Ecological Risk Assessment of Diffuse and Local Soil Contamination Using Species Sensitivity Distributions - Dealing with Contaminated Sites: From Theory Towards Practical Application - page 645

SSD-modeling and HCp, by focusing specifically on this ranking interpretation.
When focusing on HC5s for ranking compounds, one benchmark contaminant (for
example: cadmium) can be chosen, and all other compounds can be characterized by
a statement on howmany times more or less toxic another contaminant is for (tested)
soil organisms relative to cadmium. That is: if the HC5 of cadmium is 5mg/kg dw
soil and that of contaminant X it is 15mg/kg dw , then the contaminant is estimated to
be three times less toxic. An even better way of such scaling could be provided when
having two benchmarks (such as on the scale of degrees Centigrade for temperature,
freezing and boiling of water). Such a ranking in itself tells us something about the
relative environmental burden of a contaminant when released in the environment,
as compared to the benchmark(s) (all relative to the tested sets of species).
The ranking of the relative environmental burden of a set of contaminanats
according to their HCp is illustrated in Fig. 14.6 , and is shown to illustrate the
robustness of this way of looking at HCp-estimates. Two independently collected
datasets of tested contaminants were chosen from literature. The sets were used to
obtain two HCp-estimates for each of the contaminants according to independent
protocols. The two assessments were done by Jänsch et al. ( 2007 ), using EC50s
as input data, and Swartjes ( 1999 ), using NOECs as input data. Hence, the assess-
ments differed by the choice of input data in a major way. The figure shows that both
assessments resulted in a similar rank order of the HC5-NOECs and HC5-EC50s for
soil organisms. There is also some variability (distances of point estimates from the
fitted rank order lines). As a consequence of the different input data, the lines dif-
fer from an equality line (slope 1, intercept 0). The HC5-NOECs of Swartjes are
generally lower than the HC5-EC50s from Jänsch et al. The findings imply that the

Comparisons of two independent data sets

R 2 = 0.96

100

10

1

0.1

0.01

R 2 = 0.85

0.1 1 10 100

HC5 (Jansch et al.) of EC50's

Fig. 14.6 Comparison of HC-values for a set of metals ( black circles ) and organic contaminants
( white squares ) from SSDs for soil organisms. The fitted regression lines (continuous and broken,
respectively) suggest a high similarity in ranking results between both methods. Deviations of the
markers from the lines identify statistical issues, like different choices of data selection criteria
and model choices. Data from Swartjes ( 1999 ) pertain to HC5-NOECs and of Jänsch et al. ( 2007 )
pertain to HC5-EC50s. Therefore, the estimates generally differ by approximately one order of
magnitude (in accordance with an often-found acute-to-chronic ratio of 10)