Evaluating the Biogeochemical Cycle of Selenium in San Francisco Bay Through Modeling

A biogeochemical model was developed to simulate salinity, total suspended material, phytoplankton biomass, dissolved selenium concentrations (selenite, selenate, and organic selenide), and particulate selenium concentrations (selenite + selenate, elemental selenium, and organic selenide) in the San Francisco Bay estuary. Model-generated estuarine profiles of total dissolved selenium reproduced observed estuarine profiles at a confidence interval of 91–99% for 8 different years under various environmental conditions. The model accurately reproduced the observed dissolved speciation at confidence intervals of 81–98% for selenite, 72–91% for selenate, and 60–96% for organic selenide. For particulate selenium, model-simulated estuarine profiles duplicated the observed behavior of total particulate selenium (76–93%), elemental selenium (80–97%), selenite + selenate (77–82%), and organic selenide (70–83%). Discrepancies between model simulations and the observed data provided insights into the estuarine biogeochemical cycle of selenium that were largely unknown (e.g., adsorption/desorption). Forecasting simulations investigated how an increase in the discharge from the San Joaquin River and varying refinery inputs affect total dissolved and particulate selenium within the estuary. These model runs indicate that during high river flows the refinery signal is undetectable, but when river flow is low (70day residence time) total particle-associated selenium concentrations can increase to .2 mg g21. Increasing the San Joaquin River discharge could also increase the total particle-associated selenium concentrations to .1 mg g21. For both forecasting simulations, particle-associated selenium was predicted to be higher than current conditions and reached levels where selenium could accumulate in the estuarine food web. Extensive research has been done on modeling how physical, biological, or chemical parameters in an estuary individually affect the distribution and speciation of a trace element (e.g., Paucot and Wollast 1997; Mwanuzi and De Smedt 1999), but little work has been done using models to simulate the complete biogeochemical cycle of a trace element (i.e., coupling physical, biological, and chemical processes). With recent advances in estuarine modeling, more extensive simulations of biogeochemical cycles of an element are now possible. Coupling empirical observations with modeling enables estuarine processes to be more completely elucidated than using either approach individually. In this respect, selenium presents some compelling reasons for use as the ‘‘test’’ element. Human activities (e.g., irrigation, petroleum refining, power production, and mining), have increased the input of selenium to some aquatic systems. This mobilization has been implicated in elevated concentrations of selenium in waterfowl, fish, and bivalves of some estuaries like the San Francisco Bay (Ohlendorf et al. 1986). Selenium exists in four oxidation states (II, 0, IV, and VI), and in different chemical forms (i.e., organic and inorganic) within these oxidation states. In oxygenated marine and fresh waters, dissolved selenium is found as selenite (Se+IV, ca. 35% of the total selenium; Measures et al. 1980), selenate (Se+VI), and organic selenides (Se-II), with some of the organic species as selenium-containing amino acids and peptides (Cutter and Bruland 1984) and methylated forms (Cooke and Bruland 1987). Significantly, the biotic uptake and toxicity of dissolved selenium depends not only on its concentration, but also on its chemical speciation (Riedel et al. 1996). Thus, any modeling efforts with selenium must include the capability for accurate speciation predictions. The biogeochemical cycle of dissolved selenium in estuarine waters (Takayanagi and Cossa 1985; Cutter 1989; Cutter and Cutter 2004) and sediments (Belzile and Lebel 1988; Velinsky and Cutter 1991) and its bioavailability in the food web (Doblin et al. 1999) have all been examined. The biogeochemical cycle of dissolved selenium in an estuary (Fig. 1) includes inputs via rivers, anthropogenic sources, and exchange with the open ocean. Advection and diffusion move selenium through the estuary, while internal transformations occur through biotic and abiotic reactions during transport. The transformation reactions (biotic and abiotic) include the oxidation of dissolved organic selenide to selenite and selenite to selenate. Biotic reactions affecting selenium in an estuary include dissolved selenite, selenate, and organic selenide uptake by phytoplankton, and incorporation into various biochemical components (Fig. 1). The sources of particulate selenium to an estuary are particles from rivers (biogenic and mineral detritus), biogenic particles produced in the water column (phytoplankton detritus), and sediment resuspension. Suspended particulate organic selenide can undergo remineralization 1 Present address: U.S. Department of Commerce/National Oceanic & Atmospheric Administration/National Marine Fisheries Service, 212 Rogers Avenue, Milford, Connecticut 06460. Acknowledgments We thank Lynda Cutter and Martina Doblin for sampling and providing the dissolved and particulate selenium speciation data, and Robin Stewart of U.S. Geological Survey for the acquisition and processing of refinery samples. This project was supported by funds from the National Science Foundation Environmental Geochemistry and Biogeochemistry (EBG) Program (OCE-9707946 to G.A.C.) and CALFED (982015000-00096). Limnol. Oceanogr., 51(5), 2006, 2018–2032 E 2006, by the American Society of Limnology and Oceanography, Inc.