2.8. Coalescent simulation

To test if the discordance between plastome tree and nuclear trees could be explained by ILS alone, we conducted coalescent simulations following previous studies (García et al., 2017; Morales-Briones et al., 2018; Zhou et al., 2022). The ASTRAL-III tree was chosen as a guide tree for the gene tree simulation using DENDROPY v3.12.1 (Sukumaran and Holder, 2010). To simulate plastid gene trees, branch lengths of the ASTRAL-III tree were scaled by 12 to account for organellar inheritance as paleotropical woody bamboos are hexaploidy, and the effective population size of the plastome is generally expected to be one-twelfth that of the nuclear genome given the assumptions of equal sex ratios, haploidy (homoplasmic), and uniparental inheritance (McCauley, 1994; Stull et al., 2020). Finally, 1,000 gene trees were simulated under the coalescent model. The clade frequencies of simulated trees were summarized on the plastome tree inferred using RAxML with PHYPARTS (Smith et al., 2015). In a scenario of ILS alone, the clades in the empirical plastome tree should be present in the simulated trees with high frequency. If hybridization exists, the clades of the empirical plastome trees should be absent or at very low frequency in the simulated trees (García et al., 2017; Morales-Briones et al., 2018; Zhou et al., 2022).