Received: 13 April 2019 | Revised: 9 October 2019 | Accepted: 22 October 2019 DOI: 10.1111/jzs.12356

ORIGINAL ARTICLE

Integrative analyses on Western Palearctic reveal a mosaic of nascent species

Leonardo Platania1 | Raluca Vodă2 | Vlad Dincă3 | Gerard Talavera1 | Roger Vila1 | Leonardo Dapporto4

1Institut de Biologia Evolutiva (CSIC- Universitat Pompeu Fabra), Barcelona, Spain Abstract 2Department of Life Sciences and Systems occurring in the Mediterranean apparently have reduced gene Biology, University of Turin, Turin, Italy flow over sea straits, and for several species, recent wide-scale biodiversity sur- 3Ecology and Genetics Research Unit, University of Oulu, Oulu, Finland veys indicate the existence of divergent mitochondrial lineages. Here, we apply an 4Dipartimento di Biologia dell'Università di integrative approach and examine the phylogeography of the genus Lasiommata in Firenze, Firenze, Italy the Western Palearctic. Our research comprised molecular analyses (mitochondrial

Correspondence and nuclear DNA) and geometric morphometrics (wings and genitalia) for two main Leonardo Platania, Institut de Biologia species groups, and a comparative GMYC analysis, based on COI, of all the tribes Evolutiva (CSIC-Universitat Pompeu Fabra), Barcelona, Spain. within Satyrinae from this region. The GMYC approach revealed a particularly fast Email: [email protected] coalescence rate in the Parargina subtribe. The Lasiommata group was divided into

Funding information 12 evolutionary significant units: six clades for the L. maera species group, five for Ministerio de Economía y Competitividad, the L. megera species group, and one for L. petropolitana, with divergences of about Grant/Award Number: CGL2013-48277-P and CGL2016-76322-; Università degli Studi 1%. The patterns of COI were mirrored by ITS2 in L. maera, but the two markers di Firenze; H2020 Marie Skłodowska-Curie were generally inconsistent in L. megera. On the contrary, morphological differences Actions, Grant/Award Number: 609402- 2020 and 625997 were coherent with the results of COI for L. megera, but less clearly so for L. maera. L. paramegaera and L. meadewaldoi were considerably differentiated for all the ana- lyzed markers and likely proceeded faster in the process of speciation because of geographic isolation and reduced effective population size, rendering the rest para- phyletic. Our study illustrates the continuous nature of speciation and the difficulties of delimiting species. In Lasiommata, the recognition of taxa as diverging lineages or distinct, possibly paraphyletic species, mostly depends on the criteria adopted by different species concepts.

KEYWORDS COI, ITS2, Lasiommata, morphometrics, speciation

1 | INTRODUCTION become an urgent task in order to better preserve it (Dirzo et al., 2014; Wheeler et al., 2012). Speciation processes manifest as a con- Given alarming rates of biodiversity loss during the Anthropocene, tinuum of divergence, during which recognizing boundaries between understanding how diversity is generated and maintained has species or among populations diverging at various grades may be a difficult matter (Hendry, 2009; Hendry, Bolnick, Berner, & Peichel, Contributing authors: Raluca Vodă ([email protected]), Vlad Dincă (vlad.e.dinca@ 2009; Mallet, Beltrán, Neukirchen, & Linares, 2007; Nosil, Harmon, gmail.com), Gerard Talavera ([email protected]), Roger Vila ([email protected]), Leonardo Dapporto ([email protected]). & Seehausen, 2009; Seehausen, 2009; Shaw & Mullen, 2014).

J Zool Syst Evol Res. 2020;00:1–14. wileyonlinelibrary.com/journal/jzs © 2020 Blackwell Verlag GmbH | 1 2 | PLATANIA et al.

This is attested by the existence of a high number of species con- mitochondrial lineages within generally accepted species, suggest- cepts, which sometimes disagree in recognizing if particular taxa ing ongoing differentiation or even the presence of cryptic species should be considered as distinct species (de Queiroz, 2005, 2007). (Dincă, Dapporto, & Vila, 2011; Dincă, Lukhtanov, & Talavera, 2011; In the speciation continuum, a large number of genes, expressing Dincă et al., 2015; Dincă, Zakharov, Hebert, & Vila, 2011; Hausmann a wide array of functional/phenotypic traits, are expected to diverge et al., 2011). Some of these cases have been subsequently studied at varied rates, producing discrepancies of traits between popula- in more detail by using integrative approaches and occasionally re- tions (Hendry, 2009; Hendry et al., 2009). According to observed vealed that species diversity is higher than expected even in inten- geographic patterns, speciation events have been traditionally cat- sively studied areas such as Europe (Dapporto, Vodă, Dincă, & Vila, egorized as sympatric, allopatric or parapatric (Butlin, Galindo, & 2014; Dincă, Zakharov, et al., 2011; Habel et al., 2017; Hernández- Grahame, 2008; Coyne & Orr, 2004; Gavrilets, Li, & Vose, 2000; Roldán et al., 2016). Mayr, 1963; Via, 2001). Sympatric speciation is characterized by the In this paper, we analyze the Western Palearctic taxa of the absence of a physical separation between emerging taxa, which di- genus Lasiommata Westwood, 1841, belonging to the Parargina versify uniquely as a consequence of genetic changes affecting eco- subtribe, for which recent wide-scale biodiversity surveys indicate logical and behavioral adaptations (Barluenga, Stölting, Salzburger, the existence of several divergent mitochondrial lineages in the Muschick, & Meyer, 2006; Via, 2001). Allopatric speciation, consid- Mediterranean area (Dapporto et al., 2017; Dincă et al., 2015; Vodă ered as the main mechanism generating biodiversity, occurs in virtu- et al., 2016; Weingartner, Wahlberg, & Nylin, 2006). We hypothe- ally complete geographic separation between populations because sized that (a) the Lasiommata genus, and the Parargina subtribe in physical barriers produced by geological or climatic transformations general, have shorter coalescence rates compared with other sub- or after events of exceptional dispersal and colonization of remote tribes within the Satyrinae occurring in the Palearctic region. To test areas (Coyne & Orr, 2004; Hewitt, 1996, 2000, 2004; Mittelbach & for this hypothesis, we compared the results of a series of GMYC Schemske, 2015). Following allopatric speciation, sister species may analyses among Satyrinae subtribes based on COI data; (b) the main disperse and get into secondary contact (Pigot & Tobias, 2014), but lineages within this species group are likely in an advanced stage full secondary sympatry is frequently delayed due to density-de- of an unusually fast speciation process. To test for this hypothesis pendent phenomena or competition for resources (Waters, 2011; and clarify the systematic relationships within these taxa, we applied Waters, Fraser, & Hewitt, 2013). As a result, most sister, cryptic and an integrative approach, using molecular and morphological traits. sibling species have relatively narrow contact zones (Hewitt, 2004; We sequenced mitochondrial (cytochrome c oxidase subunit I—COI) Waters, 2011; Waters et al., 2013) even in the case of highly mobile and nuclear DNA (internal transcribed spacer 2—ITS2) and analyzed organisms such as butterflies (Vodă, Dapporto, Dincă, & Vila, 2015a, phenotypic markers consisting of wing shape (presumably evolv- 2015b). Secondary contacts can also occur before the speciation ing under strong environmental selection) and genitalia structures process is completed, but the genetic lineages still tend to have very (mostly evolving under sexual selection) (Habel et al., 2017; Hebert, narrow boundaries (Waters, 2011; Waters et al., 2013). In these Cywinska, & Ball, 2003; Hernández-Roldán et al., 2016; Nazari, cases, either lineage fusion or parapatric speciation can eventually Hagen, & Bozano, 2010; Zinetti et al., 2013). take place (Gavrilets et al., 2000). The analysis of genetic and phenotypic traits diverging among populations represents a fundamental resource to understand the 2 | MATERIAL & METHODS speciation process and the mechanisms underlying it. In addition, the results need to be translated into taxonomic hypotheses in 2.1 | Studied species order to categorize biodiversity, a necessity for many applications. Integrative , the classification of organisms by integrating In Europe and North Africa, the butterfly genus Lasiommata is rep- information from different features such as morphology, behavior, resented by five generally recognized species: cytology, DNA sequences, or ecology, allows a deeper view of the (Linnaeus, 1767), (Linnaeus, 1758), Lasiommata visible and hidden layers of diversity, providing researchers the petropolitana (Fabricius, 1787), (Hübner, means to explore and investigate groups that were difficult to study [1824]), and Lasiommata meadewaldoi (Rothschild, 1917), although with classical taxonomy and to refine strategies for species delimi- the taxonomic status of the latter two is still debated. The wall brown tation (Padial, Miralles, De la Riva, & Vences, 2010; Schlick-Steiner L. megera is distributed from North Africa, across a large part of Europe et al., 2010). to Western Asia. The northern European border of its range passes Butterflies (: Papilionoidea) represent one of the through Ireland, Scotland, southern Scandinavia, and the Baltic coun- best-known groups of invertebrates and are well studied in terms of tries. In the Mediterranean basin, it represents one of the most wide- taxonomy, distribution, population dynamics, ecology, and biogeog- spread species, being virtually present in all mainland areas and islands. raphy (Kudrna et al., 2011; Settele, Shreeve, Konvička, & van Dyck, The populations from Corsica, Sardinia, and some surrounding islands 2009; van Swaay et al., 2010). In recent years, several studies that have been attributed to a vicariant sibling species, L. paramegaera. The DNA-barcoded the butterfly species occurring in the Mediterranean taxonomic relationship between these two species has been controver- and in other regions of Europe highlighted the presence of deep sial for a long time, and L. paramegaera has been considered either as a PLATANIA et al. | 3 subspecies of L. megera (Chinery, 1989; Higgins & Riley, 1970; Leraut, For COI, rough estimates of node ages were obtained by ap- 1980) or as a distinct species based on morphological, biochemical, and plying a strict clock and a normal prior distribution centered on genetic evidence (Balletto & Cassulo, 1995; Biermann & Eitschberger, the mean between two widely used substitution rates (1.5% un- 1996; Dapporto, 2008; Jutzeler, 1998; Kudrna et al., 2011). corrected pairwise distance per million years (Quek, Davies, Itino, The general distribution of L. maera is relatively similar to that of & Pierce, 2004), and 2.3% (Brower, 1994)) and with a standard L. megera, but the former is absent from the British Isles, it occurs fur- deviation tuned so that the 95% confidence interval of the poste- ther north in Scandinavia and further east to Central , it has rior density coincided with the 1.5% and 2.3% rates. For both COI a more limited distribution in North Africa, and lacks from several and ITS2 analyses, parameters were estimated using two indepen- Mediterranean islands (Kudrna et al., 2011; Tarrier & Delacre, 2008; dent runs of 20 million generations (for COI) and 30 million gen- Tshikolovets, 2011). In the Moroccan High Atlas, the endemic taxon erations (for both ITS2 analyses), and convergence was checked L. meadewaldoi has a very restricted distribution, limited to high parts using the program TRACER 1.6 (Rambaut, Drummond, Xie, Baele, of the Toubkal Massif. Some authors classified it as a subspecies of & Suchard, 2018). L. maera (Higgins & Riley, 1970; Tolman & Lewington, 2008), while oth- Maximum parsimony haplotype networks for each species ers consider it as a distinct species, although there is very little infor- group were inferred with the program TCS 1.21 (Clement, Posada, mation about the ecology of this species to be compared with L. maera & Crandall, 2000) and then graphically improved with tcsBU (Múrias (Tarrier & Delacre, 2008; Tennent, 1996; Tshikolovets, 2011). dos Santos, Cabezas, Tavares, Xavier, & Branco, 2015) and Adobe Lastly, L. petropolitana is a specialist of colder habitats and inhab- Illustrator CC2015. its the main mountain chains of Southern Europe and Turkey and the To investigate putative species boundaries, we applied lowland areas of Scandinavia and Siberia (Bozano, 1999; Tolman & the general mixed Yule-coalescent model (GMYC, Fujisawa & Lewington, 2008; Tshikolovets, 2011). Barraclough, 2013; Pons et al., 2006) to the COI data. The addi- tion of Yule and coalescent signal into the data set (e.g., incor- porating related outgroup taxa with some extent of population 2.2 | Molecular analyses structure) has been demonstrated to be beneficial for the GMYC performance (Talavera, Dincă, & Vila, 2013). Thus, we obtained We obtained 372 COI sequences from samples stored in the collec- ultrametric phylogenetic trees based on 1,343 unique COI hap- tion of the Butterfly Diversity and Evolution Lab at the Institute of lotypes of Satyrinae sequences retrieved from GenBank and the Evolutionary Biology (CSIC-UPF) in Barcelona, Spain. Other 196 COI Lasiommata COI sequences produced for this study. In order to sequences were mined from BOLD. ITS2 was sequenced from 150 highlight possible differences in coalescence time among tribes, specimens that were selected to represent the main COI lineages phylogenetic trees and GMYC analyses were also inferred sepa- and spatial distribution (Data S1). Instances of intraindividual nuclear rately on the subtribes counting more than five species variation were coded as ambiguities and gaps were treated as miss- in the Palaearctic area (Coenonymphina, Erebiina, Maniolina, ing data in the ITS2 alignment. All new sequences obtained for this Melanargina, Parargina, Satyrina, following systematics by Peña study have been deposited in GenBank and are also available in the et al. (2006)). Bayesian trees were constructed using BEAST as data set DS-Lasio (BOLD process ID in Data S1) from the Barcode of described above, and single-threshold GMYC was evaluated using Life Data System (http://www.bolds​ystems.org/), where additional the R package SPLITS with default settings. information for the specimens is also available. The DNA extraction, amplification, sequencing, and alignment protocols are described in Data S2 (Method S1, Table S1). 2.3 | Morphological analyses Phylogenetic relationships were inferred using Bayesian Inference (BI) through the CIPRES Science Gateway (Miller, Pfeiffer, A total of 262 males of the L. megera group (European n = 98, African & Schwartz, 2010). BI analyses were done separately for COI (based n = 22, and Asian mainland n = 1, Sardinia n = 17, Corsica n = 13, on 107 haplotypes plus three aegeria haplotypes used as Montecristo n = 9, Capraia n = 7, and other smaller Mediterranean outgroup and inferred as described below) and ITS2 (for this marker, islands n = 95) and 49 males of the L. maera group (European n = 33, one analysis included only L. maera and L. meadewaldoi, while an- African n = 8, and Asian mainland n = 6, Sicily = 1, Lesvos = 1) were other included L. megera and L. paramegaera). Both BI analyses and examined for genitalia morphology (Data S2: Figure S1). Genitalia the estimation of node ages (done for COI) were run in BEAST 1.8.0 were dissected using standard procedures: boiling abdomens in (Drummond & Rambaut, 2007). The substitution models (GTR+G for 10% potassium hydroxide, subsequently cleaning the chitinous COI and ITS2-L. maera and HKY+I+G for ITS2-L. megera) were chosen structures, and removing the aedeagus. The lateral side of tegumen according to the values of the Akaike information criterion (AIC) ob- and valvae and the dorsal side of the aedeagus were photographed tained with JMODELTEST 2.1.3 (Darriba, Taboada, Doallo, & Posada, under a stereomicroscope. 2012). For all three analyses, base frequencies were estimated, six A combination of landmarks and sliding semilandmarks gamma rate categories were selected, and a randomly generated ini- (Bookstein, 1997) was applied to the outlines of tegumen (13 semi- tial tree was used. landmarks and 4 landmarks), brachia (10 semilandmarks and 3 4 | PLATANIA et al. landmarks), and valva (17 semilandmarks and 3 landmarks). We con- above, and their individual RGB colors were plotted on a map using sidered as landmarks points that could be precisely identified, while pie charts. the semilandmarks were allowed to slide along the outline trajectory In a second phase, we tested the existence of a signature of diver- (Data S2: Figure S1). sification in genitalia and wing shape among the clades highlighted by Wing shape was analyzed from a total of 189 males. Samples of COI (Hernández-Roldán et al., 2016). We carried a series of PLSDA the L. maera group originated from European n = 41, African n = 8, with shape variables (PCs) and the hypothesis for clade attribution and Asian mainland n = 5, Sicily n = 3, and Lesvos n = 1. Samples of as grouping variable. As relative warps can be particularly numerous L. megera group originated from European (n = 61), African (n = 17), (2 × number of landmarks-4), overfitting had to be avoided. We thus and Asian mainland (n = 1), Sardinia n = 8, Corsica n = 5, Montecristo applied a sparse PLSDA (Lê Cao, Boitard, & Besse, 2011) by includ- n = 4, Capraia n = 3, and other smaller Mediterranean islands n = 31. ing five shape variables in each component. Finally, to evaluate the Thirteen homologous landmarks were identified, representing wing degree of diversification as a percentage of cases that can be blindly vein intersections or points where the veins meet the edges of the attributed to their group, we applied a jackknife (leave-one-out) algo- wing (Data S2: Figure S2). rithm and individually classified each specimen. These analyses were Landmarks of both genitalia and wings were digitized with the carried out with the “splsda” and “predict” functions in the mixOmics program TPS-DIG 2.32 (Rohlf, 2008). In order to remove non-shape R package (Lê Cao et al., 2011). R scripts are available in Data S3. variation and to superimpose the objects in a common coordinate system, a generalized procrustes analysis (GPA) was applied to the landmark data (Adams, Rohlf, & Slice, 2004). Partial warps were cal- 3 | RESULTS culated using the shape residuals from GPA. By applying principal component analyses (PCA) to partial warps, relative warps (PCs) were 3.1 | Molecular analyses obtained and used as variables in subsequent analyses (Bookstein, 1997). For the aedeagus, we calculated the ratio between the length 3.1.1 | GMYC assessment of the lateral cornuti and the length of the aedeagus. GMYC analyses returned different values for the Yule-coalescent threshold among the tested trees of the subtribes and all Satyrini 2.4 | Analysis of configurations for genetic and (the trees are provided in Data S2: Figures S3-S9). The threshold for phenotypic markers evolutionary significant units (ESU) was higher when all Satyrinae were analyzed together (1.091%), and lower for individual sub- We performed a series of exploratory analyses to understand the tribes, the Parargina being the lowest (Parargina 0.612%, Maniolina main variation patterns of the genetic and phenotypic markers and 0.696%, Erebiina 0.782%, Satyrina 0.833%, Coenonymphina their degree of concordance. The p-distance dissimilarity matrices 0.865%, Melanargina 0.918%). for COI and ITS2 were projected in two dimensions by Principal Coordinate Analysis (PCoA) using the “cmdscale” R function. Bidimensional representations were also provided for the two phe- 3.1.2 | Phylogenetic results notypic characters (morphology of wings and genitalia) by applying Partial Least Square Discriminant Analyses (PLSDA) to the relative According to the overall analysis of Satyrini, GMYC highlighted four warps (and the ratio of cornuti/aedeagus for genitalia). As a grouping main COI lineages in the L. megera group, which occur in parapatry variable, we aggregated specimens into geographic clusters either along the geographic barriers of the region: (a) a continental clade on the basis of the continent (Europe, Africa, Asia) or island where (European clade) occurring from Spain to the Black Sea, reaching the they occur. For continents and large islands (Sardinia, Corsica, Sicily, extreme northern European area of distribution of the species, and Great Britain), specimens were pooled when they belonged to the stopping in the Danube valleys in Croatia, Hungary, and Romania, same square of 2 × 2 degrees of latitude and longitude. To facilitate without reaching the Balkan highlands (the maps of the entire sam- a direct comparison of the patterns of different markers, we elimi- pled distribution area are provided in Data S2: Figures S10-S13). This nated the effect of location and rotation among bidimensional repre- lineage is further divided into an Italian lineage showing an abrupt sentations with procrustes analyses, using the COI configuration as boundary of distribution in the northwest of the peninsula (Italian– reference. Because different markers were represented by different European subclade; Figures 1a, 3a and Data S2: Figure S14); (b) the sets of specimens, we used the “recluster.procrustes” function of the lineage represented by L. paramegaera, restricted to Sardinia, Corsica, recluster R package, which maximizes similarities among configura- and the smaller Capraia and Montecristo islands (L. paramegaera tions on the basis of partially overlapping data sets (Dapporto, Vodă, clade) which is sister to the first (Figures 1a, 3a and Data S2: Figure et al., 2014). After procrustes, the aligned bidimensional configura- S14); and two sister lineages represented by (c) an African lineage tions for specimens were projected in the RGB color space using present exclusively in the Maghreb and in Lampedusa and Pantelleria the same package (Dapporto, Fattorini, Vodă, Dincă, & Vila, 2014). islands (African clade) and (d) a lineage occurring in the Balkans, Specimens were grouped according to the geographic areas defined Turkey, Iran and Ukraine (Balkan–Oriental clade). The lineages differ PLATANIA et al. | 5

FIGURE 1 Results of the analyses for COI (a), ITS2 (b), genitalia (c), and wings (d) for the L. megera group. The specimens have been projected in the RGB color space after a Principal Coordinates Analysis for COI and ITS2, and a Partial Least Squares Discriminant Analysis for genitalia and wing shape. The resulting colors were plotted in pie charts grouping specimens from the same island or continent within 2 degrees latitude–longitude squares from each other by 6–9 mutations (around 1%) (Figures 1a, 3a and for L. megera, and the relationships among clades are complex to Data S2: Figure S14). interpret (Figure 3c). The nuclear marker ITS2 did not reveal any clear geographic pat- The Bayesian analysis based on ITS2 highlighted six well-sup- tern within L. megera (Figure 1b), but L. paramegaera was recovered ported main lineages (pp values of one; Figure 4b and Data S2: Figure as a well-supported clade (posterior probability, pp = 1) sister to all S20) partially corresponding to those highlighted by GMYC based L. megera (Figures 1b, 4a and Data S2: Figure S15). However, two on COI (Figure 4b): (a) A western European clade confined to Spain specimens concordant in morphology with L. paramegaera from the and France; (b) An Alpine–Italian clade occupying the Alps, the Italian island of Capraia clustered for ITS2 with L. megera, suggesting the Peninsula and Sicily; (c) A clade widely distributed from central-east- presence of introgression (Figure 1b). ern Europe (Romania) and Scandinavia to Russia and Kazakhstan; (d) Regarding L. maera, GMYC identified six main COI clades (Figures One clade represented by a few specimens from Greece and Turkey; 2a, 3c): (a) A clade confined to Spain and southwestern France (e) A cluster with north African L. maera and (f) A clade represented by (western European), which in the southwestern part of France is L. meadewaldoi (Figure 3c, Figure 4b and Data S2: Figure S20). sympatric with the (b) Alpine–Italian clade. This clade occupies the showed a single COI haplogroup Alps, the Italian Peninsula and Sicily (Italian clade). (c) The north– in Europe and Asia resulting in a single GMYC entity (Figure 3b). central–eastern clade widely distributed from central-east Europe For this reason, this species has not been investigated for other and Scandinavia, the Balkan Peninsula to Kazakhstan (the map x of markers. the entire sampled distribution area are provided in Data S2: Figures S16-S19). This group also reaches northeastern Italy, where it is sympatric with the Italian clade; (d–e). Two clades represented by 3.2 | Morphological analyses a few specimens, one belonging to Lesvos island and West Turkey, and the other one presents in Romania with one sample, eastern 3.2.1 | Male genitalia Turkey and Russia (Oriental clades) and (f) a clade represented by Moroccan specimens (African clade). This clade grouped together The analysis of the genitalia pattern resulted in 88 relative warps specimens attributable to L. maera and to L. meadewaldoi, although (PCs) (30 for tegumen, 22 for brachia and 36 for valva). The popu- the two taxa are separated by a minimum of five COI substitutions lations of L. megera/paramegaera and L. maera are geographically (Figure 3c). The distances among clades are similar to those found well structured for this marker as shown in Figures 1c and 2c. For 6 | PLATANIA et al.

FIGURE 2 Results of the analyses for COI (a), ITS2 (b), genitalia (c), and wings (d) for the L. maera group. The specimens have been projected in the RGB color space after a Principal Coordinates Analysis for COI and ITS2, and a Partial Least Squares Discriminant Analysis for genitalia and wing shape. The resulting colors were plotted in pie charts grouping specimens from the same island or continent within 2 degrees latitude–longitude squares

L. megera, the only groups showing incongruence with the GMYC Regarding L. maera/meadewaldoi, the specimens belonging to assessment are the African and central European clades, which, de- the western European, Italian and the African L. meadewaldoi clade, spite the differentiation in COI, do not show any clear diversifica- were correctly assigned to the three groups in more than 75% of the tion in genitalia shape (Figure 1c, Table 1). The jackknife PLSDA cases (Table 2). showed that L. paramegaera and the Balkan clade of L. megera could be blindly attributed to their group with more than 70% accuracy (Table 1). Within L. maera/meadewaldoi, the specimens belonging to 4 | DISCUSSION the western European, the eastern and the Italian clades were cor- rectly attributed in more than 80% of cases, while L. meadewaldoi Previous studies demonstrated that butterflies in the Satyrinae received a correct attribution in 100% of the cases (Table 2). The subfamily have experienced a strong radiation during the relationship between phylogenetic structure in COI and ITS2 and Oligocene (Peña, Nylin, & Wahlberg, 2011; Peña & Wahlberg, the morphology of genitalia and wings, shown as thin plate spline 2008). Our comparative GMYC analysis revealed a particularly representations, is presented in Figure 4a, b. fast coalescence rate in the Parargina subtribe with respect to the other Satyrinae. The tendency of European Parargina to have a strong or fast divergence is reflected by the presence of 3.2.2 | Wing shape several species and ESU found in various regions, notably in the Mediterranean and Macaronesian islands and might be partly due The morphometric analysis of the forewings resulted in 22 relative to an acceleration of molecular evolution in small island popula- warps (PCs) in both L. megera and L. maera groups. The geographic tions (Dapporto, 2008, 2010; Weingartner et al., 2006). Our GMYC structure is relatively well defined in both cases (Figures 1d, 2d). The approach based on COI revealed that the European Lasiommata jackknife PLSDA for the L. megera group revealed that more than has 12 ESU, divided into six clades for L. maera/meadewaldoi, five 75% of specimens belonging to the Balkan–Oriental clade and the for L. megera/paramegaera, and one for L. petropolitana. These ESU taxon paramegaera could be blindly attributed to their group (Table 1). display divergences among them of about 1% and also show varied Differences in shape among groups are shown as thin plate splines degrees of diversification in morphological and nuclear DNA traits representations in Figure 4a. (Tables 1 and 2). PLATANIA et al. | 7

FIGURE 3 Maximum parsimony haplotype networks based on the COI gene for the three taxa under study 8 | PLATANIA et al.

FIGURE 4 Comparison of the topology of the COI and of the ITS2 trees obtained for L. megera (a) and L. maera (b). Lasiommata meadewaldoi, monophyletic according to COI but not recognized as an evolutionary significant unit, is also included. The thin plate splines of the average configurations of landmarks of the specimens belonging to each lineage for tegumen and wings in L. megera and tegumen and brachia in L. maera are also shown. Support for nodes is provided in Data S1.

The 11 ESU recovered by GMYC for L. megera and L. maera could along suture zones often corresponding to geographic barriers (sea be regarded as potentially emerging species (Dincă et al., 2015; straits or mountain chains, notably the sea channels separating North Pons et al., 2006). However, as reported in Sukumaran and Knowles Africa and Sardinia–Corsica from Europe, or the Alps and the Pyrenees). (2017), the ESU provided by multispecies coalescent analyses are The tendency for sister taxa and COI lineages to be highly segregated not necessarily taxa diversified at species level. Because of the grad- in space with narrow contact areas is an increasingly known phenom- ual nature of the speciation process, discriminating between struc- enon for Mediterranean butterflies (Cesaroni, Lucarelli, Allori, Russo, ture due to population isolation and structure due to speciation is & Sbordoni, 1994; Habel et al., 2017; Vodă, Dapporto, Dincă, & Vila, not always easy. The fact that GMYC is frequently based on a single 2015a, 2015b). The same barriers segregating genetic lineages also marker, weakens the reliability of the results produced and may lead seem to separate zoogeographic regions identified by the beta-diver- to an inflation of species (Sukumaran & Knowles, 2017; Talavera et sity patterns displayed by butterfly communities (Dapporto, Fattorini, al., 2013). Thus, as done in previous studies where the existence of et al., 2014). The segregation among lineages (and species) and their cryptic taxa has been documented (Dincă, Dapporto, et al., 2011; apparent inability to disperse across relatively narrow sea straits and Hernández-Roldán et al., 2016; Nazari et al., 2010; Zinetti et al., mainland barriers represent a seeming incongruence, given the gen- 2013), we used the GMYC approach to recognize units that were erally high dispersal capacity of butterflies and their ability to track later tested using nuclear DNA sequences and morphological data. suitable areas (Devictor et al., 2012). However, not all butterfly species Our results show a clear parapatric/allopatric distribution of the have the same dispersal capability over sea, as reflected in the patterns different L. maera and L. megera COI lineages. Their populations gener- of genetic diversification in western Mediterranean islands, which are ally have little variability within regions but experience abrupt changes largely deterministic and depend on environmental constraints (e.g., PLATANIA et al. | 9

TABLE 1 The correct attribution by Jackknife analysis of specimens to their clade based on wing and genitalia shape, the existence of distinct COI and ITS2 clades, and the recognition of COI clades as independent evolutionary units by GMYC for L. megera group.

Clade Wings Genitalia COI clade ITS2 clade GMYC

African clade 0.619 0.444 Yes No Yes Balkan–Oriental clade 0.760 0.750 Yes No Yes European clade 0.297 0.236 Yes No Yes L. paramegaera clade 0.800 0.739 Yes Yes Yes

TABLE 2 Correct attribution by the Jackknife analysis of specimens to their clade based on wing and genitalia shape, the existence of distinct COI and ITS2 clades, and the recognition of COI clades as independent evolutionary units by GMYC for the L. maera group.

Clade Wings Genitalia COI clade ITS2 clade GMYC

African clade NA NA Yes No Yes W European clade 0.769 0.8 Yes Yes Yes Oriental clades 0 0.6 Yes Yes Yes L. meadewaldoi clade 0.857 1 Yes Yes With African N-C-Eastern clade 0.5 0.8 Yes Yes Yes Italian clade 0.866 0.857 Yes Yes Yes winds, environmental suitability) and intrinsic species traits (abun- females, probably as an evasive reaction to high population densities dance, mobility, specialization, phenology) (Dapporto & Dennis, 2008, in optimal habitats (Elligsen, Beinlich, & Plachter, 1997). It is also plau- 2009; Dapporto, Bruschini, Dincă, Vila, & Dennis, 2012; Dapporto et sible that the different lineages of Lasiommata are not (completely) al., 2017). The high dispersal ability of L. megera is unequivocal since reproductively isolated as shown for another Parargina butterfly, P. this species has colonized virtually every island in the Mediterranean aegeria where a high vitality of F1 and F2 crosses among lineages has and possesses the typical species traits characterizing island dispers- been demonstrated (Livraghi et al., 2018). With these premises, it is ers, such as a high frequency on mainland and a long flight period possible that male-biased dispersal could maintain gene flow in nu- (Dapporto & Dennis, 2008, 2009). Nevertheless, genetic segregation, clear loci across the relatively fixed boundaries of mtDNA lineages reduced gene flow, and delayed secondary sympatry (Pigot & Tobias, and produce discrepancies among nuclear and mitochondrial markers 2014) may be maintained across relatively weak physical barriers by (Sielezniew et al., 2011). The occurrence of male-biased dispersal is mechanisms other than reduced dispersal, such as a combination of suggested by the existence of L. megera ITS2 in some specimens of competitive exclusion, local adaptation, and incompatibility between L. paramegaera from Capraia. Apart from this case, the nuclear DNA different nuclear and mitochondrial DNA variants (Hewitt, 1996, marker revealed patterns that are relatively consistent with those 2000, 2004; Waters, 2011; Waters et al., 2013). Mitochondrial pat- highlighted by mtDNA in L. maera and largely inconsistent in L. megera. terns in may also be determined by infection with Wolbachia. The four GMYC lineages of L. megera belonging to European, Asiatic, These bacteria can manipulate the reproductive dynamics of species and African mainland were not mirrored by the topology of the ITS2 through phenomena such as male-killing and cytoplasmic incompati- tree, which did not reveal any clear geographic pattern (Figure 1b). bility and can play a role in speciation (Hernández-Roldán et al., 2016; Besides the aforementioned introgression in Capraia, the only GMYC Kodandaramaiah, Simonsen, Bromilow, Wahlberg, & Sperling, 2013). lineage recovered with good support by ITS2 is L. paramegaera. Whatever the mechanisms determining the apparent reduced gene In L. maera, all the clades identified as ESU by GMYC were also flow at suture zones, this phenomenon is widespread in Mediterranean recovered as well-supported clades by ITS2, including the clade cor- Satyrinae (Cesaroni et al., 1994; Dapporto, Habel, Dennis, & Schmitt, responding to L. meadewaldoi. However, COI and ITS2 recovered 2011; Dapporto, Vodă, et al., 2014; Nazari et al., 2010; Vodă et al., different relationships among these clades as revealed by the com- 2016) and it can be at the basis of the relatively quick accumulation parison of the phylogenetic trees (Figure 4b and Figures S15, S18, of genetic differences among populations which, in turn, can promote S20, Data S2) and this can be due to past events of introgression fast speciation processes. among lineages or even to stochastic fixation events, which could It must be noted that mtDNA, being maternally transmitted, only have produced the observed incongruences. reflects the matriline distribution. In many butterfly species, although Morphological traits showed a reversed pattern compared to it is not always the case, males are more active than females, thus more ITS2 in L. megera and L. maera. Even though in many cases speci- prone to dispersal (Bennett, Pack, Smith, & Betts, 2013; Sielezniew et mens belonging to different mitochondrial lineages can be blindly al., 2011). Accordingly, in an experiment that marked and recaptured attributed to their group based on wing and, especially, geni- specimens of L. megera in different habitat patches, it has been found talic morphology, the genitalia of L. megera appear much more that males are more prone to disperse to suboptimal patches than diversified than those of L. maera. Previous papers (Biermann 10 | PLATANIA et al.

& Eitschberger, 1996; Dapporto, 2008; Dapporto et al., 2012) recent divergence (Hörandl & Stuessy, 2010), it is not surprising that highlighted clear differences in genitalia morphology among paraphyly is detected in the speciation process of Lasiommata but- three groups of L. megera in the study area: (a) the European and terflies. They show a divergence coherent with phenomena of insu- Maghreb group, (b) the Sardinian–Corsican group (L. paramegaera), larity and segregation of different southern refugia during repeated and (c) the Balkan–Eastern group. In this study, we confirm these climatic pulses in the Pleistocene (Schmitt, 2007) and produce a strong differences in genitalic morphology, which proved to have mosaic of lineages separated by similar genetic distances (Figure 3). an almost perfect spatial congruence with the distribution of the The diversification in COI within both taxa is not extremely high, al- mitochondrial lineages, including in the mainland contact zones though there are other cases of Satyrinae species characterized by between the eastern Alps, Romania, and Bulgaria. Lasiommata low diversification in mtDNA that show a clear distinction at the spe- paramegaera also has a clear difference in wing shape (Figure 1d), cific level in morphological traits (e.g., Dincă et al., 2015; Habel et al., in addition to the well-known difference in wing pattern (Biermann 2017; Kreuzinger, Fiedler, Letsch, & Grill, 2015; Nazari et al., 2010). & Eitschberger, 1996). Both L. paramegaera and L. meadewaldoi occur in isolated areas, the Conversely, although differences in genitalic and wing morphol- first in Capraia, Montecristo, Corsica, Sardinia, and circum-Sardinian ogy among L. maera, mtDNA clades can be retrieved based on a nu- islands, which are well-known areas of endemism for European but- merical approach, upon a visual inspection they appear only slightly terflies (Dennis, Williams, & Shreeve, 1991), and the second in the different (see thin plate spline representations in Figure 4b). A clear High Atlas, about 130 km far from the closest L. maera population exception is the wing shape and, even more so, the wing pattern of (according to Tarrier & Delacre, 2008). Reduced population size and L. meadewaldoi, based on which it has been recognized as a distinct isolation could have accelerated the speciation process compared to species (Tarrier & Delacre, 2008). larger populations showing parapatric distributions on the mainland. The differences among lineages in morphological structures The occurrence of introgression in Capraia located at an intermedi- are encoded in nuclear DNA (Liu et al., 1996) and probably mirror a ate distance between Corsica (where L. paramegaera occurs) and Elba long-term isolation of European butterfly populations, experienced (where L. megera occurs), does not necessarily disprove the species during the cold periods of the Pleistocene. Throughout most of the status of L. paramegaera, and can be rather attributed to a historical Pleistocene, nearly all the central and northern parts of Europe were introgression event. Indeed, although two of the three specimens inhospitable for butterflies, and therefore, they were mainly confined from Capraia show L. megera ITS2, their COI and morphology are to refugia in the Iberian, Italian, and Balkan Peninsulas and on islands identical to the other L. paramegaera populations (Dapporto, 2007, (Hewitt, 1996, 2000, 2004; Schmitt, 2007). The genetic and morpho- 2008). It must be noted that about 12.4% of European butterfly spe- logical variation we found are highly concordant with this hypothesis, cies are known to produce fertile hybrids (Mallet, 2005). Similar evi- and an average divergence of 1% in mtDNA fits accurately with a dence in many other taxonomic groups led to the formulation of the diversification that occurred in Pleistocene climatic pulses. The fact differential fitness species concept according to which some genes that L. megera did not show a diversification in the ITS2 marker does can be exchanged between species provided that there is a series not necessarily mean that there are no differences in many other nu- of peculiar “speciation genes,” such as those involved in the repro- clear loci, and further research based on a high number of nuclear ductive output (e.g., in shaping the male genitalia), which cannot be DNA markers could provide definitive evidence in this direction. exchanged without fitness costs (Hausdorf, 2011). Lasiommata petropolitana showed a completely different pattern, The taxonomic status of the other lineages of L. megera and but still congruent with the hypothesis of Pleistocene differentiation L. maera is less clear. None of the three L. megera COI lineages (not of this genus. Among the three taxa analyzed here, L. petropolitana counting here L. paramegaera) showed a diversification in ITS2; al- inhabits the coldest regions and is limited to Scandinavia and southern though at the contact between the European and the Balkan lin- European mountain areas. It also has very limited diversification with eages, there is a congruent morphology-COI pattern. An in-depth a single haplogroup (five closely related haplotypes) across the spe- study of the contact zone and the analysis of more nuclear genes cies´ range (Figure 3c). Species that live at high latitudes actually tend could provide definitive evidence confirming or disproving the exis- to show an exiguous genetic differentiation after the fast postglacial tence of two (or three, including the North African lineage) mainland colonization processes, probably due to a wider distribution during species in the L. megera group. Until further data become available, the long cold phases of the Pleistocene with a consequent higher gene we consider the mainland COI haplogroups of L. megera as highly di- flow among populations (Hewitt, 2000; Wallis & Arntzen, 1989). verged parapatric lineages. Based on genetic, morphological, and distributional patterns, we The decision is even more difficult for the lineages of L. maera, for confirm that L. paramegaera and L. meadewaldoi represent distinct which, based on concordant patterns of COI, ITS2, and morphology, groups for all the markers under study. Although both taxa appear five groups occur, in addition to L. meadewaldoi. However, the differ- to generate paraphyly for L. megera and L. maera, respectively, for at ences in wing and genitalic morphology, although detectable with nu- least one marker (COI in L. paramegaera and ITS2 and COI for L. meade- meric analyses, are not evident and would indicate the existence of five waldoi), we do not exclude the possibility of paraphyletic speciation. cryptic taxa. Definitive identification of cryptic butterfly populations Considering that paraphyletic groups are commonly formed during as distinct species requires more in-depth analyses, ideally assessing the evolutionary process and are conspicuous particularly in cases of a higher number of nuclear markers and cytological characteristics, PLATANIA et al. | 11

Wolbachia infection, possible host plant specialization, mate recogni- Brower, A. (1994). Rapid morphological radiation and convergence among races of the butterfly Heliconius erato inferred from pat- tion during courtship, and the comparison of different characters in terns of mitochondrial DNA evolution. Proceedings of the National areas of contact (Dincă, Dapporto, et al., 2011; Dincă, Lukhtanov, et Academy of Sciences, 91(14), 6491–6495. https​://doi.org/10.1073/ al., 2011; Dincă et al., 2013; Hernández-Roldán et al., 2016; Zinetti et pnas.91.14.6491 al., 2013). For this reason, in this case as well, we prefer to postpone a Butlin, R. K., Galindo, J., & Grahame, J. W. (2008). Sympatric, parapa- definitive recognition of L. maera lineages as species. Nevertheless, all tric or allopatric: The most important way to classify speciation? Philosophical Transactions of the Royal Society B: Biological Sciences, these lineages at least seem to represent cases of incipient speciation. 363(1506), 2997–3007. https​://doi.org/10.1098/rstb.2008.0076 In conclusion, our study exemplifies the continuous nature of Cesaroni, D., Lucarelli, M., Allori, P., Russo, F., & Sbordoni, V. (1994). evolutionary processes leading to speciation and the difficulties of Patterns of evolution and multidimensional systematics in graylings delimiting species. Through the comparison of nuclear, mitochon- (Lepidoptera: ). Biological Journal of the Linnean Society, 52(2), 101–119. https​://doi.org/10.1111/j.1095-8312.1994.tb009​ drial, and morphological markers at a relatively high spatial resolu- 82.x tion, we showed that L. megera and L. maera groups have experienced Chinery, M. (1989). Butterflies and day-flying moths of Britain and Europe. a fast divergence in different Mediterranean areas during the long London, UK: Collins Professional and Technical Books. cold periods of the Pleistocene. Undoubtedly, we are dealing with Clement, M., Posada, D., & Crandall, K. A. (2000). TCS: A computer pro- gram to estimate gene genealogies. Molecular Ecology, 9(10), 1657– a fast diverging group, composed by a high number of evolutionary 1659. https​://doi.org/10.1046/j.1365-294x.2000.01020.x units located in the gray zone of an advanced speciation process. Coyne, J. A., & Orr, H. A. (2004). Speciation. Sunderland, MA: Sinauer The recognition of taxa as diverging lineages or distinct, possibly Associates. paraphyletic, species, mostly depends on the criteria adopted by dif- Dapporto, L. (2007). Cuticular lipid diversification in Lasiommata megera and Lasiommata paramegaera: The influence of species, sex, and pop- ferent species concepts. ulation (Lepidoptera: ): Cuticular mixture diversification in l. Megera and l. Paramegaera. Biological Journal of the Linnean Society, ACKNOWLEDGEMENTS 91(4), 703–710. https​://doi.org/10.1111/j.1095-8312.2007.00833.x Funding for this research came from the Spanish MINECO and AEI/ Dapporto, L. (2008). Geometric morphometrics reveal male geni- talia differences in the Lasiommata megera / paramegaera com- FEDER, UE (CGL2013-48277-P and CGL2016-76322-P to RVi), plex (Lepidoptera, Nymphalidae) and the lack of a predicted hy- Marie Skłodowska-Curie Train2Move to RVo (grant 609402–2020), bridization area in the Tuscan Archipelago. Journal of Zoological Marie Sklodowska-Curie IOF grant (project 625997) to VD, from Systematics and Evolutionary Research, 46(3), 224–230. https​://doi. Tuscan Archipelago National Park and Florence University to LD and org/10.1111/j.1439-0469.2007.00453.x from the project “Barcoding-Italian-Butterflies.” Dapporto, L. (2010). Satyrinae butterflies from Sardinia and Corsica show a kaleidoscopic intraspecific biogeography (Lepidoptera, Nymphlidae). Biological Journal of the Linnean Society, 100(1), 195– ORCID 212. https​://doi.org/10.1111/j.1095-8312.2010.01434.x Leonardo Platania https://orcid.org/0000-0002-1289-5564 Dapporto, L., Bruschini, C., Dincă, V., Vila, R., & Dennis, R. L. H. Raluca Vodă https://orcid.org/0000-0003-1970-5628 (2012). Identifying zones of phenetic compression in West Mediterranean butterflies (Satyrinae): Refugia, invasion and hy- Vlad Dincă https://orcid.org/0000-0003-1791-2148 bridization: Phenetic ‘compression’ in Western Mediterranean. Gerard Talavera https://orcid.org/0000-0003-1112-1345 Diversity and Distributions, 18(11), 1066–1076. https​://doi. Roger Vila https://orcid.org/0000-0002-2447-4388 org/10.1111/j.1472-4642.2012.00903.x Leonardo Dapporto https://orcid.org/0000-0001-7129-4526 Dapporto, L., Cini, A., Menchetti, M., Vodă, R., Bonelli, S., Casacci, L. P., … Vila, R. (2017). Rise and fall of island butterfly diversity: Understanding genetic differentiation and extinction in a highly REFERENCES diverse archipelago. Diversity and Distributions, 23(10), 1169–1181. Adams, D. C., Rohlf, F. J., & Slice, D. E. (2004). Geometric morphomet- https​://doi.org/10.1111/ddi.12610​ rics: Ten years of progress following the ‘revolution’. Italian Journal of Dapporto, L., & Dennis, R. L. H. (2008). Species richness, rarity and en- Zoology, 71(1), 5–16. https​://doi.org/10.1080/11250​00040​9356545 demicity on Italian offshore islands: Complementary signals from is- Balletto, E., & Cassulo, L. A. (1995). Lepidoptera Hesperioidea, Papilionoidea. land-focused and species-focused analyses. Journal of Biogeography, Edizione Calderini. 35(4), 664–674. https​://doi.org/10.1111/j.1365-2699.2007.01812.x Barluenga, M., Stölting, K. N., Salzburger, W., Muschick, M., & Meyer, A. Dapporto, L., & Dennis, R. L. H. (2009). Conservation biogeography of large (2006). Sympatric speciation in Nicaraguan crater lake cichlid fish. Mediterranean islands. Butterfly impoverishment, conservation pri- Nature, 439(7077), 719–723. https​://doi.org/10.1038/natur​e04325 orities and inferences for an ecological “island paradigm”. Ecography, Bennett, V. J., Pack, S. M., Smith, W. P., & Betts, M. G. (2013). Sex-biased 32(1), 169–179. https​://doi.org/10.1111/j.1600-0587.2008.05600.x dispersal in a rare butterfly and the implications for its conser- Dapporto, L., Fattorini, S., Vodă, R., Dincă, V., & Vila, R. (2014). vation. Journal of Conservation, 17(5), 949–958. https​://doi. Biogeography of western Mediterranean butterflies: Combining org/10.1007/s10841-013-9577-5 turnover and nestedness components of faunal dissimilarity. Journal Biermann, H., & Eitschberger, U. (1996). Bemerkungen zu Lasiommata of Biogeography, 41(9), 1639–1650. https​://doi.org/10.1111/jbi.12315​ megera (Linnaeus, 1767) und Lasiommata paramegaera (Hübner Dapporto, L., Habel, J. C., Dennis, R. L. H., & Schmitt, T. (2011). The bio- 1824)(Lepidoptera Saryridae). Atalanta, 27(1/2), 253–319. geography of the western Mediterranean: Elucidating contradictory Bookstein, F. L. (1997). Morphometric tools for landmark data: Geometry distribution patterns of differentiation in jurtina (Lepidoptera: and biology. Cambridge, UK: Cambridge University Press. Nymphalidae): BIOGEOGRAPHY OF MEDITERRANEAN Bozano, G. C. (1999): Satyridae part I. In: G. C. Bozano (Ed.), Guide to the BUTTERFLIES. Biological Journal of the Linnean Society, 103(3), 571– butterflies of the Palearctic region. (58 pp). Milano, Italy: Omnes Artes. 577. https​://doi.org/10.1111/j.1095-8312.2011.01639.x 12 | PLATANIA et al.

Dapporto, L., Vodă, R., Dincă, V., & Vila, R. (2014). Comparing popu- Habel, J. C., Vila, R., Vodă, R., Husemann, M., Schmitt, T., & Dapporto, L. lation patterns for genetic and morphological markers with un- (2017). Differentiation in the marbled white butterfly species com- even sample sizes. An example for the butterfly Maniola jurtina. plex driven by multiple evolutionary forces. Journal of Biogeography, Methods in Ecology and. Evolution, 5(8), 834–843. https​://doi. 44(2), 433–445. https​://doi.org/10.1111/jbi.12868​ org/10.1111/2041-210X.12220​ Hausdorf, B. (2011). Progress toward a general species concept. Evolution, Darriba, D., Taboada, G. L., Doallo, R., & Posada, D. (2012). jModelT- 65(4), 923–931. https​://doi.org/10.1111/j.1558-5646.2011.01231.x est 2: More models, new heuristics and parallel computing. Nature Hausmann, A., Haszprunar, G., Segerer, A. H., Speidel, W., Behounek, G., Methods, 9(8), 772. https​://doi.org/10.1038/nmeth.2109 & Hebert, P. D. N. (2011). Now DNA-barcoded: The butterflies and de Queiroz, K. (2005). Ernst Mayr and the modern concept of species. larger moths of Germany. Spixiana, 34(1), 47–58. Proceedings of the National Academy of Sciences, 102(Supplement 1), Hebert, P. D., Cywinska, A., & Ball, S. L. (2003). Biological identifications 6600–6607. https​://doi.org/10.1073/pnas.05020​30102​ through DNA barcodes. Proceedings of the Royal Society of London B: De Queiroz, K. (2007). Species Concepts and Species Delimitation. Biological Sciences, 270(1512), 313–321. Systematic Biology, 56(6), 879–886. https​://doi.org/10.1080/10635​ Hendry, A. P. (2009). Ecological speciation! Or the lack thereof? Canadian 15070​1701083 Journal of Fisheries and Aquatic Sciences, 66(8), 1383–1398. https​:// Dennis, R. L. H., Williams, W. R., & Shreeve, T. G. (1991). A mul- doi.org/10.1139/F09-074 tivariate approach to the determination of faunal structures Hendry, A. P., Bolnick, D. I., Berner, D., & Peichel, C. L. (2009). Along the among European butterfly species (Lepidoptera: Rhopalocera). speciation continuum in sticklebacks. Journal of Fish Biology, 75(8), Zoological Journal of the Linnean Society, 101(1), 1–49. https​://doi. 2000–2036. https​://doi.org/10.1111/j.1095-8649.2009.02419.x org/10.1111/j.1096-3642.1991.tb008​84.x Hernández-Roldán, J. L., Dapporto, L., Dincă, V., Vicente, J. C., Hornett, Devictor, V., van Swaay, C., Brereton, T., Brotons, L., Chamberlain, D., E. A., Šíchová, J., … Vila, R. (2016). Integrative analyses unveil specia- Heliölä, J., … Jiguet, F. (2012). Differences in the climatic debts of tion linked to host plant shift in Spialia butterflies. Molecular Ecology, birds and butterflies at a continental scale. Nature Climate Change, 25(17), 4267–4284. https​://doi.org/10.1111/mec.13756​ 2(2), 121–124. https​://doi.org/10.1038/nclim​ate1347 Hewitt, G. M. (1996). Some genetic consequences of ice ages, and their Dincă, V., Dapporto, L., & Vila, R. (2011). A combined genetic-mor- role in divergence and speciation. Biological Journal of the Linnean phometric analysis unravels the complex biogeographical his- Society, 58(3), 247–276. https​://doi.org/10.1111/j.1095-8312.1996. tory of Polyommatus icarus and Polyommatus celina Common Blue tb014​34.x butterflies. Molecular Ecology, 20(18), 3921–3935. https​://doi. Hewitt, G. M. (2000). The genetic legacy of the Quaternary ice ages. org/10.1111/j.1365-294X.2011.05223.x Nature, 405(6789), 907–913. https​://doi.org/10.1038/35016000 Dincă, V., Lukhtanov, V. A., Talavera, G., & Vila, R. (2011). Unexpected Hewitt, G. M. (2004). Genetic consequences of climatic oscillations in layers of cryptic diversity in wood white Leptidea butterflies. Nature the Quaternary. Philosophical Transactions of the Royal Society of Communications, 2, 324. https://doi.org/10.1038/ncomm​ s1329​ ​ London. Series B: Biological Sciences, 359(1442), 183–195. https​://doi. Dincă, V., Montagud, S., Talavera, G., Hernández-Roldán, J., Munguira, org/10.1098/rstb.2003.1388 M. L., García-Barros, E., … Vila, R. (2015). DNA barcode reference Higgins, L. G., & Riley, N. D. (1970). A field guide to the butterflies of Britain library for Iberian butterflies enables a continental-scale preview of and Europe. Collins, London. potential cryptic diversity. Scientific Reports, 5, 12395. https​://doi. Hörandl, E., & Stuessy, T. F. (2010). Paraphyletic groups as natural units org/10.1038/srep1​2395 of biological classification. Taxon, 59(6), 1641–1653. https​://doi. Dincă, V., Wiklund, C., Lukhtanov, V. A., Kodandaramaiah, U., Norén, K., org/10.1002/tax.596001 Dapporto, L., … Friberg, M. (2013). Reproductive isolation and pat- Jutzeler, D. (1998). Lasiommata paramegaera (Huebner, 1824) of terns of genetic differentiation in a cryptic butterfly species com- the Tyrrhenian isles: A good species (Lepidoptera: Nymphalidae, plex. Journal of Evolutionary Biology, 26(10), 2095–2106. https​://doi. Satyrinae). Linneana Belgica (Belgium), 16(7), 267–276. org/10.1111/jeb.12211​ Kodandaramaiah, U., Simonsen, T. J., Bromilow, S., Wahlberg, N., & Sperling, Dincă, V., Zakharov, E. V., Hebert, P. D. N., & Vila, R. (2011). Complete F. (2013). Deceptive single-locus taxonomy and phylogeography: DNA barcode reference library for a country’s butterfly fauna re- Wolbachia-associated divergence in mitochondrial DNA is not reflected veals high performance for temperate Europe. Proceedings of the in morphology and nuclear markers in a butterfly species. Ecology and Royal Society B: Biological Sciences, 278(1704), 347–355. https​://doi. Evolution, 3(16), 5167–5176. https​://doi.org/10.1002/ece3.886 org/10.1098/rspb.2010.1089 Kreuzinger, A. J., Fiedler, K., Letsch, H., & Grill, A. (2015). Tracing the ra- Dirzo, R., Young, H. S., Galetti, M., Ceballos, G., Isaac, N. J., & Collen, diation of Maniola (Nymphalidae) butterflies: New insights from phy- B. (2014). Defaunation in the Anthropocene. Science, 345(6195), logeography hint at one single incompletely differentiated species 401–406. complex. Ecology and Evolution, 5(1), 46–58. Drummond, A. J., & Rambaut, A. (2007). BEAST: Bayesian evolutionary Kudrna, O., Harpke, A., Lux, K., Pennerstorfer, J., Schweiger, O., Settele, analysis by sampling trees. BMC Evolutionary Biology, 7(1), 214. https​ J., & Wiemers, M. (2011). Distribution atlas of butterflies in Europe. ://doi.org/10.1186/1471-2148-7-214 Gesellschaft für Schmetterlingsschutz eV. Halle: Elligsen, H., Beinlich, B., & Plachter, H. (1997). Effects of large-scale cattle Lê Cao, K.-A., Boitard, S., & Besse, P. (2011). Sparse PLS discriminant grazing on populations of glycerion and Lasiommata analysis: Biologically relevant feature selection and graphical dis- megera (Lepidoptera: Satyridae). Journal of Insect Conservation, 1(1), plays for multiclass problems. BMC Bioinformatics, 12(1), 253. https​ 13–23. ://doi.org/10.1186/1471-2105-12-253 Fujisawa, T., & Barraclough, T. G. (2013). Delimiting Species Using Leraut, P. (1980). Liste systématique et synonymique des Lépidoptères de Single-Locus Data and the Generalized Mixed Yule Coalescent France, Belgique et Corse. Paris: Alexanor. Approach: A Revised Method and Evaluation on Simulated Data Liu, J., Mercer, J. M., Stam, L. F., Gibson, G. C., Zeng, Z. B., & Laurie, C. Sets. Systematic Biology, 62(5), 707–724. https​://doi.org/10.1093/ C. (1996). Genetic analysis of a morphological shape difference in sysbio/syt033​ the male genitalia of Drosophila simulans and D. mauritiana. Genetics, Gavrilets, S., Li, H., & Vose, M. D. (2000). Patterns of parapatric speciation. 142(4), 1129–1145. Evolution, 54(4), 1126–1134. https​://doi.org/10.1111/j.0014-3820.2000. Livraghi, L., Vodă, R., Evans, L. C., Gibbs, M., Dincă, V., Holland, P. W., … tb005​48.x Breuker, C. J. (2018). Historical and current patterns of gene flow in PLATANIA et al. | 13

the butterfly Pararge aegeria. Journal of Biogeography, 45(7), 1628– to Exploring Biodiversity. Annual Review of Entomology, 55(1), 421– 1639. https​://doi.org/10.1111/jbi.13354​ 438. https​://doi.org/10.1146/annur​ev-ento-112408-085432 Mallet, J. (2005). Hybridization as an invasion of the genome. Trends Schmitt, T. (2007). Molecular biogeography of Europe: Pleistocene cy- in Ecology & Evolution, 20(5), 229–237. https​://doi.org/10.1016/j. cles and postglacial trends. Frontiers in Zoology, 4(1), 11. https​://doi. tree.2005.02.010 org/10.1186/1742-9994-4-11 Mallet, J., Beltrán, M., Neukirchen, W., & Linares, M. (2007). Natural Seehausen, O. (2009). Progressive levels of trait divergence along a ’spe- hybridization in heliconiine butterflies: The species boundary ciation transect’ in the Lake Victoria cichlid fish Pundamilia. In R. K. as a continuum. BMC Evolutionary Biology, 7(1), 28. https​://doi. Butlin, J. R. Bridle, & D. Schluter (Eds.), Speciation and Patterns of org/10.1186/1471-2148-7-28 Diversity (pp. 155–176). Cambridge, MA: Cambridge University Press. Mayr, E. (1963). Species and Evolution. Massachusetts Google Settele, J., Shreeve, T., Konvička, M., & van Dyck, H. (2009). Ecology of Scholar: Belknap Press of Harvard University Press Cambridge. butterflies in Europe. Cambridge, MA: Cambridge University Press. Miller, M. A., Pfeiffer, W., & Schwartz, T. (2010). Creating the CIPRES Shaw, K. L., & Mullen, S. P. (2014). Speciation Continuum. Journal of Science Gateway for inference of large phylogenetic trees. In 2010 Heredity, 105(S1), 741–742. https​://doi.org/10.1093/jhere​d/esu060 gateway computing environments workshop (GCE) (pp. 1–8). IEEE. https​ Sielezniew, M., Ponikwicka-Tyszko, D., Ratkiewicz, M., Dziekanska, ://doi.org/10.1109/GCE.2010.5676129 I., Kostro-Ambroziak, A., & Rutkowski, R. (2011). Divergent pat- Mittelbach, G. G., & Schemske, D. W. (2015). Ecological and evo- terns in the mitochondrial and nuclear diversity of the special- lutionary perspectives on community assembly. Trends in ized butterfly Plebejus argus (Lepidoptera: Lycaenidae). European Ecology & Evolution, 30(5), 241–247. https​://doi.org/10.1016/j. Journal of Entomology, 108(4), 537–545. https​://doi.org/10.14411/​ tree.2015.02.008 eje.2011.069. Múrias dos Santos, A., Cabezas, M. P., Tavares, A. I., Xavier, R., & Branco, Sukumaran, J., & Knowles, L. L. (2017). Multispecies coalescent delim- M. (2015). tcsBU: A tool to extend TCS network layout and visualiza- its structure, not species. Proceedings of the National Academy of tion. Bioinformatics, 32(4), 627–628. https​://doi.org/10.1093/bioin​ Sciences, 114(7), 1607–1612. https​://doi.org/10.1073/pnas.16079​ forma​tics/btv636 21114​ Nazari, V., Hagen, W. T., & Bozano, G. C. (2010). Molecular systematics Talavera, G., Dincă, V., & Vila, R. (2013). Factors affecting species de- and phylogeny of the ‘Marbled Whites’ (Lepidoptera: Nymphalidae, limitations with the GMYC model: Insights from a butterfly survey. Satyrinae, Meigen). Systematic Entomology, 35(1), 132– Methods in Ecology and Evolution, 4(12), 1101–1110. https​://doi. 147. https​://doi.org/10.1111/j.1365-3113.2009.00493.x org/10.1111/2041-210X.12107​ Nosil, P., Harmon, L. J., & Seehausen, O. (2009). Ecological explanations Tarrier, M., & Delacre, J. (2008). Les Papillons de jour du Maroc. Mèze: for (incomplete) speciation. Trends in Ecology & Evolution, 24(3), 145– Biotope. 156. https​://doi.org/10.1016/j.tree.2008.10.011 Tennent, J. (1996). The butterflies of Morocco. Co: Algeria and Tunisia. Padial, J. M., Miralles, A., De la Riva, I., & Vences, M. (2010). The inte- Gem Pub. grative future of taxonomy. Frontiers in Zoology, 7(1), 16. https​://doi. Tolman, T., & Lewington, R. (2008). Butterflies of Europe. London, UK: org/10.1186/1742-9994-7-16 Harper Collins. Peña, C., Nylin, S., & Wahlberg, N. (2011). The radiation of Satyrini but- Tshikolovets, V. V. (2011). Butterflies of Europe & the Mediterranean area. terflies (Nymphalidae: Satyrinae): A challenge for phylogenetic meth- Pardubice: Tshikolovets Publications. ods. Zoological Journal of the Linnean Society, 161(1), 64–87. https​:// van Swaay, C., Cuttelod, A., Collins, S., Maes, D., Munguira, M. L., Šašić, doi.org/10.1111/j.1096-3642.2009.00627.x M., … Warren, M. (2010). European red list of butterflies. Publications Peña, C., & Wahlberg, N. (2008). Prehistorical climate change increased office of the. European Union. diversification of a group of butterflies. Biology Letters, 4(3), 274– Via, S. (2001). Sympatric speciation in : The ugly duckling 278. https​://doi.org/10.1098/rsbl.2008.0062 grows up. Trends in Ecology & Evolution, 16(7), 381–390. https​://doi. Peña, C., Wahlberg, N., Weingartner, E., Kodandaramaiah, U., Nylin, org/10.1016/S0169-5347(01)02188-7 S., Freitas, A. V., & Brower, A. V. (2006). Higher level phylogeny of Vodă, R., Dapporto, L., Dincă, V., Shreeve, T. G., Khaldi, M., Barech, G., … Satyrinae butterflies (Lepidoptera: Nymphalidae) based on DNA Vila, R. (2016). Historical and contemporary factors generate unique sequence data. Molecular Phylogenetics and Evolution, 40(1), 29–49. butterfly communities on islands. Scientific Reports, 6(1), 28828. https​://doi.org/10.1016/j.ympev.2006.02.007 https://doi.org/10.1038/srep2​ 8828​ Pigot, A. L., & Tobias, J. A. (2014). Dispersal and the transition to sympatry Vodă, R., Dapporto, L., Dincă, V., & Vila, R. (2015a). Cryptic mat- in vertebrates. Proceedings of the Royal Society B: Biological Sciences, ters: Overlooked species generate most butterfly beta-diversity. 282(1799), 20141929. https​://doi.org/10.1098/rspb.2014.1929 Ecography, 38(4), 405–409. https​://doi.org/10.1111/ecog.00762​ Pons, J., Barraclough, T. G., Gomez-Zurita, J., Cardoso, A., Duran, Vodă, R., Dapporto, L., Dincă, V., & Vila, R. (2015b). Why Do Cryptic D. P., Hazell, S., … Vogler, A. P. (2006). Sequence-Based Species Species Tend Not to Co-Occur? A Case Study on Two Cryptic Pairs Delimitation for the DNA Taxonomy of Undescribed Insects. of Butterflies. PLoS ONE, 10(2), e0117802. https​://doi.org/10.1371/ Systematic Biology, 55(4), 595–609. https​://doi.org/10.1080/10635​ journal.pone.0117802​ 15060​0852011 Wallis, G. P., & Arntzen, J. W. (1989). Mitochondrial-DNA varia- Quek, S., Davies, S. J., Itino, T., & Pierce, N. E. (2004). Codiversification tion in the crested newt superspecies: Limited cytoplasmic in an ant-plant mutualism: Stem texture and the evolution of host gene flow among species. Evolution, 43(1), 88–104. https​://doi. use in Crematogaster (Formicidae: Myrmicinae) inhabitants of org/10.1111/j.1558-5646.1989.tb042​09.x Macaranga (Euphorbiaceae). Evolution, 58(3), 554–570. https​://doi. Waters, J. M. (2011). Competitive exclusion: Phylogeography’s ‘elephant org/10.1111/j.0014-3820.2004.tb016​78.x in the room’? Molecular Ecology, 20(21), 4388–4394. https​://doi. Rambaut, A., Drummond, A. J., Xie, D., Baele, G., & Suchard, M. A. (2018). org/10.1111/j.1365-294X.2011.05286.x Posterior Summarization in Bayesian Phylogenetics Using T racer 1.7. Waters, J. M., Fraser, C. I., & Hewitt, G. M. (2013). Founder takes all: Systematic Biology., 67(5), 901–904. Density-dependent processes structure biodiversity. Trends Rohlf, F. (2008). Tpsdig, version 2.12; tpsrelw, version 1.46. Stony Brook, in Ecology & Evolution, 28(2), 78–85. https​://doi.org/10.1016/j. NY: State University of New York at Stony Brook. tree.2012.08.024 Schlick-Steiner, B. C., Steiner, F. M., Seifert, B., Stauffer, C., Christian, E., & Weingartner, E., Wahlberg, N., & Nylin, S. (2006). Speciation in Pararge Crozier, R. H. (2010). Integrative Taxonomy: A Multisource Approach (Satyrinae: Nymphalidae) butterflies–North Africa is the source of 14 | PLATANIA et al.

ancestral populations of all Pararge species. Systematic Entomology, 2 degrees latitude–longitude squares. 31(4), 621–632. Figure S12. Uncut version of map for wings in L. megera group. The Wheeler, Q. D., Knapp, S., Stevenson, D. W., Stevenson, J., Blum, S. specimens have been projected in RGB colour space after Partial D., Boom, B. M., … Woolley, J. B. (2012). Mapping the biosphere: Exploring species to understand the origin, organization and sustain- Least Squares Discriminant Analysis and the resulting colors were ability of biodiversity. Systematics and Biodiversity, 10(1), 1–20. https​ plotted in pie charts grouping specimens from the same island or ://doi.org/10.1080/14772​000.2012.665095 continent within 2 degrees latitude–longitude squares. Zinetti, F., Dapporto, L., Vovlas, A., Chelazzi, G., Bonelli, S., Balletto, E., & Figure S13. Uncut version of map for genitalia in L. megera group. Ciofi, C. (2013). When the rule Becomes the Exception. No Evidence of Gene Flow between Two Zerynthia Cryptic Butterflies Suggests The specimens have been projected in RGB colour space after the Emergence of a New Model Group. PLoS ONE, 8(6), e65746. https​ Partial Least Squares Discriminant Analysis and the resulting colors ://doi.org/10.1371/journ​al.pone.006574 were plotted in pie charts grouping specimens from the same island or continent within 2 degrees latitude–longitude squares. SUPPORTING INFORMATION Figure S14. Phylogenetic tree based on COI for Western Palearctic Additional supporting information may be found online in the Lasiommata spp. Supporting Information section at the end of the article. Figure S15. Phylogenetic tree based on ITS2 for L. megera/ paramegaera. Data S1. Supporting Information Specimen List. Figure S16. Uncut version of map for COI in L. maera group. The Data S2. Supporting Information. specimens have been projected in RGB colour space after Principal Data S3. Supporting Information R Script.R. Coordinate analysis and the resulting colours were plotted in pie Figure S1. Landmarks (empty circles) and sliding semilandmarks (co- charts grouping specimens from the same island or continent within loured circles) used for the analysis of the different structures of 2 degrees latitude–longitude squares. Lasiommata spp. male genitalia. Figure S17. Uncut version of map for ITS2 in L. maera group. The Figure S2. Landmarks (white circles) used for the analysis of specimens have been projected in RGB colour space after Principal Lasiommata spp. wings. Coordinate analysis and the resulting colours were plotted in pie Figure S3. Phylogenetic tree based on COI for Coenonymphina charts grouping specimens from the same island or continent within group and results for GMYC solution (red clades). 2 degrees latitude–longitude squares. Figure S4. Phylogenetic tree based on COI for Erebiina group and Figure S18. Uncut version of map for wings in L. maera group. The results for GMYC solution (red clades). specimens have been projected in RGB colour space after Partial Figure S5. Phylogenetic tree based on COI for Maniolina group and Least Squares Discriminant Analysis and the resulting colors were results for GMYC solution (red clades). plotted in pie charts grouping specimens from the same island or Figure S6. Phylogenetic tree based on COI for Melanargina group continent within 2 degrees latitude–longitude square. and results for GMYC solution (red clades). Figure S19. Uncut version of map for genitalia in L. maera group. The Figure S7. Phylogenetic tree based on COI for Parargina group and specimens have been projected in RGB colour space after Partial results for GMYC solution (red clades). Least Squares Discriminant Analysis and the resulting colors were Figure S8. Phylogenetic tree based on COI for Satyrina group and plotted in pie charts grouping specimens from the same island or results for GMYC solution (red clades). continent within 2 degrees latitude–longitude squares. Figure S9. Phylogenetic tree based on COI for Satyrinae subfamily Figure S20. Phylogenetic tree based on ITS2 for L. maera/ and results for GMYC solution (red clades). meadewaldoi. Figure S10. Uncut version of map for COI in L. megera group. The specimens have been projected in RGB colour space after Principal Coordinate analysis and the resulting colours were plotted in pie How to cite this article: Platania L, Vodă R, Dincă V, Talavera charts grouping specimens from the same island or continent within G, Vila R, Dapporto L. Integrative analyses on Western 2 degrees latitude–longitude squares. Palearctic Lasiommata reveal a mosaic of nascent butterfly Figure S11. Uncut version of map for ITS2 in L. megera group. The species. J Zool Syst Evol Res. 2020;00:1–14. https://doi.​ specimens have been projected in RGB colour space after Principal org/10.1111/jzs.12356​ Coordinate analysis and the resulting colours were plotted in pie charts grouping specimens from the same island or continent within