Fig. 2.

Best-fitting parameter values under the MSci model of figure 1d when data of two sequences per locus (one per species), each of n sites, are generated under the IM model of figure 1a. Five methods (a-e) are used to fit the MSci model, estimating 2, 3, 4, 5, and 4 parameters, respectively, whereas the other parameters are fixed. In (a), ${\tau }_R$ and $\phi$ are estimated, but ${\theta }_R$ and ${\theta }_S$ are fixed at their true values in the IM model, and the introgression time ${\tau }_S={\tau }_H$ is fixed at ${\tau }_T=0$. In (b), ${\tau }_S$ is estimated as well. In (c), ${\tau }_S=0$ is fixed, whereas the other four parameters are estimated. In (d), all five parameters are estimated. In (e), the constraint ${\theta }_R={\theta }_S$ is enforced so that four free parameters are estimated. The dotted lines for $\phi$ indicate the true total amount of introgression of equation (10). The dashed lines indicate ${\phi }^{*}$ of equation (11). The true and best-fitting distributions of the coalescent time (t) are shown in supplementary figure S1, Supplementary Material online.