Abstract
Understanding the relationship between precipitation (PPT) and aboveground net primary productivity (ANPP) is essential for modeling the global carbon cycle. Across grassland to forest gradients, the PPT‐ANPP relationship is well defined and nonlinear. Temporal patterns within a site over time are more variable and nearly always linear. Linear relationships, however, are inconsistent with positive asymmetry, where increases in ANPP during wet years exceed declines in dry years. The double asymmetry model predicts that concave‐down nonlinearities will occur when extreme high and low PPT years are included in a time series. We tested this prediction using long‐term observational ANPP data along with rainfall manipulation experiments. By combining observational records with experimental treatments, including drought, water addition, and nitrogen addition, we found some support for the double asymmetry model. However, the response under high precipitation coupled with nitrogen addition was concave‐up, not down. By experimentally extending the range of monsoon precipitation, we found a weak but significant, nonlinear PPT‐ANPP relationship, but only when nutrient limitation was alleviated. Our results demonstrate that multiple interacting factors govern the PPT‐ANPP relationship within a site over time, challenging our ability to predict how ANPP will respond to changes in precipitation in the future.