Figure 4
Bias in parameter inference at intermediate levels of BGS. Bias in parameter inference for different levels of BGS, redrawn from Ewing and Jensen (2016). Posterior densities from approximate Bayesian computation (ABC) inference for population size are shown. The strength of purifying selection is given as γ, where γ=2Nes. Population A (Pop A) has a true scaled size of 1 (blue line), and population B (Pop B) a true scaled size of 5 (red line). Both population sizes are scaled relative to the size of the ancestral population. As shown, the greatest misinference occurs in the presence of weakly deleterious mutations and subsequent strong BGS effects.
