Environmental Engineering Reference
In-Depth Information
Multispecies data are problematic for use in criteria derivation as a result of its pau-
city and variability. In contrast, there is much debate in the literature about whether or
not single-species toxicity tests are good predictors of ecosystem effects. Schulz and
Liess (2001) saw significant differences in fenvalerate [cyano(3-phenoxyphenyl)methyl
4-chloro-Α-(1-methylethyl)benzeneacetate] effects on caddisflies from both inter- and
intra-specific interactions. However, as previously discussed, single-species toxicity
tests can be successfully used in various extrapolation procedures to determine concen-
trations that are protective of ecosystems (Maltby et al. 2005; Hose and Van Den Brink
2004; Okkerman et al. 1993; Versteeg et al. 1999; Emans et al. 1993; USEPA 1991).
Crane (1997) reviewed use of multispecies, model ecosystem, tests for predicting
effects of chemicals in the environment. He concluded that more information is
needed on repeatability, reproducibility and predictive ability before such tests can
be used confidently for predicting environmental effects. Kraufvelin (1999) studied
Baltic Sea hard bottom littoral mesocosms and concluded that repeatability, repro-
ducibility and ecological realism of these mesocosms were poor enough to preclude
the use of such data in predictive risk assessment, or for extrapolation to natural
ecosystems. Sanderson (2002) reviewed the replicability of micro- and meso-cosms
and found that coefficients of variation (CV) averaged 45%, with large, outdoor
mesocosms averaging 51%. Also, 88% of biotic variables measured lacked statisti-
cally significant results, even with 3-4 replicates that should have yielded results
better than reported.
Another problem with field or semi-field studies is that they often have few
replicates because of unmanageable logistics. One of the reviewed methodologies
(OECD 1995) cites a SETAC-Europe document (1992) that asserts that unreplicated
experiments may be acceptable for responses that occur in a short period of time.
However, Hanson et al. (2003) found that, to detect a ≥ 25% change from control
values in microcosm exposures of
Myriophyllum
spp. to haloacetic acids, would
require between 2 and 21 replicates, depending on what endpoint is measured.
The question of how well single-species toxicity tests predict field effects has been
addressed by many researchers. As discussed previously, water quality criteria, derived
from single-species tests are protective of ecosystems in many cases. Borthwick et al.
(1985) showed that laboratory-derived NOECs were predictive of field effects of
fenthion on pink shrimp. Similarly, Crane et al. (1999) found that the response of
freshwater amphipods to pirimiphos methyl was the same, whether exposed in 250-mL
laboratory beakers or 50,000-L pond mesocosms. However, a caveat to that study is
that the amphipods were caged in the mesocosm study and, thereby, did not experience
the full effects of the mesocosm environment. While validating field predictions that
were based on laboratory-derived NOECs, Persoone and Janssen (1994) concluded
that, in general, NOECs derived from single-species laboratory studies relate well to
single- and multispecies NOECs derived from field studies.
Field or semi-field data are used in the Dutch methodology for comparison with
ERLs derived from single-species data (RIVM 2001), but are not used as input for
ERL derivation. Nonetheless, to be usable, the data must meet specific requirements.
Studies must show a distinct concentration-effect relationship, derive a reliable
multispecies NOEC, include several taxonomic groups, include at least two test
Search WWH ::
Custom Search