A novel representation of biological nitrogen fixation in GFDL-LM4.1

Background/Question/Methods

The representation of biological nitrogen fixation (BNF), which is the primary natural flux of nitrogen to terrestrial ecosystems, is a key uncertainty in representing the nitrogen cycle in terrestrial biosphere models. Many coupled carbon-nitrogen terrestrial biosphere models use an empirical relationship of BNF with either net primary production (NPP) or evapotranspiration (ET) to estimate BNF. However, these phenomenological relationships are not based on the mechanisms underlying BNF and yield significantly different projections for the magnitude of the future terrestrial carbon sink under global change.

Here, we implement a mechanistic representation of asymbiotic BNF (by implementing nitrogen-fixing soil microbes) and symbiotic BNF (by implementing a nitrogen-fixing tree plant functional type that can compete with a non-fixing tree plant functional type) in the Geophysical Fluid Dynamics Laboratory’s Land Model 4.1 (LM4.1). We validate LM4.1-BNF (LM4.1 with updated BNF and nitrogen cycling) at Coweeta Hydrologic Laboratory, a temperate forest in North Carolina, U.S., using high-resolution published site data and U.S. Forest Inventory and Analysis data. We then compare the major carbon and nitrogen pools and fluxes estimated by LM4.1-BNF to LM4.1-BNF with BNF represented as a function of NPP (LM4.1-BNF_NPP) and ET (LM4.1-BNF_ET).

Results/Conclusions

LM4.1-BNF makes realistic predictions of observations at the individual (growth rate), population (size distribution), community (relative basal area of nitrogen-fixing and non-fixing trees), and ecosystem scales (plant and soil carbon and nitrogen pools and fluxes). LM4.1-BNF reproduces observed asymbiotic BNF rate. Observed symbiotic BNF rate is generally highest in early succession (0 - 40 years) and decreases later in succession as nitrogen limitation decreases. LM4.1-BNF accurately reproduces the temporal dynamics of symbiotic BNF rate whereas LM4.1-BNF_NPP and LM4.1-BNF_ET give constant BNF rates throughout succession. The temporal dynamics of BNF are critical to predicting how terrestrial ecosystems respond to disturbance.

Estimations of total plant biomass carbon under elevated atmospheric CO₂ concentration significantly diverge between LM4.1-BNF, LM4.1-BNF_NPP and LM4.1-BNF_ET. As such, implementing a mechanistic representation of BNF is critical to predicting how BNF will influence the magnitude of the future terrestrial carbon sink under global change.