See Guidance: Data Compilation
Data Quality Screening
The quality screening criteria as defined in Guidance: Data Quality Screening were applied to select the high quality chronic ecotoxicity data of nickel to marine organisms.
An overview of all accepted individual high quality chronic ecotoxicity data is presented in the Environmental Risk Assessment of Nickel and Nickel Compounds (see EU Risk Assessments).
No bioavailability correction tools are available for the marine aquatic compartment, and therefore, no normalization of the toxicity data are performed for the marine environment.
The selected individual high quality chronic ecotoxicity data of nickel to marine organisms are aggregated according to the criteria mentioned in Guidance: Data Aggregation. An overview of the non-normalized species mean NOEC/L(E)C10 value for the most sensitive endpoint is provided in Table 1. Notably, marine fish (Cyprinodon variegatus and Atherinops affinis) were the least sensitive group to nickel exposure. Annelids (Neanthes arenaceodentata), molluscs (Haliotis rufescens), and crustaceans (Mysidopsis sp.) were among the most sensitive marine organisms.
SSD Construction and Median HC5 Derivation
The species mean NOEC/L(E)C10 values in Table 1 were further ranked from low to high. Subsequently, the conventional log-normal distribution was fitted to the ranked toxicity data using the ETx model. However, no significant fitting of the toxicity data was achieved using this model. Therefore, an alternative approach was elaborated to evaluate several other curve fitting functions for the SSD. In this “weight-of-evidence” approach, only the statistically significant parametric distribution functions for the available toxicity data set and the non-parametric “flexible kernel density estimation” were selected for the final PNEC derivation. Subsequently, the median HC5 values were calculated from the different parametric and non-parametric distribution functions, resulting in a range of median HC5 values between 5.3 and 25.4 μg/L (mean value of 19.9 μg/L) for the parametric distribution functions and a median HC5 value of 14.5 for the non-parametric distribution function. From the discussions with the Member States and the Rapporteur, it was decided to use the mean value of 19.9 μg/L and 14.5 μg/L, i.e., 17.2 μg/L as the final median HC5 value for PNEC derivation. The parametric distribution functions and the mean of median HC5 value, calculated using the @Risk software (Version 5), for the non-normalized ecotoxicity for nickel are presented in Figure 3.
The non-parametric distribution functions and the mean of median HC5 value for the kernel distribution function for the non-normalized ecotoxicity for nickel are presented in Figure 4.
An AF of 2 is applied to the mean of the median HC5 value resulting in a marine aquatic PNEC = (mean of) median HC5 / 2 = 17.2 µg/L / 2 = 8.6 µ/L.