Revisiting N_2 fixation in the North Atlantic Ocean: Significance of deviations from the Redfield Ratio, atmospheric deposition and climate variability

摘要

The average oceanic nitrate-to-phosphate molar ratio (NO_3~-:PO_4~(3-)≈16:l, referred to as the Redfield Ratio) in subsurface waters, which is similar to the average ratio of particulate nitrogen (N)-to-phosphorus (P) in phytoplankton, is the cornerstone in calculating geochemical estimates of N_2 fixation and denitrification rates. Any deviations from this canonical Redfield Ratio in intermediate ocean waters, expressed as N* (a measure of NO_3~- in excess or deficit of 16 × PO_4~(3-)), provides an integrated estimate of net N fluxes into and out of the ocean. In well-oxygenated ocean basins such as the North Atlantic Ocean, N* estimates are usually positive and can be used to infer that rates of N_2 fixation exceed rates of denitrification. We use this approach to estimate N_2 fixation over the last two decades (1988-2009) based on data collected at the Bermuda Atlantic Time-series Study (BATS) site in the North Atlantic Ocean near Bermuda. Our results indicate that interpretation of the N* tracer as an estimate of N_2 fixation should be undertaken with caution, as N_2 fixation is not the only process that results in a positive N* estimate. The impacts of a locally variable nitrogen-to-phosphorus ratio, relative to the fixed Redfield Ratio, in the suspended particulate matter as well as in the subsurface water nutrients and atmospheric N deposition on N* variability were examined. Furthermore, we explored the role of climate modes (i.e., North Atlantic Oscillation and Arctic Oscillation) on N* variability. We found that N* in the subsurface waters was significantly affected by these factors and hence previous estimates of N_2 fixation using this technique might have been substantially overestimated. Our revised estimate of N_2 fixation in the North Atlantic Ocean (0°N-50°N, 20°W-80°W) is 12.2 ± 0.9 × 10~（11） mol N yr~(-1), and based on long-term BATS data provides better constraints than both earlier indirect and direct estimates N_2 fixation.