Population genetics of neutral loci

To assess genetic population structure, we collected 63 tips of tail from geckos belonging to nocturnal/wall (N = 32), diurnal/trunk populations (N = 13) and diurnal no-trunk population (N = 18).

We analysed 8 polymorphic microsatellite loci: 6 (Tb8, Tb35, Tb71, Tb192, Tb213, Tb240) developed from Tarentola boettgeri^68,69 and 2 (Mt6 and Mt27) developed from Tarentola mauritanica⁶². Polymerase chain reaction (PCR) amplifications were carried out in 10 μl final volumes containing 20 ng of genomic DNA, 0.50 μM of each primer, 10X PCR buffer (160 mM (NH4)₂SO₄; 670 mM Tris-HCl pH 8.8; 15 mM MgCl₂; 0.1% Tween 20), 0.2 U Taq polymerase (SIGMA Dream Taq), 0.25 mM each dNTP. The thermocycler profile started with an initial denaturation step at 94 °C for 3 min, followed by 35 cycles at 94 °C for 30 s, temperature annealing at 50–55 °C for 1 min, 72 °C for 1 min followed by 72 °C for 5 min. A negative control was run with each round of PCR.

Polymorphisms of microsatellite were determined using forward primers, end-labelled with a fluorescent dye group (FAM or HEX, MWG Biotech). An internal size standard (LIZ500) was run with our samples. Amplified DNA fragments were electrophoresed using an ABI 3100 automated sequencing instrument (Perkin-Elmer) sequencer, and their genotypes were analysed with GeneMarker Software, version 4.0.

Microsatellite genotypes and sample spatial location data were analysed for the 8 loci in GENELAND package in R⁷⁰. K was set at a minimum of 1 and a maximum of 6. An allele frequency uncorrelated model was set, with 1,000,000 Markov Chain Monte Carlo (MCMC) iterations and thinning of 10, and null allele frequencies were explicitly considered.

To identify the population affinities of individual samples, we used a Bayesian clustering method implemented in STRUCTURE 5^71,72, running using the following parameters: admixture model of ancestry, correlated allele frequency model, a burn-in period of 10,000 simulations followed by a run length of 2,000,000 MCMC simulations. K, ranging from 1 to 5, was tested in three independent runs to establish consistency and the probability of data (Ln) was estimated in each without using any prior population information, so that each individual was assigned to a cluster based upon its multilocus genotype profile.

Molecular variance among individuals within and between phenotype groups was visualized in three-dimensional factorial correspondence analysis implemented in GENETIX 4.05⁷³.

Global population differentiation was estimated using Wright’s Fst⁷⁴, and pairwise Fst values were estimated using θ⁷⁵ in GENETIX 4.05. Fst p-values were computed using a permutation approach (2,000 iterations) as implemented in GENETIX 4.05.