4.2. A phylogeographic scenario of P. lilfordicolonization of the Balearic Islands
P. lilfordi genetic diversity was largely structured according to the main geographic distribution, with all analyses supporting the existence of two major genetic lineages separating the large archipelagos of Menorca from Mallorca/Cabrera (Figure 5 and 6). The tree topology was largely consistent with the ones previously reported based on mtDNA and SNPs data (Bassitta et al., 2021; Brown et al., 2008; Pérez‐Cembranos et al., 2020; Terrasa et al., 2009). While Mallorca/Cabrera lineage is further sub-structured in four well-supported clades (bs>80%, Figure 6), phylogenetic relationships among Menorcan populations remain largely unresolved, as also indicated by their low degree of differentiation (Figure 4, Fst values < 0.1). Menorcan islands were once largely panmictic (Brown et al., 2008; Terrasa et al., 2009) and a signature of past admixture can still be observed among most populations of this archipelago (Figure 5B), with the exclusion of the small islet of Porros, now representing an independent lineage (Figure 6).
Admixture analysis (K=2, Figure 5B), root inference by IQ-tree (Figure S2) and root testing all strongly suggest that the origin of the species occurred in the main island of Mallorca, with most ancestral populations in our dataset being identified in the North of Mallorca. The highest rootstrap support (86.13%) specifically points to the small island of Colomer as the population retaining nowadays most ancestral polymorphisms, in line with previous proposals (Bassitta et al., 2021; Terrasa et al., 2009). This root placement is consistent with the geographic history of both Dragonera and Colomer islands which were once connected to the Serra of Tramuntana, a mountain chain running up North of Mallorca, before their detachment from the main Mallorca Island about 2.3 Mya (Brown et al., 2008; Terrasa et al., 2009). The Serra of Tramuntana is considered to be the last refuge of P. lilfordi(Bailón, 2004) before its extinction in the main Mallorca Island, as testified by fossil records dating around 2000 years old found in the Muleta cave, in the central part of the Tramuntana (Alcover, 2000; Kotsakis, 1981). The Colomer islet is characterized by high lizard density (Pérez-Mellado et al., 2008) and limited accessibility (due to high altitude and coastal shape). Although introduction of lizards from other islands cannot be ruled out, the islet has reduced chances of external gene flow, compatible with the retention of ancestral polymorphisms (Bassitta et al., 2021; Terrasa et al., 2009).
The following process of colonization within the archipelago of Mallorca/Cabrera did not follow an isolation by distance model (Mantel test, p>0.1). According to admixture, phylogenetic analyses and Treemix reconstructions of past migration events, North of Mallorca was then the source of colonization of main Cabrera Island, with which they form a unique genetic clade (see Admixture results, Figure 5B). This closed genetic relationship was also reported by Bassitta et al., 2021, although it partly differs from mtDNA results (Terrasa et al., 2009).
Following this scenario, species expansion within the archipelago of Cabrera proceeded with the colonization of small surrounding islets (Esclatasang and Foradada), as supported by the observed coancestry with Cabrera Island (see Figure 5B). This suggests a founder effect in these small islands, followed by independent evolution leading to the nowadays well-differentiated genetic clade of Esclatasang and Foradada (Figure 5 and 6). Similarly, admixture and ML tree analyses indicated that islets on North of Cabrera could have been a source of colonization for Southern Mallorca islets, specifically the islet of Moltona, with which they still share a small proportion of coancestry (Figure 5B and 6), while the geographically closed sister population of Na Guardis has currently lost all signature of coancestry with Northern Cabrera (Figure 5B). The observed clustering of Cabrera with South Mallorca islets confirms previous haplotype grouping based on mtDNA data, which also suggested a directionality of gene flow from Cabrera to South Mallorca (Network III; Terrasa et al., 2009). Moltona would have then seeded the small islet of en Curt, with which they form a unique clade (bs=99%, Figure 6), as evidenced by their shared coancestry (Figure 5B). The tiny islet of en Curt (0,44 ha) would have followed a process of accelerated differentiation presumably driven by a founder effect, extensive genetic drift, and intense density-dependent selection with more than 1500 ind/ha (Ruiz De Infante Anton et al., 2013). As an alternative scenario, the three islets on the South of Mallorca directly derived from the ancestral mainland Mallorca population, now extinct, a scenario that we presently cannot exclude.
According to bathymetric, geological and genetic data subsequent colonization of Menorca occurred around ~ 2.8 Mya (Brown et al., 2008; Terrasa et al., 2009). We were not able to assess the root within this archipelago due to high levels of admixture. Menorcan populations were highly panmictic during a large period of time and the colonization and subsequent isolation of most Menorcan islands was gradual (Pretus et al., 2004), following a vicariance type of colonization, with an asymmetric flow from south to north of the distribution of the species (Terrasa et al., 2009).