Simultaneous Inductive and Deductive Modeling of Ecological Systems via Evolutionary Computation and Information Theory

Abstract

The addition of two emerging technologies (evolutionary computation and ecoinformatics) to computational ecology can advance our ability to build better ecological models and thus deepen our understanding of the mechanistic complexity of ecological systems. This article describes one feasible approach toward this goal–the combining of inductive and deductive modeling techniques with the optimizing power of simple algorithms of Darwinian evolution that include information-theoretic model selection methods. Specifically, the author shows a way to extend classic genetic algorithms beyond typical parameter fitting of a single, previously chosen model to a more flexible technique that can work with a suite of possible models. Inclusion of the Akaike information-theoretic model selection method within an evolutionary algorithm makes it possible to accomplish simultaneous parameter fitting and parsimonious model selection.Experiments with synthetic data show the feasibility of this approach, and experiments with time-series field data of the zebra mussel invasion of Lake Champlain (United States) result in a model of the invasion dynamics that is consistent with the known hydrodynamic features of the lake and the motile life history stage of this invasive species.The author also describes a way to extend this approach with a modified genetic programming algorithm.