Identification of ancient evolutionary lineages and areas of natural admixture between them can have important implications for practical conservation. The island gecko Tarentola boettgeri is threatened by the introduced California king snake (Lampropeltis californiae) within Gran Canaria, Canary Islands. We sampled T. boettgeri from its natural range: Gran Canaria, El Hierro and the Selvagens archipelago. A 1.9 Gb genome was assembled de novo from PacBio HiFi reads. Sequences obtained by genotyping-by-sequencing (GBS) were aligned to this reference and provided over 440 thousand SNPs in 134 geckos from 40 sample sites. Admixture analyses supported five genomic groups within Gran Canaria and two groups from the other islands. Phylogenomic and divergence time analyses of both GBS sequences and SNPs revealed lineage divergence within Gran Canaria, starting 1.5-2.9 Ma, and also between-island divergence due to colonization of both the Selvagens and El Hierro around 1 Ma. Lineage divergence within Gran Canaria was followed by secondary contact and admixture. Clustering and Bayesian multispecies coalescent analyses allowed estimation of admixture between spatially adjacent lineages. As for hybrid zones in other vertebrates, admixture was limited to a few kilometres around contact zones. This spatial limitation is typical in cases of lower relative fitness of hybrids which, in turn, suggests that the genomic integrity of the five Gran Canarian lineages should be conserved. We therefore argue that our identification and designation of sites/areas with no admixture should form the basis for practical conservation actions, including locations of snake exclusion areas and sampling for ex situ captive breeding.

The evolutionarily significant unit (ESU) is a conservation-oriented concept based on the principle of independently-evolving lineages but introgression may hamper definition of ESUs. Spatial proximity may lead to high levels of introgression and therefore affect identification of ESUs in the gecko Tarentola boettgeri within Gran Canaria, Canary Islands, where it is threatened by the introduced California king snake (Lampropeltis californiae). We sampled T. boettgeri from across Gran Canaria, El Hierro and the Selvagens archipelago. A 1.9 Gb T. boettgeri genome was assembled de novo from PacBio HiFi reads. Genotyping-by-sequencing (GBS) provided genomic sequences which were aligned to this reference genome to provide over 440 thousand SNPs in 134 geckos from 40 sample sites. Admixture analyses supported five genomic groups within Gran Canaria and two additional groups corresponding to the Selvagens and El Hierro. Phylogenomic analyses of both GBS sequences and SNPs revealed patterns consistent with ancient divergence followed by secondary contact and admixture within Gran Canaria. Unlike previous mtDNA estimates, our analyses suggest divergence within Gran Canaria began around the late Pliocene/early Pleistocene and colonization of the Selvagens and El Hierro occurred independently from north-western and western Gran Canarian lineages, respectively, during the Pleistocene. Bayesian multispecies coalescent analyses of Gran Canarian populations supported bidirectional introgression among all geographically adjacent lineages. Nonetheless, predominantly single ancestry populations were identified for all genomic groups within Gran Canaria despite their spatial proximity. We argue that these single ancestry metapopulations, and not introgressed populations, should be recognized as ESUs and be the focus of future efforts to conserve T. boettgeri biodiversity within Gran Canaria.

Morphological divergence under gene flow was investigated in the wall lizard Teira dugesii from the Atlantic island of Madeira island. Lizards (n=334) were sampled using a matched pairs design at four distinct coastal localities. Matched pairs comprised adjacent (<1 km) grey shingle beach and inland sites. Luminances of specific dorsal areas were recorded for each RGB channel from digital photographs taken in the field. Lizards were found to be significantly darker at beach sites than inland sites. Geometric morphometric analyses using 35 landmarks placed on dorsal photographs of the head revealed significant divergence between beach/inland habitats: wider snouts were found at beach sites. Genotyping-by-sequencing of 93 individuals provided 19311 cross-genomic SNPs. A spatial principal components analysis showed significant genomic divergence across the four sampled localities and within these localities. However, there was no evidence that beach and inland populations formed distinct lineages. Patterns of genomic divergence were compared with those generated from simulations under three models. Primary findings were repeated across all four localities. The model of divergence without gene flow was rejected, while the most strongly supported model incorporating two periods of gene flow: an early period of lower gene flow followed by a period of higher gene flow. Gene flow from inland to beach was greater than that in the opposite direction. This study demonstrates ecologically significant morphological divergence in the face of gene flow and adds to understanding of how divergence and speciation may occur within small islands.