Incorporating uncertainty in predictive species distribution modelling.

Area of Science:

• Ecology
• Environmental Science
• Conservation Biology

Background:

• Species distribution models (SDMs) are increasingly used to address ecological challenges like climate change and biological invasions.
• SDM predictions inform critical conservation and management decisions, including policy, invasive species control, and disease management.
• A significant challenge is the unrecognized degree and source of uncertainty inherent in SDM predictions.

Purpose of the Study:

• To review the species distribution model literature concerning uncertainty.
• To identify sources of uncertainty across three main classes of SDMs: niche-based, demographic, and process-based models.
• To discuss methods for minimizing or incorporating uncertainty to provide realistic confidence measures for predictions.

Main Methods:

• Literature review of species distribution models (SDMs).
• Focus on niche-based, demographic, and process-based model classes.
• Analysis of uncertainty sources and quantification methods.

Main Results:

• Uncertainty in SDMs is often underestimated, leading to a false sense of precision in geographical distribution predictions.
• Specific sources of uncertainty were identified for niche-based, demographic, and process-based models.
• Current methods often fail to adequately incorporate or quantify prediction uncertainty.

Conclusions:

• There is a critical need for improved statistical tools, such as hierarchical models, to better assess and integrate uncertainty across different SDM types and spatial scales.
• Developing more robust methods for evaluating predictive performance, model fit, and covariate significance is essential.
• Accurate quantification of uncertainty is vital for reliable ecological predictions and informed conservation strategies.