Environmental Engineering Reference
In-Depth Information
To use an SSD method for criteria derivation requires a policy decision about
what percentile of the distribution translates to a protective concentration. The SSD
methodologies reviewed here, for example, derive criteria using the 5th percentile
of the distribution (Fig. 1). Some methodologies refer to this concentration as an
HC 5 (hazardous concentration affecting 5% of species). This is often interpreted to
mean that species lying above the 5th percentile in the distribution will be pro-
tected, and those species lying below this point will be harmed (e.g., Lillebo et al.
1988, in reference to the USEPA methodology 1985). However, Van Straalen and
Van Leeuwen (2002) provide a more accurate interpretation. They assert that the
HC 5 does not mean that 5% of species will be harmed. Rather, the HC 5 approach is
one for deriving a PNEC, and although the choice of the 5 th percentile is purely
pragmatic, it has been validated by field studies. Solomon et al. (2001), believe that
any percentile is appropriate if it can be validated against knowledge and under-
standing of ecosystem structure and function. Following is a discussion of percentile
cutoff values used by selected methodologies, why they were chosen, and whether
they have been validated.
The USEPA rationale for choosing the 5th percentile is that criteria values
derived using the 10th or 1st percentiles were thought to be too high and too low,
respectively; the 5th percentile was selected because it falls between 1 and 10
(Stephan 1985). Nonetheless, there is good agreement between USEPA criteria and
no-effect concentrations determined in experimental stream studies (USEPA 1991).
The Dutch guidelines (RIVM 2001) use the 5th percentile for derivation of MPC
values, and the 50th percentile for calculation of the SRC ECO . The reasons for these
choices are not given, but the 5th percentile has been validated against field NOECs,
in studies by Emans et al. (1993) and Okkerman et al. (1993). The Australia/New
Zealand guidelines (ANZECC and ARMCANZ 2000) consider the question more
rigorously, but still arrive at the 5th percentile level, because it works well in
the Dutch guidelines (RIVM 2001) and it gives TVs that agree with NOEC values
from multispecies tests. The reason for not regularly using a lower percentile is that
uncertainty is very high in the extreme tail of the distribution, and such uncertainty
may contribute more to the derived TV than does the data. However, the Australia/
New Zealand guidelines do use the 1st percentile as a default value for high con-
servation ecosystems, for bioaccumulative substances, and for cases in which an
important species is not protected at the 5th percentile level. To provide further
information to water quality managers in Australia/New Zealand, other percentile
levels are also calculated so that criteria are given based on the 1st, 5th, 10th, and
20th percentiles.
Other researchers have also found good correlation between criteria derived from
the 5th percentile of single-species SSDs and NOECs determined in multispecies
tests (Maltby 2005; Hose and Van Den Brink 2004; Versteeg et al. 1999). In con-
trast, Zischke et al. (1985) found that a laboratory-derived criterion concentration
of pentachlorophenol was not protective of invertebrates and fish in outdoor experi-
mental channels. Using only arthropod species, Maltby et al. (2005) determined that
concentrations of pesticides derived from the 5th percentile of SSDs (with 95%
confidence) were protective of freshwater ecosystems whether pesticide applica-
Search WWH ::




Custom Search