2.6 Molecular dating of chlorotypes and population historical dynamics
Divergence dating of the chlorotypes in a Bayesian framework was performed in BEAST v.1.7.5 (Drummond et al. , 2012). Owing to the absence of fossil records of Gesneriaceae with known ages, we usedP. pteropoda (with an estimated age of c. 1.31 Myr; Gao et al. , 2015; Genbank accession number KF498219) to calibrate a molecular clock, on the grounds that it has the closest phylogenetic relationship with P. heterotricha (Gao et al. , 2015). Four independent Markov Chain Monte Carlo (MCMC) runs were analyzed for 80 million generations and sampled every 1000 steps with the nucleotide substitution model GTR+I+G, as selected by MRMODELTEST 2.3 (Nylander, 2004) based on the Akaike Information Criteria (AIC). The Yule speciation tree priors and the relaxed clock model with an uncorrelated lognormal distribution were followed. The resulting trees from the four runs were combined by LOGCOMBINER (Bouckaert et al. , 2014), discarding the first 25% of generations as burn-in. We specified a normal distribution with standard deviation of one million years for all the time priors. TRACER 1.5 (Rambaut & Drummond, 2007) was used to access the chain convergence and ensuring ESS values was well above 200. The maximum clade credibility (MCC) tree was summed up using TREEANNPTATOR in the BEAST package (Bouckaert et al. , 2014). The averages and PP of age estimates were visualized by FIGTREE v.1.4.2 (Rambaut, 2009).
The probable ancestral biogeographical scenarios in the phylogeny ofP. heterotricha chlorotypes were reconstructed separately by a Statistical Dispersal-Vicariance Analysis (S-DIVA) by using the S-DIVA (Yu et al. , 2010), which is parsimony-based and favours vicariance events without any prior assumptions about area relation (Yu et al. , 2010). We chose to be cautious in our definition of regions, and arranged three areas based on the geographic barriers and contiguities: (1) NW-Northwest of Changhua River, including the populations BW, YJ, EX, WX and YG; (2) SE-Southeast of Changhua River, including the populations WZ, QX, XA and QL; (3) SW-Southwest of Changhua River, consisting only of the population JF. The chlorotype trees for S-DIVA analysis were generated in BEAST v.1.7.5, including the outgroup P. pteropoda (see above).
In order to detect possible recent range expansions, Tajimaʼs D(Tajima, 1989) and Fuʼs F s (Fu, 1997) were calculated for testing the deviations from the null hypothesis of constant population size and neutral evolution for each DNA fragment, pairwise mismatch distribution and neutrality tests for all populations and clades (NW, SE and SW) were conducted in DNASP v. 6.12.01 based on nrITS and cpDNA separately. These tests show a unimodal shape in the mismatch distribution of populations that experienced historical expansion.