3.4 | Genetic changes and migration events
We used the δaδi to further explore the demographic history prior to the separation of the three populations. Following the method of Portik (Portik et al., 2017), We employed a four-round optimization technique to ensure that all final optimizations resulted in a similar log-likelihood score (Table S4). In order to better validate the possible divergence patterns between populations, we first constructed ten 3D models for three independent populations (Fig. S7). Models with the lowest scoring log-likelihoods was “Adjacent ancient migration, shorter isolation” (Fig. 5a, Table S5). This ancient migration model involving early gene flow with symmetric migration was supported as the best fit for the three populations. Next, in our eight 2D simulations (Fig. S8), the ancient migration or secondary contact plus instantaneous size change model involving divergence with ancient continuous symmetrical migration, isolation with instantaneous size change provided the best fit regarding the PEc and PEz&PEa lineages (Fig. 5b, Table S5). By comparing the best 2D and 3D models, it was found that the best 2D models have larger log values ​​and lower residuals, which coincided with the results of our phylogenetic tree. Furthermore, the best model revealed a possible divergence of roundworm populations, that is, there was bidirectional gene flow in the early ancient periods, but in the middle and current stages, the gene flow between the populations almost ceased.