2.5 Genetic diversity, population assignment and admixture
The number of alleles and allele frequencies for the selected SNPs were calculated with vcftools 0.1.16 (Petr et al. , 2011). To measure genetic diversity, we estimated expected heterozygosity (He ) and observed heterozygosity (Ho ). We used Arlequin v3.5 (Excoffier & Lischer, 2010) to estimate genetic differentiation by calculating pairwise values of differences among populations (Fst ). To compare molecular diversity between and within populations, we used analysis of molecular variance (AMOVA) and a hierarchical analysis of subdivision (Excoffier, Smouse, & Quattro, 1992; Weir, 1996; Weir & Cockerham, 1984).
We estimated population genetic structure with a Bayesian Markov Chain Monte Carlo model (MCMC) implemented in FastStructure v1.0 (Raj, Stephens, & Pritchard, 2014). We used the default setting with 10-fold cross-validation on the 112 individuals of C. chuniana , testing for subpopulations (K ) ranging from 1 to 11. The python script Choose K in FastStructure was used to choose the optimal K , i.e., the value that maximizes the marginal likelihood. Results were graphically represented and edited with Adobe Illustrator. We performed principal component analysis (PCA) using the PCA function in SNPRelate (Zheng et al., 2012) and visualized the results using the scripts of Tanya Lama (https://github.com/ECOtlama/SNPRelate.git) in the R package.