Global rates of water-column denitrification derived from nitrogen gas measurements

Biologically available nitrogen (N) limits phytoplankton growth over much of the ocean. The rate at which N is removed from the contemporary ocean by denitrifying bacteria is highly uncertain1–3. Some studies suggest that N losses exceed inputs2,4–6; others argue for a balanced budget3,7,8. Here, we use a global ocean circulation model to simulate the distribution of N2 gas produced by denitrifying bacteria in the three main suboxic zones in the open ocean. By fitting the model to measured N2 gas concentrations, we infer a globally integrated rate of water-column denitrification of 66± 6 Tg N yr−1. Taking into account isotopic constraints on the fraction of denitrification occurring in the water column versus marine sediments, we estimate that the global rate of N loss from marine sediments and the oceanic water column combined amounts to around 230±60 Tg N yr−1. Given present estimates of N input rates, our findings imply a net loss of around 20± 70 Tg of N from the global ocean each year, indistinguishable from a balanced budget. A balanced N budget, in turn, implies that the marine N cycle is governed by strong regulatory feedbacks. The loss of N in the ocean occurs as bacteria reduce N to N2 gas through a combination of heterotrophic denitrification and anaerobic ammonium oxidation (hereafter collectively termed denitrification). Suboxic conditions in which O2 is scarce enough to favour these processes are found in three principal locations in the water column: the Arabian Sea, the eastern tropical South Pacific (ETSP) and the eastern tropical North Pacific (ETNP; Fig. 1). Denitrification rates in these areas estimated from nitrate deficits and water-mass age tracers yield a global rate of watercolumn denitrification in the range of 60–90 TgN yr−1 (refs 9–14), but recently rates as large as 150 TgN yr−1 have been proposed2. Denitrification also occurs in suboxic pore waters of sediments throughout the ocean, but the rates are spatially heterogeneous and have been measured at only a handful of sites15, precluding a direct global estimate. However, the global rate of sedimentary N loss (primarily denitrification, but also a small burial of organic N) is constrained by the isotopic ratio of mean ocean nitrate to be one to four times as large as the rate of denitrification in the suboxic water column6,16,17. Estimates of the total rate of N loss in the ocean therefore inherit and amplify the uncertainty in the rate of denitrification in suboxic waters, yielding a combined rate of about 270 TgN yr−1 (ref. 3) to more than 400 TgN yr−1 (refs 2,6). These rates far exceed the estimated global rate of N fixation, which is in the range of 100–160 TgN yr−1 (refs 7,18). The apparent deficit is partially offset by riverine and atmospheric inputs of about 40–125 TgN yr−1 (refs 3,15). Still, if the upper end of the denitrification-rate estimates are correct, the ocean must be rapidly losing fixed N.