Complexity and Verisimilitude: Realism for Ecology

Abstract

When data are limited, simple models of complex ecological systems tend to wind up closer to the truth than more complex models of the same systems. This greater proximity to the truth, or "verisimilitude," leads to greater predictive success. When more data are available, the advantage of simplicity decreases, and more complex models may gain the upper hand. In ecology, holistic models are usually simpler than reductionistic models. Thus, when data are limited, holistic models have an advantage over reductionistic models, with respect to verisimilitude and predictive success. I illustrate these points with models designed to explain and predict the numbers of species on islands.[PUBLICATION ABSTRACT]