Leaf Shedding During Drought Reduces Hydraulic Stress in Trees

Abstract Leaf area has long been a proxy for ecosystem function. However, it can be highly variable even in the same forest types across space and time due to variations in local ecohydrology and climatic extremes such as droughts and heatwaves. Leaf shedding in response to drought has been documented at site‐scales, theoretically to avoid hydraulic damage. Yet it is a major unknown if such leaf area declines are adaptive, in that they minimize the impacts of water limitation, or are simply diagnostic of declining ecosystem function. Here we use a trait based, hydraulically‐enabled, tree model that adaptively adjusts leaf area annually to maximize tree fitness in response to changes in water availability. We generate predictions of annual leaf area with a focus on the worst drought that occurred for points across the continental United States during the 20‐year analysis period. We compared model predictions to interannual variations in remotely sensed leaf area index (LAI). We found that a majority of ecosystems reduced LAI during drought and that the model predicted the LAI anomaly as well or better than Standardized Precipitation‐Evapotranspiration Index, a commonly used drought index. Leveraging the mechanistic insights of the model, we found that reduced leaf area during drought in order to maximize carbon gain led to an overall reduction in hydraulic stress, but with a wide range of amplitudes across climates and forest ecosystem types. These results illustrate that crown area reductions during droughts are widespread in water limited regions, and likely adaptive in nature.