Abstract
Models predicting ecosystem carbon dioxide (CO2) exchange under future climate change rely on relatively few real-world tests of their assumptions and outputs. Here, we demonstrate a rapid and cost-effective method to estimateCO2exchange from intact vegetation patches under varying atmospheric CO2concentrations.We find that net ecosys-tem CO2uptake (NEE) in a boreal forest rose linearly by 4.7 0.2% of the current ambient rate for every 10 ppmCO2increase, with no detectable influence of foliar biomass, season, or nitrogen (N) fertilization. The lack of any clearshort-term NEE response to fertilization in such an N-limited system is inconsistent with the instantaneous downreg-ulation of photosynthesis formalized in many global models. Incorporating an alternative mechanism with consider-able empirical support – diversion of excess carbon to storage compounds – into an existing earth system modelbrings the model output into closer agreement with our field measurements. A global simulation incorporating thismodified model reduces a long-standing mismatch between the modeled and observed seasonal amplitude of atmo-spheric CO2. Wider application of this chamber approach would provide critical data needed to further improvemodeled projections of biosphere–atmosphere CO2exchange in a changing climate.