Org Divers Evol (2016) 16:791–808 DOI 10.1007/s13127-016-0282-6

ORIGINAL ARTICLE

Species radiation in the Alps: multiple range shifts caused diversification in Ringlet butterflies in the European high mountains

Thomas Schmitt1,2,3 & Dirk Louy3 & Edineia Zimmermann3,4 & Jan Christian Habel5

Received: 28 December 2015 /Accepted: 11 April 2016 /Published online: 22 April 2016 # Gesellschaft für Biologische Systematik 2016

Abstract The distributions of European high mountain spe- However, the genetic differentiation within E. cassioides cies are often characterised by small and geographically iso- sensu lato into three geographically delimited groups is justi- lated populations and, in many cases, have highly complex fying species rank: E. arvernensis distributed in the Pyrenees, biogeographic histories. The butterfly genus Erebia represents Massif Central and western Alps; E. cassioides sensu stricto in one of the best examples for small-scale diversification in the the eastern Alps and Apennines; and E. neleus in the Balkan European high mountain systems and therefore to understand mountains and the south-western Carpathians. While the dif- speciation processes and associated range dynamics of high ferentiation between western Alps and Massif Central as well mountain species. In this study, we analysed 17 polymorphic as eastern Alps and Apennines was low, the Pyrenees as well allozyme loci of 1731 individuals from 49 populations as the south-western Carpathians were significantly differen- representing four species, one of which has three subspecies: tiated from the other regions within the respective taxon. In Erebia nivalis; Erebia tyndarus; Erebia ottomana;andErebia general, the differentiation among the populations of E. neleus cassioides cassioides, Erebia cassioides arvernensis,and was stronger than between populations of the other taxa. Erebia cassioides neleus. Samples were collected in the high Within E. cassioides, we found a west-east gradient of genetic mountain systems of Europe (i.e. Pyrenees, Massif Central, similarity over the eastern Alps. Based on the obtained genetic Alps, Apennines, Carpathians, Balkan high mountains). structures, we are able to delineate glacial refugia and inter- Genetic analyses supported all previously accepted species. glacial range modifications. Based on the genetic structures and genetic diversity patterns, we conclude that, triggered by the glacial-interglacial cycles, repeated range modifications Electronic supplementary material The online version of this article have taken place with subsequent differentiation and specia- (doi:10.1007/s13127-016-0282-6) contains supplementary material, tion in the region of the Alps and Balkans. Colonisations to which is available to authorized users. Pyrenees (E. arvernensis pseudomurina, E. arvernensis * Thomas Schmitt pseudocarmenta), Massif Central (E. ottomana tardenota, [email protected] E. a. arvernensis) and Apennines (E. cassioides majellana) appear to be recent and most probably not older than the last interglacial period. 1 Senckenberg Deutsches Entomologisches Institut, Eberswalder Straße 90, 15374 Müncheberg, Germany . 2 Keywords Allopatric differentiation Allozyme Department of Zoology, Institute of Biology, Faculty of Natural polymorphisms . Erebia tyndarus group . Glacial-interglacial Sciences I, Martin Luther University Halle-Wittenberg, 06099, Halle (Saale), Germany cycles . Ice ages . Phylogeography . Range shifts . Speciation 3 Department of Biogeography, Trier University, 54296 Trier, Germany 4 Diagnose Genética Médica e Biologia Molecular Ltda, Caxias do Introduction Sul, Rio Grande do Sul, Brazil 5 Terrestrial Ecology Research Group, Department of Ecology and The number of phylogeographic studies increased since the Ecosystem Management, School of Life Sciences Weihenstephan, foundation of this scientific discipline, almost 30 years ago Technische Universität München, 85354 Freising, Germany (Avise et al. 1987). Although phylogeographic projects have 792 T.Schmittetal. been performed all around the world (Hewitt 2004), Europe generalised phylogeographic scenario across the European (Hewitt 1999;Schmitt2007) and North America (Soltis et al. high mountain systems. In particular, the European high 2006;Shaferetal.2010;Woodetal.2013) are by far the best mountain massifs with a small proportion of high mountain studied regions. In particular, Europe and the Mediterranean habitats (i.e. the alpine zone), such as the French Massif region have been addressed in many studies (reviewed in Central, the Apennines and the high mountain ranges of the Taberlet et al. 1998; Hewitt 1999;Schönswetteretal.2005; Balkan Peninsula, are poorly studied, and their importance as Schmitt 2007, 2009; Schmitt and Varga 2012;Husemannetal. stepping stones and refugial differentiation centres is far from 2014). This plenitude of studies clearly demonstrated a rather being understood comprehensively (cf. Schmitt et al. 2014). high phylogeographic complexity of the western Palaearctic For these reasons, phylogeographic studies including entire in comparison to other regions of the world, e.g. eastern North groups of closely related high mountain species are a suitable America (reviewed in Soltis et al. 2006), the African savan- study system to enhance our knowledge of the complex bio- nahs (reviewed in Lorenzen et al., 2012) or southern Australia geographic patterns within and among the different high (reviewed in Byrne 2008). However, in contrast to an earlier mountain systems of Europe. One suitable species group with theory (e.g. Reinig 1937, 1950; de Lattin 1949, 1964, 1967), it high potential to enhance our understanding in this context is is not only the warm-adapted species showing complex pat- the Erebia tyndarus species group. This group is composed of terns. Also cold-adapted species with wide distributions at least seven species in Europe (Erebia tyndarus, Erebia (reviewed in Schmitt and Varga 2012) and high mountain taxa cassioides, Erebia nivalis, Erebia calcaria, Erebia rondoui, (reviewed in Schönswetter et al. 2005;Schmitt2009)contrib- Erebia hispania, Erebia ottomana), three species in western ute considerably to the great biogeographic complexity of this Asia (Erebia graucasica, Erebia transcaucasica, Erebia region. iranica) and the Holarctic Erebia callias with its most western With respect to the phylogeographic patterns of high old world populations in the northern Urals (Albre et al. mountain taxa, a highly repetitive pattern has been shown 2008). Many of these species are endemics with small distri- for the Alps. Here, many species have several, often bution ranges; in Europe, only E. cassioides and E. ottomana four, different genetic lineages along this mountain chain are relatively widespread. E. cassioides is found from the (Schönswetter et al. 2005). These genetic lineages have been Cantabrian Mountains in northern Spain to Rila, Pirin and suggested to have evolved in perialpine refugia at the southern Stara Planina in Bulgaria; this species is split into numerous and eastern foothills of the Alps and retreated to the high regional subspecies and some authors argue that this taxon altitudes along the postglacial warming and the melting of might be a complex of closely related allopatric species. the glaciers (Schönswetter et al. 2005). This phylogeographic E. ottomana is widely distributed in most high mountain sys- pattern is now considered the differentiation paradigm for the tems of the Balkan Peninsula, but is also found in north- Alps. Other studies, yet less in number, have suggested sur- western Anatolia, the Monte Baldo and the south-eastern vival at nunataks, i.e. mountain tops above the glaciers Massif Central. E. tyndarus, E. nivalis and E. calcaria are (Stehlik et al. 2001, 2002; Holderegger et al. 2002; Stehlik endemic to different regions of the Alps, while E. rondoui is 2002), or even north of the mountain chain of the Alps restricted to the Pyrenees and E. hispana to the Sierra Nevada (Schmitt et al. 2006). Bringing together such genetic data with in southern Spain (Albre et al. 2008 ; Tshikolovets 2011). recent distribution patterns of species, palaeoecological stud- This species group has already attracted some attention but ies, niche models, etc. allows the reconstruction of glacial mostly regarding systematics (e.g. Lorković 1953, 1957, refugia for species today distributed in the high altitudes of 1958; Popescu-Gorj 1962;Lattesetal.1994; Jutzeler et al. the Alps relatively robustly. Furthermore, the combination of 2002;Martinetal.2002; Albre et al. 2008) or the biogeogra- all these methods and data helps to understand the processes phy of single species at the regional scale (Louy et al. 2014a). involved in range dynamics triggering the observed genetic However, a comprehensive study analysing the genetic struc- differentiations. tures of all more widespread species of this group in Europe is However, much less in known about the phylogeographic missing. For this reason, we sampled 1731 individuals at 49 structures of other mountain systems of Europe. So far, bio- localities across all major European high mountain systems, geographic links between adjoining high mountain systems i.e. Pyrenees, French Massif Central, Alps, Apennines, based on morphological and distribution data (Varga and Carpathians and several high mountain systems of the Schmitt 2008) were demonstrated (reviewed in Schmitt Balkan Peninsula. These samples were analysed by allozyme 2009 ), and suitable phylogeographic data have recently been electrophoresis, a well-established genetic marker system with published on the Pyrenees (e.g. Kropf et al. 2002, 2003; high phylogeographic resolution in butterflies (cf. Louy et al. Schmitt et al. 2006; Lauga et al. 2009;Lihováetal.2008; 2014b). In this study, we aim to investigate the phylogeo- Charrier et al. 2014) and Carpathians (e.g. Pauls et al. 2006; graphic structures within the more widely distributed taxa in Mráz et al. 2007;Ronikieretal.2008;Ujvárosietal.2010; this species group and clarify some of the open systematic Theissinger et al. 2012). However, it is still difficult to create a questions within the E. tyndarus complex. These results are Complex biogeography of a mountain butterfly 793 in particular valuable for searching for general biogeographic meadows and can be classified as a typical butterfly species patterns of European high mountain biota. of the subalpine belt (Louy et al. 2014a).

Molecular analyses

Materials and methods Sampling was performed during the summers 2003, 2005– 2007, 2009–2012: Erebia c. arvernensis, 579 individuals Study species from 15 populations, covering the Pyrenees, Massif Central and the western Alps, but also one population in the eastern TheCommonBrassyRingletE. cassioides (Reiner and Alps (Passo Maghen, Italy); E. c. cassioides, 266 individuals Hohenwarth, 1792) is distributed in the eastern and western from nine populations collected in the eastern Alps and the Alps, the Apennines, the French Massif Central, the Pyrenees, Apennines; E. cassioides neleus, 167 individuals from six the Cantabrian mountains, the Retezat (south-western populations covering Shar Planina, Pirin, Rila, Stara Planina Carpathians) and in the highest parts of the Balkan mountains and Retezat; E. tyndarus, 538 individuals from 14 populations (Tolman and Lewington 1997). The taxon is found from 1400 scattered over the Central Alps; E. nivalis, 43 individuals to 2600 m a.s.l. in high mountain short-turf grasslands and at from two populations collected in the eastern Alps; and rocky slopes (Tshikolovets 2011). In the Balkan high E. ottomana, 120 individuals from three populations collected mountain systems, E. cassioides is observed only above in Ossogovo, Rila and Stara Planina. Sampling sizes ranged 1900 m a.s.l., normally above 2100 m a.s.l. (Varga 2014). from 7 to 42 individuals with a mean of 35 (±10 SD) individ- Thus, this butterfly species has to be characterised as a typical uals per population. Individuals were netted in the field, frozen alpine species. The western populations (western Alps, alive in liquid nitrogen and stored at −80 °C until further Pyrenees, Cantabrian mountains) are considered to represent analyses. The sampling locations are displayed in Fig. 1;fur- the distinct subspecies E. cassioides arvernensis Oberthür, ther details on single sampling sites are given in Table 1. 1908 and are by some authors considered to be a valid species Standard procedures of sample preparation and allozyme (Lerault 1997; Albre et al. 2008). The authors of this publica- electrophoresis were performed as described in Richardson tion therefore use the taxon arvernensis for all western popu- et al. (1986) and Hebert and Beaton (1993). We analysed the lations in the following. Several different subspecies were following 17 allozyme loci for all species: IDH1, IDH2, described for the populations from the Balkan Peninsula. MDH1, MDH2, 6PGDH, G6PDH, PGI, PGM, PEPLGG, E. cassioides neleus (Freyer, 1833) was described from the AAT1, AAT2, GPDH, FUM, ME, MPI, PK and GAPDH. Retezat (BBanater Gebirge^ in the original description) and Conditions for electrophoretic analyses were similar to was the first subspecies described from south-eastern Schmitt et al. (2006) and were identical for all taxa. Europe; it was sometimes even given species rank as by Tests on Hardy-Weinberg equilibrium and linkage disequi- Popescu-Gorj (1962). For means of simplicity, the taxon librium were performed with the program ARLEQUIN ver. 3.1 neleus is used in the following for all populations from (Excoffier et al. 2005). We calculated the following six param- south-eastern Europe. The Swiss Brassy Ringlet E. tyndarus eters of genetic diversity: allelic richness AR (on the basis of (Esper [1781]) is an endemic of the central Alps (Switzerland, the lowest number of sampled individuals, here seven individ- western Austria) and has a parapatric distribution with uals) and mean number of alleles A using the program FSTAT E. cassioides at its western and eastern distribution edge ver. 2.9.3.2 (Goudet 1995); observed and expected heterozy-

(Tolman and Lewington 1997). The habitat requirements are gosity Ho and He were estimated with the program ARLEQUIN. very similar with E. cassioides (Sonderegger 2005). De Percentage of the total number of polymorphic loci Ptot and Lesse’sBrassyRingletE. nivalis Lorković & de Lesse, 1954 the percentage of polymorphic loci with the most common has a quite limited distribution: the species is restricted to the allele not exceeding 95 % P95 were calculated with the pro- highest parts of the eastern Alps of Austria and to the gram GSTAT ver. 3.0 (Siegismund 1993). Allele frequencies for Grindelwald region in Switzerland. In general, the flight all populations and loci were calculated with the latter places of E. nivalis are higher than the ones of E. cassioides program. and E. tyndarus (Lorković 1957; Sonderegger 2005). The A neighbour-joining phenogram based on Nei’s(1972)ge- Ottoman Brassy Ringlet Erebia ottomana Herrich-Schäffer netic distances including all taxa and populations analysed [1847] is widespread in the Balkan Peninsula, but is also was constructed with PHYLIP ver. 3.5.c. (Felsenstein 1993). known from the south-eastern Massif Central, the Monte Node support was assessed by means of 1000 bootstrap rep- Baldo region and north-western Anatolia (Tolman and licates. The program STRUCTURE v. 3.1 (Hubisz et al. 2009) Lewington 1997). It occurs in different habitat types and at was used to infer the most probable number of genetic clusters lower altitudes (from 1400 to 2400 m a.s.l.) compared to without a priori definition of populations. We performed sev- E. cassioides (Varga 2014). E. ottomana prefers tall-grass eral separate runs: the first run included all samples. In further 794 T.Schmittetal.

Fig. 1 Sampling locations of all studied species, with Erebia cassioides arvernensis (grey circles), Erebia c. cassioides (black circles), Erebia c. neleus in the Balkan region (black-white circles), Erebia nivalis (white triangles), Erebia tyndarus (white circles)andErebia ottomana (grey triangles)(a). The detailed map shows the distribution of the samples in the Alps (b). Given numbers coincide with other figures and tables

runs, we analysed the species of the cassioides species group the more refined ad hoc statistic ΔK based on the rate of (i.e. E. c. cassioides, E. c. arvernensis, E. c. neleus)aswellas change in the log probability of data between successive K E. tyndarus separately. The number of K values differed de- values (Evanno et al. 2005). This has been shown to better pending the number of populations available for the respective unveil the correct number of genetic clusters and to infer the taxon (see below). Calculations were performed with ten rep- real number of groups. etitions for each cluster. The repetitions were run to calculate We analysed the genetic structuring within and among lo- means and standard deviations. For each run, the burn-in and cal populations using a non-hierarchical and a hierarchical simulation length were 150,000 and 500,000, respectively. As genetic variance analyses (AMOVA) with the program the log probability values for the different K values have been ARLEQUIN. In the non-hierarchical variance analyses, we first shown to be of little reliability in some cases, we calculated analysed the molecular variance when grouping all Complex biogeography of a mountain butterfly 795

Table 1 Overview of all samples analysed, including the taxa Erebia number coincides with other figures and tables), exact GPS coordinates, cassioides arvernensis, E. c. cassioides, Erebia cassioides neleus, Erebia altitude, date of sampling and the genetic cluster to which the population tyndarus, Erebia nivalis and Erebia ottomana. Given is the geographical is assigned according to the typology of the neighbour-joining region, exact name of sampling location including the respective country phenogram and STRUCTURE analyses of sampling and a running number for each sampling location (this

Region Country-location-no. GPS (N, E) Altitude (m) Number individuals Date of sampling Genetic cluster

Erebia cassioides arvernensis Pyrenees E-Vale de Tena-1 42° 42′;0°16′ W 1900 40 20 July 2003 West Pyrenees F-Col de Puymorens-2 42° 34′;1°49′ 2100 40 26 July 2003 West Massif Central F-Monts Dore-3 45° 31′;02°49′ 1650 39 28 July 2009 West Western Alps F-St. Martin-Vésubie-4 44° 06′;7°19′ 2150 40 30 July 2003 West Western Alps I-Colle della Lombarda-5 44° 14′;7°06′ 2150 40 03 August 2005 West Western Alps F-Col de Larche-6 44° 26′;6°53′ 2000 40 01 August 2005 West Western Alps F-Col d’Isoard-7 44° 21′;6°50′ 2450 40 01 August 2003 West Western Alps F-Col de Vars-8 44° 33′;6°42′ 2050 40 31 July 2003 West Western Alps F-Col Basset-9 44° 59′;6°52′ 2500 40 30 July 2005 West Western Alps F-Les Deux-Alpes-10 45° 01′;6°06′ 1850 40 05 August 2005 West Western Alps F-Col de Galibier-11 45° 04′;6°23′ 2400 40 01 August 2003 West Western Alps F-Petit Cenis-12 45° 15′;6°54′ 2300 20 02 August 2003 West Western Alps F-Val d’Isère-13 45° 27′;6°59′ 1850 40 04 August 2003 West Western Alps F-Valée des Glaciers-14 45° 45′;6°45′ 2200 40 03 August 2003 West Eastern Alps I-Passo Maghen-15 46° 10′; 11° 26′ 1850 40 07 July 2007 West Erebia cassioides cassioides Eastern Alps I-Sellajoch-16 46° 31′; 11° 46′ 2250 40 23 July 2006 East Eastern Alps A-Patscherkofel-17 47° 12′; 11° 30′ 2000 40 31 July 2006 East Eastern Alps I-Rein in Taufers-18 46° 58′;12°03′ 2300 28 20 July 2006 East Eastern Alps A-Staller Sattel-19 46° 54′;12°15 ′ 2450 11 14 August 2005 East Eastern Alps A-Kalkstein-20 46° 49′;12°19′ 2100 23 15 August 2005 East Eastern Alps A-Conyalm-21 46° 44′;12°36′ 2000 32 26 July 2006 East Apennines I-Majelletta-22 42° 11′;14°06′ 1600 40 20 July 2009 East Apennines I-Prati di Tivo-23 42° 29′;13°33′ 1900 12 22 July 2009 East Apennines I-Cutigliano-24 44° 07′;10°47′ 1300 40 25 July 2009 East Erebia cassioides neleus Shar Planina MK-Popova Shapka-25 42° 00′;20°51′ 2100 40 06 August 2012 SE Pirin BG-Bansko-26 41° 45′;23°23′ 2450 40 02 August 2010 SE Pirin BG-Bezbog-27 41° 43′;23°30′ 2450 13 29 July 2012 SE Rila BG-Granchar-28 42° 07′;23°35′ 2300 27 31 July 2010 SE Stara Planina BG-Botev-29 42° 43′;24°56′ 2100 7 09 August 2011 SE Retezat RO-Iorgovan-30 45° 17′;22°54′ 1400 40 01 August 2011 SE Erebia tyndarus Central Alps I-Passo del Monte Moro-31 46° 00′;7°59′ 2800 40 22 July 2005 tyndarus Central Alps CH-Simplonpass-32 46° 15′;8°01′ 2050 40 08 August 2005 tyndarus Central Alps CH-Grindelwald-33 46° 40′;8°03′ 2300 40 06 August 2003 tyndarus Central Alps CH-Gletsch-34 46° 34′;8°22′ 1800 40 09 August 2005 tyndarus Central Alps CH-Colla-35 46° 28′;8°41′ 2400 40 05 August 2003 tyndarus Central Alps CH-Klausenpass-36 46° 52′;8°52′ 2000 40 10 August 2005 tyndarus Central Alps CH-Val Maroz-37 46° 24′;9°37′ 2250 40 20 July 2005 tyndarus Central Alps CH-Partnun-38 47° 00′;9°52′ 2000 40 07 August 2003 tyndarus Central Alps CH-Pontresina-39 46° 29′;9°55′ 1900 42 18/19 July 2005 tyndarus Central Alps CH-Tschierv-40 46° 39′;10°14′ 2300 40 17 July 2005 tyndarus Central Alps I-Thanai-41 46° 45′;10°40′ 2250 40 20 July 2006 tyndarus 796 T.Schmittetal.

Table 1 (continued)

Region Country-location-no. GPS (N, E) Altitude (m) Number individuals Date of sampling Genetic cluster

Central Alps A-Verpeil-42 47° 02′;10°47′ 2350 16 20 July 2006 tyndarus Central Alps I-Kuppelwiesen Alm-43 46° 34′;10°54′ 2000 40 22 July 2006 tyndarus Central Alps A-Zwieselstein-44 46° 56′; 11° 03′ 2000 40 12 August 2005 tyndarus Erebia nivalis Eastern Alps A-Sajatmähder-45 47° 30′;12°19′ 2500 14 27 July 2006 nivalis Eastern Alps A-Großglockner-46 47° 07′;12°49′ 2300 29 01 July 2007 nivalis Erebia ottomana Ossogovo BG-Iglika-47 42° 12′;22°36′ 1850 40 30 July 2005 ottomana Rila BG-Nethenica-48 42° 06′;23°37′ 2100 40 31 July 2005 ottomana Stara Planina BG-Triglav-49 42° 71′;25°10′ 1700 40 04 August 2004 ottomana

E Spain, F France, I Italy, A Austria, MK Macedonia, BG Bulgaria, RO Romania, CH Switzerland, SE south-east populations together (all species and populations). (population 22–24); and E. c. neleus from the Balkans and Afterwards, we analysed each taxon separately, with Retezat (population 25–30). E. cassioides s.l. (including cassioides s.s., arvernensis, neleus), E. tyndarus, E. nivalis and E. ottomana. In the second approach, we pre-defined groups and analysed the genetic Results variance found among these inter-specific and intra-specific groups. Genetic diversity Finally, to test the level and direction of gene flow among species, subspecies and among regional population groups, The population genetic diversities had similar values for the we used the program BAYESASS ver. 1.3 (Wilson and Rannala four previously accepted species (Kruskal-Wallis ANOVA, 2003). This program relies on multi-locus genotypes and a P > 0.05). Mean estimates (with standard deviations among Markov chain Monte Carlo (MCMC) algorithm to estimate populations) and population-specific values of six parameters proportions of non-migrants and the source of migrants for of genetic diversity are given in Table 2. Genetic diversity each sampling site (Wilson and Rannala 2003). As this pro- within E. c. arvernensis was similar for populations located gram is a non-equilibrium Bayesian method that does not in the western Alps and in the Pyrenees (U test, P > 0.05) but require data sets to conform to HWE proportions, we included was much lower in the Massif Central (no test possible as only all loci in the analysis. Wilson and Rannala (2003)suggest one population from the Massif Central was sampled). E. c. that more accurate results are obtained from runs when the cassioides had significantly lower genetic diversity in popula- number of proposed changes for the variables M (migration tions from the Apennines in comparison with the populations rate), P (allele frequencies) and F (inbreeding coefficient) is sampled in the eastern Alps (U test, P <0.05).However,the between 40 and 60 % of the total chain length. We performed central Apennines had remarkably higher values than the runs of the algorithm for 9 × 106 iterations with 3 × 106 itera- northern Apennines (no test possible due to insufficient sam- tions discarded as burn-in. Delta values of M =0.30, P =0.15 ple sizes). E. c. neleus from the Retezat (south-western and F = 0.15 yielded an average number of changes in the Carpathians) had higher genetic diversity than the average in accepted range. We first tested for potential gene flow among the Balkan mountains for all parameters analysed (no test the six main groups (i.e. taxa) clustering according to the possible as only one populations from Retezat was sampled). neighbour-joining phenogram (see results), according the fol- Allele frequencies of all population and loci are given in lowing groupings: E. c. arvernensis (population 1–15), E. c. Electronic supplementary material Appendix S1. cassioides (population 16–24), E. c. neleus (population 25– 30), E. tyndarus (population 31–44), E. nivalis (population Genetic differentiation 45–46) and E. ottomana (population 47–49). In a second cal- culation, we only selected samples forming seven clusters The neighbour-joining phenogram (Fig. 2) distinguished six within the E. cassioides species group with the following clearly defined genetic groups: the three previously accepted groups: Erebia c. arvernensis from the Pyrenees (population species E. tyndarus (Central Alps), E. nivalis (eastern Alps), 1–2), Massif Central (population 3), western Alps (population E. ottomana (Balkans) and three groups of E. cassioides s.l.: 4–14) and Passo Maghen (population 15); E. c. cassioides western Alps, Pyrenees, French Massif Central, but also Passo from the eastern Alps (population 16–21) and Apennines Maghen in the south-eastern Alps (i.e. E. c. arvernensis), Complex biogeography of a mountain butterfly 797

Table 2 Six parameters of genetic diversity given for all species heterozygosity He and Ho, percentage of all polymorphic loci Ptot and (means with standard deviations) and populations analysed. The of loci with the most common allele not exceeding 95 % P95. Values following values are given: mean number of alleles A, allelic richness based on an insufficient number of individuals are given in parenthesis calculated based on the lowest number of individuals per population and are excluded from the calculation of means. Numbers of populations (here seven individuals) AR, percentage of the expected and observed coincide with other figures and tables

Country-location-no. AARHe [%] Ho [%] Ptot [%] P95 [%]

Erebia cassioides arvernensis E-Vale de Tena-1 2.05 1.47 11.6 11.8 64.7 35.3 F-Col de Puymorens-2 2.29 1.64 17.3 18.1 58.8 47.1 F-Monts Dore-3 1.47 1.21 4.7 4.4 29.4 23.5 F-St. Martin-Vésubie-4 2.29 1.70 19.1 18.3 70.6 52.9 I-Colle della Lombarda-5 2.18 1.51 12.4 12.9 70.6 52.9 F-Col de Larche-6 1.94 1.54 14.0 15.9 64.7 47.1 F-Col d’Isoard-7 2.12 1.59 15.7 16.5 64.7 47.1 F-Col de Vars-8 2.05 1.48 13.2 12.6 64.7 35.3 F-Colle Basset-9 2.29 1.57 14.1 14.5 64.7 41.2 F-Les Deux-Alpes-10 2.18 1.53 11.9 12.8 70.6 41.2 F-Col de Galibier-11 2.12 1.59 14.6 15.1 76.5 41.2 F-Petit Cenis-12 (1.82) 1.48 11.5 11.9 58.8 41.2 F-Val d’Isére-13 1.94 1.45 9.8 10.0 52.9 47.1 F-Valée des Glaciers-14 2.24 1.57 13.6 11.8 64.7 47.1 I-Passo Maghen-15 1.59 1.32 9.0 8.3 41.2 23.5 Mean (±SD) 2.05 (±0.25) 1.51 (±0.12) 12.8 (±3.5) 13.0 (±3.7) 61.2 (±12.1) 41.6 (±9.0) Erebia cassioides cassioides I-Sellajoch-16 1.77 1.32 9.0 7.5 52.9 17.6 A-Patscherkofel-17 1.65 1.36 10.7 9.1 35.3 23.5 I-Rein in Taufers-18 1.47 1.30 10.0 10.3 35.3 23.5 A-Staller Sattel-19 (1.65) 1.49 13.4 12.8 (41.2) 35.3 A-Kalkstein-20 (1.71) 1.44 14.5 15.9 47.1 35.3 A-Conyalm-21 1.77 1.41 10.3 9.7 52.9 35.3 I-Majelletta-22 1.77 1.27 6.0 6.3 52.9 23.5 I-Prati di Tivo-23 (1.29) 1.19 3.9 4.4 (11.8) 11.8 I-Cutigliano-24 1.35 1.15 3.4 3.4 23.5 17.6 Mean (±SD) 1.63 (±0.18) 1.33 (±0.11) 9.0 (±3.9) 8.8 (±4.0) 42.8 (±11.6) 24.8 (±8.7) Erebia cassioides neleus MK-Popova Shapka-25 1.59 1.26 7.0 6.9 41.2 17.6 BG-Bansko-26 1.59 1.33 10.8 10.3 47.1 35.3 BG-Bezbog-27 (1.41) 1.26 6.9 7.8 (29.4) 29.4 BG-Granchar-28 1.65 1.42 12.3 11.3 52.9 35.3 BG-Botev-29 (1.24) 1.21 5.2 4.2 (23.5) (23.5) RO-Iorgovan-30 1.88 1.52 13.2 10.3 52.9 41.2 Mean (±SD) 1.68 (±0.14) 1.33 (±0.12) 9.2 (±3.3) 8.5 (±2.7) 48.5 (±5.6) 31.8 (±9.0) Overall mean (±SD) 1.89 (±0.29) 1.42 (±0.15) 11.0 (±3.9) 10.8 (±4.1) 54.3 (±13.8) 34.7 (±11.5) Erebia tyndarus I-Passo del M. Moro-31 1.29 1.21 4.5 3.8 23.5 5.9 CH-Simplonpass-32 1.29 1.22 3.2 3.5 23.5 11.8 CH-Grindelwald-33 1.41 1.32 8.3 8.1 35.3 23.5 CH-Gletsch-34 1.59 1.37 6.7 6.0 41.2 17.6 CH-Colla-35 1.41 1.25 4.3 4.7 29.4 11.8 CH-Klausenpass-36 1.29 1.15 3.1 4.0 23.5 5.9 CH-Val Maroz-37 1.65 1.42 6.0 6.0 47.1 17.6 798 T.Schmittetal.

Table 2 (continued)

Country-location-no. AARHe [%] Ho [%] Ptot [%] P95 [%]

CH-Partnun-38 1.65 1.41 6.6 5.4 41.2 17.6 CH-Pontresina-39 1.24 1.24 4.7 4.2 17.6 17.6 CH-Tschierv-41 1.41 1.35 7.2 7.4 35.3 29.4 I-Thanai-42 1.82 1.48 10.1 10.0 52.9 23.5 A-Verpeil-43 (1.53) 1.51 12.0 8.5 (35.3) 29.4 I-Kuppelwiesenalm-44 1.71 1.51 10.5 9.7 47.1 23.5 A-Zwieselstein-45 1.47 1.32 7.7 9.3 35.3 17.6 Mean (±SD) 1.49 (±0.19) 1.35 (±0.12) 6.9 (±2.7) 6.6 (±2.3) 35.3 (±10.6) 19.2 (±8.4) Erebia nivalis A-Sajatmähder-46 (1.53) 1.53 14.2 16.0 (47.1) 41.2 A-Großglockner-47 1.82 1.59 11.2 11.7 58.8 41.2 Mean (±SD) 1.82 1.56 (±0.04) 12.7 (±2.1) 13.9 (±3.0) 58.8 41.2 (±0.0) Erebia ottomana BG-Iglika-48 1.41 1.39 6.2 5.1 35.3 17.6 BG-Nethenica-49 1.88 1.87 14.1 13.3 70.6 41.2 BG-Triglav-50 1.65 1.64 14.0 13.4 52.9 41.2 Mean (±SD) 1.65 (±0.24) 1.63 (±0.24) 11.4 (±4.5) 10.6 (±4.8) 52.9 (±17.7) 33.3 (±13.6)

eastern Alps with Apennines (i.e. E. cassioides s.s.) as well as The genetic patterns obtained by STRUCTURE analyses Balkans and Retezat (i.e. E. c. neleus). The genetic differenti- were congruent with the topology of the population-based ation within these groups was low, with two exceptions: a neighbour-joining phenogram. In a first run including all generally high genetic differentiation among populations species and populations, we obtained the following six within E. c. neleus; E. c. arvernensis was divided into three groups supported by a high ΔK value(seeTable3): E. c. genetic groups: western Alps including French Massif arvernensis was diverging into two clusters, Pyrenees and Central, Pyrenees and Passo Maghen. western Alps (the latter including the Massif Central and

Fig. 2 Neighbour-joining phenogram calculated with the program PHYLIP ver. 3.5.c. (Felsenstein 1993), based on Nei’s(1972) genetic distances for all samples analysed. The tree topology assigned the samples into the following six main clusters (from left to right): Erebia tyndarus (Central Alps), Erebia c. neleus (Balkans and Retezat), Erebia ottomana (Balkans), Erebia c. cassioides (eastern Alps with Apennines), Erebia c. arvernensis (western Alps, Pyrenees, Massif Central and Passo Maghen located in the south-eastern Alps) and Erebia nivalis (eastern Alps). Bootstrap values calculated with 1000 permutations are given for values exceeding 50 % probability Complex biogeography of a mountain butterfly 799

Table 3 Estimates of cluster number (K) from STRUCTURE analyses Pyrenees, most probably due to the low number of individ- using allozyme polymorphisms for (a) all species and populations uals in both groups. analysed, calculated for K =1–15, and (b) for all populations of the Erebia cassioides species group, calculated for K =1–15. Ln(Pr) is the When analysing only populations of the E. cassioides com- mean log-likelihood probability calculated by the program STRUCTURE. plex, we obtained three distinct clusters (Fig. 3), supported by SD is the standard deviation calculated from ten independent runs. The ad the highest ΔK value (see Table 3). A first cluster consisted of Δ hoc statistic K is not applicable for K = 1 and the highest K value, and populations from the western Alps, Pyrenees and Massif not proper for K = 2 (Hausdorf and Hennig 2010). Highest probability and ΔK values are indicated in italics Central (i.e. E. c. arvernensis). The second group contained the populations from the eastern Alps and also included the K Ln(Pr) ± SD ΔK three Apennines populations (i.e. E. cassioides s.s.). The pop- Including all species and populations ulation from Passo Maghen (eastern Alps) showed genetic traits of both groups; however, each individual was clearly 1 −12,703.52 ± 0.06 – assigned to one of the two clusters so that representatives of 2 −10,252.36 ± 0.71 2261.27 both groups might coexist here without major hybridisation 3 −9182.43 ± 1.93 1051.06 and introgression. The third cluster consisted of the popula- 4 −8986.26 ± 4.02 0.90 tions from the Balkans and the Retezat (i.e. E. c. neleus) 5 −8867.95 ± 16.64 0.39 (Fig. 3). 6 −8813.72 ± 37.27 0.51 Additional STRUCTURE analyses showed strong differ- 7 −8692.28 ± 9.90 1.33 entiation within single species groups, with the highest 8 −8699.85 ± 17.96 0.63 ΔK values for the following clustering: E. c. arvernensis 9 −8727.42 ± 28.49 1.12 for K =3 and 7; E. c. cassioides for K =2 and 3; E. c. 10 −8791.41 ± 25.79 0.25 neleus for K =2,3,and4;andE. tyndarus for K =2. All − 11 8840.85 ± 32.50 2.21 results are shown in the Electronic supplementary mate- − 12 8952.04 ± 26.04 0.62 rial Appendix S2 and S3. − 13 9020.78 ± 82.43 0.63 Non-hierarchical analyses of molecular variance revealed a − 14 8969.73 ± 28.68 2.33 high genetic differentiation among all taxa and populations 15 −9061.37 ± 75.72 – analysed with a molecular variance of 1.7794 (FST =0.6815, Including all populations analysed for Erebia cassioides s.l. P < 0.0001). The genetic differentiation among all populations − – 1 31,163.60 ± 0.07 of E. cassioides s.l. (including the taxa cassioides s.s., 2 −19,834.31 ± 3.39 1926.99 arvernensis and neleus) was also high, with a molecular var- 3 −16,169.26 ± 2.70 452.34 iance of 0.7090 (FST = 0.4281, P <0.0001).Alowermolecular 4 −15,344.53 ± 411.64 19.34 variance was detected within E. ottomana, with 0.2175 5 −14,891.84 ± 89.13 3.85 (FST = 0.1853, P < 0.0001), within E. tyndarus with 0.0625 6 −14,404.79 ± 325.70 1.80 (FST =0.0625, P < 0.0001) and within E. nivalis with 0.0729 7 −13,750.91 ± 409.53 13.03 (FST =0.0677,P < 0.0001). This also applies to the three taxa 8 −13,641.33 ± 420.21 1.11 within E. cassioides s.l. (cassioides s.s., arvernensis and 9 −13,267.28 ± 255.65 1.28 neleus) (see Table 4). 10 −13,179.16 ± 264.18 0.56 Hierarchical variance analyses (testing genetic variance 11 −13,024.54 ± 66.34 1.89 among pre-defined population groups) revealed a high genetic 12 −13,016.49 ± 96.47 1.63 differentiation among the six genetic clusters detected by 13 −13,068.62 ± 105.67 1.45 neighbour-joining and STRUCTURE analyses (molecular vari- − 14 13,053.74 ± 85.97 4.97 ance 2.1431, FCT =0.6881, P < 0.0001). Further hierarchical 15 −13,239.19 ± 140.99 – analyses within E. cassioides s.l. indicated a genetic link be- tween the eastern Alps and the Apennines, as well as between the Massif Central and the western Alps, but not with the the eastern Alps population 15—Passo Maghen); E. c. populations from the Pyrenees (Table 5). cassioides formed one single cluster with populations from the eastern Alps including the populations from the Gene flow Apennines; E. c. neleus represented a single cluster with the populations from the Balkans and the Retezat; Gene flow estimates calculated among all populations E. tyndarus provided one genetic cluster (obtained for and species (accepting the three major genetic groups K =7); E. nivalis and E. ottomana were assigned to two in E. cassioides s.l. as separate entities) indicated low distinct groups (Fig. 3); the E. nivalis cluster was not dis- gene flow rates, with a proportion of non-migrants of tinguished from the E. c. arvernensis populations from the 83.30 % and a mean of 3.32 % (1.58E-06, 0.160) of 800 T.Schmittetal.

Fig. 3 Bayesian STRUCTURE analysis calculated a including all species (K = 6 (upper plot) and 7 (lower plot) for a and K = 3 for b). All ΔK values and populations analysed and b including all populations being assigned are given in Table 3. Numbers given for populations coincide with other to Erebia cassioides s.l., using the program STRUCTURE ver. 3.1 (Pritchard figures and tables et al. 2000). Shown are the results supported by the highest ΔK values migrating individuals. Migration rates estimated among Discussion seven clusters within E. cassioides (assigned to the fol- lowing clusters: E. c. arvernensis with Pyrenees, Massif Genetic diversity and differentiation Central, western Alps, Passo Maghen; E. c. cassioides with eastern Alps, Apennines; E. c. neleus with The mean genetic diversity of the studied E. tyndarus species Balkans and Retezat) showed similar gene flow rates, group taxa is in the order of magnitude of other Erebia species with a mean of non-migration rates of 83.3 % and a (Schmitt and Seitz 2001; Schmitt et al. 2006, 2007, 2014; mean migration rate of 2.77 % (1.15E-07, 0.144). In this Haubrich and Schmitt 2007; Schmitt and Haubrich 2008; analysis, we obtained a strong level of migration from Vila et al. 2011;Louyetal.2014b) but tends to be moderately the western to the eastern Alps (Tables 6 and 7). lower than in other butterfly studies. Hence, the genetic

Table 4 Genetic differentiation at different spatial scales Group Among populations Among individuals within Within estimated from non-hierarchical (FST) populations (FIS) individuals analyses of molecular variance (AMOVA). Variance values with All species and populations 1.7794 (0.6815***) 0.0095 (0.0115) 0.8220 respective F statistic values (in E. cassioides sensu lato (including 0.7090 (0.4281***) 0.0064 (0.0068) 0.9408 parenthesis) arvernensis, cassioides s.s., neleus) E. cassioides arvernensis 0.15690 (0.1251***) −0.02811 (−0.0256) 1.1255 E. c. cassioides 0.2105 (0.2255***) 0.0316 (0.0437*) 0.6915 E. cassioides neleus 0.3825 (0.3108***) 0.0745 (0.0878**) 0.7738 E. tyndarus 0.0625 (0.0986***) 0.0112 (0.0196) 0.5598 E. nivalis 0.0729 (0.0677***) −0.0895 (−0.0892) 1.0930 E. ottomana 0.2175 (0.1853***) 0.0645 (0.0675*) 0.8917

*P <0.05;**P < 0.01; ***P <0.001 Complex biogeography of a mountain butterfly 801

Table 5 Genetic differentiation at different spatial scales estimated from hierarchical analyses of molecular variance (AMOVA). Variance values with respective F statistic values (in parenthesis). Populations used are given as population numbers in parenthesis, in the first row

Group Among groups or Among populations Within species (FCT) within groups (FSC) individuals

Six groups according to Fig. 2 (1–15, 16–24, 25–30, 31–44, 2.1431 (0.6881***) 0.1400 (0.1441***) 0.8220 45–46, 47–49) E. cassioides s.l., western Alps vs Balkans (4–14, 25–30) 1.4929 (0.5515***) 0.1583 (0.1303***) 1.0569 E. c. arvernensis, Pyrenees vs Massif Central vs western Alps vs 0.1131 (0.0891***) 0.0825 (0.0714**) 1.0967 Passo Maghen (1–2, 3, 4–14, 15) E. c. arvernensis, Pyrenees vs western Alps (1–2, 4–14) 0.1874 (0.1318***) 0.0816 (0.0661*) 1.1860 E. c. arvernensis, Pyrenees vs Massif Central (1–2, 3) 0.1845 (0.1495***) 0.1025 (0.0977***) 0.9706 E. c. arvernensis, Massif Central vs western Alps (3, 4–14) 0.0416 (0.0347) 0.0806 (0.0698***) 1.1013 E. c. arvernensis, western Alps vs Passo Maghen (4–14, 15) 0.4699 (0.2783***) 0.0798 (0.0655***) 1.1682 E. c. cassioides, eastern Alps vs E. c. arvernensis Passo Maghen −0.0881 (−0.0863) 0.2211 (0.1993***) 0.8341 (16–21, 15) E. c. cassioides, eastern Alps vs Apennines (16–21, 22–24) 0.1075 (0.1143***) 0.0986 (0.1184*) 0.7068 C. c. arvernensis, western Alps vs C. c. cassioides, Apennines 0.4793 (0.3179***) 0.0823 (0.0799***) 0.9662 (3–14, 22–24)

*P <0.05;**P < 0.01; ***P <0.001

diversity is considerably lower than in most common and wide- differentiated by their numbers of chromosomes (40, 10 and spread species (e.g. Besold et al. 2008; Habel et al. 2009, 2010, 11, respectively; reviewed in Descimon and Mallet 2009). The 2011a, 2013;deKeyseretal.2012; Schmitt and Zimmermann three well-distinguished genetic groups within E. cassioides 2012; Louy et al. 2013) but higher than in rare or geographi- s.l. also should be considered as independent species. cally highly restricted taxa (e.g. Britten et al. 1994, 1995; Consequently, the western species E. arvernensis is distribut- Debinski 1994; Figurny-Puchalska et al. 2000; Gadeberg and ed in the Cordilliera Cantabrica, the Pyrenees, the Massif Boomsma 1997; Habel et al. 2011b;Diekeretal.2013). Central and the western Alps, where it reaches the western The level of genetic differentiation among the above de- parts of Switzerland. E. cassioides s.s. is found in the eastern fined six major taxa of the E. tyndarus group is in the order of Alps but also in the highest parts of the northern and central magnitude found among other Erebia and Satyrinae species Apennines. The populations in the Balkan high mountain sys- (Haubrich and Schmitt 2007; Schmitt and Besold 2010; Habel tems, but also in the Retezat (south-western Carpathians), rep- et al. 2011a; Schmitt et al. 2014). This supports E. ottomana, resent the species E. neleus. Comparing the genetic differen- E. tyndarus and E. nivalis as different species, which are also tiation between these six species against other allozyme data

Table 6 Results of five independent runs of a non-equilibrium 15), Erebia c. cassioides (16–24), Erebia cassioides neleus (25–30), Bayesian assessment of migration proportion by population calculated Erebia tyndarus (31–44), Erebia nivalis (45–46), Erebia ottomana (47– with the program BAYESASS (Wilson and Rannala 2003). Gene flow esti- 49). Values in columns (populations of origin in first column) represent mates calculated among six taxonomic groups (respective population migrant genes. Values >15 are marked in italics numbers are given in parenthesis): Erebia cassioides arvernensis (1–

Migrating to Coming from

arvernensis cassioides s.s. neleus tyndarus nivalis ottomana arvernensis 99.2292 0.1591 0.1415 0.1545 0.1597 0.1559 cassioides s.s. 3.0358 83.9035 3.3331 3.1080 3.2423 3.3773 neleus 3.0685 3.2908 83.5413 3.4867 3.1144 3.4983 tyndarus 0.0136 0.0158 0.0159 83.9992 0.0141 0.0143 nivalis 3.2133 3.1234 3.1459 3.2719 83.8344 3.4112 ottomana 0.2805 0.1206 0.1575 0.0727 0.1638 99.2050 802 T.Schmittetal.

Table 7 Results of five independent runs of a non-equilibrium Central (3), western Alps (4–14) and Passo Maghen (15); Erebia c. Bayesian assessment of migration proportion by population calculated cassioides from the eastern Alps (16–21) and Apennines (22–24), with the program BAYESASS (Wilson and Rannala 2003). Gene flow esti- Erebia cassioides neleus from the Balkans and Retezat (25–30). Values mates calculated among seven geographic groups within Erebia in columns (populations of origin in first column) represent migrant cassioides sensu lato (respective population numbers are given in paren- genes. Values >15 are marked in italics thesis): Erebia cassioides arvernensis from the Pyrenees (1–2), Massif

Migrating to Coming from

Pyrenees Massif Central Western Alps Passo Maghen Eastern Alps Apennines Balkans, Retezat

Pyrenees 83.3804 2.8209 2.6782 2.7666 2.8383 2.7584 2.7571 Massif Central 2.8883 82.7101 2.9542 2.8570 2.6544 2.8382 3.0978 Western Alps 0.0228 0.0228 99.8647 0.0208 0.0197 0.0253 0.0238 Passo Maghen 2.9707 2.8303 2.5657 83.2439 2.7072 2.8929 2.7893 Eastern Alps 2.0020 1.9346 19.6043 1.8899 70.7585 1.8363 1.9743 Apennines 2.7177 2.7086 2.8791 2.6239 3.0609 83.3090 2.7008 Balkans, Retezat 0.0340 0.0349 0.0311 0.0347 0.0343 0.0326 99.7984

(e.g. Haubrich and Schmitt 2007; Schmitt and Besold 2010; eastern E. ottomana group (with ottomana, graucasia, Habel et al. 2011a; Schmitt et al. 2014), an evolutionary age transcaucasica and iranica) and the western group including within the entire species group of more than 500 ky is plausi- all other taxa. These two groups are also well distinguished by ble, a reliable scenario as even the differentiation within the range of their chromosome numbers (eastern group, 40– E. arvernensis (between populations from Pyrenees and 52; western group, 8–25; review in Descimon and Mallet Alps) is sufficiently pronounced to accept at least a separation 2009). This split should have taken place around one million for one full glacial cycle. years ago (following the calibration of Albre et al. 2008)butin The general topology of the neighbour-joining any case before 500 ky BP (comparison with allozyme data of phenogram obtained from our allozyme data is similar other species groups, references see above). The much higher with allozyme data presented by Lattes et al. (1994), with age of this first split given by Peña et al. (2015)at7.57MyBP exception for the Apennines, and Martin et al. (2002), but is little likely as the necessary cold stages allowing the be- based on a much larger number of samples and a wider tween mountain dispersal pivotal for this radiation did not geographic scope. This may be the reason why Martin exist during Miocene, even not during the already cooler et al. (2002) could not distinguish E. tyndarus as strongly Pliocene. from the other species as we. Furthermore, due to smaller As the region of the Alps has the highest species diversity sample numbers and geographic coverage, Martin et al. within the entire tyndarus group (Albre et al. 2008; (2002) also could not detect the distinct genetic lineages Tshikolovets 2011), this region might represent the primary within E. cassioides s.l. as we did in our range-wide anal- centre of survival and speciation within the western group. If ysis. Data on mitochondrial DNA (mtDNA) was analysed accepting this, one might assume a cold stage distribution of by Martin et al. (2002) and Albre et al. (2008)forthis an ancestor of the entire E. tyndarus group including the re- species group as well as by Peña et al. (2015) investigat- gions of the Alps and the Balkan Peninsula with gene flow ing the entire genus Erebia. All three publications obtain- over this entire area in the early Pleistocene. This genetic ed similar results, but Albre et al. (2008) included consid- exchange was stopped by a vicariance event triggered by erably more samples and taxa of the E. tyndarus group follow-up temperature increases, restricting populations to than the other two so that we base our further discussion the Alps and the mountains of the Balkan Peninsula, resulting on this data set. in the onset of the evolution of the western and the eastern group, respectively. Refugia and range dynamics Range dynamics and differentiation in the eastern group Combining all available data allows a reconstruction of differ- entiation centres, the subsequent range dynamics and, even if Based on mtDNA data presented by Albre et al. (2008), the being hypothetical, an assignment of the respective events to eastern group must have expanded eastwards, and the specific glacial or interglacial periods. Based on mtDNA Caucasus must have been reached via the Anatolian mountain (Albre et al. 2008), a first split has taken place between the systems during a glacial range expansion. Vicariance during a Complex biogeography of a mountain butterfly 803 warm period (necessarily prior to the Günz ice age) resulted in subspecific status (E. ottomana tardenota Praviel, 1941) is the separation between a group evolving to E. ottomana (most debatable. The low level of differentiation all over the likely in the Balkan Peninsula) and a group being the ancestor Balkan Peninsula supports that E. ottomana was generally of all other species in the eastern group, with both groups widespread there even in lowlands during cold stages, at least considerably differing in their numbers of chromosomes (40 during the last ice age (Louy et al. 2014a). and 51–52, respectively; Descimon and Mallet 2009). If the sequence of basal splits presented by Peña et al. (2015)is Range dynamics and differentiation in the western group correct, this event might have happened even prior to the Balkans-Alps vicariance discussed above. The evolutionary history in the western main group is more The Caucasus region subsequently became a secondary complex than in the eastern E. ottomana group. If accepting centre of differentiation and speciation with a first split be- the calibration of the split between the four groups within the tween E. iranica and the E. graucasica group, maybe due to western main group (E. hispana, E. rondoui, E. callias and all geographic separation by the glaciated Caucasus mountains other taxa, i.e. E. cassioides s.l., E. tyndarus, E. nivalis, triggered by the onset of the Mindel ice age or earlier, as E. calcaria) at 375 ky BP (Albre et al. 2008) spreads out of supported by the data of Albre et al. (2008). In the the Alps in south-western and eastern direction during Günz is E. graucasica group, separation at the southern foot hills of a likely scenario. The south-western expansion reached as far the Great Caucasus and the more southern Transcaucasian south as the Sierra Nevada in southern Spain, the highest mountains by the onset of the Riss glaciation might have mountain range of the Spanish mainland. During the follow- resulted in the speciation into E. graucasica s.s. and ing interglacial, populations of this advance only survived in E. transcaucasica, fully being in concordance with the molec- the Sierra Nevada and the Pyrenees, remaining isolated in ular clock presented by Albre et al. (2008). these regions until today and evolving in geographic isolation The range dynamics within E. ottomana is remarkable. into the species E. hispania and E. rondoui, respectively. This After the evolution of this species, it must have colonised hypothesis is supported by considerably larger eyespots at the the Anatolian Alps in northern Turkey. However, as no fore-wings and a brighter orange of the submarginal band of Turkish material was included in any of the available studies, these two species (Tolman and Lewington 1997)aswellasa it is unknown when this expansion to Anatolia took place. It higher number of chromosomes (E. hispania,25;E. rondoui, must have happened during glacial conditions, but most likely 24) than all other taxa in the western group (8–16) (reviewed not before Günz; hence, expansion during Günz, Mindel, Riss in Descimon and Mallet 2009). or Würm is feasible, even with the possibility of cryptic spe- An eastern expansion gave birth to the ancestor of ciation in case of early range dynamics. As the species was E. callias, which is distinguished by its chromosome numbers described from Turkey, at least subspecific status of the (14–16) from all other members of the group (Descimon and Balkan populations (i.e. E. ottomana balcanica Rebel, 1903) Mallet 2009). As the interior differentiation within this species is likely. was dated at 150 ky BP (Albre et al. 2008), the Altai-Sayan Expansion out of the Balkan Peninsula in north-western region, today the most important region for this species, must direction must have taken place more than once. The Monte have been reached at least during Riss if not earlier. Due to the Baldo population (north-eastern Italy), described as shallow differentiation of the nominotypic E. callias in west- E. ottomana benacensis Dannehl, 1933, apparently represents ern North America (Albre et al. 2008), colonisation of the a relict of a glacial advance, maybe Riss, surviving only in this Rocky Mountains via the Bering land bridge, always existing region known for the survival of several relicts (Louy et al. during Pleistocene cold stages (Hewitt 2004), might have oc- 2013). Due to its considerable morphological (wing patterns, curred recently, maybe by the end of the Riss or even during male and female genitalia; Jutzeler et al. 2002; Tshikolovets the Würm glaciation. This species therefore is a rather new 2011) and genetic (Albre et al. 2008) differentiation as well as element of the North American fauna. These data on E. callias diverging chromosome numbers (Descimon and Mallet support the extraordinary range dynamic capacity within the 2009), this taxon even might represent a distinct species en- E. tyndarus species group in particular and in little mobile demic to the Monte Baldo (Albre et al. 2008). butterflies in general. The situation of the populations today isolated in the south- If the calibration by Albre et al. (2008) is correct, all taxa eastern French Massif Central is different as mtDNA (Albre constituting the forth western group should not be older than et al. 2008) and allozymes (Louy et al. 2014a) do not support a the onset of Mindel, which might have caused vicariance significant differentiation from Balkan populations. These da- events due to intensive glaciation of the Alps with different ta support a young (most likely Würm) expansion wave out of population groups surviving in different regions around these the Balkan Peninsula reaching as far westwards as the Massif mountains (cf. Schönswetter et al. 2005;Schmitt2009). Central, where the species survived as a relict of the last ice However, in case of differentiation at the species rank, the age till today. Whether this short isolation is justifying onset of vicariance should not be younger than Riss because, 804 T.Schmittetal. in the large majority of cases, one single glacial cycle appar- also Coenonympha rhodopensis widespread all over the ently is insufficient to evolve separate butterfly species by Balkan mountains (Louy et al. 2013). allopatric distribution and thus independent random evolution If looking at the STRUCTURE analysis for E. neleus alone, at (e.g. Schmitt et al. 2005; Habel et al. 2009, 2011a). least four core areas in south-eastern Europe have to be pos- A plausible scenario therefore is that the species more tulated during Riss and Würm: western Balkans (represented strongly differentiated from all others (i.e. E. tyndarus and by Shar Planina), Rila, Pirin and Retezat. The grouping of E. nivalis; for the latter species also see Lattes et al. (1994) Stara Planinia together with Shar Planina, both geographically and Martin et al. (2002)) and an ancestor of the four remaining far away from each other, most likely has to be interpreted as species E. cassioides s.s., E. arvernensis, E. neleus and an artefact due to the small sample size analysed from Stara E. calcaria survived Mindel in three different refugia. This Planina. Hence, Stara Planina most likely represents a fifth division into three groups is also supported by the morphology core area. If looking at the sequence of splits among regional of the valvae, wing profiles and wing patterning (Lorković groups, Rila and Pirin are only separated at K =4.Hence,the 1953, 1957). Note that the chromosome numbers of populations in these neighbouring mountains might not be E. calcaria (8) and E. nivalis (11) differ from all other species separated since Riss, but only since Würm. of this group (10) (Lorković 1958). Likely refuge areas of The remarkable differentiation within E. neleus might jus- these three groups might have been located north to north- tify the subspecific status assigned to various regional popu- west (E. tyndarus), north-east to east (E. nivalis) and south lation groups (cf. Varga 1998, 2014; Albre et al. 2008). At to south-east of the Alps’ glaciers (E. cassioides s.l.). The least three different morphology-based subspecies should be low genetic diversities in E. tyndarus support its northern accepted: E. neleus illyromacedonica Lorković,1953inthe survival during Mindel as strong genetic bottlenecks in these western Balkan mountains; E. neleus macedonica Buresch, less suitable refugia north of the Alps are more frequent than 1919 in Rila and Pirin; and the nominotypic E. neleus in in refugia east and south of these mountains (cf. Haubrich and Retezat. The subspecific status of Stara Planina populations Schmitt 2007). needs further investigation. During the transition from Mindel to the subsequent Coming back to the evolutionary processes within the Holstein interglacial, these three population groups all must Alps, the continuous distribution of the southern Alps group have shifted their distributions from the foot hills to high alti- was most likely disrupted into three refugial groups during the tudes of the Alps. As demonstrated by Schmitt et al. (2006), Holstein-Riss transition: one in the vicinity of the south- Haubrich and Schmitt (2007) and Alvarez et al. (2012), western Alps evolving into E. arvernensis, one in the vicinity refugia north of the Alps have mostly been less successful in of the eastern Alps evolving into E. cassioides s.s. and most their interglacial colonisation of high mountain areas than likely a third one in the unglaciated parts of Slovenia those south of the Alps. This most likely also applies to the representing the ancestor of E. calcaria (data on the latter E. tyndarus group. While the southern to south-eastern group species refer to Lattes et al. (1994), Martin et al. (2002)and (E. cassioides s.l.) colonised over most of the southern Alps, Albre et al. (2008)). The ancestors of modern E. tyndarus and hereby also reaching the western Alps, the two northern E. nivalis again survived this ice age in their previous refugia groups (i.e. the ancestors of E. tyndarus and E. nivalis)might north of the Alps’ glaciers. have colonised more restricted regions of the northern Alps, During Riss-Eem transition, the western E. arvernensis hence not translocating far away from their glacial retreats. retreated to the western Alps, but also to the Pyrenees. Since By the end of Mindel, the refugial group from the southern then, gene flow between these two groups has been disrup- to south-eastern Alps foot hills apparently not only retreated ted thus explaining the genetic differentiation between into the Alps, but via the mountains of Slovenia and Croatia E. arvernensis from the Alps and the Pyrenees. also to the higher mountain ranges of the Balkan Peninsula. The genetically strongly impoverished Apennines popula- Here, it started evolving into E. neleus. However, this taxon tions, which otherwise are similar with eastern Alps only could spread widely over the Balkan Peninsula and to the E. cassioides s.s., might have split from these also during the Retezat during Riss. Since then, these populations apparently Riss-Eem transition. The subspecies E. cassioides majellana remained trapped in their mountain refugia where surviving Fruhstorfer, 1909 might have evolved in situ since then. the Eem interglacial (between Riss and Würm) but also the Therefore, Eem survival in just one rather restricted area of Würm. This might be caused by the not that cold temperatures the Apennines (most likely the highest parts of the central as in the previous glaciations, competition with the wide- Apennines) might be responsible for the uniform genetic im- spread E. ottomana (cf. Louy et al. 2014a) or even by a com- poverishment all over the Apennines today (with locations in bination of both. This trapping within different Balkan high the northern and central Apennines analysed here). mountain systems since the Eem might well explain the con- The E. cassioides s.s. population in the eastern Alps appar- siderably stronger genetic differentiation within E. neleus if ently have not derived from one single perialpine Würm refu- compared with the other species analysed in this study, but gium because STRUCTURE analysis outlining the Apennines Complex biogeography of a mountain butterfly 805 populations as one group also distinguished two groups in the from the Alpes Maritimes from all others. They might have eastern Alps. These most likely derived from one refugium survived Würm at the unglaciated south-western tip of the north and another one south-east of the glaciated region. Alps, a typical glacial refuge region for Alpine species Postglacially, as often (Schmitt 2009), the southern group was (Schönswetter et al. 2005). If accepting subspecific differenti- more successful in the recolonisation of high mountain ranges. ationofthisgroup,theyshouldbeincludedinE. arvernensis Consequently, of the six E. cassioides s.s. populations analysed murina Reverdin, 1909. The other E. arvernensis populations here, only the most northern one (i.e. Patscherkofel, 17) exclu- from the western Alps show no major interior genetic struc- sively derived from the northern refugium. Nevertheless, both turing. Hence, they should be derived from a single Würm proveniences have mixed by secondary contact in some regions refuge located either west or east of the glaciated western as in the central eastern Alps (e.g. Rein in Taufers, 18), hereby Alps and might be included in the subspecies E. arvernensis clearly demonstrating their conspecific status. carmenta Fruhstorfer, 1909. During Eem, the geographic distribution of most taxa of the The population collected at Passo Maghen (eastern Italian E. tyndarus group might have been mostly similar to today’s Alps) genetically also belongs to E. arvernensis. Apparently, a settings. This implies that E. tyndarus s.s. should have been relict of a formerly more extended distribution of this taxon has more successful in its colonisation of major parts of the central survived the last ice age in a small refuge area east of the Monte Alps than during the previous interglacial. The differentiation Baldo region; however, its current distribution in the eastern of E. tyndarus into a geographically restricted eastern (Fig. 3; Alps is unknown. Furthermore, the STRUCTURE analysis includ- populations 41 to 44, all in the catchment area of the Adige ing the closely related species E. cassioides, E. arvernensis and river) and a more widespread western group calls for more E. neleus (Fig. 3) supports the idea of local co-occurrence of than one Würm glacial core area. These might have been E. arvernensis and E. cassioides, with the latter being less located south of the Alps during the Würm (and not north of common than the former, in the same habitat (i.e. syntopy) at the Alps as most likely during previous ice ages) with the Passo Maghen without major hybridisation. This strongly sup- western one maybe in the region of the western upper Italian ports the species status of E. arvernensis. Furthermore, the two lakes and the eastern one in the Lake Garda region. The latter eye spots of the fore-wing of E. arvernensis are bigger and might have used the Adige valley as postglacial colonisation often fuse what is much less common in E. cassioides s.s. corridor into the Alps, a corridor frequently used by Alpine (own observations). Furthermore, the height index of the valvae species (Cupedo 2010). on average is larger in E. cassioides s.s. than in E. arvernensis Würm refugia of E. tyndarus south of the Alps also might (Lorković 1953). explain the extant disjunct distribution of E. nivalis in the The Massif Central data, from where E. arvernensis is de- highest parts of the eastern Alps and the Grindelwald region. scribed, do not allow a final decision of its biogeographic This might be due either to two different core areas north of history. Two mutually exclusive scenarios are possible. The theAlpsduringWürm(cf.alsodataonRumex nivalis,Stehlik topology of the neighbour-joining phenogram and the 2002) or just one that in the West exclusively retreated to the STRUCTURE analysis for K = 7 support an independent Würm Grindelwald region, a phenomenon also found in other Alpine refugium, as repeatedly observed for other animal and plant species (Schmitt et al. 2006; Alvarez et al. 2012). However, as species (Pauls et al. 2006; Triponez et al. 2011; Kropf et al. no genetic data of E. nivalis from the Grindelwald region is 2012). However, the STRUCTURE analysis for K = 3 gives evi- available, this question needs further investigation. dence for a postglacial colonisation of the Massif Central The considerable differentiation within E. arvernensis re- mainly from the western Alps, but also from the Pyrenees with vealed by neighbour-joining and STRUCTURE analysis (Fig. 3) in situ admixture of both genetic groups. Close phylogeo- calls for more than just two Würm refugia (in geographic graphic relatedness of populations from the Massif Central vicinity to Pyrenees and Alps). The genetic differentiation with such of the Alps or Pyrenees have frequently been re- between the two populations from the western and eastern corded (Descimon 1995; Ronikier et al. 2008; Kramp et al. Pyrenees might indicate more than one centre of survival in 2009; Triponez et al. 2011; Schmitt et al. 2014). Therefore, the vicinity of these mountains as observed for a number of further analyses are necessary to clarify the biogeographic other species (Kropf et al. 2002, 2003; Schmitt et al. 2006; history of this region for E. arvernensis. Lihová et al. 2008; Lauga et al. 2009). Based on morpholog- ical differences, E. arvernensis pseudomurina de Lesse, 1952 Acknowledgments This study was supported by the Konrad Adenauer was described from the western Pyrenees and E. arvernensis Foundation (PhD fellowship to DL). We acknowledge sampling permits pseudocarmenta de Lesse, 1956 from the eastern Pyrenees, to collect the samples, given from the respective institutions where nec- thus supporting the genetic findings presented here. essary. We thank many colleagues for field assistance, especially László The geographic pattern of differentiation in the Alps even Rákosy (Cluj), Zoltán Varga (Debrecen) and Matthias Weitzel (Trier) as well as Leonardo Dapporto (Barcelona) and Zoltán Varga (Debrecen) for implies at least three Würm core areas. STRUCTURE analysis important constructive suggestions which helped improve the manuscript (Fig. 3) distinguishes well the three southernmost populations considerably. 806 T.Schmittetal.

References 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. Albre, J., Gers, C., & Legal, O. L. (2008). Molecular phylogeny of the Felsenstein, J. (1993). PHYLIP (Phylogeny inference package) ver. 3.5.c. Erebia tyndarus (Lepidoptera, Rhopalocera, Nymphalidae, Seattle: Department of Genetics, University of Washington. Satyrinae) species group combining Cox II and ND5 mitochondrial Figurny-Puchalska, E., Gadeberg, R. M. E., & Boomsma, J. J. (2000). genes: a case study of a recent radiation. Molecular Phylogenetics Comparison of genetic population structure of the large blue butterfly – and Evolution, 47,196 210. Maculinea nausithous and M. teleius. Biodiversity and Conservation, Alvarez, N., Manel, S., Schmitt, T., & the IntraBioDiv Consortium. 9, 419–432. (2012). Contrasting diffusion of Quaternary gene pools across Europe: Gadeberg, R. M. E., & Boomsma, J. J. (1997). Genetic population struc- the case of the arctic–alpine Gentiana nivalis L. (Gentianaceae). Flora, ture of the large blue butterfly Maculinea alcon in Denmark. Journal 7,408–413. of Insect Conservation, 1,99–111. Avise, J. C., Arnold, J., Ball, R. M., Bermingham, E., Lamb, T., Neigel, J. Goudet, J. (1995). FSTAT (version 1.2): a computer program to calculate E., et al. (1987). Intraspecific phylogeography: the mitochondrial F-statistics. Heredity, 86,485–486. DNA bridge between population genetics and systematics. Annual Habel, J. C., Dieker, P., & Schmitt, T. (2009). Biogeographical connec- Review of Ecology, Evolution, and Systematics, 18,489–522. tions between the Maghreb and the Mediterranean peninsulas of Besold, J., Schmitt, T., Tammaru, T., & Cassel-Lundhagen, A. (2008). southern Europe. Biological Journal of the Linnean Society, 98, Strong genetic impoverishment from the centre of distribution in 693–703. southern Europe to peripheral Baltic and isolated Scandinavian pop- Habel, J. C., Lens, L., Rödder, D., & Schmitt, T. (2011a). From Africa to ulations of the pearly heath butterfly. Journal of Biogeography, 35, Europe and back: refugia and range shifts cause high genetic differ- 2090–2101. entiation in the Marbled White butterfly Melanargia galathea. BMC Britten, H. B., Brussard, P. F., Murphy, D. D., & Austin, G. T. (1994). Evolutionary Biology, 11,215. Colony isolation and isozyme variability of the western seep fritil- Habel, J. C., Rödder, D., Lens, L., & Schmitt, T. (2013). The genetic lary, Speyeria nokomis apacheana (Nymphalidae), in the western signature of ecologically different grassland lepidopterans. Great Basin. Great Basin Naturalist, 54,97–105. Biodiversity and Conservation, 22,2401–2411. Britten, H. B., Brussard, P. F., Murphy, D. D., & Ehrlich, P. R. (1995). A Habel, J. C., Rödder, D., Schmitt, T., & Nève, G. (2011b). Global test for isolation-by-distance in Central Rocky Mountain and Great warming will affect genetic diversity and uniqueness of Lycaena Basin populations of Edith’s Checkerspot Butterfly (Euphydryas helle populations. Global Change Biology, 17,194–205. editha). JournalofHeredity,86,204–210. Habel, J. C., Rödder, D., Stefano, S., Meyer, M., & Schmitt, T. (2010). Byrne, M. (2008). Evidence for multiple refugia at different time scales Strong genetic cohesiveness between Italy and North Africa in four during Pleistocene climatic oscillations in southern Australia in- butterfly species. Biological Journal of the Linnean Society, 99, ferred from phylogeography. Quaternary Science Reviews, 27, 818–830. 2576–2585. Haubrich, K., & Schmitt, T. (2007). Cryptic differentiation in alpine-en- Charrier, O., Dupont, P., Pornon, A., & Escaravage, N. (2014). demic, high-altitude butterflies reveals down-slope glacial refugia. – Microsatellite marker analysis reveals the complex phylogeographic Molecular Ecology, 16,3643 3658. history of Rhododendron ferrugineum (Ericaceae) in the Pyrenees. Hausdorf, B., & Hennig, C. (2010). Species delimitation using dominant – PLoS ONE, 9, e92976. and codominant multilocus markers. Systematic Biology, 59,491 Cupedo, F. (2010). Geographische Variabilität und spätglaziale 503. Einwanderungswege von Erebia pluto (de Prunner, 1798) in der Hebert, P. D. N., & Beaton, M. J. (1993). Methodologies for allozyme Ortlergruppe und den Ötztaler Alpen (Nymphalidae). Nota analysis using cellulose acetate electrophoresis. Beaumont: Helena lepidopterologica, 26,137–152. Laboratories. Hewitt, G. M. (1999). Postglacial recolonization of European biota. De Keyser, R., Shreeve, T. G., Breuker, C. J., Hails, R. S., & Schmitt, T. – (2012). Polyommatus icarus butterflies in the British Isles: evidence Biological Journal of the Linnean Society, 68,87 112. for a bottleneck. Biological Journal of the Linnean Society, 107, Hewitt, G. M. (2004). Genetic consequences of climatic oscillation in the 123–136. Quaternary. Philosophical Transactions of the Royal Society of London B, 359,183–195. De Lattin, G. (1949). Beiträge zur Zoogeographie des Mittelmeergebietes Holderegger, R., Stehlik, I., & Abbott, R. J. (2002). Molecular analysis of (pp. 143–151). Kiel: Verhandlungen der deutschen Zoologischen the Pleistocene history of Saxifraga oppositifolia in the Alps. Gesellschaft. Molecular Ecology, 11,1409–1418. De Lattin, G. (1964). Die Verbreitung des sibirischen Faunenelements in Hubisz, M. J., Falush, D., Stephens, M., & Pritchard, J. K. (2009). der Westpaläarktis. Natur und Museum, 94 ,104–125. Inferring weak population structure with the assistance of sample De Lattin, G. (1967). Grundriß der Zoogeographie. Jena: Gustav Fischer. group information. Molecular Ecology Resources, 9,1322–1332. Debinski, D. M. (1994). Genetic diversity assessment in a metapopulation Husemann, M., Schmitt, T., Zachos, F. E., Ulrich, W., & Habel, J. C. – of the butterfly Euphydryas gillettii. Heredity, 70,25 30. (2014). Palaearctic biogeography revisited: evidence for the exis- Descimon, H. (1995). La conservation des Parnassius en France: aspects tence of a North African refugium for western Palaearctic biota. zoogéographiques, écologiques, démographiques et génétiques. Journal of Biogeography, 41,81–94. OPIE, 1,1–54. Jutzeler, D., Leestmans, R., Daydé, S., Lafranchis, T., Sala, G., & Volpe, Descimon, H., & Mallet, J. (2009). Bad species. In J. Settele, T. Shreeve, G. (2002). Comparaison de deux sous-espèces d’Erebia ottomana M. Konvička, & H. Van Dyck (Eds.), Ecology of butterflies in HERRICH-SCHÄFFER (1847): la ssp. tardenota PRAVIEL (1941) du Europe (pp. 219–249). Cambridge: Cambridge University Press. sud-est du Massif central (France) et de la ssp. benacensis Dieker, P., Drees, C., Schmitt, T., & Assmann, T. (2013). Low genetic DANNEHL (1933) du Mt Maldo (Italie) (Lepidoptera: Nymphalidae, diversity of a high mountain burnet moth species in the Pyrenees. Satyrinae). Linneana Belgica, 18,377–390. Conservation Genetics, 14,231–236. Kramp, K., Huck, S., Niketić,M.,Tomović, G., & Schmitt, T. (2009). Evanno, G., Regnaut, S., & Goudet, J. (2005). Detecting the number of Multiple glacial refugia and complex postglacial range shifts of the clusters of individuals using the software STRUCTURE: a simula- obligatory woodland plant Polygonatum verticillatum tion study. Molecular Ecology, 14,2611–2620. (Convallariaceae). Plant Biology, 11,392–404. Complex biogeography of a mountain butterfly 807

Kropf, M., Comes, H. P., & Kadereit, J. W. (2012). Past, present and Pauls, S. U., Lumbsch, H. T., & Haase, P. (2006). Phylogeography of the future of mountain species of the French Massif Central—the case montane caddisfly Drusus discolor: evidence for multiple refugia of Soldanella alpina L. subsp. alpina (Primulaceae) and a review of and periglacial survival. Molecular Ecology, 15,2153–2169. other plant and animal studies. Journal of Biogeography, 39,799– Peña, C., Witthauer, H., Klečková, I., Fric, Z., & Wahlberg, N. (2015). 812. Adaptive radiations in butterflies: evolutionary history of the genus Kropf, M., Kadereit, J. W., & Comes, H. P. (2002). Late Quaternary Erebia (Nymphalidae: Satyrinae). Biological Journal of the Linnean distributional stasis in the submediterranean mountain plant Society, 116,449–467. Anthyllis montana L. (Fabaceae) inferred from ITS sequences and Popescu-Gorj, A. (1962). Revision des especes du genre Erebia Dalm. amplified fragment length polymorphism markers. Molecular des Carpathes de la Roumanie (groupes pluto et tyndarus). Traveaux Ecology, 11,447–463. du Muséum National d’Histoire Naturelle BGrigore Antipa^,3, Kropf, M., Kadereit, J. W., & Comes, H. P. (2003). Differential cycles of 205–223. range contraction and expansion in European high mountain plants Pritchard, J. K., Stephens, M., & Donnelly, P. (2000). Inference of pop- during the Late Quaternary: insights from Pritzelago alpina (L.) O. ulation structure using multilocus genotype data. Genetics, 155, Kuntze (Brassicaceae). Molecular Ecology, 12,931–949. 945–955. Lattes, A., Mensi, P., Cassulo, L., & Baletto, E. (1994). Genetic variability Reinig, W. F. (1937). Die Holarktis. Jena: Gustav Fischer. in western European members of the Erebia tyndarus species group Reinig, W.F. (1950). Chorologische Voraussetzungen für die Analyse von – (Lepidoptera, Satyridae). Nota lepidopterologica Supplement, 5,93 Formenkreisen. Syllegomena Biologica, Festschrift für O. 104. Kleinschmidt, 346–378. Lauga, B., Malaval, S., Largier, G., & Regnault-Roger, C. (2009). Two Richardson, B. J., Baverstock, P. R., & Adams, M. (1986). Allozyme lineages of Trifolium alpinum (Fabaceae) in the Pyrenees: evidence electrophoresis: a handbook for animal systematics and population from random amplified polymorphic DNA (RAPD) markers. Acta studies. Sydney: Academic. – Botanica Gallica, 156,317 330. Ronikier, M., Costa, A., Fuertes Aguilar, J., Feliner, G. N., Küpfer, P., & Lerault, P. J. A. (1997). Liste des lépidoptères de France, Belgique et Mirek, Z. (2008). Phylogeography of Pulsatilla vernalis (L.) Mill. Corse (2e edn). Paris: Supplement Alexanor. (Ranunculaceae): chloroplast DNA reveals two evolutionary lineages Lihová, J., Carlsen, T., Brochmann, C., & Marhold, K. (2008). across central Europe and Scandinavia. Journal of Biogeography, 35, Contrasting phylogeographies inferred for the two alpine sister spe- 1650–1664. cies Cardamine resedifolia and C. alpina (Brassicaceae). Journal of Schmitt, T. (2007). Molecular biogeography of Europe: Pleistocene cy- Biogeography, 36,104–120. cles and postglacial trends. Frontiers in Zoology, 4,11. Lorenzen, E. D., Heller, R., & Siegismund, H. R. (2012). Comparative Schmitt, T. (2009). Biogeographical and evolutionary importance of the phylogeography of African savannah ungulates. Molecular Ecology, European high mountain systems. Frontiers in Zoology, 6,9. 21, 3656–3670. Schmitt, T., & Besold, J. (2010). Upslope movements and large scale Lorković, Z. (1953). Spezifische, semispezifische und rasische expansions: the taxonomy and biogeography of the Coenonympha Differenzierung bei Erebia tyndarus Esp. II. Differenzierungsgrad arcania - C. darwiniana - C. gardetta butterfly species complex. und verwandtschaftliche Verhältnisse der europäischen Formen von Zoological Journal of the Linnean Society, 159,890–904. Erebia tyndarus Esp. Bulletin International de l’Académie Yougoslave des Sciences et des Beaux-Artes, 10,193–224. Schmitt, T., & Haubrich, K. (2008). The genetic structure of the mountain Lorković, Z. (1957). Die Speziationsstufen in der Erebia tyndarus forest butterfly Erebia euryale unravels the late Pleistocene and Gruppe. I. Die morphologischen, ökologischen und chorologischen postglacial history of the mountain coniferous forest biome in – Merkmale der alpinen Formen cassioides, nivalis, tyndarus und Europe. Molecular Ecology, 17,2194 2207. calcarius. Biološki Glasnik, 10,61–110. Schmitt, T., & Seitz, A. (2001). Intraspecific allozymatic differentiation Lorković, Z. (1958). Exchange of genetic material: mechanisms and con- reveals the glacial refugia and the postglacial expansions of sequences. Cold Spring Harbour Symposia on Quantitative Biology, European Erebia medusa (Lepidoptera: Nymphalidae). Biological – 23,319–325. Journal of the Linnean Society, 74,429 458. Louy, D., Habel, J. C., Abadjev, S., Rakosy, L., Varga, Z., Rödder, D., et Schmitt, T., & Zimmermann, M. (2012). To hybridize or not to hybridize: al. (2014a). Molecules and models indicate diverging evolutionary what separates two genetic lineages of the Chalk-hill Blue effects from parallel altitudinal range shifts in two mountain Ringlet Polyommatus coridon (Lycaenidae, Lepidoptera) along their sec- butterflies. Biological Journal of the Linnean Society, 112,569–583. ondary contact zone throughout eastern Central Europe? Journal – Louy, D., Habel, J. C., Abadjiev, S., & Schmitt, T. (2013). Genetic legacy of Zoological Systematics and Evolutionary Research, 50,106 115. from past panmixia: high genetic variability and low differentiation Schmitt, T., & Varga, Z. (2012). Extra-Mediterranean refugia: the rule and in disjunct populations of the Eastern Large Heath butterfly. not the exception? Frontiers in Zoology, 9,22. Biological Journal of the Linnean Society, 110,281–290. Schmitt, T., Habel, J. C., Rödder, D., & Louy, D. (2014). Effects of recent Louy, D., Habel, J. C., Ulrich, W., & Schmitt, T. (2014b). Out of the Alps: and past climatic shifts on the genetic structure of the high mountain the biogeography of a disjunctly distributed mountain butterfly, the Yellow-spotted ringlet butterfly Erebia manto (Lepidoptera, Almond eyed ringlet Erebia alberganus (Lepidoptera, Satyrinae). Satyrinae): a conservation problem. Global Change Biology, 20, Journal of Heredity, 105,28–38. 2045–2061. Martin, J.-F., Gilles, A., Lörtscher, M., & Descimon, H. (2002). Schmitt, T., Hewitt, G. M., & Müller, P. (2006). Disjunct distributions Phylogenetics and differentiation among the western taxa of the during glacial and interglacial periods in mountain butterflies: Erebia tyndarus group (Lepidoptera: Nymphalidae). Biological Erebia epiphron as an example. Journal of Evolutionary Biology, Journal of the Linnean Society, 75,319–332. 19,108–113. Mráz, P., Gaudeul, M., Rioux, D., Gielly, L., Choler, P., & Taberlet, P. Schmitt, T., Rákosy, L., Abadjiev, S., & Müller, P. (2007). Multiple dif- (2007). Genetic structure of Hypochaeris uniflora (Asteraceae)sug- ferentiation centres of a non-Mediterranean butterfly species in gests vicariance in the Carpathians and rapid post-glacial coloniza- south-eastern Europe. Journal of Biogeography, 34,939–950. tion of the Alps from an eastern Alpine refugium. Journal of Schmitt, T., Varga, Z., & Seitz, A. (2005). Are Polyommatus hispana and Biogeography, 34, 2100–2114. Polyommatus slovacus bivoltine Polyommatus coridon Nei, M. (1972). Genetic distances between populations. The American (Lepidoptera: Lycaenidae)? The discriminatory value of genetics Naturalist, 106,283–291. in taxonomy. Organisms, Diversity and Evolution, 5,297–307. 808 T.Schmittetal.

Schönswetter, P., Stehlik, I., Holderegger, R., & Tribsch, A. (2005). Triponez, Y., Buerki, S., Borer, M., Naisbit, R. E., Rahier, M., & Alvarez, Molecular evidence for glacial refugia of mountain plants in the N. (2011). Discordances between phylogenetic and morphological European Alps. Molecular Ecology, 14,3547–3555. patterns in alpine leaf beetles attest to an intricate biogeographic Shafer, A. B. A., Cullingham, C. I., Côté, S. D., & Coltman, D. W. (2010). history of lineages in postglacial Europe. Molecular Ecology, 20, Of glaciers and refugia: a decade of study sheds new light on the 2442–2463. phylogeography of northwestern North America. Molecular Tshikolovets, V. (2011). Butterflies of Europe and the Mediterranean Ecology, 19,4589–4621. area. Pardubice: Vadim Tshikolovets. Siegismund, H. R. (1993). G-Stat, ver. 3, genetical statistical programs Ujvárosi, L., Bálint, M., Schmitt, T., Mészáros, N., Ujvárosi, T., & for the analysis of population data. Denmark: The Arboretum, Popescu, O. (2010). Divergence and speciation in the Carpathians Royal Veterinary and Agricultural University Horsholm. area: patterns of morphological and genetic diversity of the crane fly Soltis, D. E., Morris, A. B., Lachlan, J. S. M., Manos, P. S., & Soltis, P. S. Pedicia occulta (Diptera: Pediciidae). Journal of North American (2006). Comparative phylogeography of unglaciated eastern North Benthological Society, 29,1075–1088. America. Molecular Ecology, 15,4261–4293. Varga, Z. (1998). Die Erebien der Balkanhalbinsel und Karpaten IV. Sonderegger, P. (2005). Die Erebien der Schweiz. Biel/Bienne: Selbstverlag. Übersicht der subspezifischen Gliederung und der verbreitung der Stehlik, I. (2002). Glacial history of the alpine herb Rumex nivalis Erebia-Arten in der Balkanhalbinsel und in den Karpaten (Polygonaceae): a comparison of common phylogeographic methods (Lepidoptera, Nymphalidae, Satyrinae). Entomologia Romania, 3, with nested clade analysis. American Journal of Botany, 89,2007–2016. 15–29. Stehlik, I., Blattner, F. R., Holderegger, R., & Bachmann, K. (2002). Varga, Z. (2014). Biogeography of the high mountain Lepidoptera in the Nunatak survival of the high Alpine plant Eritrichium nanum (L.) Balkan Peninsula. Ecologica Montenegrina, 1,140–168. Gaudin in the central Alps during the ice ages. Molecular Ecology, Varga, Z., & Schmitt, T. (2008). Types of oreal and oreotundral disjunc- 11,2027–2036. tions in the western Palearctic. Biological Journal of the Linnean Stehlik, I., Schneller, J. J., & Bachmann, K. (2001). Resistance or emi- Society, 93,415–430. gration: response of the high-alpine plant Eritrichium nanum (L.) Vila, M., Marí-Mena, N., Guerrero, A., & Schmitt, T. (2011). Some but- Gaudin to the ice age within the Central Alps. Molecular Ecology, terflies do not care much about topography: a single genetic lineage 10,357–370. of Erebia euryale (Nymphalidae) along the northern Iberian moun- Taberlet, P., Fumagalli, L., Wust-Saucy, A.-G., & Cosson, J.-F. (1998). tains. Journal of Zoological Systematics and Evolutionary Comparative phylogeography and postglacial colonization routes in Research, 49,119–132. Europe. Molecular Ecology, 7,453–464. Wilson,G.A.,&Rannala,B.(2003).Bayesianinferenceofrecent Theissinger, K., Bálint, M., Feldheim, K. A., Haase, P., Johannesen, J., Laube, migration rates using multilocus genotypes. Genetics, 163, I., et al. (2012). Glacial survival and post-glacial recolonization of an 1177–1191. arctic–alpine freshwater insect (Arcynopteryx dichroa, Plecoptera, Wood, D. A., Vandergast, A. G., Barr, K. R., Inman, R. D., Esque, T. C., Perlodidae) in Europe. Journal of Biogeography, 40,236–248. Nussear, K. E., et al. (2013). Comparative phylogeography reveals Tolman, T., & Lewington, R. (1997). Field guide butterflies of Britain deep lineages and regional evolutionary hotspots in the Mojave and and Europe. London: Harper Collins. Sonoran desert. Diversity and Distributions, 19,722–737.