Amplification of expected selection differentials in small populations. Fitness distributions for two phenotypes; the size of the dot indicates the probability of that fitness value. In (A) and (B), Individuals with phenotypic value 0 leave either 0 or 1 descendant with equal probability, and those with phenotype 1 leave either 1 or 2 descendants with equal probability. In (A), there is one individual with each phenotype. The lines show the four possible (and equally probable) combinations of fitness values, with the corresponding Δ. The average change is E(Δϕ) = . (B): In an infinite population evenly divided between the two phenotypes, the total contribution of individuals with phenotype ϕ = 1 will always be 3 times greater than the total contribution of individuals with ϕ = 0, yielding , which is the prediction of classical theory. (C): Another example of fitness distributions leading to directional selection. Numbers adjacent to dots are probabilities. (D): Results of monte-carlo simulations using the fitness distributions in (C). The dashed line is the value for N = ∞.