Bottleneck event revealed by simulations
We employed the approximate Bayesian computation (ABC) method to test 12 possible models (Figure 6). By simulations, model 8 acquired the highest posterior probability, which assumed that the “s-SCS” populations shared a common ancestor with “Australasia”, then “s-SCS” diverged with “Australasia”, accompanied by a severe bottleneck, and recently expanded to its current population size (Figure 6). Migrations between the “Indo-Malayan” and “s-SCS” or between “Australasia” and “s-SCS” have occurred in recent times. This model choice is robust, with a posterior probability larger than 0.6 in all three repeated computations (Table 3).
Under the optimal demographic history model, we estimated the posterior distribution of the demographic parameters. The “Indo-Malayan” lineage and the ancestor of the “Australasia” and “s-SCS” populations (the “Pan-Australasia” cluster) diverged approximately 2.7 million years ago (95% confidence interval, i.e. CI: 2.6~3.0). Next, the “s-SCS” lineage diverged from the “Australasia” populations approximately 1.5 million years ago (95% CI, 1.2~1.9). N0 was estimated as 1916. Accordingly, the population size of the ancestral “s-SCS” group was estimated to have reduced from 10538 to 824 (5.5 N0 to 0.43 N0) in the bottleneck event. Since approximately 0.13 million years ago (95% CI 0.1~0.2), theNe of the “s-SCS” group has increased gradually to 1917 (95% CI, 1812~1986). TheNe values of the “Indo-Malayan” and “Australasia” groups were relatively constant, at 9580 and 10538, respectively. Gene flow was asymmetric, that the values are 0.258 per generation (95% CI, 0.118~0.688) from the “Indo-Malayan” to the “s-SCS” populations and 0.007 (95% CI, 0.002~0.021) in the reciprocal direction. Similarly, the gene flow was 0.403 (95% CI, 0.216~0.775) per generation from the “Australasia” to the “s-SCS” populations, and the reciprocal flow was 0.005 (95% CI, 0~0.075) per generation (Table 4).