• Genetic algorithms;
  • Mexican birds;
  • niche evolution;
  • predictive distributional models


The development of quantitative models of species’ distributions has largely ignored the potential for intraspecific variation in species’ niche requirements. Application of such models may nevertheless provide a rich, untapped opportunity to address the basic issue of niche conservatism vs. evolution. We illustrate this potential using genetic algorithms coupled with geographical information systems, which provide a powerful and novel approach to characterizing species’ ecological niches and geographical distributions. Our example consists of several species of Mexican birds with recognized subspecies, and associated climatic and vegetation data. Our basic protocol is to develop an ecological niche model for each subspecies, and use this model to predict distributions of other subspecies. In some cases, the ecological niche model inferred for one subspecies provides an excellent descriptor of other subspecies’ ranges, whereas in other cases the prediction is rather poor. We suggest that the latter may reveal the potential existence of evolved, intraspecific niche differentiation. We discuss alternative, non-evolutionary explanations, and point out potential implications of our results for predictive models of species’ invasions.