Will Biotic Interactions Ever be Predictable: Insights from Combining Correlational and Process-Based Tree Species Distribution Models

Document Type


Publication Date



: Predicting how climate change and changes to disturbance regimes will influence forest systems requires understanding abiotic constraints to species fitness as well as biotic interactions such as competition and disease. So far, predictions have been largely based on abiotic conditions because biotic interactions are notoriously difficult to describe and predict. We combine four different distribution modeling approaches on a spectrum from correlational to process-based models to elucidate general patterns of biotic interaction strength that may lay the basis for including biotic interactions in predictive models based on future conditions. We use the phenology-based model Phenofit to derive climatic suitability predictions that are closely related to the fundamental niche of 13 eastern US tree species. Differences between Phenofit models, actual tree distributions from the Forest Inventory and Analysis (FIA) data, and predictions from correlation-based species 10 distribution models using soil, landscape, and climate data (DISTRIB) indicate areas of biotic constraints. Further support for the identification of relative biotic interaction strength comes from the Leaf Area Index of the MAPSS model and from constraints-based distribution models. Finally, we relate the identified distribution and characteristics of putative biotic interactions back to climate so that they become predictable under future conditions.


Paper presented at the 27th Annual Landscape Ecology Symposium Informing Decisions in a Changing World Newport, Rhode Island, April 8 – 12, 2012