8281 - Laserfiche WebLink

Home Browse Search

232 COMMENTARY component. This new approach to mass balance recognizes the cumulative loading in water, sediment, and biota. Furthermore, a well-matched dataset of simultaneous measures of selenium in water, sediment, and a relevant biological response variable is needed to derive a predictor of adverse effects. The authors are aware of only one well- matched dataset that meets this criterion. In a study of evaporation ponds in the Tulare Lake basin, California, selenium in water was a better predictor of selenium in eggs of black-necked stilt (Himantopus mexicanus) (r2 = 0.81; Ohlendorf et aI., 1993) than was selenium in sediment (r2 = 0.60; Skorupa et al., 1996). Van Derveer and Canton (1997) concluded the opposite without citing any matched sets of data that support such a conclusion. Morevoer, the use of Hill's criteria (Hill, 1965) by Canton and Van Derveer (1997) to support their conclusion that sediment selenium is better for predicting adverse biological effects than waterborne selenium is inaccurate and lacks adequate development. Canton and Van Derveer (1997) do not adequately present the support data from the references they cite. It is unclear how they can make conclusions about how sediment organic carbon can be used to predict the bioavailability and toxicity of selenium to biota because they did not present, and apparently did not measure, sel- enium bioaccumulation at the study sites. Their analysis of linkages between sediment selenium and biological effects (Table 3 in Canton and Van Derveer, 1997) is not supported for several reasons: (1) almost half the sediment selenium concentrations are reported as'less than values (below limit of detection), yet these data were apparently used to esti- mate thresholds; (2) data for sites where effects were pre- dicted by other authors, yet they do not assess the basis or accuracy of those predictions; and (3) their toxicity thresh- olds were based on the 10th percentiles of observed effects concentrations from a dataset of only seven values. 4.5. In-Stream Considerations The unsupported logic used to justify the sediment-based water quality criteria presented in Canton and Van Derveer (1997) is cited at least seven times as support in Van Derveer and Canton (1997). Based on the model to calculate a site- specific chronic dissolved selenium stream standard, they calculated an instreaD1 standard of 31Ilg/L. However, two stream studies with to Ilg/L selenium and fathead minnow and bluegill (Lepomis macrochirus) demonstrated adverse biological effects on adults and progeny (Schultz and Her- manutz, 1990; Hermanutz, 1992; Hermanutz et al., 1992). These studies do not support the conclusion by Van Der- veer and Canton (1997) that higher water quality standards for selenium are justified in streams. The error in proposing an instream standard 31 Ilg/L is not considering the consequence to offstream waters, such as backwaters. oxbows. and reservoirs. There may be no readily observed impacts from selenium in lotic systems. due to relatively low productivity compared to [entic systems that would potentially lower the bioaccumulation of selenium in the lotic environment. However. high in- stream selenium concentrations in an offstream environ- ment, such as a backwater or reservoir. would undoubtedly result in substantial selenium bioaccumulation in the food web and potentially severe effects on fish and wildlife popu- lations. Lemly (1999) addressed this concern by proposing a hydrological unit approach for setting selenium criteria. This method recognizes that all of the hydrologically con- nected parts of a watershed basin must be considered to- gether and given the same criterion in order to protect against toxic impacts at sites where bioaccumulation is greatest. Skorupa (1998) has also addressed this concern and uses the example of California's Salton Sea. which already exhibits selenium threshold toxicity. He notes that a seemingly trivial increase of 1-2 Ilg/L in Colorado River water, as a result of implementing site-specific selenium criteria within the Colorado River and its tributaries, would double the selenium load delivered to the Salton Sea. which in turn could put millions of fish and birds in serious jeopardy. Proposing high instream selenium standards must be balanced by considering offstream consequences. Selenium will move throughout the length of the channel of the affected stream or river, and with its terminal destination the ocean. Sedimentaion does not remove selenium entirely from lotic ecosystems. For example. the major finding of the NIWQP investigation of the lower Colorado River between Davis Dam and the U.S.-Mexico border was that consis- tently higher selenium concentrations occurred in back- waters and oxbows that received water from the Colorado River than in canals returning irrigation water to the river (Radtke et aI., 1988). The investigation concluded that sel- enium concentration in the lower Colorado River was not from local agricultural sources, but rather from upstream sources, i.e., the upper Colorado River basin. even though river water had to pass through Lakes Powell. Mead. and Havasu. About 82% of the selenium entering Lake Powell from the Colorado and San Juan rivers passes through to the lower basin downstream of Lake Powell (Engberg. 1999). Selenium contamination has also been documented in the lower end of the upper Colorado River basin in fish from Lake Powell (Waddell and Wiens, 1993) and in fish from the 24-km river reach between Glen Canyon Dam and Lee's Ferry on the Colorado River (A. Ayres, AZ Dep. Game & Fish, personal communication. 1997). The basis for Skorupa's concern for the Salton Sea is sound (Skorupa. 1998). Prior to selenium arriving at the ocean or terminal destination of a river, protection of offstream effects from selenium toxicity must be the paramount decision point for deriving instream selenium standards. especially