Modeling the effect of nitrogen fixation on carbon and nitrogen fluxes at BATS

Recent geochemical estimates of N₂-fixation in the North Atlantic ocean indicate rates that are significantly higher than those derived from direct observations. In this paper different N₂-fixation rate scenarios are explored using a one-dimensional, biogeochemical model that includes an explicit representation of Trichodesmium. This model reproduces most of the observed interannual variability in phytoplankton production and generates seasonal Trichodesmium biomass and N₂-fixation cycles similar to those observed at BATS. Two solutions are presented, one where the N₂-fixation rate is increased enough to reproduce the observed summertime drawdown of DIC, and a second where it is tuned to reproduce the observed sediment trap fluxes. The high N₂-fixation solution reproduces the seasonal and interannual variability in DIC concentrations quite accurately and generates N₂-fixation rates that agree with direct rate measurements from 1990 and recent geochemical estimates. However, this solution generates export fluxes that are more than 4 times higher than those observed, and predicts the development of DON and DOC anomalies in late summer/early fall that have not been observed. In contrast, the low N₂-fixation solution generates trap fluxes that are approximately correct, but overestimates the summertime DIC concentrations by 20-30 μmol/kg^-1. Both solutions indicate that there is significant interannual variability in N₂-fixation at BATS and that the rates were much lower in 1995-1996 than in the previous six years. It is suggested that this variability is linked to decadal-scale fluctuations in the North Atlantic climate.