Global Carbon Sequestration Is Highly Sensitive to Model‐Based Formulations of Nitrogen Fixation

Biological nitrogen fixation (BNF) is the largest nitrogen (N) input pathway in natural terrestrial ecosystems at present, fueling the drawdown of atmospheric CO 2 in vegetation and soil on decadal to century time scales. Here, we use a global land‐surface model (CABLE) with three different approaches for estimating the responses of BNF to CO 2 , climate, and N deposition through Year 2100, particularly the linear versus nonlinear dependence of BNF on C investment and the temperature dependence of BNF. From 1900 to 2100, the cumulative rise in BNF varied from 1.6 to 3.0 Pg N, translating to an increase in terrestrial carbon (C) storage between 33 and 68 Pg C. This range reflects the different approaches used to model BNF (i.e., C‐limited vs. resource optimization approaches), indicating major uncertainties in C‐climate‐N interactions in Earth system model forecasts. The differences among different approaches were most significant at high or low latitudes. At high latitudes, accounting for temperature dependence of BNF resulted in 53% and 79% additional BNF increases and 14% and 21% additional land C accumulation in evergreen needle leaf forest and tundra, respectively, from 1901 to 2100. At low latitudes, resource optimization approaches using linear dependence of BNF on C investment estimated more rapid increase in BNF and therefore greater C accumulation than the C‐limited approach using nonlinear dependence. The difference in the estimated additional C accumulation due to varying BNF was as much as 24% for deciduous broadleaf forest. Our findings highlight the need for more field measurements of BNF, particularly at high latitudes to better constrain the projected BNF under future conditions, and also the fundamental importance of BNF in determining the pattern, response, and magnitude of terrestrial C accumulation through 2100.