2.4 Phylogenetic analysis and divergence time estimation
To reconstruct phylogenetic relationships among the 11 populations ofC. chuniana , we employed a maximum likelihood (ML) analysis to generate phylogenetic trees using the SNPs extracted based on the GBS dataset. Cercis chingii was used to root the trees, as based on the results of Fritsch & Cruz (2012). Analyses were performed on the high-performance computer cluster available in the CIPRES Science Gateway 3.3 (www.phylo.org; Miller et al. , 2015). The ML analyses were performed simultaneously with 1000 ML bootstrap pseudoreplicates in RAxML v8 (Alexandros, 2014). The model of nucleotide substitution was selected with the Akaike information criteria (AIC; Akaike, 1974) in PhyML-SMS (http://www.atgc-montpellier.fr/phymL/; Lefort, Longueville, & Gascuel, 2017).
Divergence times within Cercis based on the fossil record suggested that C. chuniana originated 2.4 Ma (Fritsch et al. , 2018; Liu et al. , unpublished data, 2020). Therefore, to estimate the divergence time within C. chuniana , we used BEAST v2.4.7 (Bouckaert et al. , 2014) and applied the age of 2.4 Ma as the secondary calibration point with a normal prior distributions and standard deviations of 0.2 Ma, which covered the 95% HPD range. The divergence time analyses were conducted with the GTR + G + I model and four rate categories, a Coalescent Constant Population prior, and the Strict Clock setting with uncorrelated and log-normal-distributed rate variation across the branches. We ran the MCMC simulations in BEAST for 10 million generations with parameters sampled every 1000th generation. We used Tracer v1.6 (Rambaut, Suchard, Xie, & Drummond, 2014) to assess convergence and to check that the effective sample size (ESS) was > 200 for each parameter. We discarded the first 10% of trees as burn-in with the mean node heights option, and then generated the maximum clade credibility (MCC) chronogram from the remaining trees with nodal mean heights and 95% confidence time intervals with TreeAnnotator v2.4.7 (Bouckaert et al. , 2014) in BEAST. The final trees were edited with FigTree v.1.4.3 (http://tree.bio.ed.as.uk/software/figtree/; Rambaut, 2014).