Environmental Engineering Reference
In-Depth Information
kind of mechanistic justification for using the model: a statistical test may just sig-
nal misfit, in the case of which the outcomes of SSD-modeling might be practically
doubted for that reason alone.
In the case of misfit, there is a scientific trigger to reconsider the data. Are there,
for example, two subsets of species (a sensitive and a non-sensitive subset), as in
the case of bimodal SSD for an insecticide, which was tested for a wide array of
species? In such cases, the misfit might trigger deliberate choices of the use of
data. Eventually, this may imply deriving an HC5-risk limit on the basis of “target
species” only, or the assessment of risks for various subgroups of species. Examples
of sensitivity differences between subgroups of species are provided by Frampton
et al. ( 2006 ). In this case, a refined Risk Assessment with SSDs is possible (see
Section 14.10.7 ).
Alternatively, the lack of fit may be due to one or more low quality data. In par-
ticular, the set of test data may contain tests in which the tested species are exposed
in conditions that are unnatural and stressful for the test species (Jänsch et al. 2005 ).
For example, when earthworms are tested in a sandy soil at low pH, the sensitivity
for the contaminant may be increased due to the extra stress from the acidic soil
conditions. It may be wise in such cases to remove the “extra stress” data before
making an SSD for the Risk Assessment.
14.10 Other Issues in SSD Modeling and Interpretation
14.10.1 Comparison of Hazard Indices and PAF
After setting a quality standard, sites may be evaluated against that standard, yield-
ing insight in the presence or absence of exceedances in different soil samples (a
dichotomous outcome). In soil quality assessment, one can moreover derive a quan-
titative Hazard Index, as ratio of the actual soil concentration and the standard.
When the index exceeds the value of unity, then this indicates a regulatory prob-
lem, signifying potential risks, but the resulting number (Hazard Index >1) does not
provide information beyond the number of times the criterion is exceeded, usually
interpreted as degree of seriousness of contamination. Especially if one compares
sites with different contaminants on the basis of quantitative indices, one should be
aware of the underpinning of the indices. Are they based on an SSD, or on e.g. a
lowest NOEC divided by an uncertainty factor of 10, 100 or 1000? Usually, this is
information that is hidden when quality standards are used by practitioners. This
matters a lot, since the latter implies that the indices for different compounds are
of different meaning. Moreover, since the maximum impact is 100% of species
affected, index values are not a good approach to quantify expected impact mag-
nitudes. The reasons for this are illustrated in Fig. 14.10 , and further discussed by
Klepper et al. ( 1998 ) and Solomon and Takacs ( 2002 ).
As an improvement, PAF-values may be used instead of Index values. The advan-
tages of using PAF instead of an Index in Conventional environmental assessments
Search WWH ::




Custom Search