Page 697 - The Toxicology of Fishes

Fish Toxicity Studies 677

randomly assigned, including a control (and solvent or reference, if applicable) treatment, with repli-
cations at each treatment level. Observations are made for each treatment for each effect criterion and
are assumed to be independent. The assumptions of ANOVA (i.e., equal variances among treatments
and normally distributed data) must be met. Transformations of quantal data, proportions (e.g., percent-
age larvae deformed), or other response expressions are sometimes necessary to meet the requirements
of homogeneity of variance and normality. If the ANOVA results lead to rejection of the null hypothesis
(H = no difference between treatment means), the treatment (i.e., exposure concentration) had a
0
significant influence on mean response (i.e., biological endpoints measured), but we do not know which
means were significantly different. A series of post-ANOVA methods can be used to identify which
treatment concentrations differ from each other. When data or transformations of data violate assump-
tions of ANOVA, statistical power may be sacrificed by using nonparametric, post-ANOVA tests. For
a complete review of statistical procedures used for analyses of chronic, sublethal effects data, see
Newman (1995).
At the end of a chronic test, the LOEC and NOEC are determined for each endpoint measured. In
addition, the maximum acceptable toxicant concentration (MATC) is estimated for the endpoint with
the lowest NOEC and LOEC. The MATC is the threshold concentration of a chemical within a range
bounded by the NOEC and LOEC. For regulatory purposes, it is calculated as the geometric mean of
the LOEC and NOEC. The MATC has no statistical confidence interval because the LOEC and NOEC
are used to define it. The MATC should not be extrapolated to predict a safe concentration because it
is a reflection of test design, species, and exposure duration; however, a range of MATCs from chronic
tests with different fish species will provide more supportive data to extrapolate to biologically safe
concentrations, especially when natural waters are used for chronic exposures.
The MATC can also be used to generate application factors (AFs) for chemicals. The AF is derived
as the numerical value of the ratio of the MATC to the LC . The assumption is made that the AF for
50
a given chemical is constant over a range of test species; therefore, if an AF is derived for one species
with actual MATC and LC data, then the MATC could be derived for a second species. The AF for
50
the first species and the LC for the second species could be used to estimate the MATC for the second
50
species. The use of arbitrary application factors with LC values to predict chronic toxicity and safe
50
concentrations must be approached with care. The acute-to-chronic toxicity ratio (ACR) is a variant of
the AF and is the inverse of the AF and is also used to estimate an MATC for species when only acute
toxicity data are available.

Short-Term Sublethal Effects

In the development of test procedures to evaluate the toxicity of whole effluents (e.g., municipal or
industrial wastewaters) to aquatic organisms, the U.S. EPA developed short-term sublethal tests (i.e., 7
to 9 days or less; often misleadingly called short-term chronic tests) that focus on the most sensitive
life-cycle stages. See Dorn and van Compernolle (1995) for a summary of short-term toxicity tests for
whole effluents with fish. Fish are exposed to five different effluent dilutions (e.g., 100, 50, 25, 12.5,
and 6.25%), including untreated controls, with replication to determine magnitude of toxicity; however,
in compliance monitoring (i.e., in discharge permits) an option is to choose five concentrations that
bracket the receiving water concentration (above and below). This monitoring would determine compli-
ance status (i.e., meets or exceeds permit requirements) as well as estimate the NOEC.
In short-term sublethal tests, the LC , EC , NOEC, and LOEC are reported based on percent effluent
50
50
to which organisms are exposed. Effects may include changes in growth, survival, or percent hatch, for
example. To overcome some of the problems in statistically deriving the NOEC and LOEC, the inhibition
concentration (IC) may be used which is a point estimate interpolated from the effluent concentrations
at which measured effects occurred in the sublethal test. The IC is an estimate of the effluent concentration
(i.e., percent effluent) that would cause a given percent reduction in a biological endpoint of the test
organism. An IC for growth, for example, would represent the percent effluent at which a 10% reduction
10
in growth occurred. This approach is similar to determining the LC or EC when organisms are exposed
10
10
to a single chemical. Because the IC is a point estimate, a confidence interval can be calculated.