Trait-environment relationships are predictive, but not general across species

Understanding the relationships between organisms and their environments is increasingly important given human impacts on global conditions. However, predicting how community diversity and composition will change in the future remains challenging (Mouquet et al 2015). One recent approach is to use traits to mechanistically inform how environmental conditions affect performance (i.e., trait-environment relationships), under the assumptions that these measures relate to each other in predictive and general ways. Unfortunately, results have been inconsistent, ignore phenotypic plasticity, and rely heavily on observational data (Shipley et al 2016). We evaluated the predictability and generality of trait-environment relationships in a controlled experimental microcosm system of four daphniid species. We cultured each species along a stressful gradient (conspecific density), measuring performance (fecundity) and traits related to performance (body length, 2nd antenna length, eye diameter, relative growth rate, and age at first reproduction). Using structural equation models, we evaluated the role of traits in mediating changes in individual fecundity in response to conspecific density. We built models for each species separately considering within-species trait variation, and for all species together by considering all trait variation across the four species. Results from this controlled system highlight that the relationship between individual traits and the environment (conspecific density) is strong and predictive of performance (fecundity), both within- and across-species. However, the specific trait-environment relationships which predicted fecundity differed for each species and differed from the relationships observed in the interspecific model, suggesting a lack of generality. These results will inform and improve the use of traits as a tool for predicting how changing environments will impact species abundances and distributions.