A comparative test of the adaptive plasticity hypothesis: relationships between habitat and phenotype in anuran larvae.

The hypothesis that phenotypic plasticity is maintained by divergent natural selection acting across different environments predicts that populations and species exposed to highly variable environments will express high levels of plasticity. I tested this prediction by measuring the behavioral and morphological responses to aeshnid dragonfly larvae of 16 tadpole species and asking whether predator-induced plasticity is greater in species that experience more variable densities of predators in nature. Tadpole phenotypes were measured in a series of similar experiments in outdoor artificial ponds carried out over a 9-yr period. I quantified tadpole habitats by soliciting evaluations by seven to 36 experienced field observers for each species. There were large differences among species in phenotype, mostly in agreement with earlier descriptions. Nearly all species responded to dragonflies by decreasing activity and body length relative to overall body size and by increasing relative tail fin depth, although the magnitude of the responses differed among species. There was a significant positive phylogenetic correlation between morphological plasticity and variability in exposure to predators, thus upholding the adaptive hypothesis. The correlation between behavioral responses and habitat variability was not significant, and there was little relationship between behavioral and morphological plasticity, raising the possibility that behavioral responses evolve under different scales of environmental variation than morphological responses.