Widespread temporal and spatial variability in net ecosystem productivity under climate change

The balance between net carbon assimilation by plants and carbon losses through heterotrophic respiration plays an important role in regulating the terrestrial carbon balance and the Earth's climate. The current high uncertainty in the magnitude of these carbon fluxes in Earth system models (ESMs), however, limits our ability to project the dynamics of net ecosystem productivity (NEP), which primarily determines to what extent ecosystems can store carbon. Here, we apply a data-model fusion approach to constrain global carbon flux estimates from ESMs. The analysis reveals that the spatial and temporal variability of NEP may be higher than currently expected from ESM simulations. Particularly, the NEP of Amazon forests is projected to decline considerably by the end of the century under SSP126 but will likely increase under higher emission scenarios due to the CO2 fertilization effect, highlighting the need for effective actions to maintain their C storage capacity under climate change.