Rear edges versus continuous areas of distribution in M. arion
Perfectly matching rear edge theory , this study revealed great genetic differentiation among M. arion Italian populations, both at the COI mtDNA and the nuclear EF-1α genes. This is in contrast to the almost total absence of mtDNA polymorphism and the low to moderate genetic differentiation at the nuclear gene found for M. arion populations inhabiting Poland, a recently deglaciated area. Moreover, the rear edge populations (Italy) possessed as many as 11 out of the 12 mtDNA haplotypes recorded in this study; thereby, emphasising their relevance in the context of biodiversity conservation and, most importantly, from the evolutionary perspective.
The mtDNA diversity in Italy may be thus an example of 'southern richness’ as opposed to ‘northern purity' as defined by Hewitt [1, 3, 48], e.g. high polymorphism and divergence in the south and low variation and lack of divergence in the north. “Southern richness” was also visible for the nuclear gene EF-1α as the average number of alleles per population and the genetic differentiation among them was significantly larger than among Polish populations.
Thus, our data do not support the alternative scenario that suggests lower genetic diversity in populations at the edge of a species’ distribution compared to that in the centre [26, 27]. The high genetic diversity found in Southern Europe can be attributed to prolonged demographic stability in the populations occurring in these areas  and by the high geographic structuring of populations evolving in allopatric conditions [49, 50]. In Italy, prolonged demographic stability was assumed to create a common history for many taxa and, recently, allopatric differentiation (refugia-within-refugia) was described in Bombina pachypus. The identification of four distinct genetic groups for mtDNA as well as similar N
values emphasize the role of the genetic drift as the main force responsible for the differentiation among Italian populations. In Italy, we found more structuring and greater genetic differentiation for the COI gene than for the EF-1α gene. Mitochondrial markers are characterised by much faster lineage sorting rates and more frequent haplotype extinctions than the bi-parental nuclear markers. Consequently, as is widely recognised for many Lepidoptera e.g., [52, 53], differentiation in mtDNA among populations is usually much stronger than in nuclear DNA (e.g., EF-1α).
Population structure and colonisation patterns
Although our data on mitochondrial genetic material suggested a significant population genetic structure, they did not support the subdivision of the populations into macrogeographic groups (Poland versus Italy). The most common mtDNA haplotype in our samples (H1) is indeed the most widespread throughout the species’ range because it was recorded also from Catalonia, Germany, Denmark, Sweden, Estonia, Finland, the Czech Republic, Slovakia, Romania, Hungary, Ukraine, Russia and Kazakhstan. Moreover, the second most common haplotype (H5), which we found in five Italian sites (CER, AUR, VAL, CDF, VFE), is known to occur in Catalonia, the Czech Republic and Sweden . As a result, our mtDNA data do not fully support the colonisation pattern hypothesized by Ugelvig . The author suggests the existence of three different genetic groups in the Western Palaearctic originating from different southern refugia in the Iberian Peninsula, the Balkans and Asia and thus three postglacial colonisation routes. The Alps are known to have acted as an initial barrier to the possible expansion of Italian genetic pools towards Northern Europe for many species [3, 9], but our data do not allow us to exclude Italy as a possible source area for Central and Northern European M. arion populations. In Poland, we almost exclusively found a single haplotype, which was also the most frequent in northern Italy (H1). Of course, a “European” origin of north Italian populations cannot be ruled out either. This would mean that the dispersion from the East has reached northern Italy. Our results have indeed shown that the ancestral haplotype (H1) is dominant in northern Italy, but is rare (AUR) or absent (CET, CER) in central and southern Italy. Ugelvig  suggests that, in the case of M. arion, Italy was colonised from Iberia. It should be noted, however, that only two samples from Italy were analysed, while our extensive study showed that Italy was rather colonised from two different routes.
We detected three unique and closely related mtDNA haplotypes (H3, H4, H6) in southern Italy. A similar pattern was found in Italy for the Euscorpio carpathicus complex, where Salomone et al.  concluded that the presence of ancestral haplotypes in the northern part of the country meant that other refugia, outside of the Apennine, may have existed for these taxa. Possible sources are the Balkans, which northern Italy was connected to by the Adriatic bridge during the quaternary cold periods , and the Iberian Peninsula perhaps via Col di Tenda, as was the case for the western lineage of Melanargia galathea. However, both directions (i.e., also from Italy towards Iberia) are possible, as it was suggested for Polyommatus coridon. This conclusion is supported by our data showing the presence of one haplotype (H5) in Catalonia, in the Alpes Maritimes (VAL is 20 km away from Col di Tenda) and all along the Italian latitudinal gradient.
On the other hand, the presence of unique haplotypes in southern Italy concurs with the hypothesis that, contrary to common belief, most relic rear edge populations may not have been the source of major postglacial re-colonisations, thereby preserving their high genetic distinctiveness . Rear edges, as ‘relic hotspots’, may be considered as regions of special conservation concern, as they often coincide with centres of high biodiversity and endemism. To fully resolve M. arion phylogeographical pattern further studies should be based on a much higher number of specimens, especially coming from other South European areas as well as central populations.
M. arion showed moderate to high levels of diversity in the nuclear gene EF-1α. Altogether, 30 haplotypes were found, 18 occurring in Italy and 17 in Poland. Interestingly, a study carried out on M. alcon in a closely overlapping area found only six haplotypes among the 179 individuals sampled . Our results from the EF-1α gene support the allozyme and microsatellite data [56, 57] by showing that M. arion exhibits a rather high nuclear genetic variability when compared to other members of the genus Maculinea[33, 35, 58]. It has been suggested that the specialised parasitic life style of Maculinea butterflies makes them prone to genetic drift and therefore to a loss in genetic variation . However, M. arion is considered a less advanced social parasite because it preys on ant larvae and shows a lower level of host ant specificity. In this case, it is possible that the effective population size of M. arion is higher than in other members of the genus, especially M. alcon and M. ‘rebeli’. In this regard, it is interesting to observe that recent genetic data obtained on Danish and Swedish populations suggest that M. arion is more dispersive and less sensitive to genetic erosion than was previously assumed [59, 60].
The AMOVA analysis of nuclear genes, through EF-1α, showed significant genetic differentiation between the Italian and Polish populations. This may indicate a different population origin occurred in both countries and some indications of this can be found among 24 sequences available in Genbank. A total of twelve EF-1α haplotypes in addition to our samples (30 haplotypes) were deposited by Als et al. . In the pooled sample, the most common haplotype (H1) was shared by all populations in Poland and Italy and has been found in Slovakia, Sweden and Spain. The H5 haplotype was the most common haplotype throughout Poland and network suggests its expansion here. It was also present in six out of nine Italian populations, although never as a dominant haplotype, and has been identified in Slovakia and Sweden as well. This suggests that Poland was not colonised exclusively from northern Italy. Poland and Italy shared five out of the 30 EF-1α haplotypes and one of them (H8) was also found in Slovakia . The presence of two common haplotypes in Poland may indicate that this part of Europe was colonised from two different routes (from the east and west). Finally, the H15 haplotype, which we identified only in northern Italy (CDF), was also found in Spain. The latter finding (and the presence of H1 in both countries that seems to be ancestral for the majority of Italian haplotypes; see network), may corroborate the existence of relationships between the Iberian and the Italian peninsula, as indicated by mtDNA and postulated by Ugelvig .
The low to moderate population differentiation observed in Poland for the EF-1α gene concurs with results from a previous study involving microsatellite markers that was conducted on the same DNA material; in which, rather low levels of differentiation for the species were determined, although some populations were highly isolated and the genetic differentiation was indicated to be mainly shaped by habitat fragmentation . Interestingly one population (HUT), where suitable biotope for M. arion was extensive, yielded the highest level of polymorphism in both EF-1α and microsatellite markers .
We did not find any evidence for IBD for any population grouping or for molecular marker. The lack of any correlation between geographic and genetic distances can be explained by historical demographic processes or cycles of colonisation and extinction. A similar pattern was identified for some Apennine populations of Bombina pachypus by Canestrelli et al. .
Putative subspecies divisions
Our genetic data do not support any subspecies divisions in M. arion. Among the Italian samples there were representatives of both M. a. ligurica (LOA, CUN, BDR) and M. a. obscura (VFE, CDF, VAL), while the remaining Italian populations (CET, AUR, CER), as well as all Polish populations belonged to M. a. arion. The analysis of the bar code sequence showed that two of the three populations belonging to M. arion ligurica were characterised by fixed or almost fixed haplotype H1, which was also fixed in 10 of 11 Polish populations and almost fixed in the remaining population. No species or subspecies-level differentiation was detected also as far as EF-1α gene is concerned. Hence, the results of our study are in agreement with the allozyme analysis of Hungarian populations indicating no differences between Origanum and Thymus dependent populations . This suggests that, despite the high ecological and morphological variability which led to the description of several subspecies, M. arion is a species with an absence of deep historic population isolation. Sielezniew & Dziekańska  hypothetise that clinal adaptation is a more likely explanation for the observed wing pattern variation in this butterfly than incipient speciation.
It is worth noting that in the case of two other Maculinea taxa (M. alcon and M. ‘rebeli’), molecular analysis revealed no differences supporting division into separate species [33, 62] and possible selective sweeps cannot be ruled out as well . A lack of host-associated genetic differentiation in the barcoding gene COI is reported by Craft et al.  from New Guinea, where 7 out of 28 analysed Lepidoptera species exhibited the same haplotypes despite the use of multiple LHPs. A similar pattern was observed by Hulcr et al.  in the moth Homona mermerodes.
We can speculate that use of different LHPs by a locally monophagous butterfly species, such as M. arion, has not been a mean to promote speciation, but can be ascribed to local adaptations in the species’ phenology that have evolved recently. For instance, when one of the glacial refugia was localised at the southern base of the Alps, after deglaciation, some populations could expand their range to higher altitudes tracking hosts re-colonisation, while others were able to survive in the lowlands by shifting to a novel host plant . If this scenario is true, the M. a. ligurica populations may represent a relic of M. arion populations that survived at the base of the Alps and were a source for the re-colonisation of the Alps.