Population differentiation and clustering
After stringent filtering, the average number of SNPs reduced from 52,693 to 3,729 for Carlia (N=135 across four lineages), from 150,178 to 7,472 for Diporiphora (N=147 across eight lineages), from 182,688 to 11,531 for Gehyra (N=214 across ten lineages) and from 185,704 to 10,017 for Heteronotia (N=83; three lineages). Values per lineage, the region of occurrence and habitat specialization are given in Table 1. After partitioning samples into major lineages, mean sample size was 23 individuals and, post-filtering, no lineages had more than 6% of missing data. As expected from prior studies, most species had deeply divergent intraspecific lineages, reflected in largely coincident patterns across the PCo A, fastStructureand either of the spatial or non-spatial ConStruct graphs (see Figure 2 for an example with Carlia amax; data for the other species are available in the Suppl. Mat. Figure S5-S8). The number of discrete lineages identified per species varied from one to three and these supported historically isolated lineages recognized in previous studies, mostly using different datasets (Suppl. Mat. S1). Our subsequent landscape genetic studies focused on variation within these lineages. However, some of these lineages as defined here could still comprise multiple less divergent but historically isolated populations that could confound some landscape genetic analyses but could not confidently resolved with the sampling here. These includeDiporiphora bilineata and D. perplexa (Fenker et al,unpublished data ), Gehyra gemina and G. koira(Oliver et al., 2020), G. nana 4 (Moritz et al., 2018), H. binoei TE (Moritz et al., 2016) and Carlia munda (Potter et al., 2018) (Suppl. Mat. S1).