Figure 3.

Mean ML estimates for Δ obtained from simulated data. In all cases, the effective population size is N_e = 10⁶. On the left, it is shown that the mean ML estimates of Δ are independent of the level of sequence coverage per site; 200 replicate runs were made per condition, with 10⁵–10⁶ independent pairs of sites sampled in each case. The horizontal lines denote the expected values given by Equation 6. On the right, for 10× coverage, it is shown that mean estimates of Δ are asymptotically unbiased with increasing numbers of sites; as on the left, varying combinations of the per-site parameters θ and ρ underlie the individual points.