ARTIFICIAL SELECTION ON A FITNESS‐COMPONENT IN DROSOPHILA MELANOGASTER

Recently there have been a number of studies concerning the quantitative genetics of components of fitness (e.g., Istock et al., 1976; Derr, 1980; review by Istock, 1983). This constitutes a return to an experimental approach which was once widespread (e.g., Robertson, 1957; review by Falconer, 1960), but fell from favor with the adoption of electrophoretic and related techniques in population genetics (Lewontin, 1974). Unfortunately, frustration has greeted most attempts to relate electrophoretic genotypes to fitness or fitness-component phenotypes, with only a few exceptions (e.g., Richmond et al., 1980). Taking a different approach, evolutionary ecology has endeavoured to dispense with genetic information in treating fitness-components such as fecundity and survivorship, usually referred to as life-history (Stearns, 1977). But this research strategy also has attendant shortcomings, especially in the difficulty of inferring genetic constraints on evolution from phenotypic correlations (cf. Charlesworth, 1980; Stearns, 1980; Rose and Charlesworth, 1981a). The present return to quantitative genetics research on fitness-components can to some extent be seen as an attempt to find a middle ground on which genetic variation and fitness-phenotype variation can be related to one another. But this new return to the quantitative genetics approach has by no means been entirely free of difficulties. Perhaps the most puzzling development is the apparent ubiquity, though not universality,