Abstract (from http://link.springer.com/article/10.1007%2Fs10980-015-0217-1):
Context
Tree species distribution and abundance are affected by forces operating across a hierarchy of ecological scales. Process and species distribution models have been developed emphasizing forces at different scales. Understanding model agreement across hierarchical scales provides perspective on prediction uncertainty and ultimately enables policy makers and managers to make better decisions.
Objective
Our objective was to test the hypothesis that agreement between process and species distribution models varies by hierarchical level. Due to the top-down approach of species distribution models and the bottom-up approach of process models, the most agreement will occur at the mid-level of the hierarchical analysis, the ecological subsection level, capturing the effects of soil variables.
Methods
We compared projections of a species distribution model, Climate Change Tree Atlas, and a process model, LINKAGES 2.2. We conducted a correlation analysis between the models at regional, ecological subsection, and species level hierarchical scales.
Results
Both models had significant positive correlation (ρ = 0.53, P < 0.001) on the regional scale. The majority of the ecological subsections had greater model correlation than on the regional level when all climate scenarios were pooled. Correlation was poorest for the analysis of individual species. Models had the greatest correlation at the regional scale for the GFDL-A1fi scenario (the scenario with the most climate change). Species near their range edge generally had stronger correlation (loblolly pine, northern red oak, black oak).
Conclusion
Our general hypothesis was partly accepted. This suggests that uncertainties are relatively low when interpreting model results at subsection level.