Continental scale variability in ecosystem processes: Models, data, and the role of disturbance

termination of the vulnerability of ecosystems to large-scale changes in climate or atmospheric chemistry require understanding how ecosystem processes are governed at large spatial scares. A collaborative project, the Vegetation and Ecosystem Modeling and Analysis Project (VEMAP), addressed modeling of multiple resource limitation at the scale of the conterminous United States, and the responses of ecosystems to environmental change. In this paper, we evaluate the model- generated patterns of spatial variability within and between ecosystems using Century, TEM, and Biome-BGC, and the relationships between modeled water balance, nutrients, and carbon dynamics. We present evaluations of models against mapped and site-specific data. In this analysis, we compare model-generated patterns of variability in net primary productivity (NPP) and soil organic carbon (SOC) to, respectively, a satellite proxy and mapped SOC from the VEMAP soils database (derived from USDA-NRCS [Natural Resources Conservation Service] information) and also compare modeled results to site-specific data from forests and grasslands. The VEMAP models simulated spatial variability in ecosystem processes in substantially different ways, reflecting the models' differing implementations of multiple resource limitation of NPP. The models had substantially higher correlations across vegetation types compared to within vegetation types. All three models showed correlation among water use, nitrogen availability, and primary production, indicating that water and nutrient limitations of NPP were equilibrated with each other at steady state. This model result may explain a number of seemingly contradictory observations and provides a series of testable predictions. The VEMAP ecosystem models were implicitly or explicitly sensitive to disturbance in their simulation of NPP and carbon storage. Knowledge of the effects of disturbance (human and natural) and spatial data describing disturbance regimes are needed for spatial modeling of ecosystems. Improved consideration of disturbance is a key ''next step'' for spatial ecosystem models