Site-Specific Ecological Risk Assessment - Dealing with Contaminated Sites: From Theory Towards Practical Application - page 707

Den Besten et al. ( 1995 ) used differential weights in the ERA for aquatic systems
following a multi-criteria decision analysis. Effects on e.g. top predators and benthos
received a higher weight than parameters such as mouthpart (mentum) deformities.
This information was used to rank different sites according to their possible eco-
logical risks. For the terrestrial system, less experience is available. Semenzin et al.
( 2007 , 2008 ) and Critto et al. ( 2007 ) developed tabular decision matrices to address
the issue of weighting.
15.4.5 Reference Information
A crucial issue when analyzing the results of bioassays or ecological field obser-
vations in different tiers of ERA is the reference information . This information can
be gathered from reference sites, reference samples, or literature data. Of course,
analysis of reference sites and reference samples is preferred, since this optimizes
the site-specificity in ERA. Due to a lack of sites and samples, literature data may
partially substitute a lack of suitable references. Rutgers et al. ( 2008a ) recently pub-
lished reference data of soil system attributes for ten common combinations of land
use and soil type in the Netherlands, which may be used as a source of reference
information for ERA.
The issue of reference data is relevant for any line of evidence in the Triad
approach, i.e. chemical characterization (i.e. background levels in that region), tox-
icological data from bioassays (i.e. reference soil for quantification of the no-effect
level and control soil in order to verify the test performance) and ecological field
surveys (i.e. the ecological status of reference sites). The perfect reference response
resembles the response from the contaminated soil in all relevant aspects, besides
the effects from soil contamination. When a site contains gradients in soil character-
istics, also multiple references have to be gathered in order to reflect this gradient. To
reach this goal, parameters that may affect test performance, like the soil's texture,
pH, organic matter, humidity and available nutrients, should be verified between
contaminated and reference soils. Sometimes, information is available about the
influence of soil characteristics on test performance (e.g. Natal-da-Luz et al. 2008 ).
It often is a practical problem to identify matching soil samples. This problem has to
be tackled in a sensible way and hence should be considered and discussed in detail
before initiating the ERA. The lack of suitable reference sites in field surveys may,
however, statistically be solved by the use of multivariate techniques (e.g. Kedwards
et al. 1999 ), which relate the species composition and abundance to gradients of con-
taminant concentrations in soil, taking into account possible effects of other factors
('confounders'). However, such an approach needs the analysis of large numbers of
samples in order to account for all possible gradients that may shape the ecologi-
cal parameters in the survey (Rutgers 2008 ). Many software tools are available and
have increased the possibility to use powerful multivariate analysis, which use all
collected data to evaluate effects at a higher level of organization. Of course, in a
strict sense, causal inference of field effects from contaminants is impossible, due to
imperfect reference information (Boivin et al. 2006 ; Everitt and Dunn 2001 ; Jensen
and Pedersen 2006 ; Rutgers 2008 ).