The Evolution of Conspicuous Coloration

Abstract

I introduce three scenarios for the evolution of conspicuous coloration associated with unpalatability in prey animals, based on the ways in which selection for the two character types (unpalatability and coloration) may interact. I consider only the simple two-character-state cases. I use these scenarios as a framework for an analysis of two issues in the evolution of warning coloration. First, I discuss the generality of warning color as an explanation of the coincidence between conspicuous coloration and unpalatability. I present an alternative explanation that does not involve aposematic coloration: conspicuously colored animals may be unpalatable because they are so vulnerable to predation. I caution against using the warning-color explanation without care, and I outline two instances in which the non-aposematic explanation is at least as plausible. These can involve evolution under scenario 1 (prior evolution of conspicuousness followed by unpalatability) or scenario 3 (prior evolution of unpalatability followed by conspicuousness). Second, I analyze the selective conditions favoring the evolution of warning coloration. Five main points arise from this analysis. (1) Under the conditions of scenario 3 (the evolution of warning coloration in an already unpalatable species), classical kin selection does not provide an accurate description of the selective conditions involved. Although it can be useful to think of warning coloration as having an altruistic element, the altruism is blind to kinship relations. The recipients of altruism are not relatives of the altruist in the normal sense, but individuals with similar phenotypes. "Green-beard selection" therefore provides a more accurate description. In this case, the green-beard trait is also not cheatable. (2) Under scenario 2 (the joint evolution of unpalatability and warning coloration), kin selection may be involved because cheating is possible. This will be the case if predators can use proximity cues in discriminating between cheats and non-cheats. Green-beard selection will also be involved to the extent that predators generalize by coloration. (3) In both scenarios 2 and 3, a second and distinct effect of family grouping may be important because of its effects on the initial viability of traits having density-dependent benefit and a threshold for favorability. This effect, the aggregation of similar phenotypes, has been confused with kin selection in the literature. I explain why they are distinct. (4) Because a green-beard trait such as warning coloration recognizes only individuals with similar phenotypes, the presence of Mullerian mimics has the same effect on the trait's spread as that exerted by members of the same species. Aggregating with Mullerian mimics could take the place of family grouping as a means to overcome the initial viability problem. (5) I liken Batesian mimics to cross-species cheats. Just as kin selection acts as a check against cheats within a species, family grouping in a model species may allow predators to use proximity cues in discrimination, thereby reducing the effectiveness of Batesian mimicry.