Sensitivity and uncertainty in habitat suitability modeling: guidance and pitfalls for ecological applications

Habitat suitability index (HSI) models are common tools for predicting species’ habitat requirements and informing management actions. Application of HSI models is widespread in ecological research and ecosystem restoration, but many HSI models were developed prior to the establishment and use of modern model evaluation methods, and few have been evaluated for uncertainty or sensitivity. We performed global sensitivity and uncertainty analyses on 349 blue book HSI models available in the ecorest R-package. Our results suggest that many HSI models in ecorest are not sensitive to all model inputs. Further, the complexity and structure of HSI models and their parameters influenced both model sensitivity and uncertainty. Our results emphasize the importance of rigorously testing HSI models to ensure that model structure facilitates rather than hinders model performance. Thorough evaluation of uncertainty and sensitivity can help ensure that models produce reliable and informative results to guide conservation and management decisions. • Sensitivity and uncertainty analysis enhances use of habitat suitability models. • Model structure influences model uncertainty and input sensitivity. • Increasing the number of parameters often degrades model performance. • Simple HSI models tend to outperform more complex models.