A Cautionary Tale of Caterpillars and Selectional Interference

Evolutionary theory yields many important insights into why organisms are the way they are. However, for any given problem, there is significant uncertainty about what level of abstraction is appropriate. With such issues in mind, I will discuss an evolutionary scenario concerning the size and development time of a particular species of caterpillar, the tobacco hornworm. A larva of this species grows approximately exponentially; once it reaches a critical size various hormones result in a time delay before pupation during which it continues to grow. Thus, the relationship between genotype and phenotype is non-trivial. In laboratory experiments by Nijhout and in computer simulations, one can selectively breed caterpillars based on size and development time and investigate the resulting evolutionary dynamics. Both laboratory experiments and computer simulations yield striking complexity. Selecting for size and time simultaneously yields unexpected interference, resulting in caterpillars that do not necessarily have the selected-for properties. This interference is due to constraints on the set of possible phenotypes imposed by the caterpillar’s fundamental development process. The important lesson here is that when building evolutionary models of even simple phenomena, one must be distrustful of intuition, and consider (1) how the details of the development process of individual organisms affect the ability of a population to explore the space of possible phenotypes, and (2) how selection criteria in combination can interact in unexpected ways.