Hereditas 146: 3345 (2009) Phylogeography and mitochondrial DNA divergence in Dolichopoda cave crickets (Orthoptera, Rhahidophoridae) LENE MARTINSEN, FEDERICA VENANZETTI and LUTZ BACHMANN Natural History Museum, Department for Zoology, University of Oslo, Norway Martinsen, L., Venanzetti; F. and Bachmann, L. 2008. Phylogeography and mitochondrial DNA divergence in Dolichopoda cave crickets (Orthoptera, Rhahidophoridae). * Hereditas 146: 3345. Lund, Sweden. eISSN 1601-5223. Received March 25, 2008. Accepted July 7, 2008 Dolichopoda cave crickets are found in caves in the northern Mediterranean region from the Pyrenees in the west to the Caucasus in the east. In this study we analysed the phylogeny within the genus Dolichopoda using parts of the mitochondrial cytochrome oxidase I and 16S ribosomal genes, and explored biogeographic patterns through a dispersalvicariance analysis (DIVA). Phylogenetic analyses grouped the 15 species into the four geographically restricted main lineages corresponding to the Caucasus, Greece, the Pyrenees and Italy, respectively. The species occur largely in allopatry. The Caucasian and Greek species were basal in the phylogeny, as was the clade including the nine Italian species, which grouped into two major lineages, one mainly including species from western Italian coastal regions and islands, and the other including species with a predominantly inland distribution. Thus it seems likely that there have been two main immigrations into Italy followed by multiple consecutive speciation events. The DIVA analysis supported the assumption of an east-west migration route, and indicated that there have been four major dispersal events. Since the statistical support for the basal node connecting D. remyi and D. hussoni with the west Mediterranean species is low, alternative interpretations for the colonization of the Mediterranean, namely parallel colonization of the main areas of the current Dolichopoda distribution, i.e. the Caucasus, Greece, Italy, and the Pyrenees is also possible. Particular emphasis was put on the D. geniculata-laetitiae species complex. D. geniculata included several recently diverged lineages and constitutes a paraphyletic species complex, also embracing the closely related D. laetitiae. Lutz Bachmann, Natural History Museum, Department for Zoology, University of Oslo, Norway. E-mail: lutz.bachmann@ nhm.uio.no Troglophilous, apterous insects are suitable model systems for addressing phylogeographic questions. As they are bad dispersers and dependent on cave habitats, the extent of gene flow between populations is expected low compared to other terrestrial species. The narrow geographic range that is typical for cave populations, along with their high level of endemism, usually cause distinct biogeographic patterns (PORTER 2007). In contrast, surface organisms often have more complex patterns due to recurrent episodes of extinction, recolonization, and secondary contact (SBORDONI et al. 2000). The Orthopteran family Rhaphidophoridae (camel crickets, cave crickets) consists of hydrophilous, apterous, hump-backed, cricket-like insects, usually with slender, elongated appendages (RENTZ 1991). The family has a world-wide distribution and all species are confined to wet forests, rock outcrops and caves (RENTZ 1991). Troglophily, the dependence on caves, is well developed within the family, as more than half of the 300 species are cave dwellers. The cavernicolous Rhaphidophoridae are found in the northern Mediterranean region, the Cape region of South Africa, eastern United States, Central America, Patagonia in South-America, Australia (including Tasmania), New Zealand, South-East Asia and Japan (DI RUSSO and SBORDONI 1998). The genus Dolichopoda Bolivar, I. 1880 (Dolichopodainae, Rhaphidophoridae) (ICZN 2002) is found in caves in the northern Mediterranean region from the Pyrenees in the west to the Caucasus in the east (SBORDONI et al. 1987). The highest species diversity is found in the insular and peninsular areas of Italy and Greece (SBORDONI et al. 1987). Many Dolichopoda populations depend strictly upon natural caves but some populations have also colonized cave-like habitats, such as rock crevices and ravines in mesic or moist woods, and also man-made hypogean environments, e.g. cellars, Etruscan tombs, Roman aquaducts and catacombs (SBORDONI et al. 1987, 1991). The natural and man-made caves may differ in their ecological conditions and populations inhabiting these two habitats can differ in their phenology and life history traits (BERNARDINI and DI RUSSO 2004). Although some individuals are occasionally observed outside their caves in moist or mesic woods, especially in the northern parts of Italy (CARCHINI et al. 1983; BACCETTI 1987), there is a high degree of geographical isolation between the different Dolichopoda populations. Strictly allopatric speciation processes have been assumed for Dolichopoda populations (SBORDONI et al. 1987), as is also true for many other DOI: 10.1111/j.1601-5223.2008.02068.x 34 L. Martinsen et al. cave organisms (SBORDONI et al. 2000; GIBERT and DEHARVENG 2002), and usually species show vicariant allopatric ranges (SBORDONI et al. 1991). An exception is the occurrence of D. schiavazzii in a man-made cave together with D. baccettii at the Argentario promontory. However, this is thought to be due to passive dispersal by Passionist Fathers established in this area since the 18th century (ALLEGRUCCI et al. 1982, 1997). Studies on several Dolichopoda species such as e.g. D. geniculata, D. schavaizzii and D. linderi, indicate that gene flow between populations is dependent on the habitat structure surrounding the caves (ALLEGRUCCI et al. 1997; CESARONI et al. 1997). These studies show that groups of populations from caves surrounded by xeric vegetation in coastal Mediterranean habitats are genetically more structured than groups of populations living in montane mesophilous forest. The latter show, at the same scale, higher values of Nm, i.e. higher levels of gene flow. It has been reported that populations as distant as 50 km apart might have recently exchanged migrant individuals (SBORDONI et al. 2000). Dispersal of Dolichopoda by man has occurred on several occasions. Unintentional transplantation of eggs, nymphs or even adults, is common in Dolichopoda (BERNARDINI et al. 1996). One example is the occurrence of D. schiavazzii in the Argentario Promontory. Another example is D. laetitiae in the Poscola cave, which is quite isolated from the geographic range of the Dolichopoda genus (BERNARDINI et al. 1996). The phylogenetic relationship among several western Mediterranean Dolichopoda species have been studied by applying a variety of markers such as epiphallus morphology, egg chorion structure, allozyme variability, single copy DNA-DNA hybridization, and RFLPs of mtDNA (ALLEGRUCCI et al. 1992; VENANZETTI et al. 1993). ALLEGRUCCI et al. (2005) recently published a mitochondrial DNA phylogeny of the Mediterranean species of Dolichopoda genus based on sequences of the cytochrome oxidase I (COI) and the ribosomal 16S rRNA (16S) genes. The authors suggested that the genetic diversity reflects the geographical distribution pattern, with the geographically close species being the most related ones, and that the present-day geographical distribution appears to have been influenced by glacial and interglacial cycles during the Pleistocene. The present study is a phylogenetic evaluation of Dolichopoda, using COI and 16S, significantly expanding the data set from previous studies (Allegrucci et al. 2005). Our analysis included samples from a broader geographic range, including the westernmost species D. bolivari, the Sardinia species D. muceddai Hereditas 146 (2009) and the Greek species D. hussoni. In addition, a possible historical biogeographic scenario was reconstructed by a dispersal-vicariance analysis using the DIVA software (RONQUIST 1996). MATERIAL AND METHODS Material Our study included 45 specimens of Dolichopoda spp. representing 35 populations and 12 species. Additional sequences from 43 specimens and 26 populations available in GenBank from ALLEGRUCCI et al. (2005), were also included in the analyses. In total 16 species and 45 populations were included. Troglophilus cavicola, Ceutohilus gracipiles, Hadenoecus cumberlandicus and Euhadenoecus insolitus were used in the phylogenetic analyses as outgroup taxa. All the species and populations included in the analysis are listed in Table 1. DNA extractions, PCR and sequencing DNA extractions were done with the Pure Gene Kit (Gentra Systems). The mitochondrial 16S rRNA and COI genes were PCR amplified using the primers LRJ-12887 (5?-CCGGTCTGAACTCAGATCACGT-3?) and LR-N-13398 (5?-CGCCTGTTTATCAAAAACAT-3?) for the 16S region (SIMON et al. 1994) and C1-J-1751 (5?-GGATCACCTGATATAGCATTYCC3?) (SIMON et al. 1994) and C1-N-2700 (5-AATATA ACAATAAATTGTATTTT-3?) (KNOWLES and OTTE 2000) for the COI region. The internal primers (5?GCTATTATAGCATAAATTATTC-3?) and (5-GAG ATCCAATTTTATATCAACA-3?) were used as sequencing primers for COI. The PCR conditions for the 16S region were as follows: 948C for 5 min, followed by 30 cycles of denaturation at 948C for 45 s, annealing of primers at 458C for 30 s, elongation of primers at 728C for 30 s, and one final extension step at 728C for 5 min. The PCR conditions for the COI region were as follows: 948C for 5 min, followed by 30 cycles of denaturation at 948C for 45 s, annealing of primers at 458C for 30 s, elongation of primers at 728C for 1 min, and one final extension step at 728C for 5 min. PCR products were purified using the QIAquick PCR Purification Kit (Qiagen). Sequencing was done on an ABI automated sequencer (3100), using BigDye for cycle sequencing (Applied Biosystems). GenBank accession numbers are given in Table 1. Statistical and phylogenetic analyses The sequences were aligned using Clustal X (THOMPSON et al. 1997) followed by manual adjustments. Nucleotide composition, number of variable Taxon Ingroup taxa Dolichopoda D. schiavazzii Pop. Locality Necropoli di Vetulonia, Grosseto, Toscana, Italy Grotta dei Pipistrelli, Montorsaio, Grosseto, Toscana, Italy CPS VET1,VET2 ORS1,ORS2 ORS A CPS CIS BDO CIS BDO1, BDO2 POP MRC POP MRC MRC A BSC FIC FIC1, FIC2 A CAM VET ORS D. schiavazzii caprai BSC FIC D. aegilion CAM D. linderi MTB BNP VMY SIR FRN BRA CAM A MTB BNP VMY SIR1,2 A FRN1,2 BRA1,BRA2 BRA1, 2 A SIS SIS1-3 A D. muceddai D. baccettii SAR PST SAR1,2 A PST1, 2 D. laetitiae PSC PST1, 2 A PSC1, 2 GDP FOR PSC1, 2 A GDP1, 2 A FOR D. bolivari D. bormansi Monastero dei Fratelli Passionisti, Orbetello, Grosseto, Toscana, Italy Acquedotto di Cisternino, Livorno, Toscana, Italy Grotta di Buca dell’oro, Isola d’Elba, Grosseto, Toscana, Italy Populonia, Grosseto, Toscana, Italy Marciana, Isola d’Elba, Grosseto, Toscana, Italy GenBank accession no. COI AY507638, AY507639 AY507635, AY507636 AY793633 AY507632 16S AY507601, AY507602 AY507585, AY507584 AY793573 AY507575 AY507631 AY507629, AY507630 AY507573 AY507567, AY507568 AY507637 AY507634 AY793635 Buca sopra cimitero, Orbetello, Grosseto, Toscana, Italy AY507643 Caverna di Fichino, Cascianna Terme, Pistoia, Toscana, Italy AY507633 AY793634, AY793636 Miniera di Campese, Isola del Giglio, Grosseto, Toscana, AY507652 Italy AY793600 Grotte de Montbolo, Montbolo, Eastern Pyrenees, France AY507627 Grotte de Bon Repaux, Bon Repaux, Ariege,Pyrenees, France AY507626 Grotte de Valmanya, Vinca, Eastern Pyrenees, France AY507628 Sirach Cave, Eastern Pyrenees, France AY793598, AY793599 Forat negre cueva, Serradel, Llerida, Pyrenees, Spain AY507648, AY507649 Grotta di Brando, Bastia, Corsica, France AY507646, AY507647 AY793631, AY793632, AY793627 Grotta di Sisco, Corsica, France AY793625, AY793626, AY793628 MonteLimbara, Sardinia, Italy AY793629, AY793630 Grotta di Punta degli Stretti, Orbetello, Grosseto, Toscana, AY507650, AY507651 Italy AY793639, AY793640 Grotta della Poscola, Monte di Malo, Priabona, Vicenza, AY507641, AY507642 Veneto, Italy AY793611, AY793613 Grotta della Piana, Umbria, Italy AY793612, AY793610 Ruderi di Villa Chigi, Formello, Roma, Lazio, Italy AY507589 AY507582 AY793572 AY507571 AY507577 AY793574 AY507572 AY793570 AY507583 AY507565 AY507603 AY793567 AY507579, AY507580 AY507605, AY507606 AY793578 AY793579 AY793575, AY793575 AY507594, AY507593 AY793571 AY507591, AY507592 AY793581 AY793582 Phylogeography and mitochondrial DNA divergence in Dolichopoda Individual Hereditas 146 (2009) Table 1. Species, populations, localities and accession numbers in GenBank for the Dolichopoda sequences included in this study. (The abbreviations for individuals that are followed by an A is from ALLEGRUCCI et al. 2005). 35 36 Table 1 (Continued) D. laetitiae etrusca Pop. DIA D. palpata TRE D. capreensis CPR D. geniculata PIL PAS CLP TUS ISC D. ligustica D. ligustica septentrionalis DIA DIA1, 2 A TRE1, 2 TRE1, 2 A CPR CPR1, 2 A PIL PAS CLP AUS FON PNZ TUS ISC ISC A PRA VAL VAL1, 2 A AUS FON A PNZ 1 CON PNZ1 A PNZ2 A CON SFL BOS PUG CON1, 2 A SFL BOS1, 2 PUG1, 2 PRA VAL D. geniculata pontiana Individual Locality Grotta del diavolo, Semproniano, Grosseto, Toscana, Italy AY507640 AY793614, Grotta di Tremusa, Scilla, Reggio di Calabria, Calabria, Italy AY507610, AY793608, Grotta San Michele, Isola di Capri, Napoli, Campania, Italy AY507644 AY793606, Grotta la Pila, Poggio Moiano, Rieti, Lazio, Italy AY507620 Grotta di Pastena, Pastena, Frosinone, Lazio, Italy AY507621 Grotta Regina Margherita, Collepardo, Frosinone, Lazio, AY507609 Italy Cunicolo dell’acquedotto, Frascati, Roma, Lazio, Italy AY507624 Fontana cunicoli, Isola di Ischia, Napoli, Campania, Italy AY507622 AY793595 Grotta delle Praie, Lettomanoppello, Perugia, Umbria, Italy AY507618 Grotta Valmarino, Monte S. Biagio, Latina, Lazio, Italy AY507625 AY793616, Grotta degli ausi, Prossedi, Latina, Lazio, Italy AY507611 Grotta di Fontanella, Vico Equense, Campania, Italy AY793594 Le Forme, Isola di Ponza, Latina, Lazio, Italy AY507619 Buco del Corno, Valle Cavallina, Zandobbio, Bergamo lombardia, Italy D. hussoni D. remyi VLT1, 2 A SAB 2 A NAO EDE A POZ A VOR A GOL A AY793607 AY793617 AY793596 AY793597 AY507614 Grotta di Valletto, Corsica, France Grotta di Sabara, Corsica, France Naoussa cave, Naoussa, Macedonia, Greece Edessa Cave, Macedonia, Northern-East Greece Pozarska Cave, Macedonia, Northern-East Greece Vorontzovskaya Cave, Caucasus, Russia; Golova Otapa Cave, Caucasus, Russia AY793601, AY793602, AY793603 AY793620, AY793621 AY793619 AY507615 AY793637 AY793638 AY793622 AY793623 AY507578 AY793580 AY507564 AY507598 AY793588 AY507574 AY793587 AY507587 AY507586 AY507563 AY507599 AY507581 AY793585 AY507590 AY507600 AY793583 AY507566 AY793584 AY507588 AY793586 AY507576 AY793568 AY507597 AY507569, AY507570 AY507595, AY507596 AY793569 AY793577 AY793576 AY507607 AY793589 AY793590 AY793566 AY793565 Hereditas 146 (2009) D. euxina VLT SAB NAO EDE POZ VOR GOL AY793615 AY507645 AY793609 AY793604, AY793605 Grotta Selva, Zandobbio, Bergamo, Lombardia, Italy AY507623 Grotta di Bossea, Frabosa Soprana, Cuneo, Piemonte, Italy AY507612, AY507613 Grotta del Pugnetto, Val di Lanzo, Torino, Piemonte, Italy AY507616, AY507617 PUG1, 2, 3 A D. cyrnensis GenBank accession no. L. Martinsen et al. Taxon AY793563 AY793591 IND A IND Indian Grave Point Cave, De Kalb Co., TN, USA AY793592 BAT A BAT Bat Cave, Carter Cave State Park, Carter Co., KY, USA AY793561 AY793593 Hamden, CT, USA CEU A AY793564 AY793624 TRO A CEU Ceuthophilus C. gracilipes Hadenoecus H. cumberlandicus Euhadenoecus E. insolitus AY507604 Outgroup taxa Troglophilus T. cavicola TRO TRO Grotta della Poscola, Monte di Malo, Priabona, Vicenza, Veneto, Italy Covoli di Veroli Cave, Veneto, Italy AY507653 GenBank accession no. Locality Individual Pop. Taxon Table 1 (Continued) AY793562 Phylogeography and mitochondrial DNA divergence in Dolichopoda Hereditas 146 (2009) 37 and parsimony informative sites, and the transition/ transversion ratio were calculated using MEGA version 3.1 (KUMAR et al. 2004). To assess whether nucleotide substitutions reach saturation in the comparison of ingroup species, transitions and transversions were plotted against uncorrected genetic distances (p-distances) for all pairwise comparisons in both fragments. DnaSP (ROZAS et al. 2003) was used to estimate nucleotide diversity (Pi), average number of nucleotide differences (k), and haplotype diversity (Hd). Phylogenetic analyses were conducted using Bayesian analysis, maximum likelihood (ML) and maximum parsimony (MP). The analyses were conducted on each data set separately, as well as a concatenated data set. A partition homogeneity test was performed in PAUP (SWOFFORD 1998) to test the congruence between the 16S and COI genes. The appropriate substitution model for the two alignments were determined using the program MrModeltest (NYLANDER 2004). The general time reversible substitution model (GTRGI) including invariable sites (I) and rate variation across sites (G) obtained highest likelihood score for both COI and 16S rRNA data sets, and was implemented in the Bayesian and maximum likelihood analyses. The Bayesian phylogenetic analysis was conducted using MrBayes (HUELSENBECK and RONQUIST 2001). Each analysis was run with two million generations, four chains (one cold, three heated) and a sample frequency of 100. A 50% majority rule consensus tree was made from each analysis with the first 4000 trees ignored as burn-in. The maximum likelihood analysis was done using PAUP (SWOFFORD 1998) through the Bioportal Website at the Univ. of Oslo, Norway (/<http://www. bioportal.uio.no//>). Maximum parsimony analysis was done using the program TNT-Tree analysis using New Technology (GOLOBOFF et al. 2008) made available online with the sponsorship of Willi Hennig Society (/<http://www. cladistics.org//>). Uncorrected p-distances between populations were calculated in MEGA (KUMAR et al. 2004) for species represented by more than three populations and an analysis of variance (ANOVA) with subsequent Posthoc tests Tukey honest significant difference test for unequal sample size (SPJOTVOLL and STOLINE 1973)was conducted in Statistica (STATSOFT 2005) to compare the variation within these species. The program DIVA (dispersal vicariance analysis) (RONQUIST 1996) was used to reconstruct hypothetical past biogeographic patterns of the Dolichopoda taxa during the colonization of the northern Mediterranean 38 L. Martinsen et al. region. This program gives geographic locations for common ancestors employing an approach similar to character optimization. The distributional data for the Dolichopoda genus were taken from relevant literature (BACCETTI and CAPRA 1959; SBORDONI et al. 1976, 1985, 1987, 1991, 2004; CARCHINI et al. 1983). The areas were defined as congruent areas shared by two or more species (sympatric distribution) according to the definition used by SANMARTIN (2003). The exceptions are the areas two and six (below) that are inhabited by only one Dolichopoda species each, but because of the distance to the other areas it would be artificial to include them in any of the other categories. The islands off the coast of Italy were assigned to the same category as the mainland. This is because each island inhabits only one species, and because the islands can be considered as a unit together with the continental areas due to territorial continuity as a result of drop in sea level in upper Miocene and marine regression in glacial Pleistocene periods (LA-GRECA 1990). Corsica and Sardinia are grouped together because of their closeness and their geological history occurring on the same microplate. Based on the extant distribution pattern of Dolichopoda we defined six areas: (1) the Pyrenees, (2) continental Italy north (north of Emilia Romana, in connection with the Italian Alps), (3) Corsica and Sardinia, (4) continental Italy central and south (south of Emilia Romana), (5) Greece, (6) the Caucasus. Since the DIVA program only accepts fully resolved trees, we chose the maximum likelihood tree in our DIVA analyses. The ML tree included only one polytomy, a trichotomy within the D. geniculatalaetitiae complex, compared to three polytomies in the maximum parsimony and Bayesian trees. All taxa within the trichotomy have the same distribution according to our area classification, so we randomly chose one of the three possible solutions of the trichotomy for the DIVA input file. DIVA does not perform well on the most basal nodes, because the input tree is only a small part of the tree of life, and the optimal state for the basal node depends heavily on the rest of the tree of life. This usually results in the root node distribution being large and including most or all of the areas occupied by the terminals (RONQUIST 1996). This can be solved by incorporating outgroups and/or by imposing constraints on the number of unit areas allowed in ancestral distributions. We included one of the outgroups, T. cavicola, in the DIVA analysis. The other outgroups used in the phylogenetic analyses are Nearctic species, so they cannot be used to restrict the ingroup distribution. SANMARTIN (2003) argues that one can constrain the maximum number of areas Hereditas 146 (2009) in ancestral distributions to the number of areas in the most widespread extant descendant. Since no extant Dolichopoda species are widespread among all of the defined regions, we consider it likely that the common ancestors of the genus also had a restricted distribution. We therefore tried two different values for the maxareas option, two and four, to impose a constraint on the number of unit areas allowed in ancestral distributions. We also conducted a spatial analysis of molecular variance using the software SAMOVA (DUPANLOUP et al. 2002). This program aims at clustering geographically homogeneous populations into user-defined numbers of groups (K) in order to maximize the total genetic variance observed between groups (FCT index). At the same time, the program also identifies genetic barriers between grouped populations. SAMOVA analyses were computed for K-values ranging from 4 to 15. For each K-value, 10 000 simulated annealing steps were performed starting from each of 250 sets of initial conditions. RESULTS Sequence alignments The mitochondrial DNA matrix included 94 sequences representing 45 populations and 16 species of Dolichopoda cave crickets, of which 43 sequences were taken from ALLEGRUCCI et al. (2005). A total of 1439 nucleotide characters, 496 from the 16S gene and 943 from the COI gene, were included in the alignment. There were 362 variable sites (25.2%), of which 296 (20.6%) were parsimony informative. The COI region included 282 variable sites (225 parsimony informative) and the 16S gene 80 variable sites (71 parsimony informative). The transition/transversion (ti/tv) ratios for 16S ranged from 0.27, which is in agreement with the range calculated by ALLEGRUCCI et al. (2005). The ti/tv ratios for COI ranged from 0.5 11.5, which is a broader range than reported by ALLEGRUCCI et al. (2005). Phylogenetic analyses The congruence of the 16S and COI datasets were evaluated using a partition homogeneity test, which yielded no significant difference between the data partitions (p 0.52). Thus, the 16S and COI data sets were combined and analyzed simultaneously. The phylogeny (Fig. 1) is derived from the Bayesian analysis, showing bootstrap support values from the MP and ML analyses as well. The three different phylogenetic inferences used (MP, MrBayes and ML) gave very similar tree topologies and only a few incongruences occurred. The parsimony analysis Hereditas 146 (2009) Phylogeography and mitochondrial DNA divergence in Dolichopoda 39 Fig. 1. Bayesian inference of the phylogeny of the genus Dolichopoda. The terminal taxa are populations following the abbreviations listed in Table 1. Numbers in parenthesis indicate the codes for the predefined geographic areas used in the DIVA analyses. Posterior probabilities and bootstrap support 50 in the MP analysis are given for each node. 40 L. Martinsen et al. Hereditas 146 (2009) yielded 4083 trees of equal length (1441 steps) and the 50% majority-rule consensus tree was congruent with the MrBayes phylogeny (Fig. 1). Almost all Dolichopoda species were found monophyletic with high statistical support. Only D. geniculata was found paraphyletic in relation to D. laetitiae. The Caucasian D. euxina is the sister group to a clade consisting of all other species included in this study. D. remyi and D. hussoni from Greece are found in a trichotomy with the species from Italy/Spain/France. Within this clade, a monophyletic group of the Pyrenean species (D. linderi and D. bolivari) forms a sister group to the Italian species. In the clade including the Italian Dolichopoda species, two well supported groups can be found, one including species from central and west (D. laetitiae and D. geniculata) and north Italy (D. ligustica), and the other species from central and northern Italy including Sardinia and Corsica (D. schiavazzi, D. aegilion, D. baccetti, D. bormansi, D. muceddai and D. cyrnensis) and south Italy (D. capreensis and D. palpata). The geographic distribution of the species and populations is depicted in Fig. 2. Intrapopulational sequence variation In most instances sequences from the same species cluster together in the tree, though not always with the sequences from the same population. This variation SFL PUG can be either due to lineage sorting, stochastic extinction of ancestral haplotype lineages within a population, or spatial proximity of isolated populations and gene flow between them. As the number of sequences per population is low, the intrapopulational sequence variation is not discussed in this paper. Further analysis is required to resolve these issues. Intraspecific genetic diversity The number of polymorphic (segregating) sites (S), nucleotide diversity (p) and average number of nucleotide differences (k) for species with N]7 is presented in Table 2 (N number of individuals). The genetic variation within D. geniculata was high when compared to other Dolichopoda species. In order to test whether this intraspecific variation in D. geniculata was significantly higher than in other Dolichopoda species, i.e. D. geniculata, D. laetitiae, D. ligustica, D. linderi and D. schiavazzii, we performed an ANOVA using the calculated p-distances obtained from comparisons between individuals within species. The ANOVA revealed a highly significant difference in intraspecific genetic variation between species (F34.77, p B0.001). Post-hoc tests showed that all comparisons with D. geniculata were significant (pB0.01), and that the genetic variation within D. geniculata was significantly higher than in the other species. PSC CON BOS CIS FIC GDP ORS POP SIS VET DIA FOR BRA CAM PIL PRA SAB VLT CPS PST TUS CLP BSC AUS PAS SAR VAL PNZ FON ISC CPR MRC BNP VMY FRN SIR MTB BDO VOR, GOL (Caucasus) POZ NAO TRE Fig. 2. The geographical distribution of the Dolichopoda species and populations included in this study. EDE Phylogeography and mitochondrial DNA divergence in Dolichopoda Hereditas 146 (2009) 41 Table 2. Intraspecific genetic variation of the COI and 16S target regions in various Dolichopoda species. Only species represented by n]7 individuals have been included. S number of polymorphic (segregating) sites, p nucleotide diversity, k average number of nucleotide differences. Species D. D. D. D. D. laetitiae geniculata schiavazzii ligustica bormansi No. of individuals 10 15 16 11 7 COI 16S COI16S S p k S p k S p k 16 71 31 22 10 0.0066 0.0225 0.0067 0.0085 0.005 6.2 24.193 6.4 7.636 4.667 3 22 2 1 2 0.0041 0.0139 0.001 0.0005 0.0021 1.429 6.333 0.476 0.250 1.000 19 90 32 23 12 0.006 0.0198 0.0047 0.0057 0.0037 7.667 26.819 6.617 7.718 5.238 DIVA analyses When the maxareas option was not used, and the analysis was done without the outgroup taxa, the analysis gave three solutions for the ancestral area reconstruction at the root node: i) all the areas except northern Italy (area 2) and Corsica (area 3); ii) all the areas except Corsica (area 3); and iii) all the areas in the analyses. When we included the outgroup, but still not restricting the maxareas, the analysis gave five different options for the root node: i) 16, ii) 146, iii) 56, iv) 156, and v) 1456. When restricting the maximal number of areas (maxareas) to two and four, and keeping the outgroup, the analysis gave only one solution for the root node, Greece and Caucasus (56). The two analyses gave the same results for all the nodes, except the node between D. hussoni and the Italian and Pyrenean species. Maxareas 2 gives two MP solutions: i) 15 or ii) 45; while maxareas 4 gives three MP solutions: i) 15, ii) 45, or iii) 145. The optimal reconstruction from both these analyses required four dispersal events: 1) to the Pyrenees (area 1), 2) to continental Italy central and south (area 4), 3) to continental Italy northern part (area 2), and 4) to Corsica/Sardinia (area 3). SAMOVA As expected, the FCT values increased with increasing number of user-defined groups (K), while FSC decreased (DUPANLOUP et al. 2002). The FCT estimate did not reach a clear plateau value when plotted against the K values, though the slope of the curve decreased slightly when reaching K 9 (data not shown). Not surprisingly, the Greek species D. rymei and D. hussoni group first against all other species included in this study. Already for the lowest K value (K 4), all D. schiavazzii populations are grouped. With increasing K, the new groupings followed the phylogenetic and geographic patterns. With K 7, the species assigned to the predefined biogeographic areas in the DIVA analysis group together. However, biogeographic areas 3 (Corsica and Sardinia) and 4 (continental Italy, central and south) do not group separately. Instead, there was a group including D. geniculata and D. laetitiae, a second one including D. caprensis and D. palpata, and a group containing the remaining species from biogeographic areas 3 and 4. With further increasing K more species group separately until at K15 basically all species are found in separate groups with only D. linderi and D. bolivari kept together. DISCUSSION This study focuses on the phylogeography and mtDNA variation in the Mediterranean cave cricket genus Dolichopoda. By means of a dispersal vicariance analysis we further assessed the importance of vicariance and dispersal in shaping the current distribution of the genus Dolichopoda in the northern Mediterranaen area. Phylogenetic considerations The recovered phylogenetic relationships of Dolichopoda species (Fig. 1) are to a large extent in good agreement with previously published data (ALLEGRUCCI et al. 2005). However, geographically close species did not always turn out as closest relatives. We recovered a southern Italian clade consisting of D. palpata and D. capreensis that is the sister group to the coastal western clade consisting of D. schiavazzii, D. baccettii and D. aegilion, and a clade including D. bormansi, D. muceddai and D. cyrnensis. This is in contrast to previous analyses that accepted D. palpata and D. capreensis as sister group to the central-Italian clade consisting of D. geniculataD. laetitiaeD. ligustica (ALLEGRUCCI et al. 2005). Previous morphological studies of Dolichopoda have led to the delineation of four subgenera: Capraiacris Baccetti, 1977; Chopardina Uvarov, 1921; Dolichopoda Bolivar, 1880; and Petrochilosina Boudou-Saltet, 1980. 42 L. Martinsen et al. This subgeneric division of Dolichopoda is not reflected in the mtDNA phylogeny. However, evolution in subterranean habitats is believed to be influenced by convergence and molecular phylogeny may not necessary mirror morphological differentiation (LEFEBURE et al. 2006). Bioclimatic factors may be the major determinants of the morphometric patterns observed in Dolichopoda (ALLEGRUCCI et al. 1987), and consistent with this one may expect a high degree of convergence in Dolichopoda. There is little resolution of the basal nodes in the phylogeny of Dolichopoda. The relationships between the Caucasian clade, the two Greek species, and the major clades of the Italian and Pyrenean Dolichopoda species, are not properly resolved. The split of the Caucasian species from the rest of Dolichopoda is only weakly supported. The topology within the two Italian groups mainly supported earlier findings (ALLEGRUCCI et al. 2005), but provides additional information regarding intraspecific relationships of D. schiavazzii, the Corsica Sardinian species, and the D. geniculatalaetitiae complex as well. Little genetic structure was found within D. schiavazzii which might have been caused by relatively recent dispersal or gene flow between the cave populations. Accordingly, all D. schiavazzii populations are grouped in the SAMOVA already with K4, and kept as a separate group with increasing K. Previous allozyme analyses have revealed substantial genetic differentiation between populations of D. schiavazzii (ALLEGRUCCI et al. 1997). It was previously suggested that D. baccettii and D. aegilion were sister taxa, but statistical support values were low (ALLEGRUCCI et al. 2005). In the phylogenetic analysis presented here, D. aegilion is a sister group to D. schiavazzii, though also with low statistical support. However, there is strong evidence from the phylogenetic analyses that D. baccettii, D. aegilion and D. schiavazzii, form a monophyletic group. However, this is not reflected in the SAMOVA, which grouped D. aegilion and D. baccetti together with D. bormansi, D. cyrnensis and D. muceddai (K 46) after D. schivazzii grouped separately. The Corsican species D. bormansi is more closely related to the Sardinian species D. muceddai than to the other Corsican species D. cyrnensis. This was also reported earlier by SBORDONI et al. (2004), but was not addressed in other analyses (ALLEGRUCCI et al. 2005). This may indicate that either two Dolichopoda lineages evolved in allopatry on Corsica and Sardinia, and the Sardinian lineage later recolonized Corsica, or that two species developed on Corsica with one subsequently spreading to Sardinia. Hereditas 146 (2009) We noticed significantly higher sequence diversity within the D. geniculatalaetitiae species complex than in the other species. The D. geniculatalaetitiae complex is characterized by high degree of dispersal and subsequent isolation of populations. D. schiavazzii is also isolated in several populations, but the sequence divergence is much lower among these populations. However, genetic drift may have been a stronger force in D. geniculataD. laetitiae than in other Dolichopoda species. The D. geniculatalaetitiae lineage is found in parts of Italy with a very heterogenous landscape that is fragmented by mountain ranges and bordered by the sea. As recently suggested, such landscapes can foster organismal diversity even though they were more climatically stable than regions frequented by glaciers and tundra-like conditions (WEISS and FERRAND 2007). Earlier studies of the D. geniculata D. laetitiae complex have shown that peripheral populations or groups of populations that are found close to the Tyrrhenian and the Adriatic coast, show higher degrees of genetic differentiation among each other and from the central inland populations than is observed among central inland populations (SBORDONI et al. 1987). However, this is not reflected in our mtDNA phylogeny. Phylogeography The present distribution of Dolichopoda is congruent with what is known about refugia for temperate European species during climatic fluctuations of the Pleistocene. The three main refugia that have been deduced from phylogenetic studies for most temperate species in Europe are Iberia, Italy and the Balkans (HEWITT 2000). This is exactly where most Dolichopoda species are distributed today. In addition there are four Dolichopoda species in Turkey and two in the Caucasus (DI RUSSO et al. 2007). The DIVA analyses indicated that there has been an east-west migration of Dolichopoda (or its ancestor) into the Mediterranean area, since the most eastern lineage from Caucasus branches off basally in the tree and is most closely related to the outgroup species. However, the support value for this is low in the tree topology. On the other hand this result is in agreement with previous observations on the migration of Dolichopoda from the East reviewed by HUBBEL and NORTON (1978). The authors claimed that no rhaphidophorid species survived the Pleistocene glaciations north of the Alpine-Carpathian barrier, and that the post-Miocene immigration of Dolichopodini into the northern Aegeid and Tyrrhenid areas came from the region north of the Caspian Sea (HUBBELL and NORTON 1978). The center of origin for Dolichopoda seems to be in the eastern part of its range, probably in Phylogeography and mitochondrial DNA divergence in Dolichopoda Hereditas 146 (2009) the Caucasian area or in Greece. Although not included in this study, the number of Dolichopoda species in Greece is very high indicating a center of origin in this area with spreading westward and eastward to Turkey and the Caucasus. Since the statistical support for the basal node connecting D. remyi and D. hussoni with the west Mediterranean species is low, alternative scenarios for the colonization of the Mediterranean area also need to be considered. Parallel colonization of the main areas of the current Dolichopoda distribution, i.e. the 43 Caucasus, Greece, Italy, and the Pyrenees may also be possible. The phylogenetic analyses suggests two possible scenarios for the invasion of Italy. The most parsimonious is that Italy was colonized once, since all the Italian species are found in one monophyletic clade. Within Italy an early vicariance would then have split the ancestor into two lineages. These two lineages would have led to the two present Italian groupings in the phylogenetic tree; one group mainly including species from western Italian coastal regions and islands, and the second group including species with T. cavicola Greece (5), Continental Italy North (2), Continental Italy central and south (4) D. euxina Caucasus (6) D. remyi Greece (5) D. hussoni Greece (5) D. baccettii Mainland Italy Central (4) D. aegilion Mainland Italy Central (4) 4 4 3+4 D. schiavazzii Mainland Italy Central (4) D. bormansi Corsica (3) 4 3 D. muceddai Sardinia (3) D. cyrnensis Corsica (3) D. palpata Mainland Italy South (4) 4 5+6 D. capreensis Mainland Italy D. ligustica Mainland Italy North (2) 5 D. geniculata pontiana (PNZ) Mainland Italy Centra, Ponza Islandl (4) D. geniculata (VAL, AUS, TUS, PAS) Mainland Italy Central (4) 4 2:1+5/4+5 4:1+5/4+5/1+4+5 1+4 2+4 4 4 4 4 4 4 D. geniculata (CLP, PIL) Mainland Italy Central (4) D. geniculata (PRA) Mainland Italy Central (4) D. laetitiae laetitiae Mainland Italy Central (4) D. laetitiae etrus c a Mainland Italy Central (4) D. geniculata (FON) Mainland Italy Central (4) D. geniculata (ISC) Mainland Italy Central, Ischia Island (4) D. linderi Pyrenees/France (1) D. bolivari Pyrenees/Spain (1) Fig. 3. The results from the dispersal-vicariance analysis of the genus using the program DIVA. The pre-assigned distribution areas used in the analysis were: 1 The Pyrenees, 2 continental Italy north, 3 CorsicaSardinia, 4 continental Italy central and south, 5 Greece, 6 Caucasus. The optimal reconstructions at each node are shown for two different values, i.e. for the maxareas options 2 and 4. The dispersal events are depicted on the figure as blue spots. The analysis is based on the maximum likelihood tree. Distribution areas are given in parentheses. This tree depicts only the tree topology, branch length is not informative. 44 L. Martinsen et al. a predominantly inland distribution. The other possible scenario is that Italy was colonized twice by two different lineages. Although not most parsimonious, this scenario more easily explains (i) the close relationship of D. capreensis with D. palpata, but not with D. geniculatalaetitiae, and (ii) that D.capreensisD. palpata is associated with the D. schiavazzii group but not with D. geniculatalaetitiae, and (iii) that D. ligustica is sister group to the D. geniculatalaetitiae complex. Support for the latter scenario can be found in the SAMOVA results and the phylogenetic trees. Area 4 as defined in the DIVA analysis (continental Italy central and south) does not correspond to a monophyletic clade in the phylogeny (Fig. 3). In the SAMOVA, area 4 was never recognized as a separate group. It is unlikely that the last common ancestor of Dolichopoda had a wide distribution with all populations choosing similar cave habitats at the beginning of the glacialinterglacial periods of Pleistocene. One can assume that the last common ancestor for the genus Dolichopoda was already to a certain degree adapted to caves. Note, the outgroup taxa in our study are also cave dwellers, though two of the four species in Euhadenoecus and several of the Ceuthopilus species are epigean. It seems reasonable to assume that the climatic fluctuations during Pleistocene forced the ancestor of the genus to be more dependent on a cavernicolous lifestyle. The caves evidently provided an environment that protected their inhabitants not only against the rigors during the ice ages but also against the aridity of the Mediterranean climate that supervened during the interglacials and in post-glacial time (HUBBELL and NORTON 1978). The dry Mediterranean climate is, together with distance, the major factor that limits the movement of the cave crickets outside the caves and therefore gene flow between populations. In summary, we suggest that a combination of vicariance and dispersal events can explain the distribution of Dolichopoda species and populations. Dispersal can explain the main distributional range of the genus as demonstrated by the dispersal events in the DIVA analysis, while vicariance due to habitat fragmentation may have been the main force for splitting the species into their current populations during the glacial-interglacial cycles. The DIVA analysis suggests an eastwest migration of the Dolichopoda into the Mediterranean area. An alternative explanation is a parallel invasion to the Mediterranean region from a northern ancestor population that was forced southward due to changing climate. Acknowledgements We want to thank Gunnhild Marthinsen, Eirik Rindal, Tor Arne Carlsen, Håvard Kauserud and Fahri Saatcioglu for advice when analyzing data and writing Hereditas 146 (2009) the manuscript. The project was supported by the ‘National Centre for Biosystematics’ (Project no. 146515/420), cofunded by the Norwegian Research Council and the Natural History Museum, University of Oslo, Norway. REFERENCES Allegrucci, G., Caccone, A., Cesaroni, D. et al. 1982. Natural and experimental interspecific hybridization between populations of Dolichopoda cave crickets. Experimentia 38: 9698. Allegrucci, G., Cesaroni, D. and Sbordoni, V. 1987. Adaptation and speciation of Dolichopoda cave crickets (Orthoptera, Rhaphidophoridae)-geographic variation of morphometric indexes and allozyme frequencies. Biol. J. Linn. Soc. 31: 151160. Allegrucci, G., Caccone, A., Cesaroni, D. et al. 1992. Evolutionary divergence in Dolichopoda cave crickets: a comparison of single copy DNA hybridization data with allozymes and morphometric distances. J. Evol. Biol. 5: 121148. Allegrucci, G., Minasi, M.-G. and Sbordoni, V. 1997. Patterns of gene flow and genetic structure in cavedwelling crickets of the Tuscan endemic, Dolichopoda schiavazzii (Orthoptera, Rhaphidophoridae). Heredity 78: 665673. Allegrucci, G., Todisco, V. and Sbordoni, V. 2005. Molecular phylogeography of Dolichopoda cave crickets (Orthoptera, Rhaphidophoridae): a scenario suggested by mitochondrial DNA. Mol. Phylogenet. Evol. 37: 153164. Baccetti, B. 1987. Evolutionary biology of orthopteroid insects. Ellis Horwood Ltd, Chichester, England. 1987: 1612. Baccetti, B. and Capra, F. 1959. Notulae Orthopterologicae XII. Revisione delle Specie Italiane del Genre Dolichopoda Bol. (Orthopt. Rhaphidophoridae). Redia 44: 165217. Bernardini, C. and Di Russo, C. 2004. A general model for the life cycle of Dolichopoda cave crickets (Orthoptera: Rhaphidophoridae). Eur. J. Entomol. 101: 6973. Bernardini, C., Di, R. C., Rampini, M. et al. 1996. A recent colonization of Dolichopoda cave crickets in the Poscola cave (Orthoptera, Rhaphidophoridae). Int. J. Speleol. 25: 1531. Carchini, G., Rampini, M., Severini, C. et al. 1983. Population size estimates of four species of Dolichopoda in natural and artificial caves of central Italy (Orthoptera: Raphidophoridae). Mem. Biospeol. 10: 341347. Cesaroni, D., Matarazzo, P., Allegrucci, G. et al. 1997. Comparing patterns of geographic variation in cave crickets by combining geostatistic methods and Mantel tests. J. Biogeogr. 24: 419431. Di Russo, C. and Sbordoni, V. 1998. Gryllacridoidea. In: Juberthie, C. and Decu, V. (eds), Encyclopaedia biospeologica. Soc. Biospel., Moulis and Bucarest, p. 979988. Di Russo, C., Rampini, M. and Landeck, I. 2007. The cave crickets of northeast Turkey and transCaucasian regions, with descriptions of two new species of the genera Dolichopoda and Troglophilus (Orthoptera, Rhaphidophoridae). J. Orthoptera Res. 16: 6776. Dupanloup, I., Schneider, S. and Excoffier, L. 2002. A simulated annealing approach to define the genetic structure of populations. Mol. Ecol. 11: 25712581. Hereditas 146 (2009) Phylogeography and mitochondrial DNA divergence in Dolichopoda Gibert, J. and Deharveng, L. 2002. Subterranean ecosystems: a truncated functional biodiversity. Bioscience 52: 473481. Goloboff, P. A., Farris, J. S. and Nixon, K. C. 2008. TNT, a free program for phylogenetic analysis. Cladistics 24: 774786. Hewitt, G. 2000. The genetic legacy of the Quaternary ice ages. Nature 405: 907913. Hubbell, T. H. and Norton, R. M. 1978. The systematics and biology of the cave-crickets of the North American tribe Hadenoecini (Orthoptera Saltatoria: Ensifera: Rhaphidoridae: Dolichopodinae). Misc. Publ. Mus. Zool. Univ. Mich. 156: 1124. Huelsenbeck, J. P. and Ronquist, F. 2001. MRBAYES: Bayesian inference of phylogenetic trees. Bioinformatics 17: 754755. ICZN 2002 Opinion 1998 (Case 3123) DOLICHOPODINI Brunner von Wattenwyl, 1888 (Insecta, Grylloptera): spelling emended to DOLICHOPODAINI, so removing the homonymy with DOLICHOPODIDAE Latreille, 1809 (Insecta, Diptera). Bull. Zool. Nomenclature 59. Knowles, L. L. and Otte, D. 2000. Phylogenetic analysis of montane grasshoppers from western North America (Genus Melanoplus, Acrididae: Melanoplinae). Ann. Entomol. Soc. Am. 93: 421431. Kumar, S., Tamura, K. and Nei, M. 2004. MEGA3: integrated software for molecular evolutionary genetics analysis and sequence alignment. Brief. Bioinf. 5: 150 163. La-Greca, M. 1990. The insect biogeography of west Mediterranean islands. Accademia-Nazionale-deiLincei-Atti-Convegni-Lincei-Rome. 85: 469491. Lefebure, T., Douady, C. J., Gouy, M. et al. 2006. Phylogeography of a subterranean amphipod reveals cryptic diversity and dynamic evolution in extreme environments. Mol. Ecol. 15: 17971806. Nylander, J. A. A. 2004. MrModeltest ver 2. Program distributed by the author. Evol. Biol. Centre, Uppsala Univ. (/<http://www.abc.se/nylander//>). Porter, M. L. 2007. Subterranean biogeography: what have we learned from molecular techniques? J. Cave Karst Stud. 69: 179186. Rentz, D. 1991. Orthoptera (grasshoppers, locusts, katydids, crickets). In: Naumann, I. D. and Carne, P. B. E. (eds), The insects of Australia (2nd ed.). Melbourne Univ. Press and UCL Press, London, pp. 369393 Ronquist, F. 1996. DIVA version 1.1. Computer program and manual available by anonymous FTP from Uppsala Univ. (Bftp.uu.se or ftp.systbot.uu.se ). Rozas, J., Sanchez-Delbarrio, J. C., Messeguer, X. et al. 2003. DnaSP, DNA polymorphism analyses by the coalescent and other methods. Bioinformatics 19: 24962497. Sanmartin, I. 2003. Dispersal vs. vicariance in the Mediterranean: historical biogeography of the Palearctic Pachydeminae (Coleoptera, Scarabaeoidea). J. Biogeogr. 30: 18831897. 45 Sbordoni, V., De Matthaeis, E. and Sbordoni, M. C. 1976. Phosphoglucomutase polymorphism and natural selection in populations of the cave cricket Dolichopoda geniculata. Z. Zool. Syst. Evolutionsforschung 14: 292299. Sbordoni, V., Allegrucci, G., Cesaroni, D. et al. 1985. Genetic structure of populations and species of Dolichopoda cave crickets-evidence of peripatric divergence. Boll. Zool. 52: 139156. Sbordoni, M. C., Allegrucci, G., Caccone, A. et al. 1987. Microevolutionary studies in Dolichopodinae cave crickets. In: Baccetti, B. (ed.), Evolutionary biology of Orthopteroid insects. Ellis Horwood, Chichester, p. 514 540. Sbordoni, V., Allegrucci, G. and Cesaroni, D. 1991. A multidimensional approach to the evolution and systematics of Dolichopoda cave crickets. NATO ASI Ser. H, Cell Biol., pp. 171179 Sbordoni, V., Allegrucci, G. and Cesaroni, D. 2000. Population genetic structure, speciation and evolutionary rates in cave-dwelling organisms. In: Wilkens, H., Culver, D. C. and Humphreys, W. F. (eds), Subterranean ecosystems. Elsevier Scientific, p. 453477. Sbordoni, V., Allegrucci, G. and Todisco, V. 2004. Il genere Dolichopoda in Sardegna: filogenesi molecolare e ipotesi sulla evoluzione del popolamento. Stud. Trentini Sci. Nat. Acta Biol. 81: 103111. Simon, C., Frati, F., Beckenbach, A. et al. 1994. Evolution, weighting, and phylogenetic utility of mitochondrial gene-sequences and a compilation of conserved polymerase chain-reaction primers. Ann. Entomol. Soc. Am. 87: 651701. Spjotvoll, E. and Stoline, M. 1973. An extension of the Tmethod of multiple comparison to include the cases with unequal sample sizes. J. Am. Stat. Ass. 68: 976978. Statsoft, I. 2005. STATISTICA (data analysis software system) ver. 7.1. /<www.statsoft.com/>. Swofford, D. L. 1998. Phylogenetic analysis using parsimony (* and other methods) ver. 4. Sinauer. Thompson, J., Gibson, T., Plewniak, F. et al. 1997. The CLUSTAL_X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res. 25: 48764882. Venanzetti, F., Cesaroni, D., Mariottini, P. et al. 1993. Molecular phylogenies in Dolichopoda cave crickets and mtDNA rate calibration. Mol. Phylogenet. Evol. 2: 275280. Weiss, S. and Ferrand, N. 2007. Current perspective in phylogeography and the significance of south European refugia in the creation and maintenance of European biodiversity. In: Weiss, S. and Ferrand, N. (eds), Phylogeography of southern European refugia: evolutionary perspectives on the origins and conservation of European biodiversity. Springer, p. 1377.