Predicting end-of-season timing across diverse North American grasslands.

Abstract

Climate change is altering the timing of seasonal vegetation cycles (phenology), with cascading consequences on larger ecosystem processes. Therefore, understanding the drivers of vegetation phenology is critical to predicting ecological impacts of climate change. While numerous phenology models exist to predict the timing of the start of the growing season (SOS), there are fewer end-of-season (EOS) models, and most perform poorly in grasslands, since they were made for forests. Our objective was to develop an improved EOS grassland phenology model. We used repeat digital imagery from the PhenoCam Network to extract EOS dates for 44 diverse North American grassland sites (212 site-years) that we fit to 20 new and 3 existing EOS models. All new EOS models (RMSE = 22-33 days between observed and predicted dates) performed substantially better than existing ones (RMSE = 43-46 days). The top model predicted EOS after surpassing a threshold of either accumulated cold temperatures or dryness, but only after a certain number of days following SOS. Including SOS date improved all model fits, indicating a strong correlation between start- and end-of-season timing. Model performance was further improved by independently optimizing parameters for six distinct climate regions (RMSE = 4-19 days). While the best model varied slightly by region, most included similar drivers as the top all-sites model. Thus, across diverse grassland sites, EOS is influenced by both weather (temperature, moisture) and SOS timing. Incorporating these new EOS models into Earth System Models should improve predictions of grassland dynamics and associated ecosystem processes.