Adaptive phenotypic plasticity may improve the response of individuals when faced with new environmental conditions. Typically, empirical evidence for plasticity is based on phenotypic reaction norms obtained in reciprocal transplant experiments. In such experiments, individuals from their native environment are transplanted into a different environment, and a number of trait values, potentially implicated in individuals’ response to the new environment, are measured. However, the interpretations of reaction norms may differ depending on the nature of the assessed traits, which may not be known beforehand. For example, for traits that contribute to local adaptation, adaptive plasticity implies non-zero slopes of reaction norms. By contrast, for traits that are correlated to fitness, high tolerance to different environments (possibly due to adaptive plasticity in traits that contribute to adaptation) may, instead, result in flat reaction norms. Here we investigate reaction norms for adaptive versus fitness-correlated traits, and how they may affect the conclusions regarding the contribution of plasticity. To this end, we first simulate range expansion along an environmental gradient where plasticity evolves to different values locally and then perform reciprocal transplant experiments in silico. We show that reaction norms alone cannot inform us whether the assessed trait exhibits locally adaptive, maladaptive, neutral or no plasticity, without any additional knowledge of the traits assessed and species’ biology. We use the insights from the model to analyse and interpret empirical data from reciprocal transplant experiments involving the marine isopod Idotea balthica sampled from two geographical locations with different salinities, concluding that the low-salinity population likely has reduced adaptive plasticity relative to the high-salinity population. Overall, we conclude that, when interpreting results from reciprocal transplant experiments, it is necessary to consider whether traits assessed are locally adaptive with respect to the environmental variable accounted for in the experiments, or correlated to fitness.

Matlab