An urgent challenge facing our discipline is predicting the impact of environmental change on many species at large spatial scales. Ecologists frequently argue that we must move beyond phenomenological models and base predictions on ecological and evolutionary processes. But which processes are essential? Population biologists believe that useful forecasts are impossible without accounting for demography. Evolutionary biologists argue that models must consider genetic variation and adaptation. Community ecologists focus on species interactions. Functional ecologists argue that physiology is essential. Must our models include all of these processes? How can we balance mechanistic detail against the simplicity that often improves the performance of predictive models?
Results/Conclusions
Although many processes are important in principle, determining which ones most improve predictions is ultimately an empirical question: we must make predictions with different models and test them against independent data. The key challenge is integrating information about different processes in a unified statistical framework. Reaction norms, which describe the change in a trait or performance across environments, may be one framework to help integrate these processes. Genetic variation can be linked to physiology, which can then be linked to reaction norms for individual vital rates, which together determine population growth. We are now pursuing this approach in a new project focused on understanding how climate change will alter the trajectory of the cheatgrass (Bromus tectorum) invasion across the Intermountain West.