Before investigating the change of C. faberi in population size, we estimated the divergence time of haplotypes firstly. About the origin of the Orchidaceae andCymbidium , previous analysis suggested that the ancestor of orchids lived in 76 – 84 million years ago (Ma), and diversification in cymbidium is estimated to have begun 33.9 – 50 Ma (Ramírez, Gravendeel, Singer, Marshall, & Pierce, 2007); other analysis suggested that major diversification of the largest orchid subfamilies perhaps occurring during the cooler period at the end of the Eocene and into the Oligocene, an age for cymbidium clade is 28 or 31Ma (Gustafsson, Verola, & Antonelli, 2010); Givnish et al. (2016) suggested that Orchidaceae have arisen in Australia 112 Ma, and Cymdibiumoriginated about 10 – 20 Ma. Ultimately, 33.9 – 50 Ma was chosen as the originated time of cymbidium .
The divergence time of C. faberi in the Qingling Mountains was estimated using BEAST Version 2.3.2 (Bouckaert et al., 2014); nucleotide substitution model (TVM+I+G) was chosen as the best fit to our data sets by AIC in Modeltest (Posada & Crandall, 1998) combined with PAUP* Version 4.0b10 (Swofford, 2002). Because the substitutions were expected to be constant and the rate of molecular evolution in the plastome is notably high in Orchidaceae, strict clock and a rate of 2.1×10−9 substitutions per site per year were implemented to calibrate the tree (Gaut, 1998). Markov chain Monte Carlo (MCMC) chain was run 100 million generations, then we combined the three independent runs using LogCombiner and check the effective sampling size values by Tracer; the first 25% of trees were discarded as burn-in using TreeAnnotator and the maximum credibility tree was generated by FigTree. Finally, we generated an Extended Bayesian Skyline Plot using BEAST.