The Intriguing Biogeographic Pattern of the Italian Wall Lizard Podarcis siculus (Squamata: Lacertidae) in the Tuscan Archipelago Reveals the Existence of a New Ancient Insular Clade

Simple Summary Islands represent hotspots of biodiversity and ideal places to study evolutionary processes. Wall lizards of the genus Podarcis are widespread and well-studied species in Europe, and they are present on several islands and islets across the Mediterranean. They show a remarkable hidden insular biodiversity that has often revealed the presence of undescribed biological entities. In this study, both the genetic and morphological diversity of the populations of P. siculus of the Tuscan Archipelago were investigated. Our results revealed an intriguing biogeographical pattern which could be the result of at least two different colonization waves. Particularly, we identified a new ancient insular clade from Giglio and Capraia islands which surely deserves conservation efforts and further taxonomic investigation. Abstract The Tuscan Archipelago is one of the most ancient and ecologically heterogeneous island systems in the Mediterranean. The biodiversity of these islands was strongly shaped by the Pliocene and Pleistocene sea regressions and transgression, resulting in different waves of colonization and isolation of species coming from the mainland. The Italian wall lizard, Podarcis siculus, is present on the following islands of the Tuscan Archipelago: Elba, Giglio, Giannutri, Capraia, Montecristo and Cerboli. The species in the area displays a relatively high morphological variability that in the past led to the description of several subspecies. In this study, both the genetic and morphological diversity of P. siculus of the Tuscan Archipelago were investigated. Specifically, the meristic characters and the dorsal pattern were analyzed, while the genetic relationships among these populations were explored with mtDNA and microsatellite nuclear markers to reconstruct the colonization history of the Archipelago. Our results converge in the identification of at least two different waves of colonization in the Archipelago: Elba, and the populations of Cerboli and Montecristo probably originate from historical introductions from mainland Tuscany, while those of Giglio and Capraia are surviving populations of an ancient lineage which colonized the Tuscan Archipelago during the Pliocene and which shares a common ancestry with the P. siculus populations of south-eastern Italy. Giannutri perhaps represents an interesting case of hybridization between the populations from mainland Tuscany and the Giglio-Capraia clade. Based on the high phenotypic and molecular distinctiveness of this ancient clade, these populations should be treated as distinct units deserving conservation and management efforts as well as further investigation to assess their taxonomic status.


Introduction
Island systems are recognized as hotspots of biodiversity across the world, and they can be seen as a real cauldron of evolution. The long-time isolation and the peculiar environmental conditions can trigger deep intraspecific and interspecific biological diversification, especially in small islands [1]. The Tuscan Archipelago is one of the most ancient and ecologically heterogeneous island systems in the Mediterranean Sea. It is composed of seven main islands and several islets lying between Corsica and the Italian Peninsula with a great diversity of biotopes (from deciduous Mediterranean oak forest to bush-shrub lands). The biodiversity of these islands was strongly shaped by the Pliocene and Pleistocene sea regressions and transgression [2][3][4]. This phenomenon is well known as a source of biological evolution, allowing waves of colonization of species from the mainland during periods of glacial marine regression, followed by isolation due to the rising sea levels. In this context, many different organisms were able to evolve, diversify, and become extinct [5][6][7][8]. While some islands of the Tuscan Archipelago were connected to the mainland during the Pleistocene (land bridge islands), others remained isolated. For example, the low sea level during the last glacial maximum probably led Elba, Pianosa and Giannutri to be connected with mainland Tuscany, while Montecristo, Capraia, Gorgona and Giglio Islands remained isolated ( Figure 1).

Introduction
Island systems are recognized as hotspots of biodiversity across the world, and they can be seen as a real cauldron of evolution. The long-time isolation and the peculiar environmental conditions can trigger deep intraspecific and interspecific biological diversification, especially in small islands [1]. The Tuscan Archipelago is one of the most ancient and ecologically heterogeneous island systems in the Mediterranean Sea. It is composed of seven main islands and several islets lying between Corsica and the Italian Peninsula with a great diversity of biotopes (from deciduous Mediterranean oak forest to bushshrub lands). The biodiversity of these islands was strongly shaped by the Pliocene and Pleistocene sea regressions and transgression [2][3][4]. This phenomenon is well known as a source of biological evolution, allowing waves of colonization of species from the mainland during periods of glacial marine regression, followed by isolation due to the rising sea levels. In this context, many different organisms were able to evolve, diversify, and become extinct [5][6][7][8]. While some islands of the Tuscan Archipelago were connected to the mainland during the Pleistocene (land bridge islands), others remained isolated. For example, the low sea level during the last glacial maximum probably led Elba, Pianosa and Giannutri to be connected with mainland Tuscany, while Montecristo, Capraia, Gorgona and Giglio Islands remained isolated ( Figure 1).  The islands of the Tuscan Archipelago host a rich and peculiar biological community, strongly influenced by historical, biological and current eco-geographical factors [2,4,9]. Furthermore, since ancient times, humans have favored the spread of species between the various islands and between them and the continent by altering the community composition [10][11][12][13]. Some examples are the presence of Vipera aspis hugyi on Montecristo or Zamenis longissimus on Elba [14,15]. All these factors contribute to the islands' complex and diverse biological community, which in many cases has remained little studied or completely unknown.
The wall lizards of the genus Podarcis are widespread and well-studied species in Europe, and they are present on several islands and islets across the Mediterranean. The remarkable hidden insular biodiversity of this group in terms of deep genetic and phenotypic divergence has often revealed the presence of unique biological entities that may be important with regard to their conservation [16][17][18][19][20][21][22][23][24][25][26][27][28]. Podarcis siculus is present on the following islands of the Tuscan Archipelago: Elba, Giglio, Giannutri, Capraia, Montecristo and Cerboli. The species displays in the area a relatively high morphological variability that in the past led to the description of several subspecies [29]. Despite this, only one study has been conducted on the genetic variability of P. siculus of the Tuscan Archipelago [17] using allozymes, while no molecular analysis based on mtDNA and nuclear markers is available yet.
In this study, both the genetic and morphological diversity of P. siculus in the Tuscan Archipelago were investigated in order to shed new light on the history of the colonization of the Archipelago by this species. Specifically, the meristic characters and the dorsal pattern of a large number of specimens of the islands of Elba, Giglio, Giannutri, Capraia, Cerboli and Montecristo were analyzed. In addition, the genetic relationships among these populations were explored with mtDNA and microsatellite nuclear markers to reconstruct the colonization history of the Archipelago.

Dorsal Pattern Diversity
Differences in dorsal pattern (DP) were studied on 630 adult specimens (393 males and 237 females; details in Table 1) kept at the "La Specola" Museum. Specimens belonging to both mainland Tuscany and six different islands of the Tuscan Archipelago (Giglio, Capraia, Giannutri, Montecristo, Cerboli, Elba) were analyzed. Each specimen was assigned to one of the six defined categories of DPs as shown in Figure 2. The categories are named as follows: "campestris", showing a brown and/or black continuous or partially interrupted occipital stripe bordered on both sides by a relatively large brown or green parietal band, at least in the posterior half of the body; "calabresiae" (from the name of the subspecies described for Montecristo Island-P. s. calabresiae [30]), similar to "campestris" but with very thin stripe-like parietal bands; "reticulated", showing a complete or partially complete coalescence of the elements composing the occipital stripe and the ones composing the parietal bands; "concolor", showing no DP; "intermediate 1", an intermediate phenotype between "campestris" or "calabresiae" and "reticulated"; "intermediate 2", an intermediate phenotype between "campestris" or "calabresiae" or "reticulated" and "concolor". To test possible association between phenotypes and islands, a χ 2 test was conducted separately for males and females.

Morphological Analysis
A subset of 237 adult specimens (151 males, 86 females; see Table 1 for details) belonging to six islands of the Tuscan Archipelago (Giglio, Capraia, Giannutri, Montecristo, Cerboli, Elba) and mainland Tuscany, preserved at the "La Specola" Museum of the Natural History Museum of the University of Florence (MZUF), were morphologically analyzed. The snout-vent length (SVL) was measured at the nearest 0.1 cm and the following meristic characters were counted: (a) number of midbody scales, (b) ventral scales, (c) collar scales, (d) gular scales, (e) femoral pores and (f) subdigital lamellae on the right leg, (g) supraciliary scales, (h) supraciliary granules, (i) supratemporal scales (Table S1). We then performed a PCA on meristic characters to identify the groups that better explained the morphological diversity in the Archipelago. Successively, we tested the presence of statistically significant differences between sexes and between islands performing an ANOVA on the SVL, and a MANOVA on the meristic data.

Morphological Analysis
A subset of 237 adult specimens (151 males, 86 females; see Table 1 for details) belonging to six islands of the Tuscan Archipelago (Giglio, Capraia, Giannutri, Montecristo, Cerboli, Elba) and mainland Tuscany, preserved at the "La Specola" Museum of the Natural History Museum of the University of Florence (MZUF), were morphologically analyzed. The snout-vent length (SVL) was measured at the nearest 0.1 cm and the following meristic characters were counted: (a) number of midbody scales, (b) ventral scales, (c) collar scales, (d) gular scales, (e) femoral pores and (f) subdigital lamellae on the right leg, (g) supraciliary scales, (h) supraciliary granules, (i) supratemporal scales (Table S1). We then performed a PCA on meristic characters to identify the groups that better explained the morphological diversity in the Archipelago. Successively, we tested the presence of statistically significant differences between sexes and between islands performing an ANOVA on the SVL, and a MANOVA on the meristic data.

Genetic Analysis
29 tissue samples were collected from 2014 to 2021 on Giglio, Capraia, Giannutri, Montecristo, Cerboli and Elba islands. Tissues were obtained from small tail tips by inducing autotomy after light pressure, all the individuals were released at the capture site. Tail tissues were stored in 96% pure ethanol. Genomic DNA was extracted from all the tissue samples by means of the universal extraction protocol [32]. We analyzed the mitochondrial gene cytochrome b (cytb) and eight microsatellite loci. Cytb was amplified using the primer employed by Podnar et al. [33] Cyt F (5 -TTTGGATCACTATTRGGCCTCTGCC-3 ) and H15425 (5 -GGTTTACAAGACCAGTGCTTT-3 ). Amplification was carried out using the following protocol: an initial denaturation at 94 • C for 2 min followed by 35 cycles with denaturation at 95 • C for 10 s, annealing at 55 • C for 20 s, and elongation at 72 • C for 90 s and a final extension at 72 • C for 7 min. Microsatellite loci were selected from among those available in literature [34][35][36][37] and amplified in three different mixes (mix 1: Pb73, C9; mix 2: Pli24, Pli4, Pli18, Pb10; mix 3: Lv-4-a, Lv-4-19; see Table S4 for details). Amplifications were carried out using different protocols for mix 1-2 and mix 3. For mix 1 and 2, PCR Animals 2023, 13, 386 5 of 16 was performed with an initial denaturation step at 94 • C for 5 min, followed by 32 cycles composed of a denaturation at 95 • C for 30 s, annealing at 57 • C for 90 s, elongation at 72 • C for 30 s and a final extension at 60 • C for 30 min. For mix 3, PCR was performed with an initial denaturation at 94 • C for 5 min followed by 32 cycles of denaturation at 95 • C for 30 s, annealing at 59 • C for 90 s, elongation at 72 • C for 30 s and a final extension at 60 • C for 30 min. Each individual was genotyped with the ABI3130 system and results were visualized with Thermo Fisher Peak Scanner Software.
The obtained cytb dataset was completed by adding 340 sequences from those available in GenBank from previous studies [25,33] covering the entire distribution area of Podarcis siculus (i.e., continental Italy, Croatia, Sicily, Sardinia, Corsica and some smaller islands). No sequences from the Tuscan Archipelago were available in GenBank. Sequences of cytb were aligned with MUSCLE [38] in MEGA (version 11.0.13; [39]). We then built a median joining network using the software PopArt (version 1.7; [40,41]). Successively, we calculated p-distances among the clades/lineages retrieved with the network. Population structure based on microsatellite loci was reconstructed using the spatial PCA (sPCA) as implemented in the R package adegenet (version 2.1.5; [42]). In addition, the genetic clustering was investigated using an admixture analysis with k values from 1 to 4 and the cross-validation score for each k value was calculated. To do that, we used the tess3r package (version 1.1.0; [43]) in R.

Dorsal Pattern
Considering males only, there is a high frequency of the "calabresiae" phenotype in Montecristo, while mainland Tuscany, Elba and Cerboli islands show a high frequency of the "campestris" phenotype. In the other islands there is a higher diversity of patterns with "reticulated", and "concolor" phenotypes together with their intermediates. Females follow a similar pattern with the notable difference consisting of the fact that almost all the females of Capraia show the "campestris" phenotype. The χ 2 test confirmed the statistical significance of the differences in the phenotypic frequencies reported above (p-value < 0.05) in males and females. DP frequencies are shown in Figure 3 and in Tables S2 and S3.

Morphological Analysis
The PCA results are slightly different between sexes. In males, there are two very distinct groups: one being composed of Giglio, Capraia and Giannutri and one of Elba, Cerboli, Montecristo and mainland Tuscany. In females, we can still distinguish two groups but less distinct than in males: one that includes Giglio and Capraia which are quite similar to each other, and one that includes Elba, Montecristo, Giannutri and mainland Tuscany with Cerboli being intermediate between the groups (Figure 4). MANOVA confirms the significance of these differences in morphological traits between sexes and between localities. On the contrary, ANOVA shows the presence of statistically significant differences in SVL between sexes only, with males being larger than females. There are not statistically significant differences in SVL between the different sampling localities (between islands and between islands and mainland Tuscany). MANOVA and ANOVA results are reported in Tables 2 and 3, respectively. SVLs are shown in Figure 5.

Genetic Analysis
We successfully obtained 24 cytb sequences from our samples. Elba island was excluded from the analysis due to the very low quality of its cytb sequences. The medianjoining network ( Figure 6) revealed the presence of the main mitochondrial clades already described by Senczuk et al. [25]: a Siculo-Calabrian lineage and a central-northern lineage splitting into Adriatic and Tyrrhenian clades. Specimens from Cerboli, Giannutri and Montecristo belong to the Tyrrhenian clade. Specifically, four haplotypes are found on these islands: two from Montecristo, one from Giannutri and one from Cerboli. Furthermore, a new undescribed clade, well differentiated from the previous ones, emerged from the analysis. This clade includes all specimens from the islands of Giglio and Capraia. The newly identified Giglio-Capraia clade has eight unique haplotypes found nowhere else in the range of P. siculus. These haplotypes appear to be very close to each other and related to the Adriatic clade. It is worth noting that Giglio and Capraia also show a high genetic distance (p-distance) from the mainland populations of P. siculus (7.3-7.5% from the Adriatic clade and 8.7-9.1% from the Tyrrhenian clade; Table 4) and that there is also some divergence between the haplotypes of the two islands (5.4%).

Morphological Analysis
The PCA results are slightly different between sexes. In males, there are two ver distinct groups: one being composed of Giglio, Capraia and Giannutri and one of Elba Cerboli, Montecristo and mainland Tuscany. In females, we can still distinguish tw groups but less distinct than in males: one that includes Giglio and Capraia which ar quite similar to each other, and one that includes Elba, Montecristo, Giannutri and main       All 29 sampled individuals were successfully genotyped using our microsatellites panel (see Table S5 for details). However, only seven out of the eight loci were successfully obtained. In the sPCA (Figure 7), all the islands appear distinct from each other with the sole exception of Elba and Cerboli, which show full overlap.

Genetic Analysis
We successfully obtained 24 cytb sequences from our samples. Elba island was excluded from the analysis due to the very low quality of its cytb sequences. The medianjoining network ( Figure 6) revealed the presence of the main mitochondrial clades already described by Senczuk et al. [25]: a Siculo-Calabrian lineage and a central-northern lineage splitting into Adriatic and Tyrrhenian clades. Specimens from Cerboli, Giannutri and Montecristo belong to the Tyrrhenian clade. Specifically, four haplotypes are found on these islands: two from Montecristo, one from Giannutri and one from Cerboli. Furthermore, a new undescribed clade, well differentiated from the previous ones, emerged from the analysis. This clade includes all specimens from the islands of Giglio and Capraia. The newly identified Giglio-Capraia clade has eight unique haplotypes found nowhere else in the range of P. siculus. These haplotypes appear to be very close to each other and related to the Adriatic clade. It is worth noting that Giglio and Capraia also show a high genetic distance (p-distance) from the mainland populations of P. siculus (7.3-7.5% from the Adriatic clade and 8.7-9.1% from the Tyrrhenian clade; Table 4) and that there is also some divergence between the haplotypes of the two islands (5.4%). Adriatic clade 7.3% 7.5% 5.8% 5.3% 6.5% 5.5% 6.9% -  All 29 sampled individuals were successfully genotyped using our microsatellites panel (see Table S5 for details). However, only seven out of the eight loci were successfully obtained. In the sPCA (Figure 7), all the islands appear distinct from each other with the sole exception of Elba and Cerboli, which show full overlap.   All 29 sampled individuals were successfully genotyped using our microsatellites panel (see Table S5 for details). However, only seven out of the eight loci were successfully obtained. In the sPCA (Figure 7), all the islands appear distinct from each other with the sole exception of Elba and Cerboli, which show full overlap.  The results of the admixture analysis are shown in Figure 8. For k = 2 Giglio and Montecristo show no signs of admixture with each other or with the other islands. Conversely, Capraia, Cerboli, Elba and Giannutri, show a signature of admixture of the two genetic clusters. For k = 3, the specimens from Giglio and Capraia, form two separate genetic clusters. The third cluster includes all the Montecristo specimens. The specimens of Cerboli, Elba and Giannutri show signs of admixture between Giglio and Montecristo clusters. For k = 4, Giglio, Capraia and Montecristo keep on being well distinct, while Elba, Cerboli and Giannutri show admixture between clusters. Particularly, Giannutri shows signs of admixture with Giglio. According to cross-validation scores, k = 4 is the best fitting k value (Figure 9). netic clusters. The third cluster includes all the Montecristo specimens. The specimens of Cerboli, Elba and Giannutri show signs of admixture between Giglio and Montecristo clusters. For k = 4, Giglio, Capraia and Montecristo keep on being well distinct, while Elba, Cerboli and Giannutri show admixture between clusters. Particularly, Giannutri shows signs of admixture with Giglio. According to cross-validation scores, k = 4 is the best fitting k value (Figure 9).

Discussion
The final picture that emerges from our results describes a very complex and intriguing biogeographic pattern with a high morphological, phenotypic, and genetic diversity between islands that certainly deserves further investigation.
The previous allozyme analysis by Corti et al. [17] showed that the populations of Giglio, Capraia and Giannutri differ from those of the other islands of the Archipelago which are genetically close to those of the mainland. The morphological distinctiveness of the lizards of Giglio, Giannutri and Capraia compared to those of the remaining islands had already been observed by Mertens [44] and Corti [16]. Mertens defined the lizards of these islands as a race of Lacerta sicula joining them under the name L. s. tyrrhenica, thus highlighting the distinctiveness of the populations of Giglio and Capraia and considering that of Giannutri in part similar to them.
The results of the present study match in part with these earlier observations. In fact, similarities between lizards from Giglio, Capraia and Giannutri are highlighted by morphological data in males and by DP frequencies in both sexes. The discrepancy consists in the fact that, while Giglio and Capraia belong to a well differentiated mtDNA clade, Giannutri belongs to the Tyrrhenian clade. Moreover, considering nuclear microsatellite markers, while for Giglio and Capraia there is a clear isolation that shows no signs of admixture, specimens from Giannutri seem to be admixed with Giglio, Elba and Cerboli. Therefore, it seems reasonable that the population of Giannutri could represent the result of hybridization/introgression between the ancient Giglio-Capraia clade and the Tyrrhenian clade which may have reached Giannutri with historical introductions or during the LGM, when the island was connected to mainland Tuscany. However, the low sample size of Giannutri used for molecular analysis does not allow conclusions to be drawn on this point. Apart from the case of Giannutri, our results converge in the identification of a clearly phenotypically and genetically distinct lineage of lizards on the islands of Giglio and Capraia. While lepidosis is able to separate the populations belonging to the different clades, SVL does not show a great sign of variation across the sampled localities, and this

Discussion
The final picture that emerges from our results describes a very complex and intriguing biogeographic pattern with a high morphological, phenotypic, and genetic diversity between islands that certainly deserves further investigation.
The previous allozyme analysis by Corti et al. [17] showed that the populations of Giglio, Capraia and Giannutri differ from those of the other islands of the Archipelago which are genetically close to those of the mainland. The morphological distinctiveness of the lizards of Giglio, Giannutri and Capraia compared to those of the remaining islands had already been observed by Mertens [44] and Corti [16]. Mertens defined the lizards of these islands as a race of Lacerta sicula joining them under the name L. s. tyrrhenica, thus highlighting the distinctiveness of the populations of Giglio and Capraia and considering that of Giannutri in part similar to them.
The results of the present study match in part with these earlier observations. In fact, similarities between lizards from Giglio, Capraia and Giannutri are highlighted by morphological data in males and by DP frequencies in both sexes. The discrepancy consists in the fact that, while Giglio and Capraia belong to a well differentiated mtDNA clade, Giannutri belongs to the Tyrrhenian clade. Moreover, considering nuclear microsatellite markers, while for Giglio and Capraia there is a clear isolation that shows no signs of admixture, specimens from Giannutri seem to be admixed with Giglio, Elba and Cerboli. Therefore, it seems reasonable that the population of Giannutri could represent the result of hybridization/introgression between the ancient Giglio-Capraia clade and the Tyrrhenian clade which may have reached Giannutri with historical introductions or during the LGM, when the island was connected to mainland Tuscany. However, the low sample size of Giannutri used for molecular analysis does not allow conclusions to be drawn on this point. Apart from the case of Giannutri, our results converge in the identification of a clearly phenotypically and genetically distinct lineage of lizards on the islands of Giglio and Capraia. While lepidosis is able to separate the populations belonging to the different clades, SVL does not show a great sign of variation across the sampled localities, and this is in accordance with the low effect of insularity on size variation observed for P. siculus by Avramo et al. [45]. Another interesting observation is the DP of the lizards of Capraia where males are mostly "reticulated" and females mostly show the striped "campestris" phenotype (the most common in continental Tuscany and on Elba and Cerboli islands). This sex-related difference could be easily explained, as the DP and coloration of P. siculus is known to be sexually dimorphic [46]. On the other hand, we didn't recognize the same sexual dimorphism in the DP frequencies on the other islands. Moreover, the DP of females in Capraia might reflect some form of introgression from the mainland or nearby islands, but such an event should be detected through mtDNA and microsatellites analyses which is not our case at all. Thus, it is more likely that this may represent a case of convergent evolution of the DP, since striped patterns such as the "campestris" phenotype are very common within the genus Podarcis with different species showing similar patterns [47]. However, according to mtDNA, the Giglio-Capraia clade is related to the Adriatic clade (sensu Senczuk et al. [25]), which also share some phenotypic characteristics with these two islands, such as the relatively high frequencies of "reticulated" individuals even if "campestris" is still the most common phenotype within the clade [48]. This could suggest that the different DP frequencies of females observed on Capraia may reflect some dimorphic adaptive process. In fact, context-dependent expression of sexual dimorphism has already been observed in the Tuscan Archipelago for P. muralis [49]. One more possible explanation could be the mate-choice effect, but females of P. muralis have been observed to not actively choose males based on their color morph [50]. Anyway, further investigation is surely needed to shed light on this case of sexual dimorphism.
Montecristo almost exclusively shows the unique "calabresiae" phenotype and this agrees with the lack of nuclear microsatellites admixture with the other islands. However, this distinctiveness is not supported by morphological and mitochondrial data, and this could be due to the fact that the "calabresiae" phenotype can be considered as a subtype of "campestris", a phenotype commonly found on islands belonging to the Tyrrhenian clade, the same as Montecristo. Moreover, variations in phenotypic frequencies may occur very rapidly in island-dwelling reptiles [51,52]. In fact, the population of this island was probably founded by a small number of individuals introduced from the mainland [53] and the uniqueness of the "calabresiae" phenotype is probably founder effect induced.
The fact that Giglio and Capraia show a peculiar composition of different DP, compared with the almost monomorphic condition of the other islands, would reinforce the hypothesis of a different origin of the populations of P. siculus in the Tuscan Archipelago. In fact, it has already been observed that the distribution of DP phenotypes is generally associated with mitochondrial genetic clades in P. siculus [48] and this is in agreement with the mtDNA data.
Therefore, a possible explanation for the high diversity of the Italian wall lizard of the Tuscan Archipelago is a scenario of at least two colonization events of the Archipelago. Partially convergent evidence from molecular data, morphology and DPs would suggest an independent origin of the Giglio-Capraia mtDNA clade with respect to the populations of Elba, Cerboli, Montecristo and Giannutri. Particularly relevant is the high genetic distances observed for cytb between this clade and the populations of the other islands and of mainland Tuscany (8.9-9.1% for Capraia and 8.2-8.8% for Giglio). This level of genetic distance is even higher than those observed between some other well established Podarcis species (3.3% between P. raffonei and P. waglerianus; 7.75% between P. peloponnesiaca and P. erhardii [19,54]). Following the proposed evolutionary rates for cytb in Podarcis (1.45-1.59% Myr; [55]), the divergence time from the closest continental clade ranges approximately from 5.04 to 4.59 Myr for Giglio and from 5.17 to 4.72 Myr for Capraia. Therefore, according to these evolutionary rates, the ancient Giglio-Capraia clade would have colonized these islands in the early Pliocene. The high divergence between Giglio and Capraia (5.4%) also supports an ancient insular origin of this clade.
A similar pattern of ancient colonization of an archipelago may recall that of P. latastei from the Western Pontine islands which was considered a subspecies of P. siculus and recently raised to the species rank [26][27][28]. The fact that the Giglio-Capraia clade is present in two very distant islands could suggest that other islands were also populated in the past by this ancient clade and that it subsequently became extinct and replaced by other Podarcis clades. This would strengthen the hypothesis that Giannutri may actually also belong to the ancient Giglio-Capraia clade [17,44] and that introgression/hybridization with the Tyrrhenian clade may have occurred there. Furthermore, differential patterns of colonization have already been observed in several archipelagos for reptiles [56] and, in particular, in the Tuscan Archipelago for P. muralis with an ancient clade surviving on Elba, Pianosa and Palmaiola islands and some nearby islets [57]. It is worth noting that the estimated divergence time for these island populations of P. muralis is about 2-3 Myr. This time is much more recent than the estimated time of divergence of the insular Giglio-Capraia mtDNA clade of P. siculus. Therefore, the most ancient colonization of the islands of the Tuscan Archipelago is that of P. siculus and not that of P. muralis as previously believed [58].
Given their similarity with the continental populations, it is likely that the presence of P. siculus on the islands of Cerboli and Montecristo is the result of historical introductions, as already observed by Capula [53] for Elba Island. This may raise some concerns about the conservation of the endemic Giglio-Capraia lineage which may suffer from genetic introgression from accidental introductions of new lizards. This is probably what happened in Giannutri which belongs to the Tyrrhenian clade but also shows signs of admixture with Giglio and some distinctive morphological and phenotypic characteristics typical of the Giglio-Capraia clade.

Conclusions
In conclusion, our results shed light on the origin of the Italian wall lizard populations in the Tuscan Archipelago. The remarkable high morphological and genetic diversity found in the Archipelago is probably the result of at least two different waves of colonization: Elba, and the populations of Cerboli and Montecristo probably originated from historical introductions of lizards belonging to the Tyrrhenian clade, while the populations of Giglio and Capraia islands are the last surviving populations of an ancient lineage that colonized the Tuscan Archipelago during the Pliocene which shares a common ancestry with the Adriatic clade (currently present in south-eastern Italy). Giannutri perhaps represents an interesting case of hybridization between the Tyrrhenian clade and that of Giglio-Capraia. Based on the high phenotypic and molecular distinctiveness of this ancient clade, these populations should be treated as distinct units deserving conservation and management efforts as well as further investigation to assess their taxonomic status.