Self-organization and competition in evolution: a conceptual problem in the use of fitness landscapes.

Recently it has been claimed that certain macroevolutionary patterns, such as the contrast between the Cambrian explosion and the Permian quiescence, can be explained as generic properties of selection processes, i.e. as features that are not caused by specific selection pressures, but that are to be expected in any selection process. The explanations are based on models of motions on randomly generated fitness landscapes, which are assumed to be representative of fitness landscapes in general. In this paper I show that such explanations of macroevolutionary patterns are conceptually flawed. If the concept of fitness used in the model is not the customary biological notion of fitness, there is no reason why organisms with higher fitness should displace organisms with lower fitness, and as a result the motion of a point representing a population is no longer determined by the shape of the fitness landscape. Nothing can be derived about the motions on the landscape, and the explanation collapses. If the model is based on the customary fitness concept, non-competing species have to be assigned the same fitnesses. As most species belonging to a radiation such as the Cambrian explosion are assumed to coexist, the majority of species will have the same fitness. A radiation of species can therefore not be modeled meaningfully on a fitness landscape.