The contribution of ancestry, chance, and past and ongoing selection to adaptive evolution.

The relative contributions of ancestry, chance, and past and ongoing selection to variation in one adaptive (larval feeding rate) and one seemingly nonadaptive (pupation height) trait were determined in populations of Drosophila melanogaster adapting to either low or high larval densities in the laboratory. Larval feeding rates increased rapidly in response to high density, and the effects of ancestry, past selection and chance were ameliorated by ongoing selection within 15-20 generations. Similarly, in populations previously kept at high larval density, and then switched to low larval density, the decline of larval feeding rate to ancestral levels was rapid (15-20 generations) and complete, providing support for a previously stated hypothesis regarding the costs of faster feeding in Drosophila larvae. Variation among individuals was the major contributor to variation in pupation height, a trait that would superficially appear to be nonadaptive in the environmental context of the populations used in this study because it did not diverge between sets of populations kept at low versus high larval density for many generations. However, the degree of divergence among populations (F(ST)) for pupation height was significantly less than expected for a selectively neutral trait, and we integrate results from previous studies to suggest that the variation for pupation height among populations is constrained by stabilizing selection, with a flat, plateau-like fitness function that, consequently, allows for substantial phenotypic variation within populations. Our results support the view that the genetic imprints of history (ancestry and past selection) in outbreeding sexual populations are typically likely to be transient in the face of ongoing selection and recombination. The results also illustrate the heuristic point that different forms of selection-for example directional versus stabilizing selection-acting on a trait in different populations may often not be due to differently shaped fitness functions, but rather due to differences in how the fitness function maps onto the actual distribution of phenotypes in a given population. We discuss these results in the light of previous work on reverse evolution, and the role of ancestry, chance, and past and ongoing selection in adaptive evolution.