Page 250 - Materials Chemistry, Second Edition

236 R.K. Rosenbaum et al.

providing cardinal impact measures. Among these methods are IMPACT 2002
(used in IMPACT 2002+) and USES-LCA (used in CML and ReCiPe). All these
methods adopt environmental multimedia, multipathway models employing
mechanistic cause–effect chains to account for the environmental fate, exposure and
effects processes. However, they do not necessarily agree on how these processes
are to be modelled, leading to variations in results of LCA studies related to the
choice of LCIA method. Based on an extensive comparison of these models fol-
lowed by a consensus-building process, the scientific consensus model USEtox
(UNEP/SETAC toxicity consensus model) was developed with the intention to
solve this situation by representing a scientifically agreed consensus approach to the
characterisation of human toxicity and freshwater ecotoxicity (Hauschild et al.
2008; Rosenbaum et al. 2008; Henderson et al. 2011). It has been recommended
and used by central international organisations like the United Nations Environment
Program UNEP, Society of Environmental Toxicology and Chemistry SETAC, the
European Commission and US-EPA to characterise human and ecotoxicity in
LCIA.
Among the existing characterisation models on midpoint level, three main
groups can be distinguished: (1) mechanistic, multimedia fate, exposure and effect
models, (2) key property-based partial fate models and (3) non-fate models
(EC-JRC 2011). According to ISO 14044 (2006b) “Characterisation models reflect
the environmental mechanism by describing the relationship between the LCI
results, category indicators and, in some cases, category endpoints. […] The
environmental mechanism is the total of environmental processes related to the
characterisation of the impacts.” Therefore, ecotoxicity characterisation models
falling into categories (2) and (3) do not completely fulfil this criterion. Caution is
advised regarding their use and most importantly the interpretation of their results,
which should not be employed without prior in-depth study of their respective
documentation. Having said that, depending on the goal and scope of the LCA, they
may still be an adequate choice in some applications, and indeed may agree quite
well with the more sophisticated multimedia-based models.
Ecotoxicity endpoint modelling is still in an early state and much research needs
to be performed before maturity is reached. The authors of the ILCD LCIA
handbook concluded that “For all the three evaluated endpoint methods (EPS2000,
ReCiPe, IMPACT 2002+), there is little or no compliance with the scientific and
stakeholder acceptance criteria, as the overall concept of the endpoint effect factors
is hardly validated and the endpoint part of the methods is not endorsed by an
authoritative body. […] No method is recommended for the endpoint assessment of
ecotoxicity, as no method is mature enough.” (EC-JRC 2011).
When interpreting the results of existing methods, it is important to keep in mind
that many aspects are not or only very insufficiently covered. This includes ele-
ments like terrestrial and marine ecotoxicity as well as toxicity of pesticides in
pollinators.
For further details see Chap. 40 and Hauschild and Huijbregts (2015).