Methods for Deriving Pesticide Aquatic Life Criteria - Reviews of Environmental Contamination and Toxicology - page 96

Environmental Center (GLEC 2003). In Appendix C of that report, the authors present
results of various manipulations of a basic data set. First, with no change to the four
values (data points) used to calculate the FAV, simply increasing the number of samples
( N ), always increases the FAV, as a consequence of the variability in P values of the four
data points being reduced. Second, as the range of the four values increases (i.e., the
variability among the four data points increases) the FAV decreases, because of the
increased variability around the 5th percentile. The problem with the first of these kinds
of data set manipulations is that, in an effort to derive higher criteria by the USEPA
method, one could simply conduct more tests with insensitive species. Aside from caus-
ing the set to violate the log-triangular distribution assumption, such data would drive
the criterion upward in a predictable manner, based solely on N , because the new data
would not be near the 5th percentile. With other SSD methodologies (i.e., those that do
not ignore the upper part of the distribution), the best way to drive a criterion higher is
to have a large, balanced data set, such that the variability in the whole set is reduced.
By these other methods, if a data set were “padded” with extremely high or low values,
outliers and bimodal distributions would be detected, and the set would be modified to
fix these problems before the SSD analysis (ANZECC and ARMCANZ 2000; RIVM
2001; ECD 2003). To encourage generation of balanced data sets, SSD methods that
utilize all data (RIVM 2001; ANZECC and ARMCANZ 2000) are preferable to those
that focus on a limited range of the distribution.
Manufacturers and other waste dischargers have little incentive to generate data
if an AF method is used to derive criteria, because new data that show no or low
effects are ignored, while those that show high sensitivity drive a standard lower
(Whitehouse et al. 2004). This is because only data from the most sensitive species
are used to set criteria by AF methods. However, to the extent that AF methods
utilize variable factors, they do foster data generation, in that factors are smallest
for the most complete data sets, and smaller factors yield higher criteria values.
According to the Australia/New Zealand methodology, a high reliability TV can
be established on the results of three high quality field or mesocosm studies. There
is no stipulation that such TVs will only be used if they are lower than those derived
by extrapolation methods; multispecies research is therefore encouraged (ANZECC
and ARMCANZ 2000). In contrast, if adequate field or mesocosm data are avail-
able that indicate a FCV should be lower than the one calculated by the USEPA
methodology (1985), then the FCV can be adjusted. However, this does not encour-
age generation of field or semi-field data by all stakeholders, because the FCV can
only be adjusted downward, in this scenario.

8 Summary

Environmental regulators charged with protecting water quality must have scientifi-
cally defensible water quality goals. For protection of aquatic life, regulators need
to know what levels of contaminants a water body can tolerate, without producing
adverse effects. The USEPA has developed water quality criteria for many chemicals,