Progress in the detection of costs of phenotypic plasticity in plants.

Among the most interesting biological phenomena is the fact that a genotype can develop different phenotypes depending on the environment in which this development takes place. Historically, however, the focus was on homeostasis by canalization of the phenotype to a presumed optimum (Waddington, 1960), and phenotypic plasticity was interpreted as deviation ftom such an optimum and therefore considered a nuisance. This has changed drastically with the insight that many plastic responses, such as stem elongation in response to shading, are actually adaptive strategies that increase fitness. Surprisingly, however, not all organisms are highly plastic, which suggests that the evolution of phenotypic plasticity is constrained either by a lack of heritable genetic variation or by limits and costs of plasticity, which outweigh its potential benefits (DeWitt et al., 1998; van Kleunen & Fischer, 2005). In this context modelling studies (van Tienderen, 1991) emphasized the role of costs of plasticity. It is therefore very astonishing that empirical studies have found little evidence for the existence of such costs (van Kleunen & Fischer, 2005). Possibly, costs of plastici ty are difficult to detect because genotypes burdened by high costs of plasticity have been purged from natural populations by natural selection (DeWitt et aI., 1998), and may only re-emerge after recombination (Fig. O. This motivated Dechaine et al. (this issue, pp. 874-882) and two further recent studies (Callahan et ai, 2005; Weinig et aI., 2006) to test costs of plasticity with recombinant inbred lines rather than with natural plant genotypes.