Skip to main content

Multi-model comparison on the effects of climate change on tree species in the Eastern U.S.: results from an enhanced niche model and process-based ecosystem and landscape models


Louis Iverson

Frank Thompson

S. Matthews

A Prasad

William Dijak

Jacob Fraser

Wen Wang

Brice Hanberry

Hong He

Maria Janowiak

Patricia Butler

Leslie Brandt

C Swanston

+8 more
Publication Type:
Journal Article
Year of Publication:
Secondary Title:
Landscape Ecology


Species distribution models (SDM) establish statistical relationships between the current distribution of species and key attributes whereas process-based models simulate ecosystem and tree species dynamics based on representations of physical and biological processes. TreeAtlas, which uses DISTRIB SDM, and Linkages and LANDIS PRO, process-based ecosystem and landscape models, respectively, were used concurrently on four regional climate change assessments in the eastern Unites States. We compared predictions for 30 species from TreeAtlas, Linkages, and LANDIS PRO, using two climate change scenarios on four regions, to derive a more robust assessment of species change in response to climate change. We calculated the ratio of future importance or biomass to current for each species, then compared agreement among models by species, region, and climate scenario using change classes, an ordinal agreement score, spearman rank correlations, and model averaged change ratios. Comparisons indicated high agreement for many species, especially northern species modeled to lose habitat. TreeAtlas and Linkages agreed the most but each also agreed with many species outputs from LANDIS PRO, particularly when succession within LANDIS PRO was simulated to 2300. A geographic analysis showed that a simple difference (in latitude degrees) of the weighted mean center of a species distribution versus the geographic center of the region of interest provides an initial estimate for the species' potential to gain, lose, or remain stable under climate change. This analysis of multiple models provides a useful approach to compare among disparate models and a more consistent interpretation of the future for use in vulnerability assessments and adaptation planning.