Site-Specific Ecological Risk Assessment - Dealing with Contaminated Sites: From Theory Towards Practical Application - page 714

test, arguments can be raised to use a different scaling, since avoidance is not
straightforwardly controlling earthworm populations in the field.

Approximation of Effects from Ecological Field Monitoring

The scaling of single variables from ecological field monitoring can follow
the same principles as bioassays (see above for an example). Typically, field
monitoring deals with multiple variables, for which the scaling issue is less
clear. The BKX-Triad algorithm to scale the results from multiple variables is
very simple and robust, and can be used for virtually any dataset (Jensen and
Mesman 2006 ), but has some drawbacks from unintended amplification in the
effect calculation when many variables are present:

log( x j )

n − 1

10

·

BKX-Triad = Effect = 1 −

i = 1 → n

With x the result of the observation i divided by the result from the ref-
erence observation and n is the number of observations at the site (or in
samples).
More sophisticated scaling is possible like the use of distance values in
multivariate space (e.g. Euclidean distance). This is a solution to the problem
of the BKX-Triad, i.e. all deviations from the references are adding to the total
calculated effect. Software tools for multi-criteria analysis are easily available,
but some training is necessary to use and interpret the data:
ED = k ( y ki − (ED R-C +ED C-C theor ) − 1 y kj ) 2 and: Effect = ED R-C ·

With ED is the Euclidian distance between site (or sample) i and j for k
dimensions. Subscripts R, C and C theor denote the reference, contaminated
and theoretically contaminated site to a 100% field effect.

After a proper scaling, the outcome of different lines of evidence should be in
balance. This balance will be theoretically demonstrated with a very large number
of tests addressing the set of end-points within each line of evidence. With a sub-
set of tests, like the test proposed in the Dutch Triad approach summarized above,
deviations from this balance should be expected and interpreted in terms of model
uncertainties. However, if the outcome of a subset already demonstrates conver-
gence of the results, then this is a strong basis for finalization of the ERA, providing
a solid advice for the Risk Assessment or Risk Management of the site. As a practi-
cal criterion for convergence, the deviation between the outcomes of different Triad
approach lines of evidence can be quantified as suggested by Jensen and Mesman
( 2006 ) and listed in Table 15.1 .