Group Selection, Altruism, and Structured-Deme Models

Abstract

Group selection is defined as a process by which traits advantageous to the group are favored because of the positive association of individuals exhibiting the traits. In addition, group selection acts to protect this positive association against cheats. This definition, unlike those in current use, incorporates the essential features of the traditional verbal arguments by excluding the effects of individual selection and incorporating the problem of cheating. Kin selection is considered an example of group selection in which the "groups" are associations of relatives and in which special mechanisms, such as individual recognition, maintain the integrity of the associations. The "groups" of group selection are quite different from the trait groups of structured-deme models, so that structured-deme models can be used to demonstrate individual or group selection. By analyzing the models in terms of group neighbors, it has been shown that whenever trait groups are formed at random only individual selection can act. The same results have emerged when there are no groups, only a continuous random array of individuals, and the analysis is in terms of neighborhoods. Individuals exhibiting traits that can evolve under these conditions of neighborhood selection often impart some benefit to their neighbors. Such traits are said to be benevolent. Altruism involves a loss of individual fitness and can evolve only by group selection. The evolution of altruism can be demonstrated in structured-deme models, provided that the altruists exhibit positive association and can occur whether the population forms isolated groups or a continuous array. Structured-deme models are a valuable tool for analyzing local interactions and the resulting neighborhood selection; it is important to note, however, that if a model incorporates isolated trait groups, then within-group comparisons are entirely inappropriate for evaluating the fate of genotypes. Comparing genotypes under conditions of equivalent neighborhoods not only gives a direct indication of relative fitnesses, but also provides the unexpected bonus of making the analysis technically much easier.