Often landmark conservation decisions are made despite an incomplete knowledge of system behavior and inexact predictions of how complex ecosystems will respond to management actions. For example, predicting the feasibility and likely effects of restoring top-level carnivores such as the gray wolf (Canis lupus) to North American wilderness areas is hampered by incomplete knowledge of the predator-prey system processes and properties. In such cases, global sensitivity measures, such as Sobol? indices, allow one to quantify the effect of these uncertainties on model predictions. Sobol? indices are calculated by decomposing the variance in model predictions (due to parameter uncertainty) into main effects of model parameters and their [...]
