UC Berkeley

Selenium (Se) is a trace element of great ecological importance whose environmental distribution is highly impacted by anthropogenic activity. In the 1980s, selenium was recognized as a major aquatic contaminant following widespread deformities and mortality among waterfowl hatchlings near the agricultural drainage evaporation ponds of the Kesterson Reservoir (CA, USA). Today, 400,000 km² in the Western United States are threatened by agricultural selenium contamination, as are parts of Canada, Egypt, Israel, and Mexico. From the soil aggregate to the watershed, from the soils of the Central Valley to the sediments of the Salton Sea, and from Environmental Science to Policy and Management, in this dissertation I explore agricultural selenium contamination across scales, ecosystems, and disciplines. I begin with a review of the science, policy, and management of irrigation-induced selenium contamination in California, the heart of worldwide research on the issue. I then delve into the physical and biogeochemical mechanisms that control selenium reduction and mobility within the structured surface soils that are the source of contamination, using an aggregate-scale combined experimental and reactive transport modeling approach. Finally, I present a diagenetic model for selenium incorporation into the sediment of the Salton Sea, which has been receiving seleniferous agricultural drainage over the last 100 years.

To extract lessons from the last 30 years of seleniferous drainage management and water quality regulation in California, I reviewed the history and current developments in science, policy, and management of irrigation-induced selenium contamination in California. Specifically, I evaluated improvements in the design of local attenuation methods and the development of programs for selenium load reductions at the regional scale. On the policy side, I assessed the site-specific water quality criteria under development for the San Francisco Bay-Delta in the context of previous regulation. This approach may be a landmark for future legislation on selenium in natural water bodies and I discussed challenges and opportunities in expanding it to other locations such as the Salton Sea. By combining proven management tools with the novel, site-specific policy approach, it may be possible to avoid future events of irrigation-induced selenium contamination. However, the majority of regional selenium load reductions in California were achieved by decreasing drainage volume rather than selenium concentrations. Thus, there appear to be opportunities for additional improvements through management practices that enhance selenium retention in source soils.

Soil aggregates are the basic structural units of soil. They are mm- to cm-sized microporous assemblages of loosely bound soil particles, separated from one another by macropores. To elucidate how aggregate-scale transport and microbial reduction affect selenium retention in surface soils, I conducted a series of flow-through experiments utilizing artificial aggregate systems. These systems mimic the dual porosity of structured soils with an artificial soil aggregate (ID 2.5 cm) contained in a flow-through reactor cell. Aggregates were composed of either pure quartz sand or ferrihydrite-coated quartz sand inoculated with selenium reducing bacteria (Thauera selenatis or Enterobacter cloacae SLD1a-1). Oxic and anoxic conditions were compared, as well as various selenate (0.25-0.8 mM) and carbon source (0.3 and 1.2 mM) input concentrations. The presence of oxygen in the input solution significantly decreased selenium reduction, however, the detection of selenite in effluent samples indicates the occurrence of anoxic microzones within aggregates. Furthermore, I found that solid phase concentrations of reduced selenium increased towards the core of aggregates across all experimental conditions. A bulk model, ignoring intra-aggregate heterogeneity in reactions and transport, misrepresented the dynamics of the aggregate systems.

To quantify the likely implications of these experimental results for soils with different degrees of aggregation, I formulated a general mechanistic framework for aggregate scale heterogeneity in selenium reduction. Specifically, I constructed a dynamic 2D model of selenium fate in single idealized aggregates, in which reactions were implemented with double-Monod rate equations coupled to the transport of pyruvate, O₂, and Se-species (selenate, selenite, and elemental selenium). The spatial and temporal dynamics of the model were validated with the experimental data and predictive simulations were performed covering aggregate sizes between 1 and 2.5 cm diameters. Simulations predict that selenium retention scales with aggregate size. Depending on aeration conditions and the input concentrations of selenate and pyruvate, selenium retention was predicted to be 4-23 times higher in 2.5-cm-aggregates compared to 1-cm-aggregates. Under oxic conditions, aggregate size and pyruvate-concentrations were found to have a positive synergistic effect on selenium retention. Promoting soil aggregation on seleniferous agricultural soils may thus help decrease the impacts of selenium contaminated drainage on downstream aquatic ecosystems receiving it.

One such ecosystem is California's largest inland water body, the Salton Sea, which is maintained entirely by agricultural runoff. Whereas elevated selenium concentrations are detected in the rivers feeding the lake, the lake's concentrations of dissolved selenium are low since most selenium entering the lake is sequestered in its sediment through microbial reduction. To predict the distribution profiles of selenium within the sediment and evaluate the factors driving them, I constructed a diagenetic model for reductive incorporation of selenate into Salton Sea sediment. The model predicts near surface (2 cm) sediment concentrations of solid phase selenium between 0.024 and 0.272 µmol/g depending on local reduction kinetics, and dissolved concentrations in the water column. This is in good agreement with the literature when considered in conjunction with the potential impact of bioturbation which according to exploratory simulations may lower near surface concentrations by around 25%. The range of modeled selenium concentrations in surface sediment crosses threshold values for which negative impacts on fish and waterfowl have been predicted or observed at other sites, suggesting that ecological impacts of selenium in the Salton Sea may depend locally on variation in the diagenetic factors here explored.

This work presents agricultural selenium contamination as a complex problem that crosses ecosystems, scales, and disciplines. From a management perspective, the tension between dispersed non-point sources and hotspots where elevated selenium concentrations and sensitive aquatic ecosystems converge is difficult to address. Differences in biogeochemical conditions and trophic transfer within food webs render traditional regulatory approaches ineffective and force regulators to engage with the science of site-specific selenium transfer between ecological compartments. At the same time, gaps still exist in our mechanistic understanding of selenium's environmental cycling and in our integration of scientific knowledge across different ecosystems and scales. Centimeter scale heterogeneity in the biogeochemical conditions within source soils may fundamentally control selenium emissions across large agricultural areas and thus determine the selenium loading of rivers, lakes, and estuaries. Within aquatic environments receiving seleniferous drainage, the first few centimeters of surface sediment may control selenium exposure for entire food webs. Improved understanding at this level holds the potential to simultaneously reduce selenium emissions and respond more effectively to pollution where it occurs. In order to preserve sensitive habitat while also meeting agricultural drainage needs in seleniferous regions we must bridge the gaps between ecosystems, scales, and disciplines.