Figure 2.

Alternative distribution of beneficial fitness effects as predicted by extreme-value theory (EVT). All distributions displayed here are specific cases of a generalized Pareto distribution⁹ that differ by their shape. A shape of 1 corresponds to an exponential distribution (gray), a shape parameter >1 will yield a DFE with a much heavier tail (so-called Frechet domain, in blue). Alternatively, a shape parameter <1 will yield distributions that decay much more rapidly than the exponential (so-called Weibull domain), and the beneficial mutations cannot exceed a maximum beneficial effect.