Methods for Deriving Pesticide Aquatic Life Criteria - Reviews of Environmental Contamination and Toxicology - page 85

A computer program is available for making these calculations (RIVM 2004).
HC 5 values are used as MPCs, which are used to derive EQSs. A NC, which
serves as an EQS target value, is equal to the MPC/2. HC 50 values are used as
SRC ECO , which are EQS intervention values (i.e., the ecosystem is seriously threat-
ened because 50% of species are adversely affected).
Denmark ( Samsoe-Petersen and Pedersen 1995). The Danish methodology
utilizes the SSD method of Wagner and Løkke (1991), which is essentially the
same as that used in the Netherlands (RIVM 2001), but is stated differently and
only calculates a lower one-sided confidence limit 5th percentile value. A value
called a protection concentration ( K p ) is calculated as follows:

K = exp( x − s k ) (2.12)

•

p

where
K p = concentration protecting (100-p)% of species with a specified level of
confidence
p = percentile cutoff level
x = mean of log EC or log NOEC data
s = standard deviation
k = one-sided tolerance limit factor for a normal distribution depending on
chosen confidence level (from Wagner and Løkke 1991)
The SSD method is only used in Denmark to estimate water quality criteria.
An AF method is preferred, and is given more weight in deriving criteria.
Australia/New Zealand (ANZECC and ARMCANZ 2000) . The Australia/New
Zealand guidelines use the same method as do the Dutch, but with a curve-fitting
procedure that overcomes the problem of data that do not fit an assumed distribution.
Using the program BurrliOZ v. 1.0.13 (CSIRO 2001; Campbell et al. 2000), data are
first fitted to one of a family of Burr distributions (Burr 1942; the log-logistic distri-
bution is in the Burr family). After an appropriate distribution is chosen, then the
calculation of the median HC 5 value is the same as shown for the Dutch methodology
but utilizes extrapolation factors ( k ) derived for each of the distributions.
EU Risk Assessment Guidelines (ECB 2003). Similar to the Australia/New
Zealand approach, the EU TGD (ECB 2003) utilizes the Dutch SSD procedure, but
with the provision that the distribution that best fits the data should be used. Either
the Anderson-Darling or Kolmogorov-Smirnov test may be used to check goodness
of fit. The PNEC is calculated as follows:

5% SSD(50% c.i.)

PNEC = (2.13)

AF

where
PNEC= predicted no-effect concentration
5% SSD = concentration determined from SSD expected to protect 5% of
species
50% c.i . = 50% confidence interval
AF= assessment factor of 1-5