The Bay-Delta Selenium Model is a systematic linked approach for conducting forecasts of selenium (Se) effects on aquatic food webs including higher trophic level animals such as birds and fish (Luoma and Presser, 2000). The methodology is presented as a new tool to predict ecological effects based on the major processes leading from loading through consumer organisms to predators. The approach is illustrated here and can be used with any set of explicitly stated conditions. Forecasts obtained from the Bay-Delta Selenium Model consider (1) loads, (2) water column concentrations, (3) speciation, (4) transformation to particulate forms, (5) particulate concentrations, (6) bioaccumulation, and (7) trophic transfer to predators, in addition to traditional considerations of water supply and drainage demand.
Major inputs used to determine a composite input load are (1) agricultural drainage via direct discharge to the Bay-Delta; (2) effluents from the North Bay oil refineries; (3) San Joaquin River inflows which include agricultural drainage; and (4) Sacramento River inflows. Historical analyses of drainage needs were used to identify the most likely Se loads that would be carried outside the San Joaquin Valley via a conveyance discharging a constant load and conveyance via the San Joaquin River. Selenium concentrations and forms in the Bay-Delta are forecast, then those concentrations are used to model bioaccumulation in invertebrates, like clams. Transfer from clams to predators is estimated from field data, and Se effects on predators are then forecast from data in the existing literature. Data gathered during the years prior to refinery cleanup helped check the model and provide a baseline for determining site-specific effects.
The Se load delivered to the Bay also depends on the amount of flow from the San Joaquin River that passes through the Delta and the amount recycled south through the Delta and Tracy pumping stations. The protocol for linking Se load and Se concentration under assigned hydraulic conditions and time duration is:
composite freshwater endmember concentration
= composite input load/composite input volume
The projections or outputs of the model are presented by season, where a season is defined as six months of predominantly high river inflows (December through May) or six months of predominantly low river inflows (June through November). Riverine influences also depend upon water year type. In combination with flow seasons, forecasts are made for critically dry years or for wet years. A wide range of agricultural Se input loads is possible, depending upon which management strategies are chosen. Potential ranges of annual input loads were derived assuming Se discharge was continuous and are presented here as discharged load per six months (i.e. one-half the annual load under a constant rate of loading).
The model allows consideration of many different drainage options (Luoma and Presser, 2000). In general, most options that meet existing demand for drainage appear to pose strong risks to the reproduction and survival of sensitive birds and fish. Threats to reproduction and survival of birds and fish are particularly severe during periods of low river flow. Vulnerable species include diving ducks, white sturgeon and Sacramento splittail.
An example forecast is shown (opposite page) for a dry year during the low flow season and with conveyance through a San Luis Drain extension directly to the Bay-Delta. The dry years and low flow seasons will be the ecological bottleneck (the times that will drive impacts) with regard to Se. Surf scoter, greater and lesser scaup, and white sturgeon are present in the estuary during the low flow season and leave before high flows subside. Animals preparing for reproduction, or for which early life stages develop in September through March, will be vulnerable.
The example forecast shows Se concentrations for each media forecast (water, particulate, invertebrate, predator), along with guidelines or concentrations where biotic effects are expected (Luoma and Presser, 2000). The forecasts show conditions at the head of the estuary for a range of inputs (6,800; 18,700; or 44,880 pounds Se released per six months) from the San Luis Drain and for a small amount of San Joaquin River inflow to the Bay. The input from oil refineries is assumed constant at 680 pounds Se per six months. We assume a partitioning coefficient (Kd) of 3 x 103 typical of Bay-Delta shallow sediment conditions and a generic bivalve assimilation efficiency (AE) of 0.55 to reflect particulate transformation and bioaccumulation potential from a sediment with a mixture of forms.
|The San Luis Drain transports selenium-laden agricltural drainage from the western San Joaquin Valley. Agricultural drainage waterways are posted with a state fish and waterfowl consumption advisory because of selenium.|
In general, the lowest guideline values for waterborne, particulate, dietary, and predator tissue Se are exceeded in every forecast considered in the figure where the input is from a proposed San Luis Drain extension. The highest guidelines from the literature are exceeded in all forecasts except that for the lowest load considered (6,800 pounds per six months) where exceedance does occur for particulates, white sturgeon, and greater and lesser scaup liver. If a San Luis Drain extension is constructed and if it discharges the quantities of Se predicted in our simulation scenarios, during low flow seasons, a high hazard seems likely, with threats to fish and bird species under the load scenarios tested here.
Forecast simulations also were conducted for loading via the San Joaquin River (Luoma and Presser, 2000). If careful management of an out-of-valley resolution to the drainage problem results in discharges of Se via the San Joaquin River to the Bay-Delta (for example at 3,500 pounds per six months), the risks are less than those forecast for a San Luis Drain extension. Under the low flow season of a dry year scenario, the Se concentrations forecast in prey and predators are similar to Se concentrations observed during conditions in the Bay-Delta prior to refinery cleanup. Selenium contamination documented from 1986 to 1996 was sufficient to threaten reproduction in key species within the Bay-Delta estuary ecosystems and resulted in human health advisories being posted for consumption of those species.
The Bay-Delta is probably best suited for site-specific Se guidelines and the Bay-Delta Selenium Model could provide a framework for developing new protective criteria. If water quality criteria are to be employed in managing Se inputs, then consideration should be given to the elevated Se concentrations currently occurring in clams and fish of the Bay-Delta, even though waterborne Se concentrations in the Bay-Delta are less than recommended criteria (Luoma and Presser, 2000).
A long-term monitoring program is crucial to understanding the fate and impact of management changes for protection of ecosystems receiving Se discharges. Monitoring, as conceptualized below, would sample critical environmental components at a frequency relevant to each process to determine trends in Se contamination or changes in processes that determine fate and effects of Se (Luoma and Presser, 2000).
A linked or combined approach would include all considerations that cause systems to respond differently to Se contamination.
1. Identification of vulnerable food webs
2. Identification of sites most at risk from impacts of agricultural drainage
3. Analysis of effects on predators that includes food web components
4. Identification of elevated risk periods for effects based on hydrodynamics
5. Calculation of protective loads/concentrations based on bioaccumulation in prey
Summary of Selenium Issues in California
Theresa S. Presser and Samuel N. Luoma
National Research Program
Water Resources Division
Menlo Park, CA 94025
Linking Selenium Sources to Ecosystems:
San Francisco Bay-Delta Model
|AccessibilityFOIAPrivacyPolicies and Notices|
|U.S. Department of the Interior, U.S. Geological Survey
Page Contact Information: Contact USGS
Last modified: Tuesday, November 29 2016, 05:59:34 PM