Winkler's method overestimates dissolved oxygen in seawater: Iodate interference and its oceanographic implications

Dissolved oxygen in seawater has been determined by using the Winkler's reaction scheme for decades. An interference in this reaction scheme that has been heretofore overlooked is the presence of naturally occurring iodate in seawater. Each mole of iodate can result in an apparent presence of 1.5 mol of dissolved oxygen. At the concentrations of iodate in the surface and deep open ocean, it can lead to an overestimation of 0.52 ± 0.15 and 0.63 ± 0.05 μmol kg− 1 of oxygen in these waters respectively. In coastal and inshore waters, the effect is less predictable as the concentration of iodate is more variable. The solubility of oxygen in seawater was likely overestimated in data sources that were based on the Winkler's reaction scheme for the determination of oxygen. The solubility equation of García and Gordon [Garcia H.E., Gordon, L.I., 1992. Oxygen solubility in seawater: Better fitting equations. Limnol. Oceanogr. 37, 1307–1312] derived from the results of Benson and Krause [Benson, F.B., Krause, D. Jr., 1984. The concentration and isotopic fractionation of oxygen dissolved in freshwater and seawater in equilibrium with the atmosphere. Limnol. Oceanogr. 29, 620–632] is free from this source of error and is recommended for general use. By neglecting the presence of iodate, the average global super-saturation of oxygen in the surface oceans and the corresponding efflux of oxygen to the atmosphere both have been overestimated by about 8%. Regionally, in areas where the degree of super-saturation or under-saturation of oxygen in the surface water is small, such as in the tropical oceans, the net air–sea exchange flux can be grossly under- or overestimated. Even the estimated direction of the exchange can be reversed. Furthermore, the presence of iodate can lead to an overestimation of the saturation anomaly of oxygen in the upper ocean attributed to biological production by 0.23 ± 0.07%. AOU may have been underestimated by 0.52 ± 0.15 and 0.63 ± 0.05 μmol kg− 1 in the surface mixed layer and deep water, while preformed phosphate and preformed nitrate may have been overestimated by 0.004 ± 0.001 and 0.06 ± 0.02 μmol kg− 1 in the surface mixed layer, and 0.005 ± 0.0004 and 0.073 ± 0.006 μmol kg− 1 in the deep water. These are small but not negligible corrections, especially in areas where the values of these parameters are small. At the increasing level of sophistication in the interpretation of oxygen data, this source of error should now be taken into account. Nevertheless, in order to avoid confusion, an internationally accepted standard needs to be adopted before these corrections can be applied.