Methods for Deriving Pesticide Aquatic Life Criteria - Reviews of Environmental Contamination and Toxicology - page 77

of advisory values, but only for low molecular weight nonionizable organic
chemicals. There is no evidence that default ACR values are appropriate for
pesticides, in general.
All of the AF methodologies, with the exception of the Great Lakes Tier II pro-
cedure (USEPA 2003a), consider data for aquatic animals and plants together in
criteria derivation. The criterion is based on the most sensitive species, regardless
of such factors as taxon or toxicant mode of action. However, separate freshwater
and saltwater criteria are typically derived. There is an issue of whether taxa should
be pooled or not and it is of higher concern in SSD extrapolation procedures.
This will be discussed, in depth, later in this section.
Assessment factors are recognized to be a conservative approach for dealing
with uncertainty when risks posed by chemicals are being assessed (Chapman et al.
1998). Chapman et al. (1998) also note that application of empirically based factors
to toxicity data neither quantifies uncertainty nor reduces the probability of under-
estimating risk. Similarly, the use of AFs also greatly increases the possibility of
overestimating risk. Chapman et al. (1998) are very concerned that AFs are typically
applied generically, when they should be derived and used when specific factors
require it, such as the scale, frequency, and severity of potential environmental
insults, or the steepness of a toxicant's dose-response curve. In their conclusion,
Chapman et al. (1998) suggest the following principles be applied for using safety
factors: (1) data supersede extrapolation; that is, if data are available, they should
be used; (2) extrapolation requires context; use of safety factors should be based on
existing scientific knowledge; (3) extrapolation is not fact; estimates of effect levels
obtained using safety factors should only be used as screening values, not as threshold
values; (4) extrapolation is uncertain; safety factors should encompass a range
rather than being a single value; (5) all substances are not the same; safety factors
should be scaled relative to different substances, potential exposures, and nature of
effects; and (6) unnecessary overprotection is not useful; safety factors for individual
extrapolation steps should not exceed 10, and may be much lower.
Specifically addressing ACRs, Chapman et al. (1998) cited studies showing that
measured ACRs can vary from 1 to 20,000. In view of this, it is unreasonable to
apply a generic factor (10 or another number) across species and across substances,
as is often done in criteria derivation if no chronic data are available. The reality
remains, though, that adequate chronic data are generally not available and some
means of extrapolation is needed. If an ACR is developed according to the princi-
ples for the use of safety factors (described above), then it will be derived in the
context of the best scientific understanding of the substance and of the species
under consideration, and should be a better predictor of chronic toxicity than a
generic factor would be.
One possibility for reducing the need to use ACRs is found in the work of
Duboudin et al. (2004) who have proposed a novel way of directly using acute
toxicity data to determine a chronic HC 5 value. By using an acute-to-chronic trans-
formation procedure, derived from comparisons of acute and chronic SSDs within
species categories, an acute data set is transformed into a chronic data set, which is
then used to determine the HC 5 value.