Accuracy of Hazard Ratings
71
taminated references sites (Stephens et al. 1992; Lemly 1997b). In 4
cases, no hazard was indicated by HQ, whereas moderate to high hazard was indicated by the Protocol. At these 4 locations, concentrations
of selenium in water were very low, but selenium was elevated in one
or more of the other ecosystem components used in the Protocol. The
Protocol method was compared with HQ and evaluated for 8 sites for
which full data sets were available (all 5 ecosystem components), as
well as 3 sites for which one component was missing (aquatic bird
eggs). The results were similar in all cases; the Protocol consistently
indicted greater hazard than HQ.
Verification of Results
The data sets used in this analysis represent a wide variety of habitat
types and ecological conditions, ranging from shallow freshwater
marshes to large riverine and reservoir systems. They also represent
levels of selenium contamination ranging from clean reference sites
to heavily contaminated wetlands. This range of conditions should
cover the spectrum of pathways by which selenium cycles, accumulates, and causes toxicity in aquatic habitats. It is important to examine different conditions because a dominant pathway in one system
may be relatively insignificant in another (Lemly and Smith 1987).
Hazard assessment methods that are overly sensitive to one particular
pathway will give inaccurate estimates. The data sets examined provide a good perspective from which to judge the performances of the
two techniques.
Hazard estimates from HQ analysis were consistently lower, usually
much lower, than those generated from the Protocol. Information on
the toxic effects of selenium on fish and aquatic birds at the field sites
can be used to check the results and determine which method produced the most valid estimates of hazard.
Example 1.
The contaminated sites in Utah (Roadside Ponds and Sheppard Bottom) have produced acute and chronic selenium toxicosis and reproductive impairment in fish and aquatic birds (Waddell and Stanger
1992; Hamilton et al. 1996). The severity of these biological effects, as
evidenced by the presence of both acute and chronic toxicity, shows
that the hazard from selenium was high. The Protocol indicated high
hazard at both sites, whereas HQ indicated low to moderate hazard.
Example 2.
Decreased survival of juvenile ducks due to selenium in the diet and
associated decreases in waterfowl production were identified at Fernley,
71
taminated references sites (Stephens et al. 1992; Lemly 1997b). In 4
cases, no hazard was indicated by HQ, whereas moderate to high hazard was indicated by the Protocol. At these 4 locations, concentrations
of selenium in water were very low, but selenium was elevated in one
or more of the other ecosystem components used in the Protocol. The
Protocol method was compared with HQ and evaluated for 8 sites for
which full data sets were available (all 5 ecosystem components), as
well as 3 sites for which one component was missing (aquatic bird
eggs). The results were similar in all cases; the Protocol consistently
indicted greater hazard than HQ.
Verification of Results
The data sets used in this analysis represent a wide variety of habitat
types and ecological conditions, ranging from shallow freshwater
marshes to large riverine and reservoir systems. They also represent
levels of selenium contamination ranging from clean reference sites
to heavily contaminated wetlands. This range of conditions should
cover the spectrum of pathways by which selenium cycles, accumulates, and causes toxicity in aquatic habitats. It is important to examine different conditions because a dominant pathway in one system
may be relatively insignificant in another (Lemly and Smith 1987).
Hazard assessment methods that are overly sensitive to one particular
pathway will give inaccurate estimates. The data sets examined provide a good perspective from which to judge the performances of the
two techniques.
Hazard estimates from HQ analysis were consistently lower, usually
much lower, than those generated from the Protocol. Information on
the toxic effects of selenium on fish and aquatic birds at the field sites
can be used to check the results and determine which method produced the most valid estimates of hazard.
Example 1.
The contaminated sites in Utah (Roadside Ponds and Sheppard Bottom) have produced acute and chronic selenium toxicosis and reproductive impairment in fish and aquatic birds (Waddell and Stanger
1992; Hamilton et al. 1996). The severity of these biological effects, as
evidenced by the presence of both acute and chronic toxicity, shows
that the hazard from selenium was high. The Protocol indicated high
hazard at both sites, whereas HQ indicated low to moderate hazard.
Example 2.
Decreased survival of juvenile ducks due to selenium in the diet and
associated decreases in waterfowl production were identified at Fernley,