Matching habitat choice and plasticity contribute to phenotype–environment covariation in a stream salamander

Populations optimize the match of phenotype to environment by localized natural selection, adaptive phenotypic plasticity, and habitat choice. Habitat choice may also be achieved by several mechanisms, including matching habitat choice, where individuals distribute themselves based on self-assessment of the phenotype–environment match. Matching habitat choice is a relatively untested concept, but one that could advance our understanding of the interplay of movement ecology and intraspecific phenotypic variation. Morphology of the salamander Gyrinophilus porphyriticus differs in riffles and pools, the dominant habitats in headwater streams where this species occurs. Specifically, individuals found in riffles have shorter limbs than those found in pools. Here, we used 4 yr of spatially explicit capture–mark–recapture data from three streams to test the contributions of phenotypic plasticity and matching habitat choice to this phenotype–environment covariation. We quantified morphological variation in G. porphyriticus with size-corrected principal component (PC) scores and assessed phenotype–environment match based on the difference between habitats in these PC scores. We found that both phenotypic plasticity and matching habitat choice contribute to phenotype–environment covariation in G. porphyriticus. The phenotypes of individuals that switched habitats (i.e., riffle→pool, pool→riffle) changed to become better matched to the recipient habitat, indicating a plastic response to local habitat conditions. Consistent with matching habitat choice, individuals were also more likely to switch habitats if their initial phenotype was a better match to the alternative habitat, independent of subsequent changes in morphology due to plasticity. Realized performance, survival adjusted for the likelihood of remaining in each habitat, was higher in individuals with phenotypes matched to each habitat than in those with mismatched phenotypes, but performance was generally lower in riffles than pools, suggesting that other factors influence the use of riffles. Our results underscore the value of considering how matching habitat choice interacts with other mechanisms that allow organisms to maximize performance when faced with environmental heterogeneity. More broadly, our study shows that it is important to account for movement in any study of the causes or consequences of intraspecific trait variation, a challenge that may require novel research approaches and experimental designs.