Regional grassland productivity responses to precipitation during multiyear above‐ and below‐average rainfall periods

AbstractThere is considerable uncertainty in the magnitude and direction of changes in precipitation associated with climate change, and ecosystem responses are also uncertain. Multiyear periods of above‐ and below‐average rainfall may foretell consequences of changes in rainfall regime. We compiled long‐term aboveground net primary productivity (ANPP) and precipitation (PPT) data for eight North American grasslands, and quantified relationships between ANPP and PPT at each site, and in 1–3 year periods of above‐ and below‐average rainfall for mesic, semiarid cool, and semiarid warm grassland types. Our objective was to improve understanding of ANPP dynamics associated with changing climatic conditions by contrasting PPT–ANPP relationships in above‐ and below‐average PPT years to those that occurred during sequences of multiple above‐ and below‐average years. We found differences in PPT–ANPP relationships in above‐ and below‐average years compared to long‐term site averages, and variation in ANPP not explained by PPT totals that likely are attributed to legacy effects. The correlation between ANPP and current‐ and prior‐year conditions changed from year to year throughout multiyear periods, with some legacy effects declining, and new responses emerging. Thus, ANPP in a given year was influenced by sequences of conditions that varied across grassland types and climates. Most importantly, the influence of prior‐year ANPP often increased with the length of multiyear periods, whereas the influence of the amount of current‐year PPT declined. Although the mechanisms by which a directional change in the frequency of above‐ and below‐average years imposes a persistent change in grassland ANPP require further investigation, our results emphasize the importance of legacy effects on productivity for sequences of above‐ vs. below‐average years, and illustrate the utility of long‐term data to examine these patterns.