Testing the adaptive significance of acclimation: a strong inference approach

Abstract

Acclimation is a common phenotypic response to environmental change. Acclimation is often thought to enhance performance and thus to be adaptive. This view has recently been formalized as the "Beneficial Acclimation Hypothesis" and predicts that individuals acclimated to one environment perform better in that environment than do individuals acclimated to a different environment. Although Beneficial Acclimation is appealing and widely supported, recent studies with E. coll and Drosophila have challenged its general validity. Although these challenges could be dismissed as mere exceptions, they encourage a re-evaluation of the adaptive significance of acclimation. Our philosophical approach differs from that of most previous studies of acclimation, in which the prediction derived from a Beneficial Acclimation perspective (e.g., heat tolerance is positively correlated with acclimation temperature) is tested against the null hypothesis ("single hypothesis approach"). Instead, we follow Huey and Berrigan (1996) in advocating a strong inference approach (sensu Platt, 1964), which recognizes that Beneficial Acclimation is actually one of a set of competing hypotheses that make different predictions as to how developmental temperature influences the thermal sensitivity of adults ("developmental acclimation"). Using this perspective, Huey and Berrigan proposed a factorial experimental design (3 developmental by 3 adult temperatures) designed to discriminate among all competing hypotheses. We now derive a formal statistical model (ANOVA with orthogonal polynomial contrasts) for this experimental design and use it to evaluate simultaneously the relative impact of each competing hypothesis. We then apply this model to several case studies (Drosophila, Volvox, Trichogramma), and we review also a recent study with E. coll. The influence of Beneficial Acclimation is supported (albeit often weakly) in most cases. Nevertheless, other hypotheses (especially the Optimal Developmental Temperature Hypothesis) often have a greater impact. Even so, however, Beneficial Acclimation usually predicts relative performance at extreme test temperatures. We conclude that, although rumors of its death are premature, Beneficial Acclimation cannot be viewed as the dominant expectation, at least with regard to developmental temperature acclimation. Moreover, our findings reinforce the view that a strong inference approach provides a more comprehensive portrait of complex biological responses than do single-hypothesis approaches.