Methods for Deriving Pesticide Aquatic Life Criteria - Reviews of Environmental Contamination and Toxicology - page 62

substance, then the geometric mean of values is used. Acute data, which are used
to derive moderate reliability TVs, are reduced using the same approach.
Data reduction may also be necessary for outliers. Although the USEPA
(1985) has vague advice regarding data that should be excluded, the Australia/
New Zealand guidelines provide clear instructions on how to deal with outlying
data (ANZECC and ARMCANZ 2000). First, for excessively high or low data
points, original papers are consulted in an attempt to explain the variation (e.g.,
differences between nominal and measured concentrations, water quality factors,
and errors). If data are bimodally distributed, only the lower of the two groups is
retained. The use of curve-fitting statistical extrapolation models reduces the
need to remove outliers when using this methodology. Data are excluded if they
are from unpublished studies or derived from studies with excessively wide test
concentration ranges.
Because toxicity data for a given chemical may be available in different forms
(i.e., NOEC, LOEC, LC x /EC x ), for different exposure durations, and for different
endpoints, it is necessary to provide some guidance for selecting or standardizing
values for use in criteria derivation. Instructions should also be provided on how to
reduce multiple data, for a given chemical/species combination, to a single value,
as well as for how to manage bimodal distributions and outliers.

7 Criteria Calculation

In this section, we address how criteria values are calculated by different meth-
odologies. Exposure factors that affect toxicity are reviewed, because they may
influence how criteria are derived or expressed. Assessment factor and statistical
extrapolation methods are described and evaluated and details of criteria calcula-
tions are given. Finally, other considerations in criteria derivation are discussed,
including chemical mixtures and multiple stressors, bioaccumulation and secondary
poisoning, threatened and endangered species (TES), harmonization of criteria
across environmental compartments and utilization of data.

7.1 Exposure Considerations

Typically, only effect assessments, and not exposure assessments, are used to derive
water quality criteria. However, an exposure component can be included with the
effects assessment.
Water quality criteria that adequately protect aquatic life must consider expo-
sures of varying magnitudes, durations and frequencies. A criterion designed to
protect against ongoing, chronic toxicant exposure will overprotect if only brief,
mild excursions are encountered above the criterion. Similarly, it will be underpro-
tective if brief, but large excursions are encountered. Such situations actually