Limnol. Oceanogr., 44(4), 1999, 1148–1154

Several researchers have proposed spectrophotometric modifications of the Winkler titrimetric method for measuring dissolved oxygen (DO). These modifications, although simple, are not widely used because of concern about accuracy, calibration, and possible sources of interference. Here we show, using natural samples from lakes and rivers as well as samples manipulated in the laboratory, that the spectrophotometric method can provide accurate and very precise measurements of DO over a wide range of concentrations (4 to ;13 mg O2 liter21). Further, interference from dissolved organic carbon (color) and turbidity are minor. We propose corrections for both color and turbidity, where necessary, that can be easily incorporated into the measurement design. Because of the speed and simplicity of the spectrophotometric method, it is easy to replicate measurements and thereby increase precision without greatly increasing analytical time. In 10 min of effort, we were able to achieve a coefficient of variation (CV) within one bottle of 0.09%, or 0.8% among different bottles. With n 5 7 bottles, one can easily distinguish changes in DO of 0.05 mg liter21 with this method, which makes it useful for metabolic studies in many environments. To achieve a comparable CV by conventional titration would require about 100 min of effort. The concentration of dissolved oxygen (DO) in aquatic systems, as well as the rate of its production and removal by metabolic and chemical processes, has proved to be a useful measurement across most branches of aquatic science. A number of different approaches have been devised to measure DO, including various kinds of electrodes (Reynolds 1969; Atwood et al. 1977; Wilcock et al. 1981), but most researchers needing high accuracy or precision still rely on the titrimetric method of Winkler (Winkler 1888; Aminot 1988; Carignan et al. 1998). Although it is very precise when replicated, this analysis is time-consuming. Also, in systems that are turbid or highly colored with organic matter, the end point is sometimes difficult to visualize. The use of a potentiometric end point for the titration overcame many of these problems (Carpenter 1965; Carrit and Carpenter 1966) and made possible automatic versions of the original Winkler titration (Bryan et al. 1976; Hartwig and Michael 1978; Williams and Jenkinson 1982; Graneli and Graneli 1991; Williams and Purdie 1991). The drawback of automatic titration for many researchers is the cost of the specialized equipment. Further, although the titration is itself automatic, it is still time-consuming, requiring some 3–5 min per sample. In the Winkler DO method, tri-iodide is ultimately formed in proportion to the DO present. In an equilibrium that is dependent on both temperature and the concentration of KI (added in excess), tri-iodide dissociates into molecular iodine and iodide. In the titrimetric method, the I –I2 pool is mea3 sured by reducing it to NaI by the addition of sodium thiosulfate. The spectrophotometric method makes use of the color of the I –I2 couple. The color of the resulting solution 3 is quantified in a spectrophotometer, which can analyze it at several wavelengths (Custer and Natelson 1949). A spectrophotometric modification of the Winkler DO method was first proposed by Broenkow and Cline (1969) for use at low DO levels. Since then, modifications of the spectrophotometric method have been used by a variety of researchers (Reynolds 1972; Duval et al. 1974; Ashton and Twinch 1985), but the method has not been standardized or widely tested in the field. There are several potential disadvantages to the spectrophotometric approach. Because one is simultaneously measuring light absorption by two species