4.2 Geographical isolation and population divergence associated with Pleistocene climatic oscillations and mountain ranges
All populations are monophyletic and closely aligned with geographical regions (Figure 1, 3), suggesting that they evolved mostly via local diversification. This is thought to occur especially when geographic isolation plays a dominant role (Harrington, Hollingsworth, Higham, & Reeder, 2018; Hughes, 2017; Hughes & Atchison, 2015; Kadereit, 2017; Nevado, Contreras-Ortiz, Hughes, & Filatov, 2018; Xing & Ree, 2017). Analysis of the molecular variance with significantly high population divergence (Fst = 0.99, P = 0.00) also indicates low inter-population gene flow (Table 2). Mountain ranges sometimes are considered as poorly conducive for facilitating long-distance dispersal, thus contributing to limited gene flow and geographic isolation (Oyama et al., 2018). In our study, isolation between WYS1 and WYS2 (V1) was attributed to the Wuyi Mts. acting as geographical barrier to separate the populations from each other (Figures 1, 3). The rising of Wuyi Mts. during the early Pleistocene has been reported to cause geographical isolation and genetic divergence for many species in southern China (Liu, 1984; Yan et al., 2013). Notably, the central Luoxiao Mts., with a north-south orientation, are assumed to serve as a geographic barrier particularly for east-west colonization. This appears to apply to LXS1 and LXS2 in the Luoxiao Mts., which are currently isolated from each of their eastern or western populations (V2 and V3) (Figures 1, 3). We infer that the geographical isolation between the populations of the Nanling Mts. and those to the east (V4) has arisen through the lack of geographical corridors with a west-to-east orientation. Vicariance events also exist between the western and eastern (V5) as well as the middle and northwest (V6) populations within the Nanling Mts. The Nanling Mts. present a general north-south orientation, which we infer as disadvantageous for east-west colonization, thus contributing to vicariance involving V5. In contrast with the populations NLW2~5, NLW1 is located alone on one ridge of the Nanling Mts. and geographically distant from the remaining populations, thus resulting in the vicariance involving V6. Therefore, the geographical barriers formed by the associated mountain ranges including the Wuyi, Luoxiao, and Nanling Mts. have directly limited long-distance colonization and are considered a major factor contributing to the historical isolation of C. chuniana populations (Jiang, Xu, & Deng, 2019; Li et al. , 2019; Yang et al. , 2019). Similar patterns have been found in many other plant species with a wide distribution range in subtropical China, such as Machilus pauhoi(Zhu et al. , 2017), Loropetalum chinense (Gong et al. , 2016) and Liriodendron chinense (Shen, Cheng, Li, & Li, 2019), all of which were analyzed with traditional molecular markers and data analysis.
Our study suggests that population divergence of C. chunianaoccurred in the Pleistocene and has been affected by the glacial cycles. These cycles periodically changed suitable habitat and are thought to have promoted range contraction and expansion coupled with geographic isolation (Knowles, 2010; Qu et al. , 2011). Based on Bayesian estimation, the time of divergence (0.68 Ma) between the populations in the Nanling Mts. and those of the east coincides with the last third glacial period in China in the Middle Pleistocene (Figure 4). The time may fall in the Naynayxungla Glacial period (0.5–0.7 Ma; Zheng, Xu, & Shen, 2002; Zhou & Li, 1998), Dagu Glacial period (0.5–0.6 Ma; Duan, Pu, & Wu, 1980), Kunlun Glacial period (0.62–0.78 Ma; Zhao, Shi, & Wang, 2011), or Poyang-Dagu Interglacial (Duan, Pu, & Wu, 1980). Although the time for the glaciations is uncertain, primarily the third (last) glaciation drove the genetic divergence between populations in the Nanling Mts. and those to the east, and shaped the geographical patterns of genetic variation. The estimated divergence time of the best fit model in FSC2 is older, i.e. 1.60 Ma (Figure 5), which overlaps with the earliest known Quaternary glacial of the Xixiabangma Glacial period ca. 1.6 Ma (Wan et al., 2016), or the Sizishan Periglacial period (Duan, Pu, & Wu, 1980), when the temperature was 10℃ lower than at present. The discrepancy between the results of Bayesian time estimation and FSC2 may be partially attributed to the wider time range under the log-uniform setting in FSC2. The secondary calibration used in BEAST is thought to generate smaller time estimates (Foster et al., 2017; Kong, 2017; Kong, Zhang, Hong, & Barker, 2017). The climate during glacial periods tended to be dry and cool, which would favor the populations shifting to lower elevations with contracted distribution ranges. The glacial period in the Middle Pleistocene have been shown to drive spruce fir forests to lowlands in northern China (Liu, 1988). In our study, the geographical distribution of C. chuniana in southern China is also associated with the Pleistocene glacial cycles (Figure 4). The Dagu Glacial period primarily affected the population divergence between the east and west, whereas the Dagu-Lushan Interglacial period and Lushan Glacial period primarily affected population diversification. The dominant role for Pleistocene glacial cycles on the geographic distribution of populations is also apparent in ENM, where several isolated glacial refugia were identified during the LGM, although the climatic conditions may not be analogous to that of other glacial cycles.