An Introspection on the Art of Modeling in Population Ecology

W hat exactly is science? This question has received considerable attention from contemporary philosophers of science, notably Popper (1968) on the central role of falsifiability in scientific theory, Kuhn (1962) on the structure of scientific revolutions, Hempel (1965, 1966) on the testability of scientific hypotheses and concept formation through operationalism, Levins (1966) and Wimsatt (1981) on the necessity for the robustness of particular theorems, and Lakatos (1976, 1978) on the progression of scientific research programs. Others have added to the debate by questioning the inherent lack of objectivity (Atkinson 1985) or the central role heuristics play (Wimsatt 1981) in the enterprise called science. In disciplines such as fundamental physics, in which mathematical models are constructed using operationally defined concepts and variables that can be measured with great accuracy (e.g., Newton's model for the gravitational attraction between bodies possessing mass), the enterprise produces testable hypotheses and theories with explanatory relevance (i.e., the phenomenon is expected to occur when specific conditions, described by the theory, are met)-two factors that are minimally required for a scientific theory (Hempel 1966). Population ecology as a scientific enterprise, however, is peppered with hypotheses that are not testable and theories that lack or have, at best, weak explanatory relevance. Two examples are the competitive exclusion principle and food web theory. The competitive exclusion principle is not testable because its formulation is confounded with the definition of a niche: If the test fails, one can recover the principle by redefining the niches for the populations