Abstract. Grasslands provide many important ecosystem services globally and forecasting grassland productivity in the coming decades will provide valuable information to land managers. Productivity models can be well-calibrated at local scales, but generally have some maximum spatial extent in which they perform well. Here we evaluate a grassland productivity model to find the optimal spatial extent for parameterization, and thus for subsequently applying it in future forecasts for North America. We also evaluated the model on new vegetation types to ascertain its potential generality. We find the model most suitable when incorporating only grasslands, as opposed to also including agriculture and shrublands, and only in the Great Plains and Eastern Temperate Forest ecoregions of North America. The model was not well suited to grasslands in North American Deserts or Northwest Forest ecoregions. It also performed poorly in agriculture vegetation, likely due to management activities, and shrubland vegetation, likely because the model lacks representation of deep water pools. This work allows us to perform long-term forecasts in areas where model performance has been verified, with gaps filled in by future modelling efforts.
Multi-scale drivers of soil resistance predict vulnerability of seasonally wet meadows to trampling by pack stock animals in the Sierra Nevada, USA
