Constraint, pathology, and adaptation: how can we tell them apart?

Adaptation is a central concept of modern evolutionary biology, but remains a difficult one nevertheless. Definitions of adaptation are often confounded with definitions of natural selection, rendering them somewhat circular and difficult to operationalize. Williams introduced a definition that avoids such tautology and a strategy for testing adaptive claims against chance as an alternative explanation for design complexity. Gould and Lewontin ([1979]: Proc R Soc Lond B Biol Sci 205:581-598) challenged this strategy for pitting adaptation against a straw alternative, and argued that constraint is often the cause of design complexity. The field of Darwinian medicine has underscored the fact that adaptation can also be difficult to discriminate from pathology, which can also produce design complexity. We suggest that an updated version of Williams' strategy is to consider any claim of adaptation against constraint and pathology as alternatives. We use an example drawn from the intersection of human reproductive ecology and developmental biology to illustrate how this updated strategy can be applied. Where we can generate distinct predictions for the three alternative hypotheses, constraint, pathology, and adaptation, we have a better situation in which to evaluate adaptive claims with a real possibility of falsification. We view this strategy as an improvement over Williams' original suggestion, but not as a definitive strategy. Further advances, however, will likely also be based on a sound understanding of the concept of adaptation and the identification of the strongest competing alternatives to it. Copyright (c) 2007 Wiley-Liss, Inc.