Methods for Deriving Pesticide Aquatic Life Criteria - Reviews of Environmental Contamination and Toxicology - page 58

model to predict time to death, mean brood size, mean total young per female,
intrinsic rate of natural increase, net reproductive rate, and growth. They found
that energy-based biomarker measurements combined with measurements of
DNA integrity produced good predictions of population-level effects. Maboeta
et al. (2003) found a link between a biomarker and population effects in earth-
worms. Teh et al. (2005) found that Sacramento splittail suffered reduced survival
and growth, as well as cellular stress, after a 3-mon recovery period, following a
96 hr exposure to runoff from orchards treated with diazinon [ O,O -diethyl
O -[6-methyl-2-(1-methylethyl)-4-pyrimidinyl] phosphorothioate] and esfenva-
lerate [( S )-cyano(3-phenoxyphenyl)methyl (Α S )-4-chloro-Α-(1-methylethyl)benzenea-
cetate]. Although no significant mortality occurred during the 96 hr exposure
period, histopathological abnormalities were observed after a 1-week recovery
period in clean water. Although it appears that the histopathology may have pre-
dicted the population level effects in this case, no mechanistic link was made, and
it is possible that the reduced growth resulted from other factors.
Studies showing a predictive relationship between biochemical, behavioral, or
other nontraditional endpoints and population, community, or ecosystem level
effects are rare. Much more research is needed before nontraditional toxicity test
endpoints will be generally useful as predictors of ecosystem no-effect levels.
One criticism of using single-species toxicity data for derivation of water quality
criteria is that such tests are performed on a very limited number of species. For the
majority of species no toxicity data exists, which can be of particular concern where
threatened or endangered species are at risk of chemical exposure. This section
presents tools that have potential for using toxicity results from tested species to
predict potential for toxicity to untested species.
The concept of quantitative species sensitivity relationships (QSSRs) was
developed by Vaal et al. (1997a). They looked for patterns in sensitivity variation
among 26 aquatic species for 21 toxicants. Although species could be qualita-
tively grouped according to sensitivities (e.g., vertebrates were different from
invertebrates), no quantitative predictive model could be derived. The authors
noted that to further develop QSSRs, their findings need to be interpreted in terms
of toxicokinetics, modes of action, and relevant species characteristics. In another
study, Vaal et al. (1997b) found that acute lethality of nonpolar and polar narcotics
is highly predictable for a broad range of aquatic species. Reactive and specifi-
cally acting chemicals tend to be much more toxic, with very high variation in
sensitivity between species, and their toxicity is not predictable with current
information and models.
The USEPA has developed interspecies correlation estimation software (ICE v
1.0), which can be used to estimate acute toxicity of a wide array of compounds to
aquatic species, genera, and families that lack such data (USEPA 2003d). Toxicity
estimates made by interspecies correlations work well within taxonomic families,
but less well as taxonomic distance increases. The ICE models generate estimated
toxicity values, with confidence limits, to quantify uncertainty.
Vaal et al. (1997a, b) does not believe that QSSRs are sufficiently well developed
to be generally useful for estimating toxicity to untested species. The EPA ICE