3.2 Population Grouping
Based on SAMOVA, we found that the F CT value was the highest when K = 3, which resulted in separation of populations into three groups. Among them, both the first group (P1-6) and the second group (P7-12) contained six populations, and the third group consisted of all other populations. However, three populations (P6-8) were highly similar in their geographic distributions and their habitats according to our field surveys, and this was why we did not adopt K = 4. When K was > 4, at least one group consisted of a single population (Table S3), indicating over-splitting and loss of geographic structural data. Thus, we thought it was the most reasonable to divide 43 populations of P. villosainto two major groups.
The first group, Group 1, identified using SAMOVA, consisted of 12 populations (P1-12), which mainly occurred in the central and eastern regions of the Inner Mongolian Plateau, while the second one, Group 2, was composed all other populations, which distributed throughout the range of the species in China. Notably, the populations of Group 1 tend to be found at lower elevations compared to those from Group 2 (Table S1 & Figure S1). Groupings of populations can facilitate detection of finer scale geographic structures (i.e., within groups) juxtaposed with broader, regional patterns (i.e., between groups) (Li et al., 2020). Therefore, we used the recovered grouping scheme for downstream data analysis in our present study.