Natural Selection in a Population of Darwin's Finches

Abstract

An exceptionally wet year followed by two very dry years severely perturbed the food supply of cactus finches (Geospiza conirostris) on Isla Genovesa, Galapagos. This provided us with a natural experiment for detecting and interpreting selection acting on bill shape in response to the near elimination of some of the food items and availability of others. Long-billed birds had a survival disadvantage associated with the absence of Opuntia flowers and fruits, which they normally exploit. This selection event was followed by another favoring birds with deep beaks capable of extracting the only available food, arthropods from beneath the bark of trees and within Opuntia trunks and old pads. For young birds, the ability to develop (learn) techniques for extracting a new food item was a better predictor of survival than was bill shape. Among the group that acquired these skills, more of the birds with deep bills survived, as was the case for the adults. During 1984, males that obtained females had deeper bills than those that failed to do so. Because this mating advantage did not occur in years without natural selection, we suggest that birds with deep bills had more energy available for courtship. Using the same approach, we also describe a parallel case of natural selection in a sympatric congener, G. magnirostris, in which birds with deep beaks capable of cracking large and hard Cordia lutea seeds were favored. The population has undergone little or no evolutionary change in bill dimensions in response to selection during the last nine years, despite the presence of high levels of additive genetic variance. Evolutionary change has not occurred because the action of selection on one trait in one direction has been counteracted by selection in the opposite direction on another, positively correlated trait. Changes in the proportions of dry-season niches among years cause changes in the proportions of birds with certain beak morphologies and feeding skills best suited to exploiting them. Therefore, the population tracks a moving peak in an adaptive landscape under environmental fluctuations, and there is more than one individual fitness optimum within the range of phenotypes in the population.