Dryness controls temperature-optimized gross primary productivity across vegetation types

Temperature response of gross primary productivity (GPP) is a well-known property of ecosystem, but GPP at the optimum temperature (GPP_T_opt) has not been fully discussed. Our understanding of how GPP_T_opt responds to warming and water availability is highly limited. In this study, we analyzed data at 326 globally distributed eddy covariance sites (79^oN-37^oS), to identify controlling factors of GPP_T_opt. Although GPP_T_opt was significantly influenced by soil moisture, global solar radiation, mean annual temperature, and vapor pressure deficit in a non-linear pattern (R² = 0.47), the direction and magnitude of these climate variables’ effects on GPP_T_opt depend on the dryness index (DI), a ratio of potential evapotranspiration to precipitation. The spatial pattern showed that soil moisture did not affect GPP_T_opt across energy-limited sites with DI < 1 while dominated GPP_T_opt across water-limited sites with DI >1. The temporal pattern showed that GPP_T_opt was lowered by warming or low precipitation in water-limited sites while energy-limited sites tended to maintain a stable GPP_T_opt regardless of changes in air temperature. Vegetation types in humid climates tended to have higher GPP_T_opt and were more likely to benefit from a warmer climate since it was not restricted by water conditions. This study highlights that the response of GPP_T_opt to global warming depends on the dryness conditions, which explains the nonlinear control of water and temperature over GPP_T_opt. Our finding is essential to realistic prediction of terrestrial carbon uptake under future climate and vegetation conditions.