Ecological Risk Assessment of Diffuse and Local Soil Contamination Using Species Sensitivity Distributions - Dealing with Contaminated Sites: From Theory Towards Practical Application - page 662

of the actual bioavailable fraction. Methods for that are described in Hodson et al.
( Chapter 16 of this topic), and in De Zwart et al. ( 2008b ).
The third step is to aggregate the estimated PAF-values of each compound
within groups of contaminants with similar Toxic Modes of Action. This is done
by a Concentration Additivity model . In this modeling approach, each contaminant
within a group is considered a “dilution” of the others, so that concentrations of all
contaminants can be expressed as toxicity-normalized concentrations. This yields
multi-substance PAF-values ( msPAFs ) for each mode of action represented in the
contaminated soil.
The fourth step is to aggregate the within-group msPAF-values and the remaining
PAF-values for contaminants not assigned to a group, to obtain an overall msPAF.
This is done by a Response Additivity model .
As a result, a Conventional Risk Assessment for a mixture assessment using
SSD-modeling yields a single estimate of the net, Potentially Affected Fraction
of species for a site . This can be presented with or without confidence intervals.
Such a single value is very important for the ranking of sites and also for the diag-
nostic analysis of Biomonitoring data (see Fig. 14.5 ). Note that this procedure, in
contrast to the derivation of standards, never imposes an uncertainty factor on a
PAF-estimate.
Many theoretical objections can be generated to this way of quantifying mixture
risks. As for SSDs themselves, there is no real mechanistic underpinning of the
approach. However, mathematical analyses of the models that are used have revealed
that:
the numerical values of mixture risks are grossly similar for the mixed-model approach,
in comparison with the Concentration- and/or Response-additivity models, as long as the
slopes of the concentration-response models are “moderate” (see Drescher and Bödeker
1995 ).
This mathematical rule of thumb suggests a certain robustness of msPAF as an
estimate of expected impacts, rather insensitive to modeling choices. In line with the
robust findings for single-species mixture tests, we object to assessing single con-
taminants one by one in retrospective Risk Assessments of contaminated sites, since
the numerically robust output of the mixture models (CA, RA, or the mixed-model
approach) seem (1) conceptually more justifiable than a “no addition” approach, and
(2) seem better linked to field effects than contaminant-by-contaminant assessments
(see Fig. 14.5 ).
14.10.7 When the Environmental Problem is Refined: Tiers
for SSDs
The concept of tiering has been introduced in Swartjes ( Chapter 1 of this topic).
Within the context of tiered assessment, the SSD approach can be considered a
lower-tier approach. A common approach that is simpler than SSDs is to use a
NOEC, and divide this by a fixed uncertainty factor (e.g. a NOEC divided by 10