La niche de certains modélisateurs : paramètres d’un monde réel ou d’un univers fictif ?

Hutchinson’s (1958) multi-dimensional niche concept is at the root of most current efforts and attempts to model the dynamics of communities. The main architects of this influential school of thought in modern ecology are MacArthur, Levins, and more recently May, Roughgarden, and Vandermeer, among authors critically reviewed in this paper, and others like Schoener, Cody, and Levin among authors whose work is cited but not reviewed here. Interspecific competition is the single most important phenomenon to affect and mold the structure of community according to ecologists who make such models at present. The region and amount of spatial overlap between competitors along one or more of the dimensions of the niche (the so-called resource axes) are the key indicator of competition. The models in use today are derived from those of pioneers like Lotka, Volterra and Kostitzin who worked between 1920 and 1940. The modeling technique and the scientific world view of modern modelers differ very little from those of the classic theoreticians. What does differ, however, is the assertion of present-day modelers that competition is the motor of community evolution. The model-making effort is not orientated toward testing this assertion, but instead toward building more models based on this “ fact”. Consequently, the work of modelers consists in defining and specifying the “sufficient parameters” of the theory of the niche, that theory being a mathematical and graphical description of how a few species coexisting in a “community” or in a “guild” subdivide their resources. According to Levins, the sufficient parameters are niche breadth, niche dimension, niche overlap, and community matrix. Two major elements of niche theory are the form of the resource utilisation function, and the definition of the competition coefficients (α’s). A considerable effort is made today to estimate these coefficients and to calculate the community matrix in natural or quasi-experimental situations, in the apparent and naive goal of directly transposing models into nature. These field measurements are compared to predictions derived from the theory of the niche, but these comparisons are only of very limited value because of the intrinsic difficulties of taking the measurements. Also, it is not possible to apply the criteria of mathematical stability required by the models to natural systems. Other problems involve the lack of precision in the currently used definition of several concepts (niche breadth ; niche dimension ; environmental grain ; resource ; environmental variance). However the most significant problem seems to be the dogmatism implicit throughout the modern theory of the niche. This theory rests on the key role of competition in community structure. But we should not forget that this role is still only a possibility, even though the reality of competition itself cannot be doubted.