Redfield revisited 2. What regulates the oxygen content of the atmosphere?

The continuous charcoal record, interpreted with the aid of the results of combustion experiments, indicates that the mixing ratio of atmospheric oxyg en has varied remarkably little over the past 350 Myr. We develop a dynamic feedback model of the coupled P, N, C, and O-2 cycles and use perturbation analysis and a case study of the past 40 Myr to test various feedback mech anisms that have been proposed to stabilize atmospheric oxygen. These mecha nisms involve alterations in nutrient driven productivity and the subsequen t burial flux of organic carbon, which provides the main source of atmosphe ric oxygen. Suppression of the burial of phosphorus sorbed to iron minerals under anoxic conditions in ocean bottom waters tends to increase the ocean nutrient inventory and provide negative feedback against declining oxygen [Holland 1994]. However, denitrification is enhanced by anoxia, tending to reduce the nutrient inventory and amplify declining oxygen [Lenton and Wats on, this issue]. If organic phosphorus removal from the ocean is also suppr essed under anoxic conditions, this improves oxygen regulation [Van Cappell en and Ingall, 1994], as does direct enhancement of organic carbon burial d ue to reduced oxygen concentration in bottom waters [Betts and Holland, 199 1]. However, all of the ocean-based feedback mechanisms cease to operate un der increases in oxygen that remove anoxia from the ocean. Fire frequency i s extremely sensitive to increases in oxygen above 21% of the atmosphere, r eadily suppressing vegetation on the land surface. This should transfer pho sphorus from the land to the ocean, causing less carbon to be buried per un it of phosphorus and providing a weak negative feedback on oxygen [Kump, 19 88]. However, a new proposal that increases in oxygen suppress the biologic al amplification of rock weathering and hence the input of phosphorus to th e Earth system provides the most effective oxygen regulation of all the mec hanisms considered. A range of proxies suggests that the input of available phosphorus to the ocean may have been significantly reduced 40 Myr ago, su ppressing new production and organic carbon burial in the model. With only ocean-based feedback, the atmospheric oxygen reservoir is predicted to have shrunk from similar to 26% of the atmosphere 40 Myr ago. However, when lan d plant mediated negative feedback on phosphorus weathering is added, oxyge n is regulated within 19-21% of the atmosphere throughout the past 40 Myr, in a manner more consistent with paleorecords.