Coalescent-Based Analyses and Isolation by Distance Tests
The MFDM test indicated that there was no significant probability (P > 0.05) of selection having occurred at each nuclear locus of the two species, respectively, suggesting that there was no unbalanced tree found. Therefore, the five low copy nuclear genes can be used in the IMa analysis. The longest non-recombination matrix for each locus obtained from IMgc analyses were used for IMa simulation. The IMa simulation analysis based on cpDNA and nuclear datasets showed that the effective population size of C. micholitzii was larger than that for C. bifida , and both were larger than that of the ancestral species (Figure 5a). The migrations in both directions were rather low, and the migration from C. bifida to C. micholitzii was higher than in the reverse direction. The effective population migration rates were 2N1m1 = 0.171 from C. bifida to C. micholitzii and 2N2m2 = 0.037 from C. micholitziito C. bifida , respectively (Figure 5b). The marginal posterior densities of divergence parameter, t, showed a peak at about 2.778 (Figure 5c), which converted into a divergence time of c. 5.605 Mya (90% HPD interval: 2.580 - 25.407 Mya), suggesting a late Miocene divergence.
Based on microsatellite data, the correlation between genetic and geographic distances was significant (P = 0.0003, < 0.001) in both species, suggesting that the two species has significant effect of isolation by distance (IBD) (Figure 6c). The IBD test results of single species showed that there was no significant IBD (P = 0.165, > 0.05) in C. bifida (Figure 6a), but significant (P = 0.014, < 0.05) in C. micholitizii (Figure 6b).