Gene flow typically hinders population divergence by homogenizing allele frequencies. However, studies on sticklebacks, periwinkle snails, and stick insects have shown divergence and speciation can occur with gene flow, often driven by strong divergent selection across environments. While neutral loci might be homogenized, selected loci can accumulate divergence. Small oceanic islands offer excellent models for studying divergence and speciation, particularly in lizards occupying diverse environments. However, clear examples of within-island adaptive divergence with ongoing gene flow are rare; most studies attribute divergence to historical or current interruptions in gene flow. Understanding within-island divergence is crucial for explaining island community development. Adaptive responses to microhabitats and historical gene flow interruptions likely contribute to the existence of ecomorphs. Population-level studies of species exhibiting incipient divergence within a single island can provide insights into whether isolation is a prerequisite for within-island evolution. Coalescent-based methods using genomic data are particularly useful for analyzing historical and current gene flow. This study aimed to test for morphological and color divergence in *Teira dugesii* between paired adjacent habitats and to determine the degree of gene flow between them. The researchers hypothesized that divergent selection between habitats was sufficient to overcome gene flow, resulting in parallel divergence patterns across multiple locations.

Several studies have demonstrated that divergence and speciation can occur in the presence of gene flow, particularly when strong divergent selection acts across different environments. Examples include sticklebacks, periwinkle snails, and stick insects, where adaptation to distinct habitats has driven divergence even with ongoing gene flow. Small oceanic islands provide ideal settings for studying this phenomenon, offering opportunities to examine divergence and gene flow between habitats within a limited geographical area. However, most previous research on island lizards has emphasized the role of historical or current interruptions to gene flow in driving population divergence and speciation, often citing volcanic events as isolating mechanisms. The current study aimed to contribute to the understanding of within-island divergence by examining a species showing incipient divergence within a single island, using genomic data to infer gene flow patterns and assess the role of divergent selection.

The study used a matched pairs sampling design focusing on the Madeiran wall lizard (*Teira dugesii*). Four localities with adjacent beach (B) and inland (I) habitats were selected. Lizards were trapped at each site, photographed for color and morphology analysis, and tail tips were collected for genomic analysis. **Lizard Dorsal Luminance:** Dorsal photographs were taken against a standardized background. Luminance values were extracted from six dorsal/head areas across three RGB channels using Micatoolbox within ImageJ. Log10-transformed RGB luminance data were analyzed using two-way MANOVA and DFA to assess habitat and locality effects. **Lizard Morphology:** Head measurements were taken from field photographs. Head size (centroid size) was analyzed using two-way ANOVA. Head shape was analyzed using geometric morphometrics (landmarks, Procrustes analysis, PCA, DFA) to assess variation between habitats and localities. **Habitat Variation:** Substrate RGB values and vegetation cover were recorded from photographs at each site and analyzed using two-way MANOVA and DFA to quantify habitat differences. **Genomic Data:** Genotyping-by-sequencing (GBS) was performed on DNA extracted from tail tips. SNPs were called and filtered. Outlier SNPs potentially under selection were identified using pcadapt, and their association with habitat was tested using bayenv2. Spatial structuring was assessed using spatial PCA (sPCA). Treemix was used to examine the hypothesis of separate lineages for beach and inland populations. Fastsimcoal2 was employed to compare three gene flow models (no gene flow, constant gene flow, and two periods of gene flow) using joint folded site frequency spectra (SFS).

Beach (*Teira dugesii*) individuals were significantly darker (lower luminance) than inland individuals across all four localities, with consistent patterns observed for both sexes. Geometric morphometric analyses revealed significant divergence in head morphology, with beach lizards having generally wider snouts, a pattern replicated across sexes and localities. Genotyping-by-sequencing (GBS) analysis rejected the hypothesis that beach populations represent a distinct evolutionary lineage. Bayesian analyses strongly supported models incorporating gene flow between beach and inland populations at all localities. The favoured model indicated an initial divergence followed by two distinct periods of gene flow, with higher recent gene flow rates compared to ancient rates. Gene flow was generally asymmetric, with higher rates from inland to beach habitats. The analyses revealed consistent patterns of beach-inland divergence in both morphology and color, despite substantial gene flow, supporting the hypothesis that divergent selection is sufficient to overcome gene flow in this system. The initial timing of divergence showed some variability between localities.

This study demonstrates morphological divergence in *Teira dugesii* between beach and inland habitats separated by less than 1 km, a pattern replicated across four geographically distinct locations. The consistent direction of morphological change (darker coloration and wider snouts in beach lizards) strongly supports the hypothesis of divergent selection overcoming gene flow. The detection of ongoing gene flow at all localities, with higher recent gene flow than historical flow, suggests a dynamic interplay between selection and gene flow. The asymmetric gene flow, with greater migration from inland to beach habitats, aligns with expectations for a peripheral habitat (beach) receiving migrants from a larger metapopulation (inland). The increase in gene flow over time may be related to changes in coastal topography or sea level, affecting habitat connectivity. The findings highlight that environmental differences between habitats can drive substantial morphological divergence within islands, even in the face of high gene flow. While the genetic basis of these morphological traits requires further investigation, the replicated patterns across localities support the hypothesis that divergent selection, rather than drift, is the primary driver of divergence.

This study provides compelling evidence that within-island divergence can originate solely from habitat differences without requiring interruptions to gene flow. The Madeiran wall lizard (*Teira dugesii*) exhibits significant morphological divergence between beach and inland habitats, despite high gene flow. This divergence is driven by divergent selection, highlighting the powerful influence of environmental pressures on phenotypic evolution even in highly connected populations. Future research should focus on identifying the specific genes underlying the observed morphological differences and exploring the mechanisms of divergent selection in more detail.

While the study provides strong evidence for divergent selection in the face of gene flow, some limitations exist. The exact genetic basis of the observed morphological differences remains to be elucidated. Further investigation is needed to determine whether the same loci are responsible for divergence across all localities or whether parallel evolution has occurred. Moreover, the study focuses on a specific lizard species and habitat type; the generalizability of these findings to other species and environments requires further research.