The common wall lizard Podarcis muralis exhibited a complex phylogeographic pattern with multiple divergent mtDNA clades across its range. In total 23 allo- parapatric clades were identified, most of them with a restricted distribution in the Iberian, Italian, and Balkan peninsulas, as well as in Western and Eastern France, Slovenia and Austria, Central Balkans, and North-Western Anatolia. The detection of previously unidentified lineages was evidently due to the extended sampling scheme of this study as compared to earlier assessments, which were either focussed on a regional scale [27, 28], or did not sample all clades endemic to the Iberian, Balkan or Anatolian peninsulas . Although the definition of clades is somewhat arbitrary, the lack of resolution of relationships between monophyletic haplogroups, their high divergence, their geographic coherence, and results from the SAMOVA analysis (Figure 1A,B; Additional file 4) indicate that a grouping into fewer lineages would not be practical and does not describe accurately the mitochondrial DNA diversity exhibited by P. muralis.
The observed phylogeographic structure of P. muralis does not match the current subspecific division of this species [22, 23], as already reported by  and  for the 11 subspecies occurring in Italy. Regarding the subspecies found outside Italy, many divergent clades fall into P. m. brogniardii (clades 1–3, 5, and 6) and P. m. albanica (clades 12–14, and 18). Though a taxonomic revision is beyond the scope of this study, our results indicate that the current intraspecific taxonomy of P. muralis reflects local phenotypic varieties rather than distinct evolutionary units.
While mitochondrial data showed high variability and a clear geographic structure of genetic diversity, nuclear data showed both relatively low genetic variability and shallow divergence among populations. This finding is in agreement with allozyme data ; see also  from which low levels of polymorphism and genetic differentiation were identified (number of alleles per locus, A = 1.07; proportion of polymorphic loci, P = 0.07; observed heterozygosity, H
o = 0.028; genetic distance D
NEI = 0.036; average values from ) among 23 populations of P. muralis attributable to eight mitochondrial lineages described in our study (clades 1, 2, 3, 5, 9, 14, 22, and 23). Despite low differentiation in nuclear markers, F
ST values between mitochondrial clades are significant for all three nuclear markers and the mean between-clade pairwise distance is significantly different from that observed within clades, indicating that they are significant evolutionary lineages sensu . Furthermore, even though overall nuclear data showed low genetic variation, a weak geographic association of nuclear haplotypes is apparent in sequence data. At the acm4 and mc1r loci, besides a common widespread haplotype, some closely related derived haplotypes, relatively frequent, were found in the Balkans, Turkey, Slovenia, and Eastern France. This geographic association was more evident at the pdc locus, where we observed two main haplotype groups: one including all haplotypes from Italy, and the other comprising haplotypes from the Balkans, Turkey, Slovenia, and Eastern France. Both haplogroups also occurred in Spain and Western France. Such shallow phylogeographic patterns at nuclear loci suggest that the divergence between mitochondrial clades has not been long enough for them to reach reciprocal monophyly at the nuclear level [50–52]. According to mitochondrial DNA TMRCA estimates, divergence among lineages is estimated to have occurred since the Early Pleistocene. Taking into account the slower evolutionary rates of nuclear genes, the genetic pattern observed at the mitochondrial and nuclear level is compatible with a scenario of Pleistocene divergence among lineages. Yet, the use of more variable markers such as microsatellites is needed for a full assessment of the nuclear pattern of variation within this species.
The finding of conspicuous genetic divergence among geographic clusters of populations has provided evidence for the existence of cryptic species within other Podarcis taxa including the P. erhardii, P. tiliguerta, and P. hispanica species complexes. Given the extensive genetic structure observed in P. muralis and the lack of evidence for syntopy between mitochondrial lineages (except in four populations in Italy, see Figure 1), it might be suggested that some form of barrier to gene exchange has existed between distinct lineages. Nevertheless, we have no evidence that P. muralis constitutes a species complex. Pairwise uncorrected p-distances between clades ranges from 1.0 to 7.4% in cytb and from 0.9% to 4.7% in nd4, which is a level of divergence typically found within (and not between) recognised species of the genus (e.g. P. vaucheri, ). Further, there is no evidence for high differentiation in nuclear markers which would be indicative of long-term independent evolutionary trajectories. While a low divergence level among genetic lineages is not necessarily a synonym of absence of reproductive isolation, the available evidence suggest that, contrary to other Podarcis forms, P. muralis is a single species and not a species complex.
This does not mean, however, that the genetic landscape of the species is uniform. On the contrary, genetic variation in P. muralis is structured in many distinct lineages across its range. The phylogeographic pattern observed, with many reciprocally monophyletic allo- parapatric lineages having a Pleistocene divergence, suggests a scenario of long-term isolation in multiple ice-age refugia [1, 55]. Therefore, the geographic distribution of these genetic lineages is likely to reflect the refugial structure of this species and indicate putative areas which allowed the long-term maintenance of its genetic diversity. Assessments of many widespread European species have recovered a classic pattern of higher genetic diversity and multiple genetic lineages within southern European peninsulas . On one level the pattern of P. muralis is not an exception: more than three quarters of all lineages are endemic to the Iberian (2 and 3), the Italian (7–10, 19–23) and the Balkan peninsulas (12–14, 18). The simplest explanation for this phylogeographic pattern is one of a large number of refugia occurring not only across southern peninsulas, as pioneer works in phylogeography have demonstrated at a continental scale (e.g. ), but also within each peninsula, conforming to the “refugia within refugia” paradigm . Such compartmentalization is particularly evident in Italy. Multiple, related lineages occur in South Italy, whereas other, unrelated lineages occur in central and northern regions. These latter lineages from central and northern regions were considered by  as belonging to a main lineage (clade “1A”) leading to the inference of a main refugial area north of Campania and Apulia. However, though this clade was not supported in any of their phylogenetic analyses (BP ≤ 56, BPP = 0.54), in our analyses its constituent lineages are statistically well supported, unrelated, well differentiated one from the other, and parapatrically distributed. Thus, a scenario of multiple independent refugia in Central and Northern Italy appears to fit the data better. A similar reasoning suggests the existence of multiple independent refugia in the southern Balkan Peninsula. On the other hand, the southern Italian Peninsula would have acted as an independent biogeographic compartment with four related lineages arising from distinct refugia within this main refugial area. The same situation is observed in the westernmost and easternmost areas of the species range - in the Iberian Peninsula and in Turkey. Here, two or three endemic related lineages, respectively, occupy geographically different regions that likely acted as distinct refugia within a main refugium. It is noteworthy that in Iberia, where the species currently has a discontinuous distribution, genetic discontinuities do not fully correspond to geographic isolates, suggesting recent fragmentation dynamics.
A scenario of survival in multiple refugia within a southern peninsula has been inferred for many species of the continental Europe fauna and detailed studies on amphibians and reptiles provide the best evidence (see  for a recent review). For example, there is evidence for multiple Iberian refugia in Salamandra salamandra, multiple Italian refugia in Triturus carnifex and various Balkan refugia in Lacerta viridis. To our knowledge, however, P. muralis is the first reported case of differentiation into multiple refugia within all Mediterranean peninsulas. In fact, within P. muralis, diversity and divergence levels are not strikingly different among these regions (Table 2), suggesting that all Mediterranean peninsulas significantly and simultaneously contributed to the long-term persistence of this species throughout the Pleistocene glaciations. Therefore, the case of P. muralis not only conforms to the scenario of “refugia within refugia” , but it also sets a new phylogeographic pattern of “refugia within all refugia”.
According to the high number of lineages found in southern European peninsulas, the phylogeographic pattern of P. muralis fits the classical southern refugia model to some extent. However, while for many species of continental Europe these Mediterranean peninsulas were both glacial refugia and source areas for northward postglacial colonization [2, 4], in the case of P. muralis peninsular lineages are not found outside of these areas. This means that southern peninsulas acted as glacial refugia but not as sources for postglacial expansion. A possible explanation for such an unusual pattern is that northern areas outside Mediterranean peninsulas were already occupied by the species. In fact, various distinct lineages of P. muralis were found outside southern peninsulas, in the Pyrenees and France (clades 1 and 6), Northern Italy (clade 5), Slovenia and Austria (clade 11) and the Central Balkans (clade 4), reaching the highest latitudes of the species’range. Phylogenetic analyses and permution tests showed that these lineages are well differentiated and evolutionarily independent from southern lineages. The occurrence of two allopatric lineages in Western and Eastern France (clades 1 and 6), which are unrelated and considerably divergent from each other and relative to lineages occurring in Iberia (clade 2 and 3), and on the other side of the Alpine arch in North Italy (clade 5), provides evidence for a long-term persistence of P. muralis in these areas during Pleistocene glaciations, within separate refugia. Considering that the highest genetic diversity of these lineages is found along the southern portion of their range in SW and SE France respectively (data not shown), we can hypothesize that the refugia where these lineages differentiated were located in proximity of the Mediterranean coast and acted as source areas for post-glacial recolonization of Northern France and Germany. This hypothesis has been tested and validated by a recent study focused on P. muralis populations from France and Luxembourg , and it is further supported by palaeoclimatic and phylogeographic studies which indicate southern France as an area of prolonged ecological stability where climatic oscillations were attenuated  and where genetic signatures of refugia have been found among multiple species, including plants and animals (e.g. [63, 64]). A similar reasoning applies to the allopatric mitochondrial lineages found in Northern Italy, Northern Adriatic, Eastern Slovenia and Austria, and Central Balkans (clades 5, 21, 11, and 4), which are allopatric and reciprocally monophyletic, thus suggesting long-term isolation in multiple ice-age refugia. The long-term persistence of temperate species in Northern Italy (including the Po Plain and the southern foothills of the Alps) and in the Northern Adriatic has been inferred for many amphibians and reptiles in recent phylogeographic studies based on the occurrence of divergent lineages endemic to these areas (e.g., Hyla intermedia, Pelobates fuscus, Pelophylax lessonae, Triturus carnifex, Podarcis sicula; see [59, 65]). Likewise, in the area spanning from the Eastern Alps to the South-Western Pannonian region, palaeobotanical, paleoclimatic, and genetic data indicate that temperate species such as the tree Fagus sylvatica and the butterfly Erebia medusa survived in isolated refugia during glacial periods [20, 21, 66, 67]. Finally, a refugial area in the central Balkans and south-western Carpathians has already been suggested for many species showing high genetic diversity and distinct lineages in this area where also broad-leaved trees survived glacial periods (e.g. [16, 66–68]).
Altogether these phylogeographic, palaeobotanical and palaeoclimatic studies provide evidence for the long-term persistence of temperate organisms outside the traditional refugia in southern peninsulas, and underline the prominent contribution of these northern refugia to the present-day genetic pools of many temperate species.