PREDICTING MICROEVOLUTIONARY RESPONSES TO DIRECTIONAL SELECTION ON HERITABLE VARIATION

Microevolution of quantitative traits in the wild can be predicted fro m a knowledge of selection and genetic parameters. Testing the predict ions requires measurement of the offspring of the selected group, a re quirement that is difficult to meet. We present the results of a study of Darwin's finches on the Galapagos island of Daphne Major where thi s requirement is met. The study demonstrates microevolutionary consequ ences of natural selection. The population of medium ground finches, G eospiza fortis, experienced size-selective mortality during a drought in 1976-1977; large birds with deep beaks survived better than small b irds. During another drought, 1984-1986, the population experienced se lection in the opposite direction on beak traits. Changes in food supp ly were the apparent causes of selection on beak traits in both episod es. As expected from the high heritabilities of all measured traits, t he effects of selection were transmitted to the next generation. Evolu tionary responses to both episodes of selection were quantitatively we ll predicted in general. This allows us to conclude that, to a first a pproximation, targets of selection were identified correctly, and gene tic parameters were correctly estimated. Nevertheless, not all respons es of individual traits were equally well predicted. A search for poss ible reasons for the largest discrepancies revealed evidence of bias c aused by environmental effects on growth and adult size of some traits , as well as possible selection on the offspring generation before the ir measurement. These findings illustrate an important assumption in t he study of microevolution: that the environments experienced during g rowth to maturity by the parental and offspring generations are the sa me, for otherwise a measured difference between generations may have a partly environmental cause, thereby giving a misleading estimate of t he evolutionary response to selection. Simple extrapolations from obse rved selection to long-term evolution may underestimate the total forc e of selection involved if it oscillates in direction or acts on the e nvironmental variance.