridanus, A. bipunctatus strymonicus and Gila robusta. ⁄, denotes lack of bootstraps originating from the difference between the BEAST and ML trees. (b) Network created from 134 phased nuclear RAG1 haplotypes (unrooted minimum-spanning network, Bandelt et al., 1999). Colours denote mt cytochrome b clades. Each line between haplotypes represents one mutation, unless indicated otherwise (in red). Size of circle corresponds to haplotype frequencies, with the biggest encompassing 25 samples. Two hybrids from Mušnica River are marked in yellow.

Table 2 Estimates of evolutionary divergence over sequence pairs between the groups (cyt b clades and sub-clades; MEGA 5.0, Tamura et al., 2011) are shown below the diagonal in bold, standard error estimates obtained by a bootstrap procedure (1000 replicates) above the diagonal.

Clade 1a Clade 1b Clade 1c Clade 1d Clade 2 Clade 3 Clade 4 Clade 5 Clade 6 Clade 7 Clade 8 Clade 1a 0 0 0 0.01 0.01 0.01 0.01 0.01 0.01 0.01 Clade 1b 0.02 0 0 0.01 0.01 0.01 0.01 0.01 0.01 0.01 Clade 1c 0.02 0.03 0 0.01 0.01 0.01 0.01 0.01 0.01 0.01 Clade 1d 0.03 0.04 0.03 0.01 0.01 0.01 0.01 0.01 0.01 0.01 Clade 2 0.05 0.05 0.05 0.06 0.01 0.01 0.01 0.01 0.01 0.01 Clade 3 0.06 0.06 0.06 0.06 0.05 0.01 0.01 0.01 0.01 0.01 Clade 4 0.06 0.06 0.05 0.06 0.05 0.05 0.01 0.01 0.01 0.01 Clade 5 0.06 0.06 0.06 0.07 0.06 0.06 0.06 0.01 0.01 0.01 Clade 6 0.08 0.08 0.08 0.08 0.07 0.08 0.09 0.09 0.01 0.01 Clade 7 0.09 0.09 0.09 0.09 0.07 0.09 0.10 0.10 0.09 0.01 Clade 8 0.09 0.10 0.10 0.10 0.08 0.09 0.11 0.11 0.10 0.09

Hence, molecular analysis confirmed the results of morphological Contrary to the notion of P. phoxinus in the Danubian watershed studies, indicating that the genus Phoxinus in the Balkans is not and P. lumaireul in the Adriatic watershed (Kottelat and Freyhof, represented by a single species. 2007), our results point to a much larger number of deeply 122 A. Palandacˇic´ et al. / Molecular Phylogenetics and Evolution 92 (2015) 118–123 divergent genetic entities of Phoxinus extending north-west to Supplementary material), may point to the possibility of subter- south-east along the western Balkan Peninsula across both water- ranean dispersion. sheds (Fig. 1). This considerably more structured phylogeographic The type locality of Cyprinus lumaireul, originally described by pattern than previously proposed may explain why taxonomists Schinz in 1840, is the River Po in Italy, and thus P. lumaireul could have been unable to clearly distinguish putative P. lumarieul from be attributed only to the linage in the northern Adriatic area. P. phoxinus (Bianco, 2014). For example, Adriatic Clade 1a is more However, no type specimens are known to be available for closely related to its neighbouring Danubian 1b–d sub-clades than investigation (Eschmeyer, 2014). According to the present study, to those along the Adriatic coast. Furthermore, Clade 2 extends P. lumaireul could coincide with Clade 1a (northern Adriatic: across both the Adriatic and Danube watersheds. Therefore, the samples from River Ticino, Po drainage), but the cyt b genetic molecular data do not support a separate single Phoxinus species distance from neighbouring Clades 1b and 1c is only 2%. Several inhabiting the whole Adriatic range, nor does it support the authors quote cyt b differences of between 2% and 11% when trying two-species view (i.e. P. phoxinus and P. lumaireul), but rather indi- to delimitate species or evolutionary significant units (Gilles et al., cates a multispecies complex. 2010; Schönhuth et al., 2012; Tsoumani et al., 2014). Nevertheless, The haplotypes detected in the southern Adriatic basin formed they agree that a number of genetic markers (mitochondrial the two most distantly related groups on the tree (Clades 7 and 8). and nuclear), as well as morphological and geographical This observation aligns with the deep phylogenetic divergence of characterisation, should be combined for final species or linage many other fish lineages populating lakes Skadar and Ohrid com- determination. Whereas the largest distances observed in the pared to their counterparts in Danubian river systems, as reported present study were from 7% to 11% for Clades 6, 7 and 8, and in morphological (Talevski et al., 2009; Bogutskaya et al., 2012) and the three clades are also supported by the network based on the molecular studies (Milosevic´ et al., 2011; Tsoumani et al., 2014). nuclear RAG1 gene and partially by their geographical distribution, Moreover, we found three additional clades along the Adriatic river taken together such findings might be sufficient evidence for system, in congruence with a detailed study on sticklebacks determination of a new species. Yet, to be absolutely certain, a (Gasterosteus aculeatus; DeFaveri et al., 2012) covering the area of detailed morphological study is required. the Adriatic Sea basin. Four genetic lineages reported from the Mirna, Krka, Neretva and Skadar watersheds match the distribu- Acknowledgments tion pattern of Clades 1a, 2, 6 and 7. A comparable pattern of dis- tribution can also be seen in Balkan trout (Kottelat and Freyhof, We thank Iain F. Wilson for improving the manuscript and Dr 2007). Ivan Bogut, Omar Bašic´, Dr Saša Maric´ and the Fisheries Research As revealed from the mtDNA data reported in the present study, Institute of Slovenia for providing samples. Our thanks extend to samples from Zalomka and Mušnica clustered into two different Dr Luise Kruckenhauser for support throughout the writing pro- clades (2 and 7), while based on the RAG1 fragment they mostly cess, the two anonymous reviewers and the editor Dr Guillermo grouped in Clade 7. According to Funk and Omland (2003) there Ortí for helpful comments on the manuscript. are three possible explanations for alterations in gene trees: (1) RS was supported by the Czech Ministry of Culture (project no. lack of signal in the chosen sequences, (2) incomplete lineage sort- DF12P01OVV021). ing, and (3) introgressive hybridisation (ancient or recent). In the present study, the first option could apply to RAG1 sequences, which exhibited limited diversity. The two other possible causes, Appendix A. Supplementary material incomplete linage sorting and introgressive hybridisation, both produce similar patterns and are hard to distinguish (Tancioni Supplementary data associated with this article can be found, in et al., 2013). In cases of sympatric sharing of geographically loca- the online version, at http://dx.doi.org/10.1016/j.ympev.2015.05. lised mtDNA sequences such as reported in the present paper, 024. Funk and Omland (2003) proposed introgressive hybridisation, which is also suggested by the presence of two hybrids detected References in Mušnica River (yellow, Fig. 2). As both localities with mixed hap- lotypes are intermittent streams in a karst area, where single Bandelt, H.J., Forster, P., Röhl, A., 1999. Median-joining networks for inferring intraspecific phylogenies. Mol. Biol. Evol. 16 (1), 37–48. (unique) or recurrent underground fish migrations have been pre- Bianco, P.G., 2014. An update on the status of native and exotic freshwater fishes of viously reported (Borowsky and Mertz, 2001; Dillman et al., 2010; Italy. J. Appl. Ichthyol. 30, 62–77. http://dx.doi.org/10.1111/jai.12291. Palandacˇic´ et al., 2012), we suggest this situation arose through Bogutskaya, N.G., Naseka, A.M., 2004. Catalogue of Agnathans and of Fresh and Brackish Waters of Russia with Comments on Nomenclature and secondary contact followed by introgression via springs connected . KMK Scientific Press Ltd., Moscow, 389 pp. (in Russian). underground. Bogutskaya, N.G., Zupancˇicˇ, P., Bogut, I., Naseka, A.M., 2012. Two new freshwater Secondary contact is also supported by the old hydrological lit- fish species of the genus Telestes (, Cyprinidae) from karst in Eastern Herzegovina and littoral ( and erature (all in Serbian, and cited in detail in Bogutskaya et al., Croatia). Zookeys 180, 53–80. 2012), according to which rivers Zalomka and Mušnica, together Bonacci, O., 1999. Water circulation in karst and determination of catchment areas: with the Trebišnjica, Bregava, and Bunica, belong to a example of the Zrmanja River. Hydrol. Sci. J. 44 (3), 373–386. http://dx.doi.org/ 10.1080/02626669909492233. Neretva tributary and are interconnected underground. Such a Borowsky, R.B., Mertz, L., 2001. Genetic differentiation among populations of the complex hydrological network is congruent with our findings, cave fish Schistura oedipus (: Balitoridae). Environ. Biol. Fishes 62, which placed samples from Bregava and Bunica to mt Clade 2 225–231. and Trebišnjica to Clade 7, while samples from Mušnica and DeFaveri, J., Zanella, L.N., Zanella, D., Mrakovcˇic´, M., Merilä, J., 2012. Phylogeography of isolated freshwater three-spined stickleback Gasterosteus aculeatus Zalomka belonged to both (Fig. 1). Even though human activities populations in the Adriatic Sea basin. J. Fish Biol. 80 (1), 61–85. cannot be ruled out as a possibility for the presence of mixed or Dillman, C.B., Bergstrom, D.E., Noltie, D.B., Holtsford, T.P., Mayden, R.L., 2010. unexpected haplotypes in certain locations, many of such occur- Regressive progression, progressive regression or neither? Phylogeny and evolution of the Percopsiformes (Teleostei, Paracanthopterygii). Zool. Scr. 40 rences in the karst area suggest otherwise (rivers CER, ZRM, RIJ, (1), 45–60. SVR, BRR). The unusual geographic distribution of some of the hap- Drummond, A.J., Suchard, M.A., Xie, D., Rambaut, A., 2012. Bayesian phylogenetics logroups (or clades), together with a comprehensive hydrological with BEAUti and the BEAST 1.7. Mol. Biol. Evol. 29 (8), 1969–1973. Eschmeyer, W.N., 2014. Catalog of Fishes, Genera, Species, References. (electronic Bonacci, 1999; Kovacˇicˇ, 2010; see also Table 1s in the version accessed 14.04.14). A. Palandacˇic´ et al. / Molecular Phylogenetics and Evolution 92 (2015) 118–123 123

Funk, D.J., Omland, K.E., 2003. Species-level paraphyly and polyphyly: frequency, Palandacˇic´, A., Matschiner, M., Zupancˇicˇ, P., Snoj, A., 2012. Fish migrate causes, and consequences, with insights from animal mitochondrial DNA. Annu. underground: the example of adspersus (Cyprinidae). Mol. Ecol. Rev. Ecol. Evol. Syst. 34, 397–423. 21 (7), 1658–1671. Gilles, A., Costedoat, C., Barascud, B., Voisin, A., Banarescu, P., Bianco, P.G., Perdices, A., Doadrio, I., Economidis, P.S., Bohlen, J., Banarescu, P., 2003. Pleistocene Economidis, P.S., Maric´, D., Chappaz, R., 2010. Speciation pattern of Telestes effects on the European freshwater fish fauna: double origin of the cobitid souffia complex (Teleostei, Cyprinidae) in Europe using morphological and genus Sabanejewia in the Danube basin (Osteichthyes: Cobitidae). Mol. molecular markers. Zool. Scr. 39, 225–242. Phylogenet. Evol. 26 (2), 289–299. Guindon, S., Dufayard, J.F., Lefort, V., Anisimova, M., Hordijk, W., Gascuel, O., 2010. Schönhuth, S., Hillis, D.M., Neely, D.A., Lozano-Vilano, L., Perdices, A., Mayden, R.L., New algorithms and methods to estimate Maximum-Likelihood phylogenies: 2012. Phylogeny, diversity, and species delimitation of the North American assessing the performance of PhyML 3.0. Syst. Biol. 59 (3), 307–321. round-nosed (Teleostei: Dionda), as inferred from mitochondrial and Harrigan, R.J., Mazza, M.E., Sorenson, M.D., 2008. Computation vs. cloning: nuclear DNA sequences. Mol. Phylogenet. Evol. 62 (1), 427–446. evaluation of two methods for haplotype determination. Mol. Ecol. Resour. 8 Stephens, M., Scheet, P., 2005. Accounting for decay of linkage disequilibrium in (6), 1239–1248. haplotype inference and missing-data imputation. Am. J. Hum. Genet. 76 (3), Karamata, S., 2006. The geological development of the Balkan Peninsula related to 449–462. the approach, collision and compression of Gondwanan and Eurasian units. Stephens, M., Smith, N.J., Donnelly, P., 2001. A new statistical method for haplotype Geol. Soc., London, Spec. Publ. 260, 155–178. http://dx.doi.org/10.1144/ reconstruction from population data. Am. J. Hum. Genet. 68, 978–989. GSL.SP.2006.260.01.07. Talevski, T., Milosevic´, D., Maric´, D., Petrovic´, D., Talevska, M., Talevska, A., 2009. Kottelat, M., 2007. Three new species of Phoxinus from Greece and southern France Biodiversity of Ichthyofauna from Lake Prespa, Lake Ohrid and Lake Skadar. (Teleostei: Cyprinidae). Ichthyol. Explor. Freshw. 18 (2), 145–162. Biotechnol. Biotechnol. Equip. 23 (1), 400–404. Kottelat, M., Freyhof, J., 2007. Handbook of European Freshwater Fishes. Steven Tamura, K., Peterson, D., Peterson, N., Stecher, G., Nei, M., Kumar, S., 2011. MEGA5: Simpson Books. molecular evolutionary genetics analysis using maximum likelihood, Kovacˇicˇ, G., 2010. An attempt towards an assessment of the Cerknica water evolutionary distance, and maximum parsimony methods. Mol. Biol. Evol. 28, balance. Acta Carsol. 39 (1), 39–50. 2731–2739. Mendel, J., Lusk, S., Vasileva, E., Vasilev, V.P., Lusková, V., Ekmekci, F.G., Erkakan, F., Tancioni, L., Russo, T., Cataudella, S., Milana, V., Hett, A.K., Corsi, E., Rossi, A.R., 2013. Ruchin, A., Košcˇo, J., Vetešník, L., Halacˇka, K., Šanda, R., Pashkov, A.N., Testing species delimitations in four Italian sympatric Leuciscine fishes in the Reshetnikovet, S.I., 2008. Molecular phylogeny of the genus Gobio Cuvier, Tiber River: a combined morphological and molecular approach. PLoS One 8 (4), 1816 (Teleostei: Cyprinidae) and its contribution to taxonomy. Mol. Phylogenet. e60392. http://dx.doi.org/10.1371/journal.pone.0060392. Evol. 47 (3), 1061–1075. Tsoumani, M., Georgiadis, A., Giantsis, A.I., Leonardos, I., Apostolidis, A.P., 2014. Milosevic´, D., Winkler, K.A., Maric´, D., Weiss, S., 2011. Genotypic and phenotypic Phylogenetic relationships among Southern Balkan Rutilus species inferred from evaluation of Rutilus spp. from Skadar, Ohrid and Prespa Lakes supports revision cytochrome b sequence analysis: micro-geographic resolution and taxonomic of endemic as well as taxonomic status of several taxa. J. Fish Biol. 79 (5), 1094– implications. Biochem. Syst. Ecol. 54, 172–178. 1110. Zardoya, R., Doadrio, I., 1999. Molecular evidence on the evolutionary and biogeographical patterns of European Cyprinids. J. Mol. Evol. 49, 227–237.