three groups: one group consisting of four populations (DL, YT, RC and LZ), one group containing another four populations (SS, LJ, QG and NJD) and with the population of GY isolated. As shown in Fig. 3, after adding the YL population, the ten populations were divided into two groups: YL, formed a separate group, distinguished from the other groups. Bootstrap support values for the NJ tree were not high for most groups. The tree topology was not highly compatible with the geographic locations of sample sites. Based on the STRUCTURE software, the greatest division between groups appears when K ¼ 2 (value of DK is greatest). YL was classified with northern populations (shown in Fig. S1).
The AMOVA analysis based on groups determined by the Neighbor-joining tree (YL as a group and the other as another group) showed that variation between individuals within populations accounted for most of the variation (70.0%) and identified significant variation among groups, explaining 14.6% of the total variance (Table S2). The variance component among populations within groups was also significant (15.4%), indicating that populations within groups should not be pooled together due to their genetic heterogeneity.