Jan C Habel, Luc Lens, Dennis Rödder and Thomas Schmitt
BMC Evolutionary Biology 2011, 11:215
"To study the biogeographical importance of the Maghreb region and its connection with Europe, we selected the Marbled White butterfly species complex Melanargia galathea (Linnaeus, 1758) and Melanargia lachesis (Hübner, 1790) as a model system using two analytical tools (allozyme polymorphisms and distribution modelling). Today, M. galathea is widely distributed from the Maghreb region (mountain ranges of Morocco, Algeria and Tunisia)[22,23] to the English Midlands , and from the Pyrenees to the Baltic Sea in Poland . On the Iberian Peninsula, M. galathea is replaced by its sibling species M. lachesis. Thus, the Italian peninsula is the only possible link between North Africa and Europe for M. galathea."
"A neighbour joining phenogram based on allele frequencies (Figure 1) showed a first split between M. galathea and M. lachesis with a genetic distance  of about 0.9. The second split between M. galathea populations from Tunisia and Sicily on the one hand and all remaining M. galathea populations on the other was in average about half of the genetic distance of the first split. The outgroup M. lachesis routing the tree also supports this split being the first one in M. galathea. The Tunisia - Sicily group showed a further genetic differentiation between these two geographic regions. All these splits are supported by bootstrap values. The remaining populations split into five groups, the populations from Morocco and four European groups: (i) mainland Italy and southern France (Condat), (ii) western Balkans and western Carpathian Basin, (iii) eastern Balkans, Romania and Central Europe, as well as (iv) south-western Alps (Col de Tende)."
Figure 2. Individual based genetic classification of Melanargia galathea and M. lachesis. Bayesian analysis of all M. galathea and M. Lachesis populations performed using the STRUCTURE software. Analyses for K = 2 to K = 8 are depicted.
"STRUCTURE plots (for K = 2 to K = 8) support the topology of the neighbor joining phenogram (Figure 2); the sequence of splits of STRUCTURE groups is mostly reflecting the genetic distances in the tree. Additionally, the samples from Morocco are consistently separated as one group from all other M. galathea populations from K = 5 onwards."
"Thus, these analyses strongly support (i) the genetic break in the Maghreb, (ii) the break between Sicily and mainland Italy, (iii) the differentiation into four genetic lineages in continental Europe, (iv) the cohesiveness between Sicily and Tunisia, (v) the lack of differentiation from the eastern Balkans via Romania to Central Europe and (vi) the strong genetic similarity between mainland Italy and southern France."
Figure 4. Biogeographic scenario of Melanargia galathea and M. lachesis. Range expansions and retractions of M. galathea and M. lachesis (on the Iberian Peninsula) during the past ice-ages (a-c) and fluctuations of marine isotope stages (d) (redrawn after Gibbard & van Kolfschoten ). Refugia are marked by grey areas, expansions/retractions by arrows. 4a) Günz and Mindel refugia in Iberia (M. lachesis) and the Maghreb (M. galathea); expansion from the eastern Maghreb to Sicily during the Mindel/Holstein transition; Holstein expansion over Europe, but retraction in the Maghreb to the west. 4b) Riss refugia in Iberia (M. lachesis) and the western Maghreb, Sicily and southern Balkans (M. galathea; Riss/Eem transition expansion from Sicily to the eastern Maghreb; Eem expansion of the southern Balkan group over major parts of Europe including peninsular Italy. 4c) Würm refugia in Iberia (M. lachesis) as well as western and eastern Maghreb, Sicily, peninsular Italy, southwestern Alps and Balkan area (M. galathea); postglacial expansion from Iberia (M. lachesis), peninsular Italy and the Balkan area (M. galathea).
"The recent geographic restriction of M. lachesis to Iberia and the highest genetic diversity of M. galathea in Morocco support the idea of a centre of origin of the entire species complex in this area. This assumption is further supported by other Melanargia species mostly endemic to the Atlantic-Mediterranean region (M. occitanica, M. ines) and other endemics to further Mediterranean refugia (M. arge: peninsular Italy; M. pherusa: Sicily, M. larissa: Pontic-Mediterranean region and Iran). The onset of the differentiation between these sister species should be due to vicariance events most likely correlated with the onset of an ice age. If giving one glacial-interglacial cycle for the lowest level of differentiation (i.e. the subgroups within the two major M. galathea lineages), the most likely time horizon of this vicariance event is the onset of the Günz glaciation some 560,000 years BP  (Figure 4a). Since then, M. lachesis most likely has never expanded out of Iberia whereas M. galathea colonised most of Europe from its Maghreb expansion centre. Similar splits between Iberia and the Maghreb are commonly observed in many species groups [e.g. [13,34-36]]."
From the Maghreb to Europe:
"The deepest split in the M. galathea populations is between the Sicily - Tunisia group and all the other populations. As this split is about twice the genetic differentiation among their subgroups and less than half of the distance against M. lachesis, the onset of the Riss glaciation (about 310 ky BP)  might be the trigger for vicariance and thus the beginning of this differentiation. As (i) Iberia was continuously blocked for the expansion of M. galathea to Europe by M. lachesis [cf. 27] and (ii) all European M. galathea populations except Sicily are more similar to populations from Morocco than from Tunisia, a scenario with this split taking place in the Maghreb is little likely. This assumption is further supported by SDMs for ice age conditions predicting mostly continuous distributions over North Africa (Figure 3b) thus allowing vicariance in this region only during the relatively short interglacial stages. For these reasons, M. galathea must have reached Europe before the Riss glaciation."
"As the region of the eastern Sahara in Egypt apparently always have been too dry for an expansion of M. galathea, this first expansion of M. galathea to Europe must have been from Tunisia to Sicily (Figure 4a), a sea strait known for biogeographical connections for many taxa [e.g. ; and references therein]. As the Strait of Sicily was considerably narrower during glacial periods due to eustatic sea level lowering, the transition from Mindel glaciation to Holstein interglacial with still low sea level but already higher temperatures might have been a suitable time period for this dispersal. After arrival to Sicily, the Holstein interglacial might have given suitable condition for the expansion of M. galathea over most parts of Europe, including the Balkans but excluding Iberia as this peninsula was already populated by M. lachesis (Figure 4a)."
"With the climatic cooling of the Riss ice age [about 200 ky BP, see Timeline of Glaciation (link)], which was considerably longer than the following Würm glaciation and had longer durations of minimum temperatures[33,37], M. galathea most probably was nearly extinct in Europe only surviving in the southernmost possible retreats in Sicily and the southern Balkan (Peleponnesos), but also in the Maghreb; M. lachesis could survive in southern Iberia (Figure 4b)."
... and back to the Maghreb
"As the time for differentiation between the four M. galathea lineages from continental Europe is assumed to be the result of one glacial cycle (see above) and as the differentiation between populations from Sicily and Tunisia are in the same order of magnitude, we assume that the onset of this differentiation is in the same time frame. As the genetic diversity is significantly higher in Sicily than in Tunisia and the warm and dry interglacial climatic conditions in Tunisia generally unsuitable for the survival of M. galathea, we assume that a colonisation most likely has taken place from Sicily to Tunisia. While the sea level was still considerably lowered at the transition from Riss to Eem [about 130 ky BP, see Timeline of Glaciation (link)], thus facilitating dispersal between these two areas, this time period might be the most likely for this expansion event. During the following Eem interglacial [130ky - 110ky BP], the Balkan refuge of M. galathea most probably could colonise most parts of Europe apart from Iberia and Sicily, which were occupied by other genetic lineages of this species complex (Figure 4b)."
The existence of extra-Mediterranean refugia for thermophilic taxa
"During the Würm ice age [110ky - 12ky BP (link)], which was not more severe than the two previous glaciations but with a shorter maximum , the Marbled White butterflies were not that much pushed to the South than in the previous cases. This is well matching the remarkable differentiation of the species in Europe allowing to distinguish five lineages (see above), which most likely are the result of survival of the Würm ice age in a larger number of different refugia."
"This pattern implies at least two different refugia at the Balkan Peninsula at the western and the eastern flank; more in detail analyses also support a third Balkan centre in the peninsula's southern parts  (Figure 4c). This pattern of multiple refugia in the Balkans was already erected by Reinig  postulating different centres of survival in the western, southern and eastern Balkans and was later supported by genetic analyses showing genetic divergences between these areas for a variety of different animal species [e.g. [18,40-42]]."
"Furthermore, different Würm refugia have to be postulated for Sicily and peninsular Italy, a pattern also repeated by other genetic analyses [e.g. [17,43]]. Furthermore, other genetic studies show a remarkable genetic differentiation in the southernmost parts of peninsular Italy [e.g. [34,44,45]]."
"The last remaining lineage of M. galathea in the south-western Alps most likely is not representing a Mediterranean refuge of this species, but an extra-Mediterranean refuge area at the southern slopes of the glaciated Alps (Figure 4c). As already shown by Steward and Lister , glacial survival of temperate species in Europe was not only possible in the classical Mediterranean refugia sensu de Lattin , but also in small climatically buffered pockets in more northern regions [8,48,49]. Recent works especially highlight the southern and south-eastern parts of the Alps of particular importance for additional Würm ice age refugia for temperate species [e.g. [42,50,51]], and also for species formerly thought to be of exclusive Mediterranean origin [e.g. [52,53]]. This apparently was also the case for the Marbled White."
"During the Postglacial [12ka BP to present (link)], several lineages of M. galathea were mostly blocked in their expansion by other lineages representing the respective leading edges [cf. 54]. In the case of M. galathea in Morocco, their northwards expansion was blocked by M. lachesis distributed in Iberia. The lineage surviving in the eastern Balkans apparently had the most important impact in the recolonisation of more northern parts of Europe as its dispersal was not hampered by any major mountain obstacle [cf. 9] so that this lineage could expand throughout Central Europe to the western parts of Germany (Figure 4c). However, the samples of north-eastern France and southern Germany show an intermediate genetic structure between this lineage and the south-western Alps lineage, making hybrid origin of these populations rather likely and thus expansion of the southern Alps lineage over the chains of the Alps."
"Also the Italian lineage could expand beyond its refugium to southern France. Therefore the entire region of northern France and southern Germany might be a zone of mixing between these three lineages. Hybrid zones between different taxa are frequently observed in this region [e.g. [9,55]]. Furthermore, the southernmost population in Calabria (southern Italy) has an intermediate genetic texture between the Italian and the Sicily group thus speaking for a postglacial contact and intermixing between these two groups in this region."