Incorporation of migration into a simulation framework of complex trait evolution
Past migrations left a distinct print on our genomes, one we have yet to fully understand. A system doubtlessly affected by migration is complex trait evolution. We investigate the effects of migration in a simulated complex trait evolution system, and adds to the relevance of the framework to actual populations. We used simuPOP, a python environment using forward-time and backward-time simulation methods, to run simulations of two diverging subpopulations as they are exposed to differing selective pressures at 50 trait associated loci. Of these 50 loci, 10 were assigned strong positive selective constraints, while the other 40 constraints were weak. Additionally, 2000 neutral markers were evenly distributed across the genome. To prevent loss of derived alleles by chance, allele trajectories were determined through backward time simulation, and used to guide the forward-time simulation of the population. After the 200 generation burn-in period, the population was allowed to grow linearly for 300 generations until the split into two populations with different selective pressures for a quantitative complex trait. Migration began 50 generations after the split. After the 700th generation, a sample of 30 individuals was taken from each population, phenotypes were recorded and several summary statistics were estimated on this sample.
As expected, our results show that migration between the two populations slows down their divergence. The difference in phenotype between the two populations in the four migration treatments was shown to be significant by Student’s t-test. However, high migration rates tended to decrease the phenotypic distance. Additionally, we see a decrease in genetic distance (FST) between the two populations as migration rate increases. This implies that at higher migration probabilities there is not enough difference to count the two groups as distinct populations. Additionally, if migration is not taken into account, it may confound the identification of TAL. Indeed, mean FST for neutral regions under no migration (FST=0.05111) and selected loci under moderate migration (FST=0.06702) were comparable. This brings to light the possibility of missing a potential TAL if migration is not taken into account.