Benchmarking historical CMIP5 plant functional types across the Upper Midwest and Northeastern United States

Centennial-scale climate-ecosystem feedbacks are a major source of predictive uncertainty for land-atmosphere fluxes of energy, carbon, and water. Accurate representations of plant functional type (PFT) distributions through time and space are required for modeling centennial-scale feedbacks within Earth system models (ESMs). We tested the ability of ESMs from the Coupled Model Intercomparison Project Phase 5 (CMIP5) to capture historical PFT distributions at the time of Euro-American settlement in the Northeastern United States against a new subcontinental-scale data set of historical tree abundances derived from forest composition surveys. To identify and diagnose errors in ESM-simulated PFT distributions and quantify impacts on modeled albedo, net primary productivity, and transpiration, we analyzed actual and modeled PFT distributions with respect to historical mean annual climate and modeled elasticity among PFTs, climate, and vegetation-atmosphere fluxes. Historical PFT distributions were poorly matched between ESMs and the settlement-era data, often due to inaccurate PFT-climate relationships within ESMs, particularly for evergreen trees. Some models exhibited large local, but regionally compensating, errors in simulated albedo, net primary productivity, and transpiration due to inaccurate PFT distributions, while others had systematic regional biases in vegetation-atmosphere fluxes. Internal variable elasticity varied among ESMs, and these differences closely corresponded to model skill in predicting PFT distributions. New historical benchmarks like the settlement-era vegetation data provide opportunities to confront ESMs, parse sources of error, and improve simulations of historical and future vegetation-atmosphere feedbacks.