Some butterflies do not care much about topography: a ... · the Iberian Peninsula. Most of them...

1Department of Molecular and Cell Biology, Faculty of Sciences, University of Coruna, A Coruna, Spain; 2Department ofPhysiology, Faculty of Medicine, University of Santiago de Compostela, Santiago de Compostela, Spain; 3Department of

Biogeography, Trier University, Trier, Germany

Some butterflies do not care much about topography: a single genetic lineage ofErebia euryale (Nymphalidae) along the northern Iberian mountains

Marta Vila1, Neus Marı-Mena

1, Ana Guerrero2 and Thomas Schmitt

3

AbstractThe phylogeography of montane species often reveals strong genetic differentiation among mountain ranges. Both classic morphological andgenetic studies have indicated distinctiveness of Pyrenean populations of the butterfly Erebia euryale. This hypothesis remained inconclusive untildata from the westernmost populations of the distribution area (Cantabrian Mountains) were analysed. In the present study, we set out todescribe the population structure of Erebia euryale in western Cantabria, where the species occurs in scattered localities. For this goal, we estimatethe genetic diversity and differentiation found in 218 individuals from six Cantabrian (North Spain) localities genotyped by 17 allozyme loci. Wealso sequence 816 bp of the cytochrome oxidase subunit I mitochondrial gene in 49 individuals from Cantabrian localities and 41 specimens fromfive other European sites. Mitochondrial data support the recognition of four major genetic groups previously suggested for the Europeanpopulations based on allozyme polymorphisms. Both mitochondrial and nuclear markers reveal genetic distinctiveness of a single Pyrenean–Cantabrian lineage of E. euryale. The lack of geographical structure and the star-like topology displayed by the mitochondrial haplotypesindicate a pattern of demographic expansion in northern Iberia, probably related to Upper Pleistocene climatic oscillations. By contrast, withinthe Pyrenean–Cantabrian lineage, Cantabrian samples are genetically structured in nuclear datasets. In particular, San Isidro is significantlydifferentiated from the other five populations, which cluster into two groups. We recognize an evolutionary significant unit for Pyrenean–Cantabrian populations of Erebia euryale. Our results also illustrate that the evolutionary history of a species may be shaped by processesundetectable by using mtDNA alone.

Key words: allozymes – phylogeography – population genetics – Lepidoptera – evolutionarily significant unit – mtDNA – cytochrome oxidase I(COI)

Introduction

The evolutionary history of many Palearctic taxa is undoubt-edly linked to the environmental changes that occurred duringQuaternary climatic oscillations (Hewitt 1996). The so-called

Iberian refuge hosted several refugial areas that likely varied inarea and location during the different glacial stages (Gomezand Lunt 2007). Some of them were used by populations thatsurmounted the Pyrenees and expanded northwards when

climatic conditions were more suitable, such as those of thezygaenid moth Aglaope infausta (Linnaeus, 1767) (Schmitt andSeitz 2004) or the southern water vole Arvicola sapidus Miller,

1908 (Centeno-Cuadros et al. 2009). In contrast, other areasdid not contribute to the current genetic pattern of populationsoutside the Iberian Peninsula, e.g. the grasshopper Chorthippus

parallelus (Zetterstedt, 1821) (Cooper et al. 1995) or the Scotspine Pinus sylvestris L. (Naydenov et al. 2007; Pyhajarvi et al.2008) and so the areas in question might therefore beconsidered as �relicts� rather than refuges (Petit et al. 2005),

reflecting a model with lack of range movement.To the best of our knowledge, the phylogeography of

terrestrial taxa currently occurring along the northern Iberian

mountain ranges (Cantabrian Mountains and Pyrenees) doesnot show such a general genetic pattern. Some investigationshave indicated that the mountains of northern Spain currently

host glacial relicts, e.g. the Convallariaceae plant Polygonatumverticillatum (L.) (Kramp et al. 2009). This is also the case forthe chamois Rupicapra pyrenaica (Linnaeus, 1758) (Perez et al.

2002) and the caddisfly Drusus discolor (Rambur, 1842) (Pauls

et al. 2006). However, some other surveys have revealed acommon lineage present all along the northern part of Spainand south-eastern France, e.g. the bank vole Myodes glareolus

(Schreber, 1780) (Deffontaine et al. 2009). In addition, boththe Pyrenean and Cantabrian Mountains currently showcontact zones for some previously separated lineages, such asthe capercaillie, Tetrao urogallus Linnaeus, 1758, in the

Pyrenees (Rodrıguez-Munoz et al. 2007) and the fire salaman-der, Salamandra salamandra (Linnaeus, 1758), in the Canta-brian Mountains (Garcıa-Parıs et al. 2003). These studies

comprise taxonomically and ecologically diverse species. Thelack of a common pattern in their genetic structure may reflectnot only different evolutionary histories but also differences in

life history traits (e.g. dispersal ability, susceptibility toextreme temperatures) that may have shaped their phylogeo-graphical pattern. The study of closely related and recentlydiverged species has proved extremely useful for incorporating

the role of the evolutionary constraints upon which intraspe-cific processes act in shaping the genetic structure of popula-tions. In fact, several investigations have already showed that

closely related taxa may also show differing populationhistories (Pauls et al. 2009 and references therein).One suitable model group for such a comparative approach

is the genus Erebia Dalman, 1816 (Nymphalidae: Satyrinae:Satyrini), which underwent a rapid diversification, like manyof its tribe (Pena et al. in press), possibly because of the spatial

and temporal restriction of its species, which has resulted in alarge number of endemic taxa (Martin et al. 2000). This genusis considered to be composed of numerous glacial relictsbecause of the typical narrow elevational distribution of its

many species in central and southern Europe (revised byTennent 2008). Some species of Erebia have already provided

Corresponding author: Marta Vila ([email protected])Contributing authors: Neus Marı-Mena ([email protected]), Ana Guer-rero ([email protected]), Thomas Schmitt ([email protected])

� 2010 Blackwell Verlag GmbHAccepted on 20 October 2010

J Zool Syst Evol Res doi: 10.1111/j.1439-0469.2010.00587.x

J Zool Syst Evol Res (2011) 49(2), 119–132

interesting scenarios for studying incipient speciation innorthern Iberia (e.g. Vila et al. 2006).For the present study, as a starting point for a comparative

phylogeographical study on northern Iberian Erebia species,we selected one of the Erebia species present in both theCantabrian and the Pyrenean Mountains: the large ringlet

Erebia euryale (Esper, 1805). As many as 19 of the approx-imately hundred recognized species of Erebia occur inthe Iberian Peninsula. Most of them (17) are present in thePyrenees, but eight Erebia species occur only along the

northern mountains of Spain (Pyrenean and Cantabrianmountain ranges) (Garcıa-Barros et al. 2004). The CantabrianMountains pose an interesting biogeographical framework for

this group, as one of the two Iberian species of Erebia absentfrom the Pyrenees inhabits the Cantabrian Mountains.The survival of Erebia euryale in mountain areas of northern

Spain during glacial stages was postulated by Schmitt andHaubrich (2008), although their work proposed the existenceof a refugial area in the northern foot hills of the Pyrenees.

Another interesting issue raised by their study was thesuggestion that the most ancient split within the EuropeanErebia euryale was one between the western Alpine lineage(E. e. adyte) and all the other surveyed populations. The same

work suggested that the Pyrenean populations might representa unique genetic entity within Erebia euryale; however, thatstudy included only one population from the eastern Pyrenees,

which limited the possible interpretations.The hypothesis of a single and distinct genetic lineage for the

Cantabrian and Pyrenean populations of Erebia euryale is

supported by classical intraspecific taxonomy (Appendix S1).However, in the light of the scattered distribution of the speciesin northern Spain and the intraspecific taxonomy, we also

expected a low but significant degree of genetic differentiationwithin that lineage.In the present study, we aimed to:

(1) describe the population structure of the westernmost part of

the distribution area of Erebia euryale (i.e. the CantabrianMountains), where the species occurs in scattered localities; and(2) determine whether the Cantabrian and Pyrenean localities

should be considered as a single evolutionarily significant unitor not (following Fraser and Bernatchez 2001).We studied the allozyme polymorphisms of six populations

from the Cantabrian Mountains and compared the resultsagainst populations from the Pyrenees, Alps and easternEuropean Mountains. Furthermore, we sequenced 816 bp ofthe mtDNA COI gene for these Cantabrian populations and

for one Pyrenean, one eastern European and three Alpinelocalities. Derived from the above-mentioned two mainobjectives, we also addressed the following questions:

(3) Did the ancestors of the Cantabrian and Pyreneanpopulations suffer drastic demographic changes during thelast glacial stages?

(4) Do the mitochondrial and nuclear data provide a concor-dant pattern?

Material and Methods

Study species

The large ringlet Erebia euryale occurs in several European mountainsystems, from the Cantabrian Mountains to the Balkans, Urals andKanin Peninsula. Alpine populations are biennial (Lafranchis 2000;Sonderegger 2005), but Spanish populations likely follow an annualcycle (H. Mortera and R. Vila, pers. comm.). According to Tolmanand Lewington (1997), the species� habitats are grassy, flowery placesin pinewood and spruce forest clearings and grassy slopes above thetreeline (elevational range = 750–2500 m a.s.l.). Larval host plantsinclude Poaceae species (genera Sesleria, Festuca, Poa and Calama-grostis) as well as several species from the family Cyperaceae, i.e.Carex flacca, Carex ferruginea. It shows marked local and regionalmorphological differences (Cupedo 2010). This has led to the definitionof several subspecies, forms and variants, and even an Erebia euryalespecies complex (as revised by Korshunov et al. http://www.jugan.narod.ru/satyridae1_eng.htm, accessed October 19, 2010). This taxonis classified as of Least Concern by the European Red List ofButterflies (van Swaay et al. 2010). Schmitt and Haubrich (2008)reported the following four major phenotypic groups, called subspeciesE. e. isarica Heyne, 1895 (widespread in western Europe as far East asthe northern Carpathians), E. e. adyte (Hubner, 1822) (restricted tothe western and south-western Alps), E. e. ocellaris Staudinger, 1861(endemic to a particular area of the south-eastern Alps), andE. e. syrmia Fruhstorfer, 1910 (widespread in the southern and easternCarpathians as well as in the Balkan Peninsula).

Sampling

A total of 218 individuals of E. euryale were collected at six localitiesfrom northern Spain (Fig. 1). For conservation purposes, only maleswere captured. Between 33 and 40 specimens per locality were caughtat these Cantabrian sites (Fig. 1, Tables 1 and 2, Table S11). Theimagos were hand-netted between July 6 and 23, 2006, and stored inliquid nitrogen until analysis.

Laboratory procedures

We genotyped the samples for 17 allozyme loci as described in Schmittand Haubrich (2008). The obtained dataset was compared against their

Fig. 1. Iberian distribution area ofErebia euryale (yellow circles,adapted from Garcıa-Barros et al.2004) and sampling locationssurveyed for the present genetic(allozymes + mtDNA) study.Allozyme data from La Glebe werepublished in Schmitt and Haubrich(2008). Readers are referred toFig. S10 for a map displaying theother European localities fromwhere mitochondrial results arepublished in the present study

120 Vila, Marı-Mena, Guerrero and Schmitt

J Zool Syst Evol Res (2011) 49(2), 119–132� 2010 Blackwell Verlag GmbH

results for 11 E. euryale populations sampled in the Pyrenees (onelocality), the Alps (seven sites), the southern Carpathians (two areas)and the Rila Mountains (one population). Data from the Alpinelocality of Schoneck (Austria) was not included in the analyses becauseof small sample size (n = 7).We also sequenced the second half of the mitochondrial cytochrome

c oxidase subunit I (COI) gene in 5–10 individuals per Cantabrianlocality. That part of the COI gene is the most variable, according toprior literature (Lunt et al. 1996) and prior experience on species ofErebia (Vila et al. 2005). Genomic DNA was extracted from adultthoracic tissue using the Genelute� Mammalian Genomic DNAMiniprep kit (Sigma-Aldrich, Saint Louis, MO, USA). Primers andPCR conditions are described in Vila and Bjorklund (2004). To obtaina wider pattern of the genetic differentiation of these samples, we alsosequenced this marker in 5–10 individuals from some of thepopulations analysed by Schmitt and Haubrich (2008) (Table 1).The total number of sequenced individuals was 90. The 816-bpsequences used for this survey start at position 706 of the COI gene(reference sequences for this alignment were GenBank accessionnumbers DQ102703 and EF621724) and correspond to positions2182–2997 of the mitochondrial genome of Drosophila yakuba Burla,1954 (accession no. X03240). New haplotypes were deposited inGenBank (Table 1).

Statistical analyses: allozyme dataset

Allozyme alleles were labelled according to their relative mobility,starting with �1� for the slowest. Allele labels were the same as thoseused by Schmitt and Haubrich (2008). Consequently, some alleles areabsent from the present dataset (e.g. alleles 1 and 2 at Apk locus).Allele frequencies and parameters of genetic diversity [i.e. meannumber of alleles per locus (A), expected and observed heterozygosity(He, Ho), total percentage of polymorphic loci (Ptot) and percentage ofpolymorphic loci with the most common allele not exceeding 95%(P95)] were computed with G-Stat (Siegismund 1993). Analyses ofHardy–Weinberg (HW) segregations were performed for all loci andpopulations. We used the score test (U-test) to evaluate whetherobserved deviations were due to heterozygote deficit both at popula-tion and at locus level (tests based on 1560 randomizations). U-testscalculated with statistica were also used to test for differences ingenetic diversity values from different populations. We performedexact probability tests to assess the adjustment to HW expectations foreach locus in each population. An unbiased estimate of the exact p-value for those 45 tests was computed using the Markov Chain MonteCarlo Method as implemented in the software genepop 4.0.10 (Rousset2008). Weighted F-statistics were calculated using the estimatorsdescribed by Weir and Cockerham (1984). Their values and signifi-cance were estimated in fstat 2.9.3 (Goudet 1995) after 10 000randomizations. We also reported estimates of the pairwise estimatorof differentiation Dest (Jost 2008) across loci calculated using smogd

2.6 (Crawford 2010).We also calculated locus-by-locus amovas, hierarchical genetic

variance analysis, as well as linkage disequilibrium as implemented inarlequin 3.5.1.2 (Excoffier and Lischer 2010). Significance wasobtained after 1023 permutations. Nei�s standard genetic distances(Nei 1972), neighbor-joining (NJ) phenograms and bootstraps basedon 1000 iterations were calculated with PHYLIP (Felsenstein 1989).We assessed population structuring using the simulated annealing

procedure developed by Dupanloup et al. (2002). This approachallows the definition of groups of geographically homogeneouspopulations that maximize the proportion of total genetic variancebecause of differences between population groups. For this, we ran asamova (Spatial Analysis of Molecular Variance) 1.0 using 100simulated annealing processes for k values from 2 to 16.The presence of limited gene flow can be indicated by a pattern of

isolation by distance across the surveyed populations. FollowingRousset (1997), we performed a Mantel test as implemented in ibdws

3.15 (Jensen et al. 2005) to assess the correlation betweenFST ⁄ (1 ) FST) and the geographical distance for both the wholeEuropean dataset and the Pyrenean–Cantabrian partition. Significanceof the Mantel test was evaluated after 10 000 randomizations. Wecalculated geographical distance (in kilometres) for each pair of

Table

1.Geographicalandfrequency

distributionofthe13mitochondrialhaplotypes

[cytochromeoxidase

I(C

OI)]ofErebia

euryale

(rows)

from

each

sampledlocation(columns).Geographical

coordinatesare

indicatedbelow

thenameofthesamplinglocality

Haplotype

name

SanGlorio

Pajares

SanIsidro

Somiedo

Leitariegos

Pandetrave

LaGlebe

Vald�Isere

Kalkstein

Obertauern

Sinaia

Genbank

accession

number

43�04¢N

,4

�62¢W

42�59¢N

,5�45

¢W43�04¢N

,5�25¢W

43

�01¢N

,6�13¢W

42�59¢N

,6�24

¢W43�06¢N

,4

�32¢W

42

�40¢N

2�13¢E

45�27¢N

6�59¢E

46

�48¢N

12

�19¢E

47�15¢N

13�34¢E

45�21¢N

25�31¢E

H01

10

87

85

710

AY346221

H02

11

HQ176010

H03

1HQ176011

H04

1HQ176012

H05

1HQ176013

H06

1HQ176014

H07

10

HQ176015

H08

5HQ176016

H09

8HQ176017

H10

1HQ176018

H11

1HQ176019

H12

5HQ176020

H13

1HQ176021

Phylogeography of Erebia euryale 121


sampling locations with Geographic Distance Matrix Generator 1.2.3(PJ Ersts, http://biodiversityinformatics.amnh.org/open_source/gdmg,accessed October 19, 2010).

To disentangle the population structure pattern both across Europeand within the Pyrenean–Cantabrian Mountains (North Spain), weused the program structure 2.3.3 (Pritchard et al. 2000). Thissoftware traditionally used model-based Bayesian clustering of indi-viduals without using prior population definitions. Version 2.3.3incorporates the possibility of using models with informative priors,i.e. allowing the software to use information about sampling locations.This may be useful to, for instance, assist the clustering when the signalof structure is relatively weak (e.g. significant FST between samplinglocations, but no detection of structure as from the standard structuremodels) (Hubisz et al. 2009). We first used this method on theEuropean allozyme dataset (592 individuals, 17 polymorphic loci, 16localities) under standard settings and the admixture model, trying theassumption of correlated allele frequencies, without consideringinformation about the population of origin of the samples. We triedthe admixture model, where the individuals may have mixed ancestry,as (i) it is recommended as starting point, and (ii) the hypothesis of oneor several hybrid zones for Erebia euryale seemed plausible at theEuropean scale. We used a burn-in of 50 000 iterations followed by100 000 iterations for parameter estimation. Each simulation was runfive times, exploring values for K (the total number of clusters to beconstructed in a given simulation) ranging from 1 to 17. To select themost appropriate number of groups, we followed Evanno et al. (2005).

We then ran structure for the Pyrenean–Cantabrian allozymedataset (258 individuals, 13 polymorphic loci, 7 locations), under theno admixture model (more powerful at detecting subtle structurestandard settings) and using the LOCPRIORS option. The rest of theparameters and interpretation of results were set as above, except forthe values of K to be explored (from one to eight).

As there were suspicions of inbreeding in the locality of Leitariegos,pairwise relatedness values between individuals were calculated for thissite using the method by Lynch and Ritland (1999), as implemented inidentix 1.1.5 (Belkhir et al. 2002). We estimated relatedness (r) usingthe multilocus genotypes from the eight loci polymorphic at Leitari-egos. After estimation of r and its variance, we performed 1000permutations to tests the null hypothesis of no relatedness. Thisprocedure can be applied to both arithmetic mean and variance of theabove-mentioned estimator.

Statistical analyses: mtDNA dataset

DNA sequences were aligned by eye. Alignments were straightforward.Haplotypes and their frequencies, standard indices of genetic variationsuch as the number of segregating sites (S), nucleotide diversity pergene (p), haplotype diversity (h) and the average number of nucleotide

differences (k) were calculated in DNAsp 5.1 (Librado and Rozas2009).

The phylogeny of the haplotypes was inferred using the statisticalparsimony (SP) criterion (Templeton et al. 1992). The 95% SPnetwork was calculated using TCS 1.2 (Clement et al. 2000). To havea reference to calibrate the extent of divergence between haplogroups,we rooted the resulting network with an homologous sequence fromErebia ligea (accession number AY346224).

The demographic history of samples belonging to the Pyrenean–Cantabrian lineage was inferred using different methods. First, amismatch distribution of the pairwise genetic differences (Rogers andHarpending 1992) was conducted, and their goodness-of-fit to asudden expansion model was tested using parametric bootstrapapproaches (1000 replicates). The sum of squared deviations (SSD)between the observed and expected mismatch distributions was used toassess the significance of the test. In addition, the significance ofpopulation expansion was tested using Tajima�s D and Fu�s Fsneutrality tests. All these calculations were performed in arlequin

3.5.1.2. Mismatch analyses were also used to estimate the approximatetiming of expansion of the Pyrenean–Cantabrian lineage of Erebiaeuryale. We used the relationship s = 2ut (Rogers and Harpending1992) and the approach taken by Horn et al. (2009) and Hammoutiet al. (2010). Therefore, we performed these calculations using bothl = 1.15*10)8 and l = 5*10)9 per year per site.

Results

Cantabrian allozyme dataset

The study of 218 adult males of Cantabrian Erebia euryalerevealed 13 of the 17 analysed loci as polymorphic. Loci Fum,

Gpdh, Mdh1 and Pk were monomorphic in the six surveyedlocalities, as they were in the Pyrenean locality of La Glebe,investigated by Schmitt and Haubrich (2008). The highest

number of six different alleles per locus was observed for twoloci (Pgi and Pgm), and the average was 2.94 (±1.75 SD). Theallele frequencies of the six Cantabrian populations analysedare shown in Table S2.

To describe the extent of genetic diversity and compare itwith the data from Schmitt and Haubrich (2008), we calculatedthe following estimators for the 218 Cantabrian individuals

based on allele frequencies of the 17 surveyed loci: (i) the meannumber of alleles per locus per population (A) ranged from1.71 to 2.18, with a mean of 1.92 (±0.19 SD), (ii) the

percentage of polymorphic loci with the most common allelenot exceeding 95% (P95) ranged from 29% to 47% with a

Table 2. Expected heterozygosity (He), observed heterozygosity (Ho), percentage of polymorphic loci [total (Ptot) and with the most commonallele not exceeding 95% (P95)], mean number of alleles per locus (A), sample size (N) for all allozyme loci (including monomorphic) and samplesanalysed of Erebia euryale. For means of comparison, the average values of ten samples of Erebia euryale from other regions of Europe are givenat the bottom of the table. The p values of U-tests between the Cantabrian samples and the other European as well as the south-eastern Europeansamples are presented

Sample site He Ho Ptot [%] P95 [%] A N

San Glorio 0.159 0.143 52.9 35.3 2.18 33Pajares 0.179 0.176 58.8 35.3 2.06 36San Isidro 0.149 0.145 52.9 47.1 1.94 37Somiedo 0.145 0.133 47.1 41.2 1.71 35Leitariegos 0.137 0.103 47.1 29.4 1.71 40Pandetrave 0.178 0.190 47.1 41.2 1.94 37Cantabrian mean 0.158 0.148 51.0 38.3 1.92 36.3±SD ±0.018 ±0.031 ±4.8 ±6.2 ±0.19 2.3Mean Schmitt and Haubrich (2008) 0.156 0.152 68.5 41.2 2.33 34.6±SD ±0.029 ±0.03 ±14.7 ±12.6 ±0.45 ±9.8p (U-test) 0.801 0.762 0.018 0.639 0.044 0.571Mean south-eastern EuropeSchmitt and Haubrich (2008)

0.178 0.173 84.3 51.0 2.84 39.7

±SD ±0.011 ±0.008 ±6.8 ±12.2 ±0.30 ±0.6p (U-test) 0.197 0.302 0.017 0.085 0.019 0.065



mean of 38% (±6 SD), and (iii) the total percentage ofpolymorphic loci (Ptot) ranged from 47% to 58%, mean 51%(±5 SD); (iv) average expected heterozygosity (He) was 0.158

(±0.218 SD) ranging from 0.137 to 0.179, (v) averageobserved heterozygosity (Ho) was 0.148 (±0.031 SD) varyingfrom 0.103 to 0.176. The means for Ptot and A were

significantly lower than the ones of the eleven populationsfrom other European mountain ranges, whereas the otherthree parameters and the mean number of individuals analysedwere not. This pattern was more pronounced when compared

against the samples from the genetically richest region ofsouth-eastern Europe found by Schmitt and Haubrich (2008).Details are shown in Table 2.

We repeated the descriptive calculations for the Cantabriandataset presented in this work and provided by the 13polymorphic loci. The average of alleles per locus (overall

populations) was 2.19 (±1.17 SD), ranging from 1.92 ± 0.95at Leitariegos to 2.54 ± 1.39 at San Glorio. Data on allelicrichness, a more accurate descriptor as it takes difference in

sample size into account, are provided in Table 3. Overall, allpopulations and loci conformed to Hardy–Weinberg expecta-tions (HWE) (exact test, v2 = 85.76.84, df = 90, p > 0.05).Using the probability tests, we found a significant deviation

from HWE in five cases combined over loci and populations.Based on chance alone, we would expect to find threesignificant results at the 0.05 level out of the 45 tests. None

of those five tests were significant after sequential Bonferronicorrection. When we tested for HWE using the score U(global) test and heterozygote deficit as null hypothesis, the

only locus globally departing from HWE after correction formultiple tests was PGM (Table S3). The only population thatsignificantly deviated from HWE was Leitariegos, which

congruently showed a significant inbreeding coefficient(FIS = 0.249, p < 0.001, Table 3, Table S4). Data fromLeitariegos did not significantly depart from the null hypoth-esis of no relatedness (average r = )0.025, variance = 0.143)

neither when the test was performed on the means nor thevariances of the relatedness coefficient (p > 0.05). The sameresult was obtained when permutation type was changed from

�genotypes� to �alleles�. No significant genetic linkage disequi-librium was detected between any pair of loci. There was somesubtle structuring within the Cantabrian dataset, as revealed

by the test for population differentiation (log-likelihood Gstatistic not assuming random mating within samples,p < 0.0001) (Goudet et al. 1996).

The genetic variance among Cantabrian localities was

relatively high (locus-by-locus amova results as a weighted

average over 13 loci (variance component = 0.111; FST

(FST) = 0.062, p = 0.003). This result agreed with the FST

(h) value = 0.077 (95% CI = 0.043–0.116). The genetic

variance within populations was moderate among individuals(variance component = 0.083, FIS 0.077, p < 0.001), butrelatively high within individuals (variance compo-

nent = 1.246). The unbiased genetic distances between thesix Cantabrian populations ranged from 0.007 to 0.038 with amean of 0.018 (±0.008 SD).

Comparison with published data: European scale

The NJ tree calculated using the Cantabrian and the samples

surveyed by Schmitt and Haubrich (Fig. 2) did not show astraightforward correspondence between its topology and thegeographical location of the samples. The most closely related

population to the Cantabrian samples was the Pyreneanlocality of La Glebe (distance: 0.031 ± 0.016 SD), followed bythe eastern Alpine sites (distance: 0.048 ± 0.017 SD) and the

south-eastern European ones (distance: 0.053 ± 0.018 SD);the western Alps samples showed the highest genetic distances

Table 3. Genetic diversity within populations of Erebia euryale as from 13 polymorphic allozyme loci. Mean n, average number of individualsanalysed; proportion of missing data per locus for each population; AR, allelic richness (based on 27 diploid individuals); PA, number of privatealleles; HWE, deviations from Hardy–Weinberg equilibrium; significance after sequential Bonferroni correction is marked in bold. NS, p > 0.05;*p < 0.05; **p < 0.01; ***p < 0.001;HE, unbiased expected andHO, observed heterozygosities; FIS, inbreeding coefficient. Standard errors aregiven after symbol ±

Mean n Missing data (locus) AR PA HWE HE HO FIS

San Glorio 32.54 ± 0.31 0.12 (6Pgdh) 2.45 ± 1.33 1 (G6P) * 0.207 ± 0.247 0.187 ± 0.218 0.096 NSPajares 36.00 ± 0.00 0 2.28 ± 1.24 0 NS 0.234 ± 0.251 0.231 ± 0.253 0.013 NSSan Isidro 36.77 ± 0.17 0 2.15 ± 1.22 0 NS 0.196 ± 0.214 0.19 ± 0.21 0.028 NSSomiedo 34.23 ± 0.23 0.08 (Apk) 1.89 ± 0.93 0 NS 0.19 ± 0.225 0.174 ± 0.2 0.088 NSLeitariegos 39.77 ± 0.12 0 1.82 ± 0.89 0 *** 0.178 ± 0.234 0.134 ± 0.162 0.2491*Pandetrave 35.23 ± 0.96 0.16 (6Pgdh),

0.24 (Mdh), 0.18 (Pgm)2.11 ± 1.11 1 (MD2) NS 0.229 ± 0.241 0.247 ± 0.257 )0.077 NS

1Significant deficit of heterozygotes, based on 1560 randomizations. Indicative adjusted nominal level (5%) was 0.00064. p = 0.0006.

Fig. 2. Neighbor-joining phenogram based on Nei�s standard geneticdistances (Nei 1972) of the six populations of Erebia euryale from theCantabrian Mts. Ten populations from the Pyrenees, the Alps, thesouthern Carpathians and the Rila Mountains (Schmitt and Haubrich2008) were included. Distances were calculated using the results pro-vided by 17 allozyme loci. Bootstrap values over 50% are given at thenodes. Cantabrian localities are marked in italics. Those localities fromwhich some individuals were sequenced for the cytochrome oxydase Igene appear underlined



against the Cantabrian Mountains (distance: 0.096 ±0.022 SD). Hierarchical variance analyses supported this

structure (FCT values against: Pyrenees 0.060, p = 0.134;eastern Alps 0.143, p < 0.0001; south-eastern Europe 0.166,p < 0.001; and western Alps 0.287, p < 0.001).

There was significant structuring for the whole Europeandataset, as revealed by the test for population differentiation(log-likelihood G statistic not assuming random mating within

samples, p < 0.0001) and the relative differentiation providedby the 95% confidence interval for FST (0.096–0.35). Thepairwise comparisons involving maximum differentiationmostly agreed between FST and Dest estimators (Val d�Isereversus San Isidro: FST = 0.509, Dest = 0.023) (Table 4). Theminimum differentiation was congruently determined by bothestimators among the eastern European localities of Sinaia,

Groapa Seaca and Rila Mountains. All sampling sitescontributed similarly to the average relative differentiation,i.e. FST indices were fairly similar for the 16 localities ranging

from 0.206 (Rila Mountains) to 0.214 (Val d�Isere) (Table 4).Therefore, no particular evolutionary constraint (e.g. selec-tion) left a deep track in the allozyme dataset (Weir and Hill

2002).The European nuclear dataset did not conform to the

isolation-by-distance model, as there was no significant corre-lation between genetic differentiation in relation to geograph-

ical distance (Z = 39647.54, r = 0.15, one-sided p = 0.104)(Fig. S5).The genetic variance among the 16 European localities was

relatively high (locus-by-locus amova results as a weightedaverage over 17 loci): variance component = 0.342; FST:0.206, p < 0.001). The genetic variance within populations

was moderate among individuals (variance compo-nent = 0.055, FIS 0.042, p = 0.002), but high within individ-uals (variance component = 1.262).The spatial analysis of molecular variance also supported

the genetic affinity of the Cantabrian and Pyrenean samples.This analysis indicated a most likely subdivision of the wholedistribution area into four groups (K = 4, Fig. 3), when (i)

FCT values approached a plateau, (ii) FST showed a steep

decrease, and (iii) DFCT was the largest (DFCT = 0.012). These

four groupings corresponded to: [La Glebe plus the sixCantabrian sites] [St. Martin-Val d�Isere-Partnun] [GroapaSeaca-Sinaia-Rila Mountains], and [Kalkstein-Obertauern-

Trauneralm]. The genetic variance among groups was 18.7%,among populations within groups 6.4% and within popula-tions 75% with FCT = 0.187 and FSC = 0.079 (allp < 0.001).

The same four groups were obtained as second most likelyclustering scheme when applying the Bayesian algorithmimplemented in structure 2.3.3 to the whole allozyme

dataset. Interestingly, the most likely number of clusters (K)was two, separating the Pyrenean–Cantabrian localities (WEurope) from the remaining surveyed localities. Two localities

showed very homogeneous gene pools: Val d�Isere (W Alps)and San Isidro (W Europe) (Fig. 4).

Significant differences in average gene diversity and allelic

richness were found among these four groupings (two-sidedtest, 5000 permutations, p = 0.001 for gene diversity, andp = 0.006 for allelic richness). The average gene diversity was

Fig. 3. Results of the spatial analysis of molecular variance showingthe genetic affinity between the Cantabrian and Pyrenean samples. Themost likely subdivision of the whole distribution area into four groups,when the increment of FCT was the largest (DFCT = 0.0124). Thesefour groupings corresponded to: [Pyrenean–Cantabrian samples] [WAlps] [SE Europe], and [E Alps]

Table 4. Below diagonal: Pairwise FST (Weir and Cockerham 1984) values between the 16 localities of Erebia euryale as revealed by the allozymedataset (17 loci). Significant FST values are displayed in bold. Indicative adjusted nominal level (5%) for multiple comparisons = 0.00042. Abovediagonal: Harmonic mean of the estimator of actual differentiation Dest (Jost 2008). Glb, La Glebe; StM, St. Martin; Vis, Val d�Isere; Pat,Partnun; Kal, Kalkstein; Obe, Obertauern; Tra, Trauneralm; Gro, Groapa Seaca; Sin, Sinaia; Ril, Rila Mountains; Glo, San Glorio; Paj, Pajares;Isi, San Isidro; Som, Somiedo; Lei, Leitariegos; Pan, Pandetrave. Diagonal: Population specific FST indices, in italics

Western Alps Eastern Alps East Europe Cantabrian Pyrenean

StM Vis Pat Kal Obe Tra Gro Sin Ril Glo Paj Isi Som Lei Pan Glb

StM 0.209 0.004 0.006 0.003 0.006 0.005 0.012 0.010 0.009 0.014 0.013 0.015 0.009 0.016 0.020 0.026Vis 0.121 0.214 0.004 0.012 0.014 0.013 0.014 0.014 0.014 0.018 0.019 0.023 0.012 0.019 0.021 0.026Pat 0.099 0.171 0.209 0.010 0.010 0.010 0.007 0.007 0.007 0.010 0.010 0.023 0.016 0.027 0.025 0.018Kal 0.205 0.407 0.294 0.209 0.003 0.001 0.011 0.012 0.011 0.007 0.009 0.009 0.008 0.012 0.014 0.013Obe 0.221 0.381 0.257 0.049 0.207 0.004 0.014 0.014 0.012 0.006 0.010 0.009 0.011 0.017 0.015 0.014Tra 0.205 0.386 0.278 0.022 0.042 0.208 0.010 0.010 0.008 0.005 0.009 0.012 0.009 0.018 0.016 0.013Gro 0.196 0.325 0.179 0.175 0.146 0.151 0.207 0.000 0.000 0.012 0.011 0.015 0.011 0.018 0.017 0.007Sin 0.182 0.315 0.177 0.163 0.131 0.147 )0.002 0.207 0.000 0.011 0.011 0.013 0.012 0.020 0.018 0.010Ril 0.179 0.315 0.177 0.156 0.133 0.137 0.003 0.003 0.206 0.010 0.009 0.016 0.012 0.021 0.018 0.009Glo 0.273 0.415 0.272 0.133 0.135 0.144 0.147 0.165 0.151 0.209 0.002 0.006 0.004 0.006 0.003 0.006Paj 0.270 0.404 0.258 0.142 0.118 0.147 0.128 0.128 0.128 0.045 0.206 0.006 0.002 0.007 0.006 0.003Isi 0.365 0.509 0.403 0.266 0.285 0.291 0.286 0.295 0.277 0.108 0.151 0.209 0.003 0.004 0.005 0.016Som 0.261 0.411 0.314 0.137 0.145 0.159 0.151 0.151 0.155 0.063 0.050 0.131 0.209 0.001 0.001 0.004Lei 0.325 0.463 0.377 0.215 0.234 0.244 0.245 0.246 0.239 0.104 0.077 0.106 0.027 0.209 0.001 0.005Pan 0.302 0.429 0.336 0.183 0.188 0.202 0.199 0.205 0.194 0.054 0.061 0.088 0.030 0.026 0.209 0.005Glb 0.331 0.456 0.320 0.222 0.185 0.193 0.148 0.167 0.165 0.099 0.070 0.238 0.097 0.140 0.104 0.208



higher in the eastern than in the western Alps groups (one-sided test, 5000 permutations, p = 0.025). It was also higher inthe eastern European gene pool than in the western Alps(p = 0.001) (Table 5). Allelic richness was significantly higher

in eastern Europe than in both the western Alps (p = 0.005)and the Pyrenean–Cantabrian gene pool (p = 0.0006)(Table 5).

Comparison with published data: Pyrenean–Cantabrian lineage

There was also a remarkable population structuring among theseven northern Iberian localities surveyed, as revealed by theFST value 0.095 (95% CI = 0.059–0.123), and the test for

population differentiation (log-likelihood G statistic notassuming random mating within samples, p < 0.001).

The Pyrenean–Cantabrian nuclear dataset did not conformto the isolation-by-distance model, as there was no significant

correlation between genetic differentiation in relation togeographical distance (Z = 649.47, r = 0.35, one-sidedp = 0.21; Fig. S6), not even when using the logarithm ofgeographical distance (Z = 5.07, r = 0.31, p = 0.23).

Regarding population-wise comparisons, San Isidroappeared as the most differentiated population. The extent ofsignificant differentiation was moderate (FST > 0.1) for the

four pairs where San Isidro was involved (i.e. San Isidro-SanGlorio, San Isidro-Pajares, San Isidro-Somiedo and SanIsidro-Leitariegos), reaching 0.238 between San Isidro and

the Pyrenean locality of La Glebe. The lowest degree ofdifferentiation resulted between Leitariegos and Somiedo (FST)or Somiedo-Leitariegos-Pandetrave (Dest) (Table 4).

Exploratory calculations to identify the genetic groupswithin the Pyrenean–Cantabrian Erebia euryale through theBayesian clustering implemented in structure 2.3.3 alwaysresulted in a most likely number of clusters (K) of two (data

(a)

(b)

(c)

(d)

Fig. 4. Two (a) and four (b) most likely groupings of 592 individuals of Erebia euryale as revealed by running the Bayesian structure 2.3.3 witha 17 allozyme loci dataset and under the admixture model and correlated allele frequencies. The Pyrenean population (La Glebe) always clusteredwith the Cantabrian localities. Bayesian clustering of 258 individuals of the Pyrenean and Cantabrian localities as revealed by 13 polymorphic lociusing the LOCPRIOR option and (c) non-admixture model, correlated allele frequencies, or (d) non-admixture model, independent allelefrequencies. Each individual is represented as a bar, divided into K colours, where K is the number of most likely clusters inferred followingEvanno et al. (2005)

Table 5. Below diagonal: Pairwise FST (Weir and Cockerham 1984) values between the four major genetic groups of Erebia euryale as revealed bythe allozyme dataset. All FST values resulted significant after correction for multiple comparisons [indicative adjusted nominal level(5%) = 0.0083 calculated after 120 permutations]. Above diagonal: Harmonic mean of the estimator of actual differentiation Dest across loci(Jost 2008). The numbers after regions correspond to the number of complete multilocus genotypes. Diagonal: average gene diversity and allelicrichness across loci

W Alps(89)

E Alps(103)

E Europe(117)

Pyrenean–Cantabrian(229)

W Alps 0.118, 1.94 0.005 0.008 0.012E Alps 0.258 0.160, 2.19 0.011 0.008E Europe 0.203 0.137 0.178, 2.56 0.013Pyrenean–Cantabrian 0.294 0.146 0.160 0.157, 1.84



not shown). This was inferred using only genetic information,dismissing the geographical origin of samples and by examin-ing the graphical output from the software and by considering

the logarithmic probability of data [lnP(D)] for differentnumbers of K as well as the statistic DK (Evanno et al. 2005),which considers the rate of change in lnP(D) among successive

K values. As some structure was expected (see significant FST

values in Table 4), and our dataset might show a relativelyweak signal of structure, we ran again the clustering algorithmbut using the sampling locations as prior information to assist

the grouping (Hubisz et al. 2009). Four most likely clusterswere inferred, regardless of using the correlated or independentfrequency models. In both cases, the localities of San Isidro

and La Glebe were clearly differentiated (Fig. 4). Interestingly,both La Glebe and San Isidro appeared as the most homo-geneous gene pools within the surveyed localities in northern

Iberia.

mtDNA dataset

Thirteen haplotypes were obtained for the COI fragmentanalysed in 90 individuals from the six surveyed Cantabrianlocalities plus one from the Pyrenees (La Glebe), another one

from western Alps (Val d�Isere), two from eastern Alps(Obertauern and Kalkstein) and one from south-easternEurope (Sinaia) (Table 1). These haplotypes were defined by

13 variable sites along the analysed COI fragment (816 bplength). Five of them corresponded to singleton variable sites,whereas the remaining eight were parsimony informative (two

variants). Eleven substitutions were transitions. Twelve sub-stitutions involved synonymous changes. The only replace-ment change was a first codon position transition that

differentiated samples from Sinaia, Obertauern and Kalksteinfrom the western populations. Overall, haplotype diversity was(h) = 0.632 ± 0.055, nucleotide diversity(p) = 0.0021 ± 0.0003, and the average number of nucleotide

differences (k) = 1.7563 (Table 6).Figure 5 shows the statistical parsimony network and

geographical occurrence of the obtained mitochondrial vari-

ants. Haplotypes differing by up to 12 mutations (including theoutgroup) could be connected with 95% confidence as multiplesubstitutions did not occur at any particular nucleotide

position. When focussed, Erebia euryale, the most common

haplotype (H01) in the Cantabrian Mountains, was the onlyone present in the Pyrenean locality (La Glebe). The Canta-

brian haplotypes grouped in a star-like shape. Pajares was themost diverse locality (four haplotypes, two of them sharedwith Somiedo), followed by San Isidro, Pandetrave andSomiedo (two haplotypes). Leitariegos (westernmost locality),

San Glorio (easternmost Cantabrian site) and the Pyreneanpopulation only yielded the most common haplotype. H01 wasonly one substitution apart from the one present in Val d�Isere(western Alps). Sinaia (SE Europe, three haplotypes) was theonly locality containing haplotypes differentiated by more thanone mutational step. One of its haplotypes (H11) was

phylogenetically closer to haplotype H13 from Obertauern(E Alps) than to the others from Sinaia. In fact, the twopopulations from the eastern Alps (Kalkstein and Obertauern)appeared closer to Sinaia than to each other.

Mismatch analyses for the Pyrenean–Cantabrian lineagewere consistent with the sudden expansion model(SDD = 0.002, p = 0.44) and showed a unimodal distribu-

tion that closely fitted the expected distributions (Fig. S6). Fu�sFs and Tajima�s D statistics were both negative and significant(Fs = )6.057, p < 0.001; D = )1.906, p = 0.002). The

approximate timing of demographic expansion t was estimatedfor the Pyrenean–Cantabrian lineage assuming an annual cyclefor the species (s = 3.0, 95% CI = 0.35–3.5), and trying two

mutation rates used by prior literature. Therefore, the expan-sion of the Pyrenean–Cantabrian lineage was estimated to dateback to either 79 923 years before present (yBP) (95%CI = 9324–93 243 yBP; calculations using l = 1.15*10)8)

or 183 823 yBP (95% CI = 21 446–214 460 yBP; calculationsusing l = 5*10)9).

Table 6. Measurements of genetic diversity for Erebia euryale as fromthe mitochondrial dataset and according to sampling location. n,number of sequenced individuals; S, number of segregating sites; NH,number of haplotypes; h, haplotypes diversity; p, nucleotide diversity;and k, average number of nucleotide differences

Population n S NH h p k

Leitariegos 5 0 1 0 0 0Somiedo 9 1 2 0.2222 0.0003 0.2222Pajares 9 3 4 0.5833 0.0008 0.6667San Isidro 8 1 2 0.25 0.0003 0.25Pandetrave 8 1 2 0.25 0.0003 0.25San Glorio 10 0 1 0 0 0La Glebe 10 0 1 0 0 0Val d�Isere 10 0 1 0 0 0Kalkstein 5 0 1 0 0 0Obertauern 6 1 2 0.3333 0.0004 0.3333Sinaia 10 3 3 0.3778 0.0007 0.6

Fig. 5. Statistical parsimony network showing abundance and occur-rence of each cytochrome oxidase I haplotype obtained for the 89individuals of Erebia euryale. Circle size reflects the frequency ofhaplotype (maximum: 53; minimum: 1). Solid lines connecting haplo-types represent a single mutational event, regardless of their length.Black rectangles represent missing or theoretical haplotypes. Leitari-egos (Lei), Somiedo (Som), Pajares (Paj), San Isidro (Isi), Pandetrave(Pan), San Glorio (Glo), La Glebe (Glb), Val d�Isere (Vis), Kalkstein(Kal), Obertauern (Obe), Sinaia (Sin). The Erebia euryale intraspecificstatistical parsimony network (95%) had a connection limit of six. Thenetwork was rooted using Erebia ligea as an outgroup



Discussion

Population structure of Cantabrian Erebia euryale

The 218 individuals from the Cantabrian Mountains showedsignificant structuring as for the nuclear dataset, but they weremonomorphic for four out of the 17 surveyed allozyme loci

(see Appendix S7 for further discussion on the levels of geneticdiversity and resolution of the allozyme dataset). Leitariegoswas the only locality departing from HWE due to a significant

deficit of heterozygotes. However, it is likely that such adeviation was not due to relatedness within the surveyedsample. We rather advocate null alleles as a more likelyexplanation for such an excess of homozygotes (Appendix S7).

Three Cantabrian clusters were inferred: Leitariegos-Somi-edo-Pandetrave, San Isidro, and Pajares-San Glorio. Indeed,there was no correlation between genetic and geographical

distance among these localities. Such a pattern of geneticdifferentiation may be the result of genetic drift acting uponseveral isolated ancestral populations. It is worth noting that

this species occurs in scattered localities along the CantabrianMountains (Fig. 1). Considering the current geographicalorientation of the sample localities, it could be argued thatthe grouping Leitariegos-Somiedo-Pandetrave (northern side,

Asturias Province) resulted from a common ancestor thatcolonized the northern slope of the Cantabrian Mountains,whereas the grouping Pajares-San Glorio (southern side, Leon

Province) might have resulted from a northwards colonizationfrom the southern slopes. Thus, the large ringlets from SanIsidro may be the descendents of a third ancestral gene pool.

Gutierrez (1997) found a higher butterfly species richness insheltered gorges of the National Park of Picos de Europa(sampled also in the present study, i.e. Pandetrave) than on

northern slopes. Differentiation depending on slope orienta-tion was found for populations of Pinus sylvestris in thenorthern Meseta and peripheral mountain chains (central andnorth-western Iberian Peninsula). Robledo-Arnuncio et al.

(2005) showed that Scots pine isolates growing on disjunctmountain blocks, but on slopes flowing to the same basin weregenetically closer than populations growing on different slopes

of the same mountain chain, flowing to different basins.Genetic differentiation between populations from northernand southern Pyrenean sides was reported for the lizard

Zootoca vivipara (Jacquin, 1787) (Guillaume et al. 2000).Gutierrez (1997) also suggested that current species richnessand composition of butterfly assemblages in the Picos de

Europa area might be the consequence of differential coloni-zation of refuges during the Quaternary climatic changes.Again, the evolutionary history of Z. vivipara (Guillaumeet al. 2000) illustrates how heterogeneous colonization of

glacial Pyrenean ⁄ Iberian refugia may have caused geneticdifferentiation within them. Other reports for the Cantabrianarea suggested the existence of at least two glacial refugia for

some species (e.g. Garcıa-Parıs et al. 2003; Vialette et al. 2008),in agreement with the fine-scale pattern of �refugia within theIberian refuge� reviewed by Gomez and Lunt (2007).

Nevertheless, we cannot rule out the possibility that thecurrent Cantabrian allozyme pattern is simply caused bystochastic effects, i.e. strong geographical isolation of all thesepopulations over the postglacial period and quick random

allele shifts in these isolates. In this context, populationssuffering stronger bottlenecks might be either genetically moredistant from the other demes and ⁄or genetically impoverished.

The evolution of genetically more distant populations typically

results after long-lasting but relatively mild bottlenecks overgenerally small populations. Such diffuse reductions in popu-lation size mostly affect allele frequencies, but much less

genetic diversity. San Isidro is a suitable candidate to matchthat case. On the other hand, genetic impoverishment isusually produced by strong, but relatively short bottlenecks,

mostly affecting the number of alleles and, to a lesser extent,polymorphic loci (England et al. 2003). This might be the casefor Somiedo and Leitariegos. Our nuclear dataset did not trackany bottleneck (data not shown), so this stochastic scenario

remains speculative. Future research surveying more northernIberian populations and using more variable genetic markers isneeded to unravel the processes (stochastic demographic

effects and ⁄or landscape features) shaping the observed geneticpattern for the Cantabrian large ringlets.

Definition of an Evolutionarily Significant Unit in northern

Iberia

Both mitochondrial and nuclear markers revealed the geneticdistinctiveness of a single genetic lineage of Erebia euryalespanning the Cantabrian Mountains and eastern Pyrenees.The mitochondrial dataset shows geographical co-occurrence

of related haplotypes, i.e. all Pyrenean–Cantabrian localitiesshare the same most common haplotype (H01), whereas theother five rare haplotypes H02–H06 are separated from H01

by only one substitution. From the nuclear perspective, wefound the maximum differentiation between the Pyrenean–Cantabrian lineage and the south-western Alps when using

Nei�s distance and FST as estimators (Fig. 2, Table 4). Bycontrast, the Pyrenean–Cantabrian grouping and eastern Alps(H11) ⁄ eastern Europe (H13) were the most differentiated

(congruently with geography) as from the mitochondrialperspective and the actual differentiation estimator (Dest)applied on the nuclear dataset (Fig. 5, Table 4). Thesefindings, together with the monophyly of these populations

inferred from the allozyme dataset, prove that Cantabrian andPyrenean E. euryale share a most common recent ancestor.We also confirmed the strong decline of the genetic diversity

from East to West detected by Schmitt and Haubrich (2008)(Tables 2 and 3). For instance, allelic richness decreased fromE Europe (2.56) to the Pyrenean–Cantabrian lineage (1.84)

(Table 5). Such a gradient has also been detected in vertebratespecies, e.g. the Pyrenean capercaillie (Tetrao urogallus,Rodrıguez-Munoz et al. 2007) and a woodland plant (Polyg-onatum verticillatum; Kramp et al. 2009). Richer East-poorer

West gradients, in terms of genetic diversity, seem to indicatelarger effective population sizes for those taxa in populationsoutside the Iberian Peninsula and within the framework of

Quaternary history and postglacial expansions (revised bySchmitt 2009).Our data support the hypothesis of a distinct Pyrenean

lineage within Erebia euryale put forward by Schmitt andHaubrich (2008). The present work extends this finding toshow that the Pyrenean locality of La Glebe and the six West

Cantabrian localities newly genotyped formed a single geneticlineage, opposite to what we expected in the light of classicaltaxonomy and genetic findings in other northern Iberianmontane species (e.g. Perez et al. 2002; Pauls et al. 2006). The

discovery of a distinct Cantabrian–Pyrenean lineage is animportant knowledge for conservation managers, as allthe Cantabrian sampling sites are within protected areas

(Ministerio de Medioambiente, Medio Rural y Marino, http://



www.mma.es/portal/secciones/biodiversidad/rednatura2000/rednatura_espana/index.htm, accessed October 19, 2010) con-sidering the general interest in knowing the extent of genetic

structuring of species occurring in different protected areas.As a distinct Pyrenean–Cantabrian lineage was clearly

revealed by both nuclear and mitochondrial results, we

support the definition of the Pyrenean–Cantabrian lineage ofErebia euryale as an ESU which is reinforced by the followingtwo arguments. First, the DK statistic of Evanno et al. (2005)showed the largest variation (Fig. S8b) when the data clustered

in two groups (K = 2). This indicates that the most likelydivision for the European allozyme dataset is the separation ofthe Pyrenean–Cantabrian lineage from all the other surveyed

samples. There was a second interesting change in DK when theclustering determined four groupings (K = 4), which alsomade geographical sense, as those four entities essentially

grouped samples according to their major regions of origin(Fig. 4b). This split between lineages endemic (or predomi-nant) at the Pyrenean–Cantabrian region and the rest of

Europe is in accordance with the evolutionary history inferredfor, e.g. Pinus sylvestris (Cheddadi et al. 2006) or Zootocavivipara (Guillaume et al. 2000). Second, there might be somedifferentiation between the habitat occupied by the Cantabrian

Erebia euryale and the populations of this species surveyed bySchmitt and Haubrich (2008). Their populations are locatednear coniferous forests, which are scarce in the Cantabrian

Mountains. Cantabrian large ringlets currently use non-conif-erous tree species to find shelter, as well as some southernmostAlpine, Apennine and Balkan populations of this species (F.

Cupedo, pers. comm.). As the Cantabrian pine tree forestsseverely declined since the Late Holocene (Rubiales et al.2008) and particularly since the 19th century (Quevedo et al.

2006), one might argue that the patchy distribution ofE. euryale in the Cantabrian Mountains relates to the intensiveuse of land and deforestation.

Population fluctuations in northern Iberia

The expansion of the Pyrenean–Cantabrian lineage of Erebia

euryale likely occurred during the Upper Pleistocene. Using thesame substitution rate as Horn et al. (2009), the pine-depen-dent beetle Tomicus piniperda (Linnaeus, 1758) and E. euryale

show a somehow coincident time (90, 80 kyBP, respectively)for expansion of their Iberian mitochondrial lineages(Appendix S7). Both dates correspond to the marine oxygen-isotope stage (MIS) 5 (Gibbard and van Kolfschoten 2004).

One might argue that these dates represent the transitionbetween the interglacial MIS5e (Eemian chronostratigraphicsubage) and the Late Glacial Maximum (MIS 2, Late

Weschelian ⁄Wurm). However, at fine scale, MIS5 also showedclimatic oscillations (Shackleton et al. 2003) and 80 kyBPlikely corresponded to a warmer stage than 90 kyBP (Potter

et al. 2004). In both species, their Iberian lineages have notcontributed to the current genetic diversity found in otherEuropean regions. However, computing the expansion time by

means of the average rate given by Hammouti et al. (2010), thegrowth of the Pyrenean–Cantabrian lineage (183 kyBP) wouldfall into the transition between MIS 7 (interglacial) and 6(glacial), both within the Saalian ⁄Riss. The temporal discrep-

ancy because of the use of different mutation rates as well asthe wide and overlapping confidence intervals for thoseestimates prevent further discussion on the palaeoenvironmen-

tal scenario when such a population expansion took place.

This is because the wide temporal intervals may imply oppositedisplacement scenarios: during glacial stages large ringletslikely sought to open wide lowlands for refuge, whereas during

interstadials populations are expected to become fragmentedas they gain elevation. We acknowledge that montane speciesmay seek refuge in three types of areas during cold stages:

nunatak, peripheral and lowland refugia (Holderegger andThiel-Egenter 2009). However, the current Cantabrian eleva-tional range (700 to 1650 m a.s.l., H. Mortera, pers. comm.)makes plausible the peripheral refugia during glacial stages.

We therefore postulate that the obtained expansion ofE. euryale corresponds to an environmental cooling eventand the down-slope shift of montane forests.

A deeper understanding of the time when Erebia euryalesplit from its sister taxon or when the splitting of the majorgenetic groups occurred is beyond the scope of this paper. Our

markers have insufficient resolution for these questions, whilea caveat is also imposed for applying a strict molecular clock atthe intraspecific level when (i) population sizes can be

demonstrated not to be constant (as in our case from theheterozygosity deficit text, data not shown, e.g. expansions inWest and East Alps, as well as East Europe), and (ii) it isinvalid to extrapolate molecular rate of change across different

evolutionary timescales because of the time dependency ofmolecular rate estimates (Ho et al. 2005). However, it isencouraging that the above-mentioned tentative time frame for

the Pyrenean–Cantabrian lineage agrees with the UpperPleistocene formation of divergent lineages within species ofErebia (Martin et al. 2002; Vila et al. 2005; Albre et al. 2008).

Mitochondrial versus nuclear patterns

Despite the continuum along the Pyrenean–Cantabrian local-ities revealed by the mtDNA, there was significant differenti-ation between some Cantabrian localities and the Pyrenean siteof La Glebe (i.e. maximum differentiation with San Isidro:

FST = 0.238, Dest = 0.016). Such a nuclear differentiationmost probably resulted from genetic drift (causing quickrandom allele shifts) acting upon different demes that became

isolated after the population expansion of the Pyrenean–Cantabrian ancestral population. A similar pattern due topostglacial events has also been observed in the violet copper

Lycaena helle (Denis and Schiffermuller, 1775) (Habel et al.2010). Another factor that may contribute to this discrepancybetween mitochondrial and nuclear results would be a sex-biased dispersal.

To the best of our knowledge, there are no published dataon the dispersal pattern of Erebia euryale. However, we cannotrule out dispersal in this species to be male-biased, as it occurs

in some congeneric species [e.g. Erebia epiphron (Knoch, 1783)and Erebia sudetica Staudinger, 1861, Kuras et al. 2003], butnot in others (e.g. Erebia epipsodea Butler, 1868, Brussard and

Ehrlich 1970).Our allozyme and mitochondrial results agree in indicating

four major European lineages (Fig. S8). The degree of

differentiation we infer among the four major groupings mustbe regarded as provisional, because genetic information fromItalian populations is lacking. Samples from the Italianpeninsula (Ligurian Alps as well as Central Apennines) will

likely reveal at least a fifth distinct genetic lineage, as suggestedby the diverging morphology of Italian large ringlets and alsostudies on other organisms (e.g. Nazari and Sperling 2007;

Ansell et al. 2008; Rousselet et al. 2010).



The nuclear differentiation of the Pyrenean–Cantabrianlineage and other major European geno-phenotypic groups ofthis species is high (Table 5). This also holds true for genetic

lineages of other Erebia species (e.g. E. epiphron, Schmitt et al.2006) and even other satyrine butterflies such as Melanargiasp. (Habel et al. 2005). However, the major genetic groups

revealed by the nuclear dataset of Erebia euryale are notlargely differentiated at the mitochondrial level (Fig. 5). Infact, only one substitution differentiates the most frequentPyrenean–Cantabrian haplotype from the one present in the

south-western Alps; a lesser extent of differentiation thanshown by the same mitochondrial marker for other subspeciesor divergent lineages of Erebia, e.g. Erebia triaria pargaponda-

lense Fernandez Vidal, 1984 (Vila et al. 2005), westernPannonian populations of Erebia medusa (Denis andSchiffermuller, 1775) (Hammouti et al. 2010) or other lepid-

opterans (Vandewoestijne et al. 2004; Nazari and Sperling2007; Kerdelhue et al. 2009).

Schmitt and Haubrich (2008) suggested that according to

their allozyme and morphological data, the most ancient splitin European E. euryale occurred between the western AlpineE. euryale adyte and all the other populations. Our mitochon-drial results rather point to a first split of an eastern group with

haplotypes currently present in East Europe (Sinaia) and theeastern Alps (Obertauern) (Fig. 5).

The population sample from Val d�Isere used for the mtDNA

analyses seems genetically impoverished at nuclear level (seeTable 3 at Schmitt and Haubrich 2008). In fact, its commonestallele is fixed at several loci, but allele frequencies are also

strongly distorted at some other loci. The signal is the same inthe other western Alps samples (St. Martin, Partnun) but to alesser extent. This pattern results in higher FST values (Table 4)

and a long branch in the NJ tree (Fig. 2). Future research shalldetermine whether the COI haplotype found in this valley isrepresentative of the surrounding populations or not, as Vald�Isere might correspond to a relatively isolated population.

Populations currently occurring in the south-western Alpstracked a glacial refuge in peripheral areas also used by otherspecies such as the European Scots pine Pinus sylvestris

(Cheddadi et al. 2006) and the caddisfly Rhyacophila pubescensPictet, 1834 (Engelhardt et al. 2008). This refugial area in somecases provided forebears of present Pyrenean populations.

Schmitt (2009) revised the links between the Pyrenees and south-western Alpine populations of several taxa, but the presentstudy reveals that this is unlikely to be the case for Erebiaeuryale. Rather, the phylogeographical pattern of the Pyre-

nean–Cantabrian lineage supports the isolation of northernIberian Erebia euryale from remaining European populations.

Comparative phylogeographical surveys are still scarce and

mostly compared distantly related species (Pauls et al. 2009).The pattern obtained for Erebia euryale can be compared to theone exhibited by E. epiphron and the closely related species pair

Erebia palarica–E. meolans distributed along the CantabrianMountains and the Pyrenees. The present work and that of Vilaet al. (2006) pave the way for a future comparative phylogeo-

graphical study aiming at disentangling the evolutionaryfactors shaping the genetic structure of populations of Erebiaspecies occurring in the northern Iberian mountain ranges.

Acknowledgements

We thank Nuria Remon and Nadja Stabenow for their valuable help inthe laboratory. Vittorio Baglione, Almudena Fernandez Briera, Eliseo

H. Fernandez Vidal, Tomas Latasa, Hugo Mortera, Gloria Robles,Hugo Robles, and Georges Verhulst contributed to fieldwork withtheir advice, assistance, and equipment. This work benefited from thediscussions held with Frans Cupedo, Graciela Estevez, and Roger Vilaas well as the revisions by Zoltan Varga and two anonymous referees.An earlier version of this manuscript was greatly improved afterreviewed by David C. Lees. David Romero assisted us in preparingartwork. This investigation was performed under permits from allenvironmental authorities concerned. It was financially supported bygrants from Xunta de Galicia (PGIDIT06PXIB103258PR) and theUniversity of A Coruna. MV was funded by Xunta de Galicia (IsidroParga Pondal programme). NMMwas funded by the Spanish Ministryof Science (FPI programme, ref. BES-2008-002571).

Resumen

A algunas mariposas no les importa demasiado la topografıa: un unicolinaje genetico de Erebia euryale (Nymphalidae) a lo largo de lasmontanas septentrionales de la Penınsula Iberica

La filogeografıa de especies de montana ha revelado frecuentementeuna fuerte diferenciacion genetica entre especies que habitan distintossistemas montanosos. Tanto la morfologıa clasica como estudiosgeneticos previos sugerıan la distincion de las poblaciones pirenaicasde la mariposa Erebia euryale, aunque esta hipotesis precisaba serconfirmada mediante el estudio de las poblaciones mas occidentales delarea de distribucion de esta especie: la Cordillera Cantabrica. En estetrabajo se pretende describir la estructura genetica de las poblacionesde Erebia euryale de dicha area, donde la especie esta distribuida enlocalidades dispersas. Para ello, se estimaron la diversidad genetica y ladiferenciacion presentadas por 218 individuos muestreados enseis localidades de la Cordillera Cantabrica (Norte de Espana) ygenotipados para 17 loci alozımicos. Asimismo, se secuenciaron 816 pbdel gen mitocondrial COI en 49 individuos de dichas localidadesCantabricas y 41 especımenes procedentes de otras cinco poblacioneseuropeas. Los datos mitocondriales apoyaron la division en cuatrogrupos geneticos principales para las poblaciones no ibericas previa-mente sugeridos por otros autores en base a datos alozımicos. Losresultados procedentes de los marcadores mitocondriales y nuclearesrevelaron la distincion de un unico linaje genetico Cantabrico-Pirenaico de Erebia euryale. La falta de estructura geografica, asıcomo la topologıa en forma de estrella mostrada por los haplotiposmitocondriales indicaron una expansion demografica en la poblacionancestral a las actualmente distribuidas en el Norte peninsular. Dichaexpansion probablemente estuvo relacionada con las oscilacionesclimaticas acaecidas durante el Pleistoceno tardıo. Dentro del linajeCantabrico-Pirenaico, los datos nucleares indicaron una clara estruc-tura genetica para las muestras procedentes de la Cordillera Canta-brica. Los individuos procedentes de San Isidro fueron los quemostraron una mayor diferenciacion con respecto a las otras cincolocalidades cantabricas, estructuradas a su vez en dos gruposdiferentes. En base a los resultados obtenidos, se propone y argumentala definicion de una unidad evolutiva significativa para el linajeCantabrico-Pirenaico de Erebia euryale. Nuestros resultados ilustranque la historia evolutiva de una especie puede sermoldeada por procesosindetectables utilizando unicamente ADNmt.

References

Albre J, Gers C, Legal L (2008) Molecular phylogeny of the Erebiatyndarus (Lepidoptera, Rhopalocera, Nymphalidae, Satyrinae)species group combining CoxII and ND5 mitochondrial genes:a case study of a recent radiation. Mol Phylogenet Evol 47:196–210.

*Ali M, Tuncman G, Cross NCP, Vidailhet M, Bokesoy I,Gitzelmann R, Cox TM (1994) Null alleles of the aldolase Bgene in patients with hereditary fructose intolerance. J Med Genet31:499–503.

*Alvarez-Lao D, Garcıa N (2010) Chronological distribution ofPleistocene cold-adapted large mammal faunas in the IberianPeninsula. Quat Int 212:120–128.



Ansell SW, Grundmann M, Russell SJ, Schneider H, Vogel JC (2008)Genetic discontinuity, breeding-system change and populationhistory of Arabis alpina in the Italian Peninsula and adjacent Alps.Mol Ecol 17:2245–2257.

Belkhir K, Castric V, Bonhomme F (2002) IDENTIX, a software totest for relatedness in a population using permutation methods. MolEcol Notes 2:611–614.

*Bjornerfeldt S, Hailer F, Nord M, Vila C (2008) Assortative matingand fragmentation within dog breeds. BMC Evol Biol 8:28.

Brussard PF, Ehrlich PR (1970) The population structure of Erebiaepipsodea (Lepidoptera: Satyrinae). Ecology 51:119–129.

Centeno-Cuadros A, Delibes M, Godoy JA (2009) Phylogeography ofSouthern Water Vole (Arvicola sapidus): evidence for refugia withinthe Iberian glacial refugium? Mol Ecol 18:3652–3667.

Cheddadi R, Vendramin GG, Litt T, Francois L, Kageyama M,Lorentz S, Laurent J-M, de Beaulieu J-L, Sadori L, Jost A, Lunt D(2006) Imprints of glacial refugia in the modern genetic diversity ofPinus sylvestris. Glob Ecol Biogeogr 15:271–282.

Clement M, Posada D, Crandall KA (2000) TCS: a computer programto estimate gene genealogies. Mol Ecol 9:1657–1660.

Cooper SJB, Ibrahim KM, Hewitt GM (1995) Postglacial expansionand genomy subdivision in the European grasshopper Chorthippusparallelus. Mol Ecol 4:49–60.

Crawford NG (2010) SMOGD: software for the measurement ofgenetic diversity. Mol Ecol Resour 10:556–557.

*Cupedo F (2007) Geographical variation and Pleistocene history ofthe Erebia pandrose – sthennyo complex (Nymphalidae; Satyrinae).Nota Lepid 30:329–353.

Cupedo F (2010) A revision of the infraspecific structure of Erebiaeuryale (Esper, 1805) (Nymphalidae: Satyrinae). Nota Lepid 33:85–106.

Deffontaine V, Ledevin R, Fontaine MC, Quere J-P, Renaud S, LiboisR, Michaux JR (2009) A relict bank vole lineage highlights thebiogeographic history of the Pyrenean region in Europe. Mol Ecol18:2489–2502.

*Dick CW, Heuertz M (2008) The complex biogeographic history of awidespread tropical tree species. Evolution 62:2760–2774.

Dupanloup I, Schneider S, Excoffier L (2002) A simulated annealingapproach to define the genetic structure of populations. Mol Ecol11:2571–2581.

Engelhardt CHM, Pauls SU, Haase P (2008) Population geneticstructure of the caddisfly Rhyacophila pubescens, Pictet 1834, northof the Alps. Fundam Appl Limnol 173:165–176.

England PR, Osler GHR, Woodworth LM, Montgomery ME, BriscoeDA, Frankham R (2003) Effects of intense versus diffuse populationbottlenecks on microsatellite genetic diversity and evolutionarypotential. Conserv Genet 4:595–604.

Evanno G, Regnaut S, Goudet J (2005) Detecting the number ofclusters of individuals using the software STRUCTURE: a simula-tion study. Mol Ecol 14:2611–2620.

Excoffier L, Lischer HEL (2010) Arlequin suite ver 3.5: a new series ofprograms to perform population genetics analyses under Linux andWindows. Mol Ecol Resour 10:564–567.

Felsenstein J (1989) PHYLIP – Phylogeny Inference Package (Version3.2). Cladistics 5:164–166.

Fraser DJ, Bernatchez L (2001) Adaptive evolutionary conservation:towards a unified concept for defining conservation units. Mol Ecol10:2741–2752.

Garcıa-Barros E, Munguira ML, Martın Cano J, Romo Benito H,Garcia-Pereira P, Maravalhas E (2004) Atlas of the Butterflies of theIberian Peninsula and Balearic Islands (Lepidoptera: Papilionoidea& Hesperioidea). Sociedad Entomologica Aragonesa, Zaragoza,Spain, pp 126–135.

Garcıa-Parıs M, Alcobendas M, Buckley D, Wake DB (2003)Dispersal of viviparity across contact zones in Iberian populationsof fire salamanders (Salamandra) inferred from discordance ofgenetic and morphological traits. Evolution 57:129–143.

Gibbard P, van Kolfschoten T (2004) The Pleistocene andHolocene Epochs. In: Gradstein FM, Ogg JG, Smith AG (eds),A Geologic Time Scale. Cambridge University Press, Cambridge,pp 441–452.

Gomez A, Lunt DH (2007) Refugia within refugia: patterns ofphylogeographic concordance in the Iberian Peninsula. In: Weiss S,Ferrand N (eds), Phylogeography of Southern European Refugia.Springer, Dordrecht, The Netherlands, pp 155–188.

Goudet J (1995) FSTAT version 1.2: a computer program to calculateF-statistics. J Hered 86:485–486.

Goudet J, Raymond M, de-Meeus T, Rousset F (1996) TestingDifferentiation in Diploid Populations. Genetics 144:1933–1940.

*Grandal d�Anglade A, Lopez-Gonzalez F, Vidal Romanı JR (1997)Condition factors in the distribution of large mammals from Galicia(NW Iberian Peninsula) during the Upper Quaternary. Cad LabXeol Laxe 22:43–66.

Guillaume C-P, Heulin B, Arrayago MJ, Bea A, Brana F (2000)Refuge areas and suture zones in the Pyrenean and Cantabrianregions: geographic variation of the female MPI sex-linked allelesamong oviparous populations of the lizard Lacerta (Zootoca)vivipara. Ecography 23:3–10.

Gutierrez D (1997) Importance of historical factors on species richnessand composition of butterfly assemblages (Lepidoptera: Rhopalo-cera) in a northern Iberian mountain range. J Biogeogr 24:77–88.

Habel JC, Schmitt T, Muller P (2005) The fourth paradigm pattern ofpost-glacial range expansion of European terrestrial species: thephylogeography of the Marbled White butterfly (Satyrinae, Lepi-doptera). J Biogeogr 32:1489–1497.

*Habel JC, Meyer M, El Mousadik A, Schmitt T (2008) Africa goesEurope: the complete phylogeography of the Marbled Whitebutterfly species complex Melanargia galathea ⁄ lachesis. Org DiversEvol 8:121–129.

Habel JC, Schmitt T, Meyer M, Finger A, Rodder D, Assmann T,Zachos FE (2010) Biogeography meets conservation: the geneticstructure of the endangered lycaenid butterfly Lycaena helle (Denis& Schiffermuller, 1775). Biol J Linn Soc Lond 101:155–168.

Hammouti N, Schmitt T, Seitz A, Kosuch J, Veith M (2010)Combining mitochondrial and nuclear evidences: a refined evolu-tionary history of Erebia medusa (Lepidoptera: Nymphalidae:Satyrinae) in Central Europe based on the COI gene. J Zoolog SystEvol Res 48:115–125.

Hewitt GM (1996) Some genetic consequences of ice ages, and theirrole in divergence and speciation. Biol J Linn Soc Lond 58:247–276.

Ho SYW, Phillips MJ, Cooper A, Drummond AJ (2005) Timedependency of molecular rate estimates and systematic overestima-tion of recent divergence times. Mol Biol Evol 22:1561–1568.

Holderegger R, Thiel-Egenter C (2009) A discussion of different typesof glacial refugia used in mountain biogeography and phylogeog-raphy. J Biogeogr 36:476–480.

Horn A, Stauffer C, Lieutier F, Kerdelhue C (2009) Complexpostglacial history of the temperate bark beetle Tomicus piniperdaL. (Coleoptera, Scolytinae). Heredity 103:238–247.

Hubisz MJ, Falush D, Stephens M, Pritchard JK (2009) Inferringweak population structure with the assistance of sample groupinformation. Mol Ecol Resour 9:1322–1332.

Jensen JL, Bohonak AJ, Kelley ST (2005) Isolation by distance, webservice. BMC Genet 6:1–6.

Jost L (2008) G(ST) and its relatives do not measure differentiation.Mol Ecol 17:4015–4026.

Kerdelhue C, Zane L, Simonato M, Salvato P, Rousselet J, Roques A,Battisti A (2009) Quaternary history and contemporary patterns in acurrently expanding species. BMC Evol Biol 9:220.

Kramp K, Huck S, Niketic M, Tomovic G, Schmitt T (2009) Multipleglacial refugia and complex postglacial range shifts of the obligatorywoodland plant Polygonatum verticillatum (Convallariaceae). PlantBiol 11:392–404.

Kuras T, Benes J, Fric Z, Konvicka M (2003) Dispersal patterns ofendemic alpine butterflies with contrasting population structures:Erebia epiphron and E. sudetica. Popul Ecol 45:115–123.

Lafranchis T (2000) Les papillons de jour de France, Belgique etLuxembourg et leurs chenilles. Editions Biotope, Meze, France, 448pp.

*Langley CH, Voelker RA, Brown AJL, Ohnishi S, Dickson B,Montgomery E (1981) Null allele frequencies at allozyme loci innatural populations ofDrosophilamelanogaster.Genetics99:151–156.



Librado P, Rozas J (2009) DnaSP v5: a software for comprehensiveanalysis of DNA polymorphism data. Bioinformatics 25:1451–1452.

Lunt DH, Zhang DX, Szymura JM, Hewitt GM (1996) The insectcytochrome oxidase I gene: evolutionary patterns and conservedprimers for phylogenetic studies. Insect Mol Biol 5:153–165.

Lynch M, Ritland K (1999) Estimation of pairwise relatedness withmolecular markers. Genetics 152:1753–1766.

Martin JF, Gilles A, Descimon H (2000) Molecular phylogeny andevolutionary patterns of the European satyrids (Lepidoptera:Satyridae) as revealed by mitochondrial gene sequences. MolPhylogenet Evol 15:70–82.

Martin JF, Gilles A, Lortscher M, Descimon H (2002) Phylogeneticsand differentiation among the western taxa of the Erebia tyndarusgroup (Lepidoptera : Nymphalidae). Biol J Linn Soc Lond 75:319–332.

Naydenov K, Senneville S, Beaulieu J, Tremblay F, Bousquet J (2007)Glacial vicariance in Eurasia: mitochondrial DNA evidence fromScots pine for a complex heritage involving genetically distinct refugiaat mid-northern latitudes and Asia Minor. BMC Evol Biol 7:233.

Nazari V, Sperling FAH (2007) Mitochondrial DNA divergence andphylogeoraphy in western Palaearctic Parnassiinae (Lepidoptera:Papilionidae): how many species are there? Insect Syst Evol 38:121–138.

Nei M (1972) Genetic distances between populations. Am Nat106:283–291.

Pauls SU, Lumbsch HT, Haase P (2006) Phylogeography of themontane caddisfly Drusus discolor: evidence for multiple refugia andperiglacial survival. Mol Ecol 15:2153–2169.

Pauls SU, Theissinger K, Ujvarosi L, Balint M, Haase P (2009)Patterns of population structure in two closely related, partiallysympatric caddisflies in Eastern Europe: historic introgression,limited idspersal, and cryptic diversity. J North Am Benthol Soc28:517–536.

Pena C, Nylin S, Walhberg N (in press) The radiation of Satyrinibutterflies (Nymphalidae: Satyrinae): a challenge for phylogeneticmethods. Zool J Linn Soc.

Perez T, Albornoz J, Domınguez A (2002) Phylogeography of chamois(Rupicapra spp.) inferred from microsatellites. Mol Phylogenet Evol25:524–534.

Petit RJ, Hampe A, Cheddadi R (2005) Climate changes and treephylogeography in the Mediterranean. Taxon 54:877–885.

Potter E-K, Esat TM, Schellmann G, Radtke U, Lambeck K,McCulloch MT (2004) Suborbital-period sea-level oscillations dur-ing marine isotope substages 5a and 5c. Earth Planet Sci Lett225:191–204.

Pritchard JK, Stephens M, Donnelly P (2000) Inference of populationstructure using multilocus genotype data. Genetics 155:945–959.

Pyhajarvi T, Salmela MJ, Savolainen O (2008) Colonization routes ofPinus sylvestris inferred from distribution of mitochondrial DNAvariation. Tree Genet Genomes 4:247–254.

*Queney G, Ferrand N, Weiss S, Mougel F, Monnerot M (2001)Stationary distributions of microsatellite loci between divergentpopulation groups of the European rabbit (Oryctolagus cuniculus).Mol Biol Evol 18:2169–2178.

Quevedo M, Banuelos MJ, Saez O, Obeso JR (2006) Habitat selectionby Cantabrian capercaillie Tetrao urogallus cantabricus at the edgeof the species� distribution. Wildl Biol 12:267–276.

*Raspe O, Jacquemart A-L (1998) Allozyme diversity and geneticstructure of European populations of Sorbus aucuparia L. (Rosa-ceaee: Maloideae). Heredity 81:537–545.

Robledo-Arnuncio JJ, Collada C, Alıa R, Gil L (2005) Geneticstructure of montane isolates of Pinus sylvestris L. in a Mediterra-nean refugial area. J Biogeogr 32:595–605.

Rodrıguez-Munoz R, Mirol PM, Segelbacher G, Fernandez A,Tregenza T (2007) Genetic differentiation of an endangered caper-caillie (Tetrao urogallus) population at the Southern edge of thespecies range. Conserv Genet 8:659–670.

Rogers AR, Harpending H (1992) Population growth makes waves inthe distribution of pairwise genetic differences. Mol Biol Evol 9:552–569.

Rousselet J, Zhao R, Argal D, Simonato M, Battisti A, Roques A,Kerdelhue C (2010) The role of topography in structuring the

demographic history of the pine processionary moth, Thaumetopoeapityocampa (Lepidoptera: Notodontidae). J Biogeogr 37:1478–1490.

Rousset F (1997) Genetic differentiation and estimation of gene flowfrom F-statistics under isolation by distance. Genetics 145:1219–1228.

Rousset F (2008) Genepop�007: a complete reimplementation of theGenepop software for Windows and Linux. Mol Ecol Resour 8:103–106.

Rubiales JM, Garcıa-Amorena I, Garcıa Alvarez S, Gomez Manz-aneque F (2008) The Late Holocene extinction of Pinus sylvestris inthe western Cantabrian Range (Spain). J Biogeogr 35:1840–1850.

Schmitt T (2009) Biogeographical and evolutionary importance of theEuropean high mountain systems. Front Zool 6:9.

Schmitt T, Haubrich K (2008) The genetic structure of the mountainforest butterfly Erebia euryale unravels the late Pleistocene andpostglacial history of the mountain coniferous forest biome inEurope. Mol Ecol 17:2194–2207.

Schmitt T, Seitz A (2004) Low diversity but high differentiation: thepopulation genetics of Aglaope infausta (Zygaenidae: Lepidoptera).J Biogeogr 31:137–144.

Schmitt T, Hewitt GM, Muller P (2006) Disjunct distributions duringglacial and interglacial periods in mountain butterflies: Erebiaepiphron as an example. J Evol Biol 19:108–113.

*Semerikov VL, Semerikov LF, Lascoux M (1999) Intra- andinterspecific allozyme variability in European Larix Mill. Species.Heredity 82:193–204.

Shackleton NJ, Sanchez-Goni MF, Pailler D, Lancelot Y (2003)Marine Isotope Substage 5e and the Eemian Interglacial. GlobalPlanet Change 36:151–155.

Siegismund HR (1993) G-Stat, Version 3, Genetical StatisticalPrograms for the Analysis of Population Data. The Arboretum,Royal Veterinary and Agricultural University, Horsholm, Denmark.

Sonderegger P (2005) Die Erebien der Schweiz (Lepidoptera: Satyri-nae, genus Erebia). Selbstverlag (=private edition), Biel ⁄Bienne,Switzerland, 712 pp.

van Swaay C, Cuttelod A, Collins S, Maes D, Lopez Munguira M,Sasic M, Settele J, Verovnik R, Verstrael T, Warren M, Wiemers M,Winhoff I (2010) European Red List of Butterflies. PublicationsOffice of the European Union, Luxembourg, pp 35.

Templeton AR, Crandall KA, Sing CF (1992) A cladistic analysis ofphenotypic association with haplotypes inferred from restrictionendonuclease mapping and DNA sequence data. III. Cladogramestimation. Genetics 132:619–633.

Tennent WJ (2008) A checklist of the satyrine genus Erebia (Lepi-doptera) (1758–2006). Zootaxa 1900:1–109.

Tolman T, Lewington R (1997) Butterflies of Britain and Europe.Harper Collins, London, pp 210.

Vandewoestijne S, Baguette M, Brakefield PM, Saccheri IJ (2004)Phylogeography of Aglais urticae (Lepidoptera) based on DNAsequences of the mitochondrial COI gene and control region. MolPhylogenet Evol 31:630–646.

*Varga Z (1998) Die Erebien der Balkanhalbinsel und Karpaten IV.Ubersicht der subspezifischen Gliederung und der verbreitung derErebia-Arten in der Balkanhalbinsel und in den Karpaten (Lepi-doptera, Nymphalidae, Satyrinae). Entomol Rom 3:15–29.

*Verity R (1927) Notes sur quelques Rhopaloceres d�Espagne [Lep.].Bull Soc Entomol France 1927:172–176.

Vialette A, Guiller A, Bellido A, Madec L (2008) Phylogeography andhistorical demography of the Lusitanian snail Elona quimperianareveal survival in unexpected separate glacial refugia. BMC EvolBiol 8:339.

Vila M, Bjorklund M (2004) The utility of the neglected mitochondrialcontrol region for evolutionary studies in Lepidoptera (Insecta). JMol Evol 58:280–290.

Vila M, Vidal-Romanı JR, Bjorklund M (2005) The importance oftime scale and multiple refugia: Incipient speciation and admixtureof lineages in the butterfly Erebia triaria (Nymphalidae). MolPhylogenet Evol 36:249–260.

Vila M, Cassel Lundhagen A, Thuman KA, Stone JR, Bjorklund M(2006) A new conservation unit in the butterfly Erebia triaria(Nymphalidae) as revealed by nuclear and mitochondrial markers.Ann Zool Fenn 43:72–79.



*Von der Goltz DH (1930) In: Seitz A (ed.), Die Gross-Schmetterlingeder Erde. Supplement 1: 147. Alfred Kerner, Stuttgart, pp 147 andplate 9.

Weir BS, Cockerham CC (1984) Estimating F-statistics for the analysisof population structure. Evolution 38:1358–1370.

Weir BS, Hill WG (2002) Estimating F-statistics. Annu Rev Genet36:721–750.

Supporting Information

Additional Supporting Information may be found in the onlineversion of this article:

Appendix S1. Phylogeography of Erebia euryale and intraspe-cific taxonomy.Table S2. Allele frequencies of all 13 polymorphic allozyme

loci of six populations of Erebia euryale from the CantabrianMts.Table S3. Single and multilocus genetic diversity estimates for

13 polymorphic loci.Table S4. FIS values per locus and sampling locality.Figure S5. Results of the Isolation-By-Distance analyses.Figure S6. Unimodal mismatch distribution for the Pyrenean-

Cantabrian lineage consistent with the expected suddenexpansion model.

Appendix S7. Genetic diversity and resolution of the allozymemarkers.Figure S8. (a) Log-likelihood of the allozyme data on 16 Erebia

euryale populations given K clusters that were obtained afterfive runs of the STRUCTURE algorithm; (b) CorrespondingDK statistic (Evanno et al. 2005) showing peaks at K ¼ 2 and

K ¼ 4.Table S9. Proportion of missing data (italics) and privatealleles (bold) for the eleven populations and 17 loci of Erebiaeuryale analysed by Schmitt and Haubrich (2008).

Figure S10. Approximate distribution area (grey surface) andsampling locatilities surveyed by Schmitt and Haubrich (2008)(bold) and the present work.

Table S11. Sampling information.Please note: Wiley-Blackwell are not responsible for the

content or functionality of any supporting materials supplied

by the authors. Any queries (other than missing material)should be directed to the corresponding author for the article.



1

Supporting Information

S1. Phylogeography of Erebia euryale and intraspecific taxonomy.

S.1.1. Iberian phylogeography

The hypothesis of a distinct genetic lineage for the Cantabrian and Pyrenean

populations of Erebia euryale is supported by classical intraspecific taxonomy. Erebia

euryale antevortes *Verity (1927) at the Hautes-Pyrénées (southwestern France); ssp.

pyreaenaeicola *von der Goltz (1930) at the eastern Pyrenees; and ssp. cantabricola *Verity

(1927) in the Cantabrian Mountains (northern Spain). Although the genus Erebia is one of the

more nomenclaturally controversial genera among the Holarctic Lepidoptera (Albre et al.

2008) due to the high number of species and intraspecific taxa, recent morphological work

supports the significant differentiation between the subspecies antevortes and pyraenaeicola.

However, individuals from some Pyrenean localities also present intermediate features.

Cantabrian populations may also be very variable concerning wing pattern, somewhat like

those intermediate populations found in the Pyrenees (F. Cupedo, pers. comm.). A recent

morphological study surveying not only wing traits, but also male genitalia, classified the

subspecies antevortes, pyraenaeicola, and cantabricola (as well as five other subspecies) as

belonging to a single group: Erebia euryale (group euryale, see below) (Cupedo 2010).

Interestingly, there is fairly high morphological variability within the Cantabrian large ringlets

from San Glorio (n = 28, F. Cupedo pers.comm.). It is worth mentioning that the description

of the subspecies cantabricola (*Verity 1927) was based on individuals from Pajares, a

locality surveyed by the present study.

West-East Pyrenean differentiation was found in several other Erebia species, either

from the morphological (e.g., Erebia pandrose – sthennyo complex; *Cupedo 2007) or

genetic perspective (e.g., Erebia epiphron; Schmitt et al. 2006). We acknowledge that our

study lacked samples from the subspecies antevortes (W Pyrenees), and therefore, some new

lineages might appear when surveying additional northern Iberian localities. In fact, the

occurrence of a Basque/SW French/W Pyrenean lineage in such divergent organisms as the

land snail Elona quimperana (Vialette et al. 2008) or the bank vole Myodes glareolus

(Deffontaine et al. 2009) opens the possibility of a second genetic lineage in the western

Pyrenees. Such a hypothesis is triggered by the classic circumscription of the morphological

subspecies E. euryale pyraenaicola in the western Pyrenees. However, in the light of ongoing

research, the intraspecific classification of the northern Iberian Erebia euryale into three

separate subspecies remains controversial (F. Cupedo, pers. comm.).

2

We postulate that the distinctiveness of the Pyrenean-Cantabrian lineage found in

Erebia euryale may have resulted from an ancestral population spreading from France into

North Spain from the West and East fringes of the Pyrenees. The Pyrenees acted as a barrier

for many species crossing from France to Iberia and vice-versa. However, the areas between

this mountain range and the Bay of Biscay and the Mediterranean were suitable habitats for

the migration of many animal taxa. This is inferred, for instance, from the type of dated

fossils found in the large number of paleontological sites in the northeastern Spanish province

of Girona and all along the so-called “northern-corridor” (strip between the Cantabrian

Mountains and the Bay of Biscay) further down to Portugal (*Grandal d’Anglade et al. 1997;

*Álvarez-Lao and García 2010). Our hypothesis of a East-West entrance of Erebia euryale

into Spain at some point in the Upper Pleistocene does not preclude the existence of one or

more ancient populations already present in Iberia, as there is a wide geographic gap between

San Glorio and La Glèbe not covered by the present study.

S.1.2. Implications for intraspecific taxonomy

Our results at European scale partially supported a four group genetic division,

congruent with the results by Schmitt and Haubrich (2008). Our allozymes and mitochondrial

results indicated (i) a Pyrenean-Cantabrian lineage (Pyrenean lineage of isarica), (ii) a

western Alps grouping (adyte), (iii) an eastern Alps cluster (isarica from Obertauern and

Trauneralm plus ocellaris from Kalkstein), and (iv) a final eastern Europe clade (syrmia).

The subspecies ocellaris was considered and ecotype of ssp. isarica by Schmitt and

Haubrich (2008), which is supported by our nuclear data (Fig. 2), but not by the mitochondrial

results (Fig. 5). In the morphology based intraspecific taxonomy suggested by Cupedo (2010)

the subspecies pyraenaeicola, cantabricola, isarica, ocellaris, and syrmia were clustered into

the euryale group. Further support to synonymise E. euryale isarica with E. euryale euryale

(*Varga 1998) comes from the fact that the nominotypic population “Riesengebirge” is only a

marginal isolate of the northern Alpine-Carpathian genetic lineage (Z. Varga, pers. comm.).

The taxonomic status of E. euryale syrmia Fruhstorfer, 1910 from eastern/southern

Carpathian – Balkan area also needs further work, as this subspecies was described from

Bosnia (Trebević) and the eastern Balkan populations show some morphological

differentiation (Z. Varga, pers. comm.).

Both Schmitt and Haubrich (2008) and Cupedo (2010) defined an adyte group. This is

in line with our nuclear data, as the sample from Val d’Isère (assigned to the subspecies adyte

by Schmitt and Haubrich 2008) showed a significant differentiation from the other two

3

localities from the West Alps (St. Martin and Partnun). However, the resolution of the

markers used in the present study prevented us from drawing any further conclusion about the

differentiation within this West Alpine group.

To conclude, the intraspecific taxonomy of Erebia euryale deserves further molecular

research, as there are several issues that need to be clarified. For instance, the degree of

differentiation of the kunzi-group, confined to a restricted Italian part of the southern Alps

(Cupedo 2010).

4

S2. A

llele

freq

uenc

ies

of a

ll 13

pol

ymor

phic

allo

zym

e lo

ci o

f six

pop

ulat

ions

of Erebia euryale

from

the

Can

tabr

ian

Mts

.

Loc

us

Alle

le

San

Glo

rio

Paj

ares

San

Isid

ro

Som

i

edo

Lei

tari

eg

os

Pand

etra

ve

6Pg

dh

1 0.

069

0 0.

216

0 0

0

2

0.46

6 0.

6

67

0 0.

647

0.46

2 0.

435

3

0.46

6 0.

3

33

0.78

4 0.

353

0.52

6 0.

548

4

0 0

0 0

0.01

3 0.

016

Apk

3

0.98

5 1.

0

00

1.00

0 1.

000

1.00

0 1.

000

4

0.01

5 0

0 0

0 0

G6P

dh

1 0.

136

0.3

06

0.21

6 0.

309

0.51

3 0.

392

2

0 0

0 0.

015

0 0

3

0.81

8 0.

6

81

0.78

4 0.

676

0.48

7 0.

608

4

0 0.

0

14

0 0

0 0

5

7

0.04

5 0

0 0

0 0

Gap

dh

1 0

0.0

83

0 0

0 0

2

1.00

0 0.

9

17

1.00

0 1.

000

1.00

0 1.

000

Aat

1 4

1.00

0 0.

9

58

0.98

6 1.

000

1.00

0 1.

000

5

0 0.

0

42

0.01

4 0

0 0

Aat

2 2

0.01

6 0

0 0

0 0.

162

5

0.54

7 0.

5

83

0.86

5 0.

853

0.85

0 0.

662

6

0.01

6 0

0 0

0 0

7

0.42

2 0.

4

17

0.13

5 0.

147

0.15

0 0.

176

Idh1

2

1.00

0 0.

9

72

0.94

4 0.

882

0.96

2 1.

000

4

0 0

0 0.

029

0 0

8

0 0.

0

28

0.05

6 0.

088

0.03

8 0

Idh2

4

0.03

0 0

0 0

0 0

6

5

0.97

0 1.

0

00

1.00

0 1.

000

0.98

8 1.

000

9

0 0

0 0

0.01

3 0

Mdh

2

1 0.

061

0.0

14

0.01

4 0

0 0.

036

2

0.84

8 0.

9

58

0.90

0 1.

000

1.00

0 0.

804

3

0.01

5 0.

0

28

0.01

4 0

0 0.

054

4

0.07

6 0

0.07

1 0

0 0.

089

5

0 0

0 0

0 0.

018

Me

2 1.

000

1.0

00

1.00

0 0.

986

1.00

0 0.

959

3

0 0

0 0.

014

0 0.

041

Pep

1 0.

015

0.0

14

0.08

1 0.

074

0.01

3 0.

081

2

0.98

5 0.

9

86

0.91

9 0.

926

0.98

8 0.

919

Pgi

4 0

0 0

0 0

0.01

4

5

0.04

5 0.

0

28

0.23

0 0

0 0

7

6

0.03

0 0.

0

69

0.01

4 0

0 0

7

0.89

4 0.

5

69

0.70

3 0.

871

0.83

7 0.

824

8

0.03

0 0.

3

06

0.05

4 0.

129

0.15

0 0.

162

9

0 0.

0

28

0 0

0.01

3 0

Pgm

1

0.01

6 0.

0

14

0.09

5 0

0 0

2

0.04

7 0.

0

28

0.04

1 0.

044

0.10

0 0

3

0.17

2 0.

2

50

0.33

8 0.

412

0.47

5 0.

500

5

0.43

8 0.

4

86

0.43

2 0.

309

0.37

5 0.

250

7

0.32

8 0.

2

22

0.09

5 0.

235

0.05

0 0.

233

8

0 0

0 0

0 0.

017

8

S3. Single and multilocus genetic diversity estimates for 13 polymorphic loci.

Calculations performed across the six Cantabrian populations of Erebia euryale. AR =

allelic richness (based on 27 individuals), HWE = deviations from Hardy-Weinberg

equilibrium (score U (global) test, H1 = heterozygote deficit) significance after

sequential Bonferroni correction is marked in bold. NS = p >0.5, * = p < 0.05, ** = p <

0.01, *** = p < 0.001. Average F-statistics and their standard errors (after symbol ±)

were calculated using a jacknifing-over-populations procedure. The G-test for

population differentiation was calculated not assuming random mating within samples.

Significance obtained after 10,000 randomisations.

Locus AR HWE FIS FST FIT G-

test

6Pgdh 3.189 * +0.130 ±

0.097

0.174 ±

0.132

+0.284 ±

0.153***

***

Apk 1.126 - -0.002 ±

0.000

0.003 ±

0.001

0.000 ± 0.000

G6Pdh 2.581 NS +0.072

±0.118

0.060 ±

0.041

+0.131 ± 0.137 ***

Gapdh 1.550 NS +0.524 ±

0.238

0.123 ±

0.056

+0.611 ± 0.278

Aat1 1.413 * -0.034 ±

0.021

0.026 ±

0.019

-0.008 ± 0.005

Aat2 2.953 NS -0.018 ±

0.072

0.094 ±

0.018

+0.078 ± 0.075 ***

Idh1 2.107 NS -0.067 ±

0.026

0.030 ±

0.024

-0.036 ± 0.015

Idh2 1.356 NS -0.016 ±

0.012

0.013 ±

0.012

-0.003 ± 0.002

Mdh2 3.133 NS +0.067 ±

0.107

0.037 ±

0.023

+0.102 ± 0.099

Me 1.412 NS -0.033 ±

0.019

0.025 ±

0.017

-0.008 ± 0.006

Pep 1.934 NS -0.063 ±

0.009

0.014 ±

0.009

-0.048 ± 0.018

9

Pgi 4.057 NS -0.101 ± 0.03 0.084 ±

0.033

-0.009 ± 0.037 ***

Pgm 4.777 *** +0.167 ±

0.054

0.037 ±

0.013

+0.198 ±

0.061***

***

ALL 2.43 ±

1.39

** +0.068 ±

0.045

0.078 ±

0.023

+0.140 ±

0.044***

***

10

S4. FIS values per locus and sampling locality. Estimation of p-values was based on

1560 randomisations. Indicative adjusted nominal level (0.05) for one table was =

0.00064. NA: non applicable. The proportion of randomisations that gave a larger FIS

(i.e., significant deficit of heterozygotes) resulted in a single significant value (in bold).

There was no significant FIS value when testing for excess of heterozygotes.

SanGlorio Pajares SanIsidro Somiedo Leitariegos Pandetrave

6Pgdh -0.087 0.014 0.216 0.496 0.259 -0.124

Apk 0.000 NA NA NA NA NA

G6Pdh -0.158 0.074 0.216 0.093 0.395 -0.292

Gapdh NA 0.286 NA NA NA NA

Aat1 NA -0.029 0.000 NA NA NA

Aat2 0.237 -0.129 -0.143 -0.158 0.032 -0.058

Idh1 NA -0.014 -0.045 -0.091 -0.027 NA

Idh2 -0.016 NA NA NA 0.000 NA

Mdh2 0.230 -0.019 0.085 NA NA -0.114

Me NA NA NA 0.000 NA -0.029

Pep 0.000 0.000 -0.075 -0.065 0.000 -0.075

Pgi -0.062 -0.095 -0.187 -0.133 0.016 -0.090

Pgm 0.267 0.118 0.059 0.060 0.367 0.121

All 0.096 0.013 0.028 0.088 0.249 -0.077

11

S5. Results of the Isolation-By-Distance analyses. (a) The European nuclear dataset

did not conform to the isolation-by-distance model, as there was no significant

correlation between genetic differentiation in relation to geographic distance (Z =

39647.54, r = 0.15, one-sided p = 0.09); (b) the Pyrenean-Cantabrian dataset did not

conform either to the isolation-by-distance model (Z = 649.47, r = 0.35, one-sided p =

0.21).

(a)

(b)

12

S6. Unimodal mismatch distribution for the Pyrenean-Cantabrian lineage

consistent with the expected sudden expansion model (SDD=0.00246, p = 0.44).

13

S7. Genetic diversity and resolution of the allozyme markers.

The six Cantabrian populations analysed in the present study were monomorphic

for four out of the 17 surveyed allozyme loci. Those markers were polymorphic for

other samples of Erebia euryale, as revealed by Schmitt and Haubrich (2008), who

reported that the total percentage of polymorphic loci ranged from 47.1 (La Glèbe,

souteastern France) to 88.2 (Groapa Seaca and Sinaia, two Romanian localities). The

Pyrenean population of La Glèbe was also monomorphic for the same four loci as the

Cantabrian samples, as well as for five other markers (Apk, Gapdh, Idh1, Idh2, and

Mdh2).

Taking into account the 13 polymorphic loci, our study on 218 adult males of

Cantabrian large ringlets revealed an average of 1.92 alleles per locus. The proportion

of missing data ranged from 0.08 (locus 6Pgdh at Somiedo) to 0.24 (locus Mdh2 at

Pandetrave). The only locus globally departing from HWE was Pgm. This pattern

indicated that the deviation from HWE was local, a locus-specific phenomenon, and

might be caused by processes such as selection, assortative mating or genotyping

problems. We cannot exclude any of those hypotheses, but we acknowledge the

presence of null alleles as the most likely explanation (see below).

Leitariegos was the only locality departing from HWE due to a significant

deficit of heterozygotes. Such an excess of homozygotes might be due to processes such

as under-dominant selection, unintentional mixing of separate breeding units (i.e.,

Wahlund effect), inbreeding or genotyping errors (e.g., null alleles). Homozygote

excess was found in two out of the eight loci that resulted polymorphic in that locality,

and none of them remained significant after Bonferroni correction, so selection does not

appear as a plausible explanation. The Wahlund effect is unlikely, as each locality was

sampled in plots less than one hectare in size and over one or two consecutive days.

However, it would be still possible to find intra-locality genetic structuring, as different

morphs (i.e., putative subspecies) of Erebia euryale occur sympatrically but do not

hybridise in some European localities (F. Cupedo, pers. comm.). The results from the

Bayesian clustering of individuals implemented in STRUCTURE 2.3.3 did not support

the sub-structuring within Leitariegos (Fig. 4). This might, in turn, favour inbreeding as

an explanation. Inbreeding should have affected similarly to all markers, causing a large

proportion of loci affected by significant positive FIS values (e.g., 74% of polymorphic

loci in *Björnerfeldt et al. 2008). This did not seem to be the case of Leitariegos (Table

S4). Nevertheless, we tested whether the surveyed individuals from Leitariegos

14

belonged to kin groups, instead of representing a random mating deme. We calculated

the average (r = -0.025) and dispersion (variance = 0.143, SD = 0.379) of the

relatedness coefficient r (Lynch and Ritland 1999) from the output provided by

IDENTIX for every possible pair of individuals sampled at Leitariegos (data available

on request). Data from Leitariegos did not significantly depart from the null hypothesis

of no relatedness neither when the test was performed on the means nor the variances of

the relatedness coefficient.

Finally, we cannot exclude the excess of homozygotes in Leitariegos to be due

to the presence of null alleles and/or genotyping errors. We consider the first possibility

as more likely. Null alleles are usually a major matter of concern when working with

microsatellites in Lepidoptera, but they also occur in allozyme loci as no protein

product or a protein product that is enzymatically non-functional (*Langley et al. 1981

and references therein). It is noteworthy that semi-null alleles have been also reported in

allozymes, as protein activity may differ when extracted from different tissues

(*Semerikov et al. 1999). Concerning lack of product, it was surprising the high

proportion of missing data for three loci at Pandetrave: 6Pgdh (0.16), Mdh2 (0.24), and

Pgm (0.18). This is a much higher level of missing data than found by Schmitt and

Haubrich (2008), which ranged from zero (most cases) to 0.12 (locus Pgm at Val

d’Isère, France) (Table S9). To the best of our knowledge, the preservation of the

Cantabrian samples should not be the reason for this allelic pattern, as individuals were

kept alive until arrival to the -80ºC freezing facilities and after five months transported

to Germany (where stored in liquid nitrogen until analysis) in dry ice in less than 16

hours. This same transport protocol and personnel produced more accurate results in

other allozymic studies in butterflies from far distant localities (e.g., *Habel et al. 2008).

We would like to highlight several facts concerning null alleles that may be of

importance when interpreting a homozygote excess such as the one found at

Leitariegos. A single null allele may lead a population to depart from HWE (e.g.,

*Queney et al. 2001). Also, the assumed codominant mode of inheritance for a set of

allozyme loci may be disturbed by recessive null alleles (e.g., *Raspé and Jacquemart

1998). An interesting idea that might be applicable in this scenario would be that the

putative null alleles causing Leitariegos not to adjust to HWE were private, exclusive to

that locality. For instance, a single highly polymorphic microsatellelite locus (out of a

panel of 18 markers) failed to amplify in two Iberian localities which belonged to a

distinct genetic lineage of the fish Gasterosteus aculeatus (Vila et al. unpublished).

15

Leitariegos and San Glorio were the only Cantabrian localities showing one private

allele each, so it might as well be that a private null allele is preventing Leitariegos to

adjust to HWE. Unfortunately, this idea remains as speculative in the present study, as

mitochondrial results do not support a distinct lineage for Leitariegos. Lastly, the

presence of allozyme null alleles has also been linked to deleterious traits expressed as a

result of inbreeding (*Ali et al. 1994). To conclude, we postulate that null alleles may

play a role in some Cantabrian localities, particularly at Leitariegos, and therefore the

results of the presented allozyme data should be interpreted with caution.

Demonstration of the presence of null alleles in the Cantabrian populations of Erebia

euryale, as well as if they are related to a putative genetic distinctiveness of the

westernmost part of the species distribution area (i.e., private null alleles) or to

inbreeding opens the possibility for a future study in combination with other genetic

markers.

16

S8. (a) Log-likelihood of the allozyme data on 16 Erebia euryale populations given

K clusters that were obtained after five runs of the STRUCTURE algorithm; (b)

Corresponding ∆K statistic (Evanno et al. 2005) showing peaks at K = 2 and K = 4,

which indicate that those are the best solutions for K given the data. Multimodal

distributions for ∆K are not unusual (e.g., *Dick and Heuertz 2008) and typically

indicate a hierarchical structure.

(a)

-8500.00

-8000.00

-7500.00

-7000.00

-6500.00

-6000.00

1 3 5 7 9 11 13 15 17

K

average L(K) +- SD

(b)

0.00

50.00

100.00

150.00

200.00

250.00

1 3 5 7 9 11 13 15 17

K

DELTA(K)

17

S9. Proportion of missing data (italics) and private alleles (bold) for the eleven

populations and 17 loci of Erebia euryale analysed by Schmitt and Haubrich

(2008). F: France, CH: Switzerland, A: Austria, Ro: Romania.

Population Proportion of missing data (locus) Locus Allele Freq

La Glèbe (F) - - - -

St Martin (F) 0.08 (6Pgdh) Aat2 4 0.015

Val d'Isère (F) 0.10 (6Pgdh), 0.12 (Pgm) Idh2 4 0.013

Partnun (CH) 0.06 (Fum) - - -

Kalkstein (A) 0.08 (Pgm) Idh1 4 0.014

Obertauern (A) - Aat1 5 0.028

Obertauern (A) Idh2 6 0.014

Schöneck (A) - - - -

Trauneralm (A) - Pgi 9 0.013

Trauneralm (A) Pgm 10 0.013

Groapa Seaca (Ro) - Idh1 5 0.013

Groapa Seaca (Ro) Idh2 9 0.013

Groapa Seaca (Ro) Mdh1 4 0.013

Groapa Seaca (Ro) Mdh1 5 0.013

Groapa Seaca (Ro) Pk 5 0.013

Sinaia (Ro) - G6Pdh 6 0.013

Sinaia (Ro) Gapdh 4 0,013

Sinaia (Ro) Idh1 7 0,013

18

S10. Approximate distribution area (grey surface) and sampling locatilities

surveyed by Schmitt and Haubrich (2008) (bold) and the present work (cf. Fig.

1). The names of subspecies of Erebia euryale cited in the text appear in italics

and are placed near the areas where they occur (adapted from Tolman and

Lewington (1997) and Cupedo (2010)). GLB: La Glèbe, STM: St. Martin, VIS:

Val d’Isère, PAT: Partnun, KAL: Kalkstein, OBE: Obertauern, TRA:

Trauneralm, GRO: Groapa Seaca, SIN: Sinaia, RIL: Rila Mountains, GLO: San

Glorio, PAJ; Pajares, ISI: San Isidro, SOM: Somiedo, LEI: Leitariegos, PAN:

Pantedrave

19

S11. Sampling information.

Sample site Geographic

coordinates

N Elevation (meters above sea

level)

Date of

collection

San Glorio 43º04’N, 4º62’W 33 1609 July 18, 2006

Pajares 42º59’N, 5º45’ W 36 1378 July 16, 2006

San Isidro 43º04’N, 5º25’W 37 1520 July 11, 2006

Somiedo 43º01’N, 6º13’W 35 1486 July 14, 2006

Leitariegos 42º59’N, 6º24’W 40 1525 July 6, 2006

Pandetrave 43º06’N, 4º32’W 37 1562 July 23, 2006

Some butterflies do not care much about topography: a ... · the Iberian Peninsula. Most of them...

Documents

Transcript of Some butterflies do not care much about topography: a ... · the Iberian Peninsula. Most of them...