Location: Range Management ResearchTitle: Multi-scale assessment of a grassland productivity model
Submitted to: Biogeosciences
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 2/11/2021
Publication Date: 4/1/2021
Citation: Taylor, S.D., Browning, D.M. 2021. Multi-scale assessment of a grassland productivity model. Biogeosciences Journal 18(6):2213-2220. https://doi.org/10.5194/bg-18-2213-2021.
Interpretive Summary: Grasslands in North America provide multiple ecosystem services and drive production for the majority of grain, beef, and other staples. Grassland productivity is strongly affected by rainfall, temperature, and management, and forecasting the effects of climate change on grassland productivity requires consideration of all these aspects. Numerical models incorporating all available information will be central in producing a long-term grassland forecast, but require independent validation to ensure the resulting forecasts are sound. We evaluated a grass forecast model using nearly 500 years of grassland camera data and found the areas where the model worked well, as well as locations where it did not. Long-term grassland forecasts for the suitable locations can be made immediately with the current model, while other areas (for example, Southwestern U.S. grasslands and grazing lands) will need further model development. The PhenoGrass model integrates camera data that offer an “on the ground” perspective that offers ease of interpretation and confidence in productivity forecasts over other products generated from broad-scale satellite remote sensing. Improved forecasts for grassland productivity at the pasture-scale will benefit producers of all types - farmers, ranchers, and land managers – to make better decisions for resource management and economic viability.
Technical 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.