2.3 Phylogenetic analyses
The phylogeny reconstruction based on two databases of haplotypes (ITS and cpDNA sequences) were performed using Bayesian inference (BI) and maximum likelihood (ML). Gaps were treated as missing data during the tree searches. The two programs MrBayes version 3.2(Ronquist et al., 2012) and RAxML v.8(Stamatakis, 2014) were used to conduct BI and ML analyses. The best model for ITS and cpDNA were both GTR+G (BI and ML) inferred by ModelFinder(Kalyaanamoorthy, Minh, Wong, von Haeseler, & Jermiin, 2017) based on the Akaike information criterion (AIC). Markov chain Monte Carlo (MCMC) method was used for BI and set to run 10 million generations, sample every 1,000 generations, discarde 25% of the trees as burn-in. The convergence of MCMC inference was evaluated using Tracer v.1.7.11 to ensure sufficient samples (ESS > 200) by checking the ESS (Effective Sample Size) value. Besides, we ran RAxML onITS and cpDNA with 1000 bootstrap replicates. Finally, BI posterior probabilities (BI PP) and ML bootstrap support (ML BS) were presented at the nodes separately.