Phylogeography of Bleaks Alburnus Spp. (Cyprinidae) in Italy, Based on Cytochrome B Data

Journal of Fish Biology (2009) 75, 997–1017 doi:10.1111/j.1095-8649.2009.02357.x, available online at www.interscience.wiley.com

Phylogeography of bleaks Alburnus spp. (Cyprinidae) in Italy, based on cytochrome b data

V. Ketmaier*†, F. Finamore‡, C. Largiader§,` M. Milone‡ and P. G. Bianco‡

*Unit of Evolutionary Biology–Systematic Zoology, Institute of Biochemistry and Biology, University of Potsdam, Karl-Liebknecht-Strasse 24-25, Haus 25, D-14476, Potsdam, Germany, ‡Department of Zoology, University of Naples, via Mezzocannone 8, I-80134 Naples, Italy and §University Hospital, University of Bern, Inselspital, CH-3010 Bern, Switzerland

(Received 29 January 2009, Accepted 13 May 2009)

Sequence variation of a fragment of the mitochondrial DNA encoding for the cytochrome b gene was used to reconstruct the phylogeography of the two species of bleaks occurring in Italy: the alborella Alburnus arborella in northern Italy and the vulturino Alburnus albidus in southern Italy. The study includes four populations of the alborella and 14 populations of the vulturino. A total of 57 haplotypes were identified; these could not be sorted into two reciprocally monophyletic clusters. Multiple phylogenetic methods and nested clade phylogeographical analysis consistently retrieved three well-supported clades, two of which contained both Northern and Southern Italian haplotypes. A third clade is limited to southern Italy. This clade is tentatively assigned to the vulturino. The placement in the same clade of northern and southern Italian haplotypes is explained in light of the introductions of fishes operated from northern to central and southern Italy. The origin of the vulturino dates back to the last two million years. This divergence time estimate identifies the Pleistocene confluences between adjacent river basins along the Adriatic slope of the Italian peninsula and their subsequent isolation as the cause that triggered the diversification of the genus in the area. The existence of a clade endemic to southern Italy supports the recognition of the area as a new peri-Mediterranean ichthyogeographic district, the borders of which correspond to the northern and southern edges of the vulturino range. © 2009 The Authors Journal compilation © 2009 The Fisheries Society of the British Isles

Key words: Alburnus; Cyprinidae; Italian peninsula; mitochondrial DNA; phylogeography.

INTRODUCTION Italy, like several other peri-Mediterranean countries, hosts a distinctive assemblage of primary freshwater fishes, mostly cyprinids. This has led to the recognition of two ichthyogeographic districts in the country, a Padano-Venetian district (north of Apennines) and a Tuscano-Latium district (south of Apennines from River Arno to River Tiber along the Tyrrhenian slope of central Italy; Fig. 1) (Bianco, 1987). The two districts can be clearly distinguished on the basis of their respective primary freshwater fish fauna. Conversely, very little is known on the freshwater fish fauna

†Author to whom correspondence should be addressed. Tel.: +49 331 9775604; fax: +49 331 977 5070; email: [email protected] 997 © 2009 The Authors Journal compilation © 2009 The Fisheries Society of the British Isles 998 V. KETMAIER ET AL. for details) Introduction )–(a) Sampling localities a Tyrrhenian slope Ionian slope Adriatic slope Southern Italy Padano–Venetian district (from 14 to 17; each locality identiﬁed by the same 7-Sinni River 8-Basento River 9-Ofanto River 10-Carapelle River 11-Fortore River 12-Biferno River 13-Treste River 14-Adda River 15-Ticino River 16-Maggiore Lake 17-Meletta Stream 4-Fiumarella River 5-Ripiti River 6-Bussento River 1-Alento River 2-Badolato River 3-Palistro River (c) (b) A. arborella PV) and Tuscano-Latium (TL) district (Bianco, 1995 7 8 9 6 14 10 , original distribution in grey, areas of introduction in pale red (see 3 5 2 4 15 11 1 16 12 nt colour)–(b) Sampling localities for 17 13 ers are also indicated. Population numbers are as in Table I. Alburnus arborella (a) , green; dashed lines indicate the Padano–Venetian ( TL PV Alburnus albidus (from 1 to 13; each locality identiﬁed by a differe A. albidus for and of the vulturino shade of grey)–(c) Slopes of the sampled Southern Italian riv . 1. (a–c) Schematic maps showing ranges of the alborella ig F

© 2009 The Authors Journal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 997–1017 PHYLOGEOGRAPHY OF THE GENUS ALBURNUS IN ITALY 999 of Southern Italy. This lack of knowledge is due to a combination of historical factors (ichthyological researches had focused mainly on central and northern Italy) and environmental conditions. Southern Italy is the poorest drained area of the peninsula; river basins exhibit a typical Mediterranean hydrological regime with wide seasonal and annual fluctuations (Bianco, 1995a). In addition, for nearly 100 years from the end of the 19th century, official massive translocations of fishes were made from northern to central and southern Italy (Bianco, 1995a). This practice has superseded the original fish fauna composition, rendering the recognition of the authochthonous southern Italian fish fauna extremely difficult. Only five cyprinid species have been consistently reported for southern Italian rivers, such as the Apennine barbel Barbus tyberinus Bonaparte, the Italian chub Squalius cephalus squalius (L.), the vairone Telestes muticellus (Bonaparte), the loach Cobitis zanandrei Cavicchioli and the vulturino Alburnus albidus (Costa). Historical records tracking translocations of the first two species across the Italian peninsula render their autochthonous occurrence in the area that is very dubious (Bianco, 1995a,b). Conversely, the other species have very limited ranges with A. albidus being by far the most widespread. In particular, Rivers Volturno and Sinni–Alento mark, respectively, the northern and southern edges of the range of A. albidus (Fig. 1) (Bianco, 1980, 1987; Kottelat, 1997). Alburnus albidus is a warm-adapted, preferentially lacustrine species, typical of the lower section of rivers; it prefers small, shallow, well-oxygenated streams, with aquatic and riparian vegetation, coarse substrata and medium to low or standing flow. The species is currently classified as vulnerable in the IUCN Red List. Indeed, its abundance has been declining steadily in the last 20 years as a consequence of habitat alteration by impoundment and waste discharge, river regulation, introduction of exotic competitive species and hybridization with S. cephalus (Bianco, 1982, 1995a). The Italian fauna includes a second representative of the genus Alburnus, the alborella Alburnus arborella (Bonaparte). Taxonomy of this species has been controversial since the end of the 19th century. Alburnus arborella has been alternatively considered a subspecies of A. albidus or of the European bleak Alburnus alburnus (L.); only recently the name A. arborella has been formally validated (Kot- telat & Bianco, 2005). The species was originally endemic to the Padano–Venetian district, but it has been largely transplanted into the Tuscano–Latium district (Fig. 1) following the practice to stock central Italian rivers with fishes of northern Italian origin (Bianco, 1995b). No precise data are available on whether and to what extent A. arborella might have been introduced south of the Tiber River. Should this be the case it would represent a further threat for the survival of the A. albidus because of possible hybridization and competition between the two species (Bianco, 1995a). The aims of this study were to produce a comprehensive phylogeographic hypothesis for the vulnerable A. albidus throughout its entire geographical range. Hence, 141 individuals from 13 southern Italian rivers (six flowing into the Tyrrhenian Sea, two flowing into the Ionic Sea and five flowing into the Adriatic Sea; Fig. 1) were collected. The sampling covers the entire species range (Bianco, 1980). Each individual included in the study was screened for sequence polymorphisms of a fragment of 565 base pairs (bp) of the mitochondrial DNA (mtDNA) region encoding for the cytochrome b gene (cyt b). This gene has proven useful in phylogeographic studies on cyprinids (Ketmaier et al., 2004, 2008a and references therein). This is the first molecular study based on an extensive sampling of an endemic Southern Italian

© 2009 The Authors Journal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 997–1017 1000 V. KETMAIER ET AL. cyprinid species; it thus might shed light on the tempo and mode of diversiﬁcation of the strictly freshwater fauna of the area. Second, given the vagaries on the taxonomic status of A. alborella, reciprocal monophyly at the cyt b level of Northern and Southern Italian bleaks was tested preliminarily in this study. To this aim, four populations (29 individuals) of the alborella from three tributaries of the Po River and from the Maggiore Lake (Fig. 1) were included in the study. Warm-adapted, preferentially lacustrine species such as those considered in the present study can only exploit coalescences of rivers on lowlands to disperse (Bianco, 1995b;Ketmaieret al., 2004). These occurred intermittently at every Quaternary glacial maximum as a consequence of the global lowering of sea level. Coalescences were particularly pronounced along the Adriatic coast of the Italian peninsula due to the southward extension of the Po River (Bianco, 1995b). They allowed warm-water-adapted species to surmount the geographical barrier represented by the Apennines and to disperse in a north–south direction. Therefore, the eventual split between Northern and Southern Italian lineages should not be older than two million years (myr).

MATERIALS AND METHODS

SAMPLING, DNA EXTRACTION, PCR AMPLIFICATION AND SEQUENCING A total of 141 individuals of A. albidus and 29 individuals of A. arborella (13 and four populations, respectively) were collected by electrofishing from 17 rivers and one lake, including the A. albidus type locality (Alento River) (Fig. 1 and Table I). A clip of the pelvic fin and several scales were removed from each sample and fixed in absolute ethanol; fishes were released later. Sampled individuals were assigned to the two species on the basis of the characters used by Bianco (1980) to revise the systematics of the genera Alburnus and Alburnoides. These are the number of branched rays of the anal fin (12–14 in A. albidus; 14–19 in A. arborella) and the position of the mouth (terminal in A. albidus; superior in A. arborella). Total genomic DNA was extracted following the protocol described by Briolay et al. (1998). A 565 bp long fragment of the cyt b gene was amplified by polymerase chain reaction (PCR) using the primer pair L15267 and H15891 (Briolay et al., 1998) for all individuals included in the study. Double-stranded PCR amplifications were performed in a 50 μL reaction volume containing 10 mM Tris–HCl (pH 8·8), 50 mM KCl, 1·5mMMgCl2, each dNTP at 2·5 mM, each primer at 1 mM, genomic DNA (10–100 ng) and 5 units of Amplitaq (Applied Biosystems; www.appliedbiosystems.com). Each PCR cycle (for a total of 30) consisted of ◦ ◦ ◦ a denaturation step at 94 C for 1 min, annealing at 55 C for 1 min and extension at 72 C ◦ for 2 min; cycles were followed by a final extension at 72 C for 7 min. PCR products were purified using the GenEluteTM PCR DNA Purification kit (Sigma; www.sigmaaldrich.com). Sequences were determined with an automated sequencer (Applied Biosystems 373A) following the manufacturer’s protocols. To promote accuracy, strands were sequenced in both directions for each individual using the same primers used for PCRs.

PHYLOGENETIC AND POPULATION GENETICS ANALYSES Sequences were aligned by eye following the guideline provided by the reading frame and analysed phylogenetically by neighbour-joining (NJ) and Bayesian methods as implemented in PAUP* 4·0b10 (Swofford, 2002) and MrBayes 3·0b4 (Ronquist & Huelsenbeck, 2003), respectively. PAUP* 4·0b10 (Swofford, 2002) was also used to calculate base frequencies and the absolute number of transitions (Ti) and transversions (Tv). The following parameters of genetic diversity for each population were calculated using Arlequin 3·0 (Excofﬁer et al.,

Table I. Sampling locations and sample sizes for the vulturino Alburnus albidus and the alborella Alburnus arborella. Numbers in the ﬁrst column correspond to those in Fig. 1. The last four columns report diversity indices for cyt b data for each population. These are the number of haplotypes (H ), haplotype diversity (h), mean number of pair-wise differences between haplotypes (π) and nucleotide diversity (πn)

No. River/lake NHhππn (%) A. albidus 1AlentoR.5 2 0·60 0·60 10·62 2 Badolato R. 18 6 0·78 5·03 89·19 3 Palistro R. 13 4 0·68 2·82 49·92 4 Fiumarella R. 17 3 0·58 0·63 11·19 5 Ripiti R. 1 1 — — — 6 Bussento R. 15 4 0·73 1·51 26·63 7 Sinni R. 6 3 0·60 1·66 29·50 8 Basento R. 8 5 0·86 7·10 125·79 9OfantoR.26190·98 8·07 142·95 10 Carapelle R. 6 4 0·80 2·06 36·58 11 Fortore R. 7 4 0·86 5·14 91·02 12 Biferno R. 7 5 0·90 6·85 121·37 13 Treste R. 12 4 0·45 0·63 11·26 A. arborella 14 Adda R. 7 6 0·95 5·52 98·64 15 Ticino R. 18 11 0·86 6·01 109·01 16 Maggiore L. 2 2 1·00 11·00 221·77 17 Meletta S. 2 1 — — —

2005): haplotype diversity (h), mean number of pair-wise differences between all pairs of haplotype (π) and the nucleotide diversity (πn). Trees were routed with Scardinius acarnanicus Economidis (GenBank Accession N. AF090775) and Anaecypris hispanica (Steindachner) (GenBank Accession N. AF045978). According to the cyt b phylogeny of the European cyprinids, Anaecypris is sister to Alburnus and Scardinius is basal to the Anaecypris–Alburnus clade (Zardoya & Doadrio, 1999). For comparative purposes, the following Alburnus species were also included in the analysis: A. alburnus, Alburnus escherichii Steindachner and Albur- nus filippi Kessler (GenBank Accession N. Y10443, AY026390 and AF095602, respectively). Modeltest (Posada & Crandall, 1998) was used to determine the model of sequence evolution that best fits the data. The output of Modeltest was implemented to calculate maximum likelihood (ML) distances for the NJ analyses. For the Bayesian approach, the same model of sequence evolution was adopted allowing site-specific rate variation partitioned by codon positions. MRBayes was run for two million generations with a sampling frequency of 100 generations. One cold and three heated Markov chains were run (two independent runs). To establish whether the Markov chains had reached stationarity, the likelihood scores of sampled trees were plotted against generation time. Trees generated before the stationarity phase were discarded as ‘burn-in’ (first 10% of the sampled trees), and posterior probability values for each node were calculated based on the remaining 90% of sampled trees. These trees were then used to construct a 50% majority rule consensus tree using PAUP* 4·0b10 (Swofford, 2002). The robustness of the NJ phylogenetic hypotheses was tested by 1000 bootstrap replicates (Felsenstein, 1985). Phylogenetic tree topologies generated with different phylogenetic methods and competing phylogenetic hypotheses were statistically evaluated with the approximately unbiased tree selection test (AU) (Shimodaira, 2002), as implemented in the software package CONSEL (Shimodaira & Hasegawa, 2001). For comparison, the more conservative Shimodaira and Hasegawa (SH) test (Shimodaira & Hasegawa, 1999)

© 2009 The Authors Journal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 997–1017 1002 V. KETMAIER ET AL. was also performed as implemented in PAUP* version 4·0b10 with the re-sampling estimate log-likelihood (RELL) technique. Tree topologies were always compared simultaneously (Shimodaira & Hasegawa, 1999). Divergence times were calculated in a Bayesian Markov-Chain Monte Carlo (MCMC) framework using Beast 1·4·6 (Drummond & Rambaut, 2007). A model of uncorrelated but log-normally distributed rates of molecular evolution was adopted (Drummond et al., 2006). By taking advantage of two published cyt b phylogenies, three calibration points were added to our phylogenetic tree: 6·5–7·5 myr for the separation of Anaecypris from Scardinius (Zardoya & Doadrio, 1999), 5·5–6·5 myr for the origin of the genus Alburnus and 3·5–2·5 M years for the split between A. alburnus and A. escherichii (Durand et al., 2002). Log-normal prior distributions for the calibration were adopted in the analysis so that 95% of the prior mass fell on the specified interval as well as a Yule prior on rates of evolution. The same model of sequence evolution as in the Bayesian searches was used in the analysis. Three independent analyses of 50 000 000 generations each were run; the corresponding outputs were analysed in Tracer 1·4, TreeAnnotator 1·4·6 and FigTree 1·0 (Drummond & Rambaut, 2007). The nested clade phylogeographical analysis (NCPA) (Templeton, 1998) was used to test the association between genealogy and geography in the data set to infer population processes. In spite of recent criticism (Petit, 2008), NCPA has still a number of advantages compared with alternative methods (Templeton, 2009). NCPA was carried out using ANeCA (Panchal & Beaumont, 2007). This software automates the complex NCPA and thus renders manual and questionable procedures (i.e. the nesting design) in the analysis unnecessary. The package includes TCS (Clement et al., 2000) and GEODIS (Posada et al., 2000) to, respectively, derive the haplotype network and to calculate distances among haplotypes and clades and their statistical significance. The statistical distribution of distances was determined by recalculating distances after 10 000 random permutations (i.e. clades against sampling localities). ANeCA automatically runs a slightly modified version of the inference keys dated 14 July 2004. Finally, NCPA was complemented by computing an index for haplotypes and n-step clades (n = 0, 1, 2, etc.) found in each sampling site (Templeton, 2001). The index is calculated by averaging the pair-wise geographical distances between the geographical centres of haplotypes and clades found in each sampling site (geographical centres are provided in the GEODIS output). For a given sampling site, the index will increase if haplotypes or clades found in that site have centres geographically away from each other. This implies that the site hosts divergent haplotypes, a phenomenon usually associated with secondary contacts of divergent lineages. Conversely, if the index decays with increase in the hierarchical level (i.e. from the haplotype to the nth step clade), this means that at that site all haplotypes are closely related and isolation by distance is the most likely process responsible for the observed pattern of haplotype occurrence. Levels of genetic diversity were tested by a hierarchical analysis of molecular variance (AMOVA) (Excoffier et al., 1992) using Arlequin 3·0 (Excoffier et al., 2005) with 1000 permutations. For this analysis, sampled drainages were grouped by slope of origin (Adriatic, Ionian, Tyrrhenian plus Northern Italy, see Table I). AMOVA was also run on populations grouped by species (A. arborella v. A. albidus). SAMOVA (Dupanloup et al., 2002) was used to determine whether the data formed significant groupings without making apriori assumptions about the location of phylogeographic breaks. SAMOVA was used to assign samples to groups that were geographically homogeneous and maximally differentiated from each other. This involved partitioning the samples into two to n putative groups, with n equal to the total number of locations sampled. AMOVA was then re-run to partition variation within and among the groups identified by SAMOVA. Arlequin 3·0 (Excoffier et al., 2005) was used to calculate pair-wise FST values for all pairs of populations; statistical significance of FST values was assessed by 1000 permutations with sequential Bonferroni correction for multiple tests. Arlequin 3·0 (Excoffier et al., 2005) was also used to run Tajima’s D (Tajima, 1989) and Fu’s Fs (Fu, 1997) tests to determine whether patterns of sequence variation were consistent with predictions of the neutral model. These tests are also informative about the demographic forces that have affected the species (Fu, 1997; Tajima, 1989). Significance of the tests was assessed by 1000 permutations.

10 55 30 Southern Italy district 87/- 32 91/- 43 1-Alento River 6 41 2-Badolato River 51 93/- 52 3-Palistro River Tyrrhenian slope 100/90 15 I 4-Fiumarella River 3 31 5-Ripiti River 53 6-Bussento River 33 56 96/- 87/- 76/- 57 7-Sinni River 99/- Ionian slope 9 8-Basento River 19 11 91/- 40 9-Ofanto River 44 14 10-Carapelle River 92/- 47 11-Fortore River 48 II Adriatic slope 7 12-Biferno River 100/100 42 13-Treste River 5 45 38 100/74 25 Padano–Venetian district 98/- 24 22 14-Adda River 26 100/98 2 15-Ticino River 23 16-Maggiore Lake 28 13 17-Meletta Stream 34 12 85/- 96/- 27 4 III 29 37 100/77 17 18 35 98/86 39 1 36 80/- 20 98/75 21 46 8 54 100/84 49 16 100/88 50 A. alburnus A. escherichii A. filippi A. hispanica S. acarnanicus 0·005 substitutions site−1

Fig. 2. Bayesian haplotype phylogram based on the GTR+G (shape parameter α = 0·1652) model of sequence evolution. Only statistical supports ≥75% for both the NJ (1000 bootstrap replicates) and Bayesian searches (2 000 000 generations) are reported on branches. Haplotype numbers are as in Table II; different colours identify the geographical origin of haplotypes and match those in Fig. 1. Bold numbers identify haplotypes from southern Italy (A. albidus); normal numbers are for haplotypes conﬁned to northern Italy (A. arborella), whereas boxed numbers identify haplotypes shared between the two areas. Clusters I, II and III are described in the text.

RESULTS

SEQUENCE DIVERSITY AND HAPLOTYPE DISTRIBUTION Cyt b sequences were globally G deficient (17·52%); the other nucleotides had similar frequencies (A 25·84%; C 27·26%; T 29·38%). This nucleotide composition has been frequently reported in cyt b-based studies on a variety of fishes, including cyprinids (Cantatore et al., 1994; Brito et al., 1997; Ketmaier et al., 2004). Most of the observed substitutions were transitions (Ti; 93·8%); the transition/transversion ratio (Ti/Tv) was 19·6. No gaps were found in the alignment; sequences have been deposited in GenBank (Accession N. FJ177362-FJ177418). Among the 170 A. alburnus included in the study, 57 haplotypes were identified; these were defined by 58 variables sites, 32 of which were parsimony informative. Table II lists the haplotypes and their absolute frequency in the populations and species included in the study. All the study populations (except those from the Ripiti River and Meletta Stream) had more than one haplotype; number of haplotypes was the highest in the Ofanto River (19). Estimates of genetic variability vary accordingly (Table I). The 58 individuals sampled in the Adriatic rivers exhibited 33 haplotypes, 24 of which were exclusive for the area. The 14 individuals from Ionian rivers carried seven haplotypes (four exclusive); 12 haplotypes (eight exclusive) were found among the 69 individuals sampled in rivers flowing into the Tyrrhenian Sea. Finally, the three north Italian populations (29 individuals) bore 18 haplotypes (14 exclusive). A high number of haplotypes is unique to single populations. Only six haplotypes (13, 18, 19, 20, 22 and 24) were present in three or more populations with haplotype 13

Table II. Absolute haplotype frequencies of all individuals of the A. albidus and A. arborella sequenced for the study. For population identiﬁers, see Fig. 1 and Table I

Haplotypes Pop.12345678910111213141516171819202122232425262728 123 2713 3 7 312 497 51 66 74 11 82 91121123121121121111 10 1 3 1 1 11 2 2 2 12 2 13 9111 14 1 15 1 1 1 16 17

© 2009 The Authors Journal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 997–1017 PHYLOGEOGRAPHY OF THE GENUS ALBURNUS IN ITALY 1005 being by far the most frequent. Four haplotypes were shared between northern and southern Italian populations (7, 10, 19 and 32) but none of them was simultaneously distributed in Adriatic, Ionian, Tyrrhenian and North Italian drainages. Haplotype 19 was found in Tyrrhenian, Adriatic and Northern Italian drainages but not in Ionian rivers.

PHYLOGENETIC SEARCHES, DIVERGENCE TIMES AND NESTED CLADE PHYLOGEOGRAPHICAL ANALYSIS Fig. 2 shows the Bayesian topology based on the GTR+G (shape parameter α = 0·1652) model of sequence evolution (as selected by the Akaike information crite- rion in Modeltest) and summarizes the result of the NJ searches. Bayesian and NJ topologies were largely congruent and indistinguishable with the AU and SH tests (P>0·6 in both cases). All Alburnus populations and species clustered in a strongly supported monophyletic clade; A. alburnus was resolved as basal to a large clade including all A. arborella and A. albidus haplotypes. Haplotypes of these two species were grouped together with maximum support for either phylogenetic method, but they were not sorted into two reciprocally monophyletic clusters as expected. Rather, it was possible to identify three distinct and supported clades (labelled I, II and III in Fig. 2). Clades I and II both contain a mixture of northern and southern Italian haplotypes, whereas clade III clusters contain haplotypes of exclusive southern Ital- ian origin. Alternative hypotheses forcing northern and southern Italian haplotypes to form reciprocal monophyletic clades were rejected by both AU and SH tests (P 0·001).

Table II. Continued

Haplotypes 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57

511

351

3111

1 1121

12111 1 1 1711111 11 2

Genetic divergence among clades I, II and III ranged between 1·9 ± 0·2and0·3% (II v. I and III, respectively) and 2·1 ± 0·3% (I v. III). Sequence divergence among A. alburnus, A. escherichii and A. filippi comprised between 3·7and7·0%, whereas the comparison between the A. albidus–A. arborella clade and all the other Alburnus species displayed an average genetic divergence of 5·8 ± 1·3%. Outgroup v. ingroup divergence was 11·4 ± 0·6%. The Yule process birth rate retrieved from the Bayesian MCMC analyses was 0·748 [95% high posterior density, HPD: 0·486–1·031; ESS: (1·081)4] with a mean evolutionary rate of 0·002 substitutions per site per million years [95% HPD: 0·001–0·843; effective sample size, ESS: (1·074)4]. The data conformed only slightly to a clock- like behaviour, the coefficient of variation being 0·887 (HPD: 0·573–1·234; ESS: 7439·67). Parent and daughter branches showed no covariation, the mean covariance being 0·069 [HPD:–0·148 to 0·237; ESS: (3·679)4]. The mean covariance value spans zero; this implies that branches with fast and slow rates are next to each other in the phylogenetic tree. There was thus no evidence of autocorrelation of rates in the tree. The age of clades I, II and III were 0·820 (HPD: 0·235–1·481), 1·340 (HPD: 0·523–2·283) and 1·768 M years (HPD: 0·860–2·800), respectively. Fig. 3 shows the haplotype network and the nesting design with four levels of nesting (first level omitted). The pattern of relationships in the network was highly congruent with that depicted in the phylogenetic reconstruction of haplotypes with three major groups (clades 3-3, 3-5 and 3-4) corresponding to the three clusters (I, II and III) of Fig. 2. Clades 3-4 and 3-5 are connected to each other through eight missing haplotypes. Clade 3-4 includes only haplotypes found in southern Ital- ian rivers, whereas northern and southern Italian haplotypes are not sorted by their respective geographic origins in the other clades. Clade 3-4 shows little structuring by geographic origin of samples (i.e. it was not possible to identify a Tyrrhenian, an Ionian and an Adriatic group). This clade is essentially star-like in shape, with a single common haplotype (13) lying in the centre and connected to a high number of more rare haplotypes. The 19 haplotypes found in the most diverse population (River Ofanto) were spread throughout the network with six of them (1, 2, 4, 12, 13, 17 and 18; all but 13 and 18 exclusive to the location) limited to clade 3-4. The nesting procedure yielded 39 clades that could be tested for geographical association, eight of which resulted in significant permutational contingency tests indicating non- random geographical distribution of haplotypes (Table III). The automated inference keys implemented in ANeCA suggested allopatric fragmentation for clade 2-8, long- distance colonization and past fragmentation for clade 2-10; restricted gene flow with isolation by distance for clade 3-3; restricted gene flow but with some long- distance dispersal over intermediate not occupied areas for clades 4-1 and 4-2. In three circumstances (clades 3-4, 3-5 and for the total cladogram) inconclusive out- comes were obtained. The analysis suggested by Templeton (2001) was therefore carried out to complement the nested clade analysis; the results are shown in Fig. 4. For seven locations, the average pair-wise geographic distances go to zero as one moves from the haplotype to the four-step level clade. This suggests that haplotypes found in those sites tend to be closely related evolutionarily. This also means that isolation by distance is the most appropriate process to explain the geographical distribution of haplotypes. Conversely, in eight locations the pair-wise location distances increase moving from the haplotype to the four-step clade level. This result suggests that these eight sites are affected by admixture; that is, the joint presence

© 2009 The Authors Journal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 997–1017 PHYLOGEOGRAPHY OF THE GENUS ALBURNUS IN ITALY 1007 dicate how many espective of the length represents the nesting level (from x Adriatic slope Ionian slope Tyrrhenian slope ,where y - Southern Italy x Padano–Venetian district 1-Alento River 2-Badolato River 3-Palistro River 4-Fiumarella River 5-Ripiti River 6-Bussento River 9-Ofanto River 10-Carapelle River 11-Fortore River 12-Biferno River 13-Treste River 7-Sinni River 8-Basento River 14-Adda River 15-Ticino River 16-Maggiore Lake 17-Meletta Stream data. The relative size of the circles is proportional to the b 36(5) 3-4 (III) 2-1 28 (1) 17(1) 2(1) 39(1) 1(1) ways one mutational step away from each other irr 2-15 23(1) is the number assigned to that particular clade. I, II and III correspond to the 34(1) derived from cyt 38(1) y 3-1 33(5) 24 (4) 18 (3) 26 (1) 13 (38) 22 (14) 2-5 30(1) 2-7 2-2 25(1) 2- 13 57(1) 35 (1) 9(2) 4(1) 56(1) 12(2) 55(1) 32 (2) 27 (1) 2- 18 6(2) Alburnus alburnus 20 (14) 29(6) 11(1) 37(1) 10 (4) 2- 16 4-1 3-6 2-17 43(1) 41(1) ting design as obtained with ANeCA. Clades are designed as 21(2) 2-14 19 (6) 2-3 3-5 (II) 47(3) 2-10 otype. Numbering and colours of haplotypes are as in Fig. 1 and Table II. Numbers in parentheses in (I) 3(2) 3-3 14(1) 48(1) 2-11 31(1) 53(1) lack dots are missing haplotype. Haplotypes are al 7 (6) 5(1) 42(2) 52(8) 15(2) 2-8 -clade phylogeographical analysis of 51(1) 44(1) 2-9 40(1) 45(2) 54(1) 2-4 8 (1) 2-6 16(1) 46(1) 4-2 49(1) 50(1) 3-2 number of individuals carrying that particular hapl individuals carried that particular haplotype. B of the branch between them. The ﬁgure also shows the nes the haplotype to theclusters fourth in level Fig. but 2. haplotype and one-step levels are not shown) and . 3. Haplotype network and nested ig F

Table III. Results of the nested-clade analysis of the geographical distribution of A. albidus and A. arborella haplotypes. Only the 10 clades with signiﬁcant χ 2 permutational contingency tests are shown. Clade names and nesting designs are as in Fig. 3. The last two columns show the sequence followed in the inference key implemented in ANeCA and the resulting biological inference

Clade χ 2 test Inference key sequence Biological inference 2-8 * 1, 19, no Allopatric fragmentation 2-10 * 1, 2, 11, 12, 13, 14, no Long-distance colonization and past fragmentation (not necessarily mutually exclusive) 3-3 * 1, 2, 3, 4, no Restricted gene flow with isolation by distance 3-4 *** 1, 2, I.O. Inconclusive outcome 3-5 * 1, 2, I.O. Inconclusive outcome 4-1 *** 1, 2, 3, 5, 6, 7, 8, yes Restricted gene flow–dispersal with some long-distance dispersal over intermediate areas not occupied by the species; or past gene flow followed by extinction of intermediate populations 4-2 *** 1, 2, 3, 5, 6, 7, 8, yes Restricted gene flow–dispersal with some long-distance dispersal over intermediate areas not occupied by the species; or past gene flow followed by extinction of intermediate populations Total cladogram *** 1, 2, I.O. Inconclusive outcome

*, P<0·05; ***, P<0·001 of haplotypes from clades 4-1 and 4-2 that mark a statistically signiﬁcant fragmentation event.

POPULATION STRUCTURE

Table IV reports pair-wise FST values among all populations included in the study. The analysis revealed a profound genetic subdivision with c. 68% of the values being significant after Bonferroni correction. Significant FST values are distributed among as well as within major geographic groups (i.e. Adriatic, Ionian, Tyrrhenian and northern Italy). SAMOVA identified maximal genetic difference when the number of groups was equal to two (FST = 0·501; P<0·001). The two groups included (1) all the north Italian populations plus Ofanto and Biferno and (2) the remaining southern Italian samples. Table V shows the results of the AMOVA based on three different groupings of populations (i.e. by geographic origin, by species and following the SAMOVA results). In all cases, high, comparable and significant (P<0·001) global FST values (0·201, 0·198 and 0·201) were obtained with most of the variation apportioned within populations regardless of the grouping scheme adopted (79·88, 80·15 and 79·86%). Variation among groups was always low (7·84, 3·44 and 5·90%).

© 2009 The Authors Journal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 997–1017 PHYLOGEOGRAPHY OF THE GENUS ALBURNUS IN ITALY 1009 500 — · 0 0. Population · 310 012 — — · · 0 0 − 287 075 033 — · · · 0 071 318 104 0 068 0 — · · · · 352 331 0 616 0 323 0 441 0 · · · · · 0 121 0 400 0 162 0 142 001 — 145 0 — · · · · · · 0 − 029 0 229 0 069 0 090 018 0 250 044 0 — · · · · · · · 0 0 095 0 044 350 0 125 0 024 102 165 0 119 0 — · · · · · · · · 0 − 096 0 067 0 342 0 133 0 169 104 0 366 0 142 119 0 · · · · · · · · · 216 0 150 0 538 0 239 0 300 0 294 0 116 — 489 0 265 0 241 0 · · · · · · · · · · 172 0 134 0 446 0 213 0 271 0 003 — 248 0 119 0 430 0 243 0 222 0 · · · · · · · · · · · 0 − 207 0 152 0 552 0 234 0 124 — 294 0 181 285 0 174 0 494 0 258 0 233 0 · · · · · · · · · · · · 0 0 0 0 0 000 0 014 000 0 000 0 400 — 320 0 040 400 0 000 142 545 0 142 095 0 — · · · · · · · · · · · · · 0 0 0 − estimates of differentiation among the populations included in the study as obtained using Arlequin 3 267 1 193 512 1 419 276 1 089 0 337 079 0 089 0 336 1 192 0 475 0 308 288 0 · · · · · · · · · · · · · · 0 0 ST − numbers are as in Table I and Fig. 1. Boldface values indicate signiﬁcance after Bonferroni correction F 140 0 104 0 368 0 062 — 222 0 177 0 053 213 0 045 0 053 0 171 099 0 368 0 189 0 168 0 — · · · · · · · · · · · · · · · 0 0 0 0 171 0 124 0 403 0 104 168 266 201 0 188 0 239 0 099 078 0 206 0 041 397 0 213 0 193 0 · · · · · · · · · · · · · · · · 0 0 0 0 0 0 0 0 0 0 0 IV. Pair-wise able 1— 2 3 40 50 6 70 80 9 14 T Pop.1234567891011121314151617 15 11 16 0 12 17 13 10

Table V. Results from the analyses of molecular variance (AMOVA). Populations have been alternatively pooled by slopes of sampled drainages (Tyrrhenian, Ionian, Adriatic, Northern Italy; see Fig. 1) or by nominal species (A. albidus v. A. arborella)

Sum of Source of squares Variation Fixation Structure variation (d.f.) component P index % Variation By slope Among groups 5·351 (2) Va = 0·036 *** FCT = 0·078 7·84 Among 9·167 (9) Vb = 0·057 *** FSC = 0·133 12·28 populations Within 48·146 (128) Vc = 0·376 *** FST = 0·201 79·98 populations By species Among groups 2·269 (1) Va = 0·016 * FCT = 0·034 3·44 Among 15·327 (12) Vb = 0·079 *** FSC = 0·169 16·41 populations Within 58·337 (151) Vc = 0·386 *** FST = 0·198 80·15 populations

*, P<0·05; P<0·001

Tajima’s D (Tajima, 1989) and Fu’s Fs (Fu, 1997) tests were generally positive and non-significant. Negative and significant Fs values were obtained for two Adriatic populations, Ofanto (Fs =−5·93; P<0·05) and Treste (Fs =−1·18; P<0·05). Negative and significant Fs values reflect an excess of rare polymorphisms and are often observed in populations that have experienced a recent expansion (Fu, 1997).

DISCUSSION

VULTURINO AND ALBORELLA: DISTINCT LINEAGES? Haplotypes from Northern and Southern Italy could not be sorted in two reciprocally monophyletic clusters as expected on the basis of their geographical origin. Enforcing this phylogeographic pattern was rejected statistically. SAMOVA consistently grouped all the northern Italian populations with those sampled in the rivers Ofanto and Biferno. The AMOVA results (Table V) indicated a high degree of genetic structuring, but only a little fraction of the overall genetic variation could be explained in light of taxonomic or geographical criteria. This leads to question whether the vulturino and the A. arborella do represent different lineages. According to Bianco (1980), the two taxa exhibit distinct morphological features and a natively disjunct distribution (Fig. 1). Our multiple phylogenetic searches and NCPA have identiﬁed a strongly supported cluster of haplotypes (clade III in Figs 2 and 3) of exclusive southern Italian origin. Genetic divergence between this clade and the other two (I and II) is relatively shallow (c. 2%), lower than the average level of sequence divergence found among the other Alburnus species included in the study (3·7–7%) and also lower than the values frequently reported in the literature for European cyprinids (Costedoat et al., 2006; Ketmaier et al., 2008a). Our ﬁgures

© 2009 The Authors Journal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 997–1017 PHYLOGEOGRAPHY OF THE GENUS ALBURNUS IN ITALY 1011 are, however, comparable with those obtained by Kotl´ık et al. (2004) among taxonomically distinct lineages of Barbus of recent origin (Middle to Late Pleistocene). In the haplotype network of Fig. 3, clade III (3-4) is respectively seven and nine mutation steps away from the two more closely related clades (3-2 and 3-5). From six to eight missing haplotypes are necessary to connect clade 3-4 to clades 3-2 and 3-5. Furthermore, this clade includes all haplotypes found in the Alento River, the type locality of the A. albidus (Bianco, 1980). Finally, all clades but 3-4 contain a mixture of southern and northern Italian haplotypes, whatever level of nesting. These considerations, along with the fact that our sampling covers the entire range of the A. albidus, lead to tentatively assign clade III to this species. No supported clade of exclusive northern Italian origin could be retrieved; therefore, no firm conclusions on the A. arborella could be derived. A more exhaustive sampling throughout the species range is necessary to address this issue. It should be kept in mind, however, that the A. arborella is one of the most heavily transplanted cpyrinid species in Italy (Bianco, 1995b). It is therefore very likely that these actions might have irreparably blurred the original phylogeographic structure of the species. A number of southern Italian haplotypes were phylogenetically more closely related to northern Italian haplotypes than they were to other southern Italian haplotypes. In addition, four haplotypes (7, 10, 19 and 32) co-occur at northern and southern Italian locations. In our opinion, translocations of fishes from northern to central and southern Italy have to be invoked to explain the observed pattern of haplotype distribution. The alternative scenario would foresee dispersal between northern and southern Italy. These movements would have lasted until very recently, as attested by the simultaneous distribution in the two areas of the same four haplotypes (7, 10, 19 and 32). This is rather an unlikely scenario, however, given the lack of any current connection between the hydrographic systems located north and south of the Apennine chain. It has also to be considered that meristic characters traditionally used to identify cyprinid species may vary clinally and that counts tend to decrease in a north–south fashion in response to the increase in water temperatures (Holcik & Jedlicka, 1994). It is possible that counts of the introduced populations of the northern Italian A. arborella might have, after a few generations, decreased and partially overlapped with those of the A. albidus as an adaptive response to warmer water temperatures and to the shift from a lacustrine to a riverine habitat. It is therefore of paramount importance to cross-validate morphological and genetic data and to avoid grounding systematics revisions on meristic counts only. Strong economic and political pressures have been, and in some circumstances still are, supporting translocations of fishes. They were aimed at replenishing the abundance and diversity of local fish fauna for the amusement of sport fishermen. This common (mal)practice has had a profound effect at zoogeographical and genetic levels (Bianco & Ketmaier, 2001, 2005) and it is important to stress how these actions could potentially bring about introgressive hybridization. This is particularly true for cyprinids, a group where this phenomenon can easily occur even among different genera (Costedoat et al., 2006). Hybridization between the A. albidus and the introduced S. cephalus squalius has been already reported for southern Italian rivers (Bianco, 1982). Given the exclusively maternal inheritance of mtDNA, it is not possible to exclude that some of the individuals carrying haplotypes placed in

© 2009 The Authors Journal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 997–1017 1012 V. KETMAIER ET AL. clade III might be of hybrid origin (i.e. A. albidus × A. arborella). A screening based on nuclear markers would be necessary to fully address the point.

SOUTHERN ITALY: A NEW, AND THREATENED, ICHTHYOGEOGRAPHIC DISTRICT Strictly freshwater fishes, being limited in their dispersal to a contiguity of suitable habitats, tend to show pronounced phylogeographic breaks (Ketmaier et al., 2004; Zaccara et al., 2007 and references therein). No exhaustive phylogeographic studies on southern Italian fishes have been carried out so far. There is, instead, robust evidence of the occurrence of distinct evolutionary lineages in northern and central Italy for the genera Telestes, Scardinius and Rutilus (Ketmaier et al., 2003, 2004, 2008a; Bianco & Ketmaier, 2005; Zaccara et al., 2007). This clear-cut distribution overlaps with the two major Italian ichthyogeographic districts (Padano–Venetian and Tuscano–Latium, Fig. 1) (Bianco, 1995). The present study has identified for the first time a mitochondrial lineage endemic to southern Italy. According to Bianco (1995a), this area would then meet the criteria to be recognized as the third Ital- ian ichthyogeographic district (Fig. 1). These conclusions are also in line with the guidelines recently proposed to delineate freshwater eco-regions at a global scale (Abell et al., 2008). Given (1) the utility of cyprinid ranges in pinpointing zoo- geographic regions, and (2) the co-occurrence in the same area of two additional endemic species (T. muticellus and C. zanandrei) even more geographically local- ized than the A. albidus, the borders of the southern Italian district should coincide with the northern and southern edges of the A. albidus range (Rivers Volturno and Sinni–Alento, respectively; Fig. 1). The time estimate for clade III is relatively recent (c.1·8 M years) and fits well into the paleogeographic reconstructions available for the area. River connections occurred repeatedly during the Pleistocene along the coasts of the Italian peninsula (Colantoni et al., 1984). In particular, at every glacial maximum (from 2·4to0·2M years) the lowering of the level of the Mediterranean Sea determined an extension of the Po basin. The lower section of rivers flowing into the Adriatic Sea coalesced into the Po basin creating the habitat of election for a southward dispersal of A. alburnus along the eastern slope of the Italian peninsula. Our dating for clade III identifies the end of the Biber ice age (when Adriatic drainages became isolated again) as the most likely event responsible for the origin of the vulturino. Similar tempo and mode of divergence had been found in another warm-water-adapted genus, Scardinius (Ketmaier et al., 2003, 2004). The data presented here also illustrate the deleterious effects of translocations on southern Italian zoogeography (Bianco, 1995b). In particular, all those southern Italian haplotypes that do not cluster in clade III should be treated as of dubious geographic origin. Although our mtDNA data do not allow to derive any conclusion on the extent of the eventual hybridization between the A. albidus and the A. arborella it is nonetheless evident that a mixing of haplotype has taken place and that this might have confounded the original biogeographic pattern. The NCPA gave an inconclusive outcome for the A. albidus clade (3-4). The analysis of distribution of average pair-wise distances for haplotypes and clades, however, revealed that for seven rivers (Alento, Fiumarella, Bussento, Sinni, Cara- pelle, Fortore and Treste) the index decreases or goes to zero when the level of

300 250 200

150 Haplotypes 100 50 0

350 300 250 200 150 100 One-step clades 50 0

400 350 300 250 200 150 Two-step clades 100 50 0

800 Average pair-wise distance (km) 700 600 500 400 300 Three-step clades 200 100 0

250

200

150

100 Four-step clades

0 Adda Ticino Sinni* Ofanto Treste* Biferno Alento* Fortore* Palistro* Basento* Badolato* Bussento* iumarella* Carapelle* L Maggiore F

Fig. 4. Average pair-wise location distances (km) for haplotypes and clades for the A. albidus clade (clade 3-4 in Fig. 3). Sampling sites are ordered as in Fig. 1 and Table I; Ripiti River and Meletta Stream are not included in the analysis because no genetic variation was found at these locations. The ﬁrst panel shows the average pair-wise distance between the geographical centres of haplotypes found at each sampling site, the second panel shows the comparable distances for one-step clades found at each site, the third panel shows two-step clade distances, the fourth panel shows three-step clade distances and the ﬁfth panel shows four-step clade distances. Asterisks indicate populations that were assigned to the Southern Italian group by SAMOVA.

© 2009 The Authors Journal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 997–1017 1014 V. KETMAIER ET AL. nesting increases (Fig. 4). All these rivers were grouped together by SAMOVA. This implies that at these locations haplotypes are all evolutionarily closely related, and that isolation by distance is the most likely process responsible for the observed pattern of haplotype occurrence (Templeton, 2001; Ketmaier et al., 2008b). Con- versely, the index for the other southern Italian rivers included in the study (Badolato, Palistro, Basento, Ofanto and Biferno) increases with increasing hierarchical level, as expected when divergent lineages co-occur. Ofanto and Biferno were placed in the same group with all the northern Italian populations by SAMOVA. It is therefore possible to distinguish southern Italian rivers that have been affected by the introduction of allochthonous haplotypes from rivers where this phenomenon is not evident at the mtDNA level. An astonishing total of 19 haplotypes were found in the River Ofanto, most of which are not phylogenetically related to each other and occur at very low frequencies (see Table II and Figs 2 and 3). Only six of these 19 haplotypes are placed in clade III. A negative and significant Fs value (Fs =−5·93; P = 0·018) was obtained for the River Ofento population; this is an indication of the occurrence of rare polymorphisms at low frequency. The hypothesis that such a pattern might be due to a recent demographic expansion is not very likely, though. A more realistic possibility based on artificial introduction of individuals of allochthonous origin carrying different haplotypes should instead be favoured. This hypothesis is further supported by the lack of genetic relatedness among these 19 haplotypes as attested by their scattered placements in the phylogenetic tree and in the network. Similarly, most of the NCPA inferences should be interpreted in light of human- induced rather than natural processes. In particular, all long-distance colonization events identified by the NCPA would imply a recent trans-Apennine dispersal, which is not plausible either biogeographically or ecologically. The most recent events that allowed primary freshwater fishes to cross the Apennines were the local plate tectonics that caused captures between the headwaters of rivers and river catch- ments (Bartolini & Pranzini, 1981; Cattauto et al., 1988). These events took place c. 200 000 years ago, well after our youngest time estimate (800 000 years for clade I). They favoured the dispersal of cold-water-adapted species ecologically limited to the upper and intermediate section of rivers. This scenario is consistent with the phylogeographic patterns observed in the reophilic T. muticellus (Ketmaier et al., 2004; Zaccara et al., 2007). These routes, however, were precluded to warm-water-adapted species such as the A. albidus and the A. arborella, which could only use lowland river confluences as possible dispersal routes.

CONCLUSIONS The data presented in this study have revealed a complex pattern of relationships for the genus Alburnus in Italy. In particular, the mixing of northern and southern Italian haplotypes found in our phylogenetic, population genetics and network analyses can be explained in light of the heavy alterations caused by human activities on the Italian freshwater ﬁsh fauna. A cluster of haplotypes limited to southern Italian rivers, however, was consistently retrieved. Introduction of allochthonous cyprinids in the area is the major threat to the survival of this vulnerable entity. If the original diversity of the single southern Italian endemic cyprinids has to be preserved, actions

We wish to thank S. Rossi (University of Pavia, Italy) for kindly providing us with the Northern Italian samples and we appreciate the technical and logistic support provided by S. Cozzolino and his research group (University of Naples, Italy) for the entire duration of the study. We also wish to thank the Assistant Editor (B. Hanﬂing)¨ and two anonymous reviewers for their useful criticisms.

References Abell, R., Thieme, M. L., Revenga, C., Bryer, M., Kottelat, M., Bogutskaya, N., Coad, B., Mandzak, N., Baldezas, S., Bussing, W., Stassny, M. L. J., Skeltom, P., Allen, G. R., Unmack, P., Nascka, A., Ng, R., Simdozf, N., Robertson, J., Azmijo, E., Higgins, J. V., Heibel, T. J., Wikzamanayake, E., Olson, D., Lopez,´ H. L., Reis, R. E., Lundberg, J. G., Sabajezez,´ M. H. & Petny. P. (2008). Freshwater eco-regions of the world: a new map of biogeographic units for freshwater biodiversity conservation. BioScience 58, 403–414. Bartolini, C. & Pranzini, G. (1981). Plio-Quaternary evolution of the Arno basin drainage. Zeitschrift für Geomorphologie 40, 77–91. Bianco, P. G. (1980). Remarks on the genera Alburnus and Alburnoides and description of a neotype for Alburnus albidus (Costa, 1838). Senior synonimus of Alburnus vulturius (Costa, 1838) (Pisces, Cyprinidae). Cybium 9, 31–44. Bianco, P. G. (1982). Hybriization between Alburnus albidus and Leuciscus cephalus in Italy. Journal of Fish Biology 21, 593–603. Bianco, P. G. (1987). L’inquadramento zoogeografico dei pesci d’acqua dolce d’Italia e prob- lemi determinati dalle falsificazioni faunistiche. Atti Convegno ALLAD 2, 41–65. Bianco, P. G. (1995a) Mediterranean endemic freshwater fishes of Italy. Biological Conser- vation 72, 159–170 Bianco, P. G. (1995b). Factors affecting the distribution of freshwater fishes especially in Italy. Cybium, 19, 241–259. Bianco P. G. & Ketmaier, V. (2001). Anthropogenic changes in the freshwater fish fauna of Italy, with reference to the central region and Barbus graellsii, a newly established alien species of Iberian origin. Journal of Fish Biology 59, 190–208. Bianco P. G. & Ketmaier, V. (2005). Will the Italian endemic gudgeon, Gobio benacensis, survive to the interaction with invasive introduced Gobio gobio? Folia Zoologica 54, 42–49. Briolay, J., Galtier, N., Brito, R. M. & Bouvet, Y. (1998). Molecular phylogeny of Cyprinidae inferred from cytochrome b DNA sequences. Molecular Phylogenetics and Evolution 9, 100–108. Brito, R. M., Briolay, J., Galtier, N., Bouvet, Y. & Coelho, M. M. (1997). Phylogenetic relationships within genus Leuciscus (Pisces, Cyprinidae) in Portuguese fresh waters, based on mitochondrial DNA cytochrome b sequences. Molecular Phylogenetics and Evolu- tion 8, 435–442 Cantatore, P., Roberti, M., Pesole, G., Ludovico, A., Milella, F., Gadaleta, M. N. & Sac- cone, C. (1994). Evolutionary analysis of cytochrome b sequences in some perciformes: evidence for a slower rate of evolution than in mammals. Journal of Molecular Evo- lution 39, 589–597. Cattauto, C., Cencetti, C. & Gregori, L. (1988). Lo studio dei corsi minori dell’Italia Appen- ninica come mezzo d’indagine sulla tettonica del Plio-Pleistocene. Bollettino del Museo di Storia Naturale della Lunigiana 6–7, 7–10. Clement, M., Posada, D. & Crandall, K. A. (2000). TCS: a computer program to estimate gene genealogies. Molecular Ecology 9, 1657–1659. Colantoni, P., Fabbri, A., Rossi, E. & Sarotori, R. (1984). Panoramica sulla geologia dei mari italiani. Acqua-Aria 8, 803–820.

Costedoat, C., Chappaz, R., Barascud, B., Guillard, O. & Gilles, A. (2006). Heterogeneous colonization patter of European cyprinids, as highlighted by the dace complex (Teleostei: Cyprinidae). Molecular Phylogenetics and Evolution 41, 127–148. Drummond, A. J. & Rambaut, A. (2007). BEAST: Bayesian evolutionary analysis by sampling trees. BMC Evolutionary Biology 7, 214. Drummond, A. J., Ho, S. Y. W., Phillips, M. J. & Rambaut, A. (2006). Relaxed phylogenetics and dating with confidence. PloS Biology 4, 699–710. Dupanloup, I., Schneider, S. & Excoffier L. (2002). A simulated annealing approach to define the genetic structure of populations. Molecular Ecology 11, 2571–2581. Durand, J. D., Tsigenopoulos, C. S., Unlu, E. & Berrebi, P. (2002). Phylogeny and biogeography of the family Cyprinidae in the Middle East inferred from cytochrome b DNA–evolutionary significance of this region. Molecular Phylogenetics and Evolu- tion 22, 91–100. Excoffier, L., Laval, G. & Schneider, S. (2005). Arlequin ver 3.0: an Integrated software package for population genetics data analysis. Evolutionary Bioinformatics Online 1, 47–50. Excoffier, L., Smouse, P. E. & Quattro, J. M. (1992). Analysis of molecular variance inferred from metric distances among DNA haplotypes: application to human mitochondrial DNA restriction data. Genetics 131, 479–491. Felsenstein, J. (1985). Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39, 783–791. Fu, Y. X. (1997). Statistical tests of neutrality of mutations against population growth, hitch- hiking and background selection. Genetics 147, 915–925. Holcik, J. & Jedlicka L. (1994). Geographical variation of some taxonomically important characters in fishes: the case of the bitterling Rhodeus sericeus. Environmental Biology of Fishes 41, 147–170. Ketmaier, V., Bianco, P. G., Cobolli, M. & De Matthaeis, E. (2003). Genetic differentiation and biogeography in southern European populations of the genus Scardinius (Pisces, Cyprinidae) based on allozyme data. Zoologica Scripta 32, 13–22. Ketmaier, V., Bianco, P. G., Cobolli, M., Krivokapic, M., Caniglia, R. & De Matthaeis, E. (2004). Molecular phylogeny of two lineages of Leuciscinae cyprinids (Telestes and Scardinius) from the peri-Mediterranean area based on cytochrome b data. Molecular Phylogenetics and Evolution 32, 1061–1071. Ketmaier, V., Bianco, P. G. & Durand, J. D. (2008a). Molecular systematics, phylogeny and biogeography of roaches (Rutilus, Teleostei, Cyprinidae). Molecular Phylogenetics and Evolution 49, 362–367. Ketmaier, V., Pirollo, D., De Matthaeis, E., Tiedemann, R. & Mura, G. (2008b). Large-scale phylogeography in the halophilic fairy shrimp Phallocryptus spinosa (Milne-Edwards, 1840) (Branchiopoda: Anostraca). Aquatic Sciences 70, 65–76. Kotl´ık, P., Bogutskaya, N. G. & Ekmekçi, G. (2004). Circum Black Sea phylogeography of Barbus freshwater fishes: divergence in the Pontic glacial refugium. Molecular Ecology 13, 87–95. Kottelat, M. (1997). European freshwater fishes. Biologia 52, 1–271. Kottelat, M. & Bianco, P. G. (2005). On the valid name of the alborella, Alburnus arborella (Teleostei: Cyprinidae). Ichthyological Explorer Freshwaters 16, 179–182. Panchal, M. & Beaumont, M. A. (2007). The automation of nested clade phylogeographic analysis. Bioinformatics 23, 509–510. Petit, R. J. (2008). The coupe de grace for the nested clade phylogeographic analysis? Molec- ular Ecology 17, 516–518. Posada, D. & Crandall, K. A. (1998). Modeltest: testing the model of DNA substitution. Bioinformatics 14, 817–818. Posada, D., Crandall, K. A. & Templeton, A. R. (2000). GEODIS: a program for the cladis- tic nested analysis of the geographical distribution of genetic haplotypes. Molecular Ecology 9, 487–488. Ronquist, F. & Huelsenbeck, J. P. (2003). MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19, 1572–1574. Shimodaira, H. (2002). An approximately unbiased test of phylogenetic tree selection. Sys- tematic Biology 51, 492–508.

Shimodaira, H. & Hasegawa, M. (1999). Multiple comparisons of log-likelihoods with appli- cations to phylogenetic inference. Molecular Biology and Evolution 16, 1114–1116. Shimodaira, H. & Hasegawa, M. (2001). CONSEL: for assessing the confidence of phylogenetic tree selection. Bioinformatics Application Notes 17, 1246–1247. Swofford, D. (2002) PAUP*: Phylogenetic Analysis Using Parsimony (* and Other Methods), version 4.0β10. Sunderland, MA: Sinauer Associates. Tajima, F. (1989). Statistical method for testing the neutral mutation hypothesis by DNA polymorphism. Genetics 123, 585–595. Templeton, A. R. (1998). Nested clade analysis of phylogeographic data: testing hypotheses about gene flow and population history. Molecular Ecology 7, 381–397. Templeton, A. R. (2001). Using phylogeographic analyses of gene trees to test species status and processes. Molecular Ecology 10, 779–791. Templeton, A. R. (2009). Statistical hypothesis testing in intraspecific phylogeography: nested clade phylogeographical analysis vs. approximate Bayesian computation. Molecular Ecology 18, 319–331. Zaccara, S., Stefani, F. & Del Mastro, G. B. (2007). Phylogeographical structure of the vairone Telestes muticellus (Teleostei, Cyprinidae) within three European peri-Mediterranean districts. Zoologica Scripta 36, 443–453. Zardoya, R. & Doadrio, I. (1999). Molecular evidence on the evolutionary and biogeograph- ical patterns of European Cyprinids. Journal of Molecular Evolution 49, 227–237.