Cross-scale integration of knowledge for predicting species ranges: a metamodeling framework

Area of Science:

• Ecology
• Computational Biology
• Environmental Science

Background:

• Species distribution models (SDMs) are crucial for forecasting range shifts.
• Existing SDMs often use limited data and ignore smaller-scale ecological processes.
• Reconciling divergent predictions from different SDM approaches is challenging.

Purpose of the Study:

• To present a flexible framework for integrating multiple SDMs at various scales using hierarchical Bayesian methods.
• To demonstrate the framework's ability to combine diverse data sources and improve uncertainty quantification.
• To provide a more robust approach for ecological forecasting and decision-making.

Main Methods:

• Developed a hierarchical Bayesian framework to build a metamodel constrained by multiple sub-models.
• Applied the framework to a simulated dataset integrating a correlative SDM and a theoretical model.
• Integrated a physiological model with presence-absence data for sugar maple (Acer saccharum) in Eastern North America.

Main Results:

• Integrated models successfully incorporated all data sources, enhancing uncertainty characterization.
• The integrated model for sugar maple outperformed source models in characterizing present-day range uncertainty.
• Future projections showed reduced uncertainty where models agreed and increased uncertainty where they diverged, reflecting a more realistic knowledge state.

Conclusions:

• The proposed framework offers a powerful tool for species distribution modeling, integrating multi-source and multi-scale data.
• It provides accessible methods for applied ecologists, potentially using off-the-shelf software.
• The approach can drive improved ecological decision-making by providing a consensus view from diverse models.