Population genetic structure of highlights the necessity and feasibility of hierarchical analysis for a highly differentiated species.

Population genetic structure can provide valuable insights for conserving genetic resources and understanding population evolution, but it is often underestimated when using the most popular method and software, STRUCTURE and delta K, to assess. Although the hierarchical STRUCTURE analysis has been proposed early to overcome the above potential problems, this method was just utilized in a few studies and its reliability needs to be further tested. In this study, the genetic structure of populations of was evaluated by sequencing 12 nuclear microsatellite loci of 905 individuals from 38 populations. The STRUCTURE analysis suggested the most likely number of clusters was two, but using multi-hierarchical structure analysis, almost every population was determined with an endemic genetic component. The latter result is consistent with the extremely low gene flow among populations and a large number of unique cpDNA haplotypes in this species, indicating one level of structure analysis would extremely underestimate its genetic component. The simulation analysis shows the number of populations and the genetic dispersion among populations are two key factors to affect the estimation of K value using the above tools. When the number of populations is more than a certain amount, K always is equal to 2, and when a simulation only includes few populations, the underestimation of K value also may occur only if these populations consist of two main types of significantly differentiated genetic components. Our results strongly support that the hierarchical STRUCTURE analysis is necessary and practicable for the species with lots of subdivisions.