The data on the toxicity of nickel to freshwater organisms were compiled from three main sources: open literature, internationally recognized databases (e.g., Science Direct, Web of Science), and industry-sponsored research programs. A large dataset on the chronic ecotoxicity of nickel to freshwater organisms was com-piled. All gathered data were further screened using the criteria as outlined in Data Quality Screening.
Data Quality Screening
Each individual ecotoxicity data point was screened for quality before incorporation in the nickel ecotoxicity database based on the following criteriai:
- data were retained for the following groups of organisms: algae, higher plants, invertebrates, amphibians, and fish;
- data covered the following relevant endpoints: survival, growth, and/or reproduction;
- Ni-only exposure data were considered relevant (studies were rejected if indications of impurities or other substances might have an effect on the toxic properties of nickel);
- the results reported measured pH, hardness (Ca and Mg concentration), and dissolved organic carbon (DOC);
- the range of the physico-chemistry of the test media (pH, hardness, DOC) were within the range of the developed/validated bioavailability models (BLMs), (Fact Sheet 4);
- the data were from studies conducted according to approved international standard test guidelines (however, data from non-standardized tests were also assessed);
- only long-term or chronic toxicity data, involving endpoints that are realized over periods of several days to years depending on the organism, were used;
- the tests were performed according to standard operational procedures, with a detailed description of the methods em-ployed during toxicity testing;
- preference was clearly given on the use of measured nickel concentrations in the test concentrations;
- a clear concentration-response was observed;
- toxicity threshold values calculated as L(E)C10 (the concentration that causes 10% effect during a specified time interval) values were preferred; however, NOEC values (No Observed Effect Concentration) were seen as equivalent;
- the toxicity tests were performed with soluble nickel salts (e.g., NiCl2 and NiSO4);
- the toxicity test results reflected dissolved nickel concentrations and were expressed as µg Ni/L; and
- ecotoxicity threshold values were derived using the proper statistical methods.
Only identified ecotoxicity data fulfilling the above mentioned criteria were used for the freshwater PNEC derivation.
Applying the above mentioned quality screening criteria to the identified ecotoxicity data resulted in the selection of an extensive high quality database on the ecotoxicity of nickel to freshwater organisms. Indeed, the database comprised 31 different “species means” for 19 different families from 214 individual high quality L(E)C10/NOEC values [58 individual L(E)C10/NOEC values for algae, 6 for higher plants; 113 for invertebrates; 37 for fish/amphibians]. An overview of accepted individual high quality chronic ecotoxicity data is presented in the Environmental Risk Assessment of Nickel and Nickel Compounds (EU Risk Assessments).
Incorporation of Bioavailability (Data Normalization)
There is extensive evidence demonstrating the importance of bioavailability and water quality conditions on the toxicity of nickel to aquatic organisms. Indeed, site-specific geochemical conditions (e.g., pH, hardness, DOC) influence the degree to which organisms take up nickel and exhibit adverse effects. From a risk assessment perspective, it is critical to consider bioavailability, as geographically distinct eco-regions, water-sheds, and sites will often show distinctive geochemical characteristics leading to different critical effects concentrations (PNECs). For further guidance, see Fact Sheet 4 entitled Bioavailability Models for the Freshwater Compartment.
Normalized high quality ecotoxicity data are grouped/aggregated in order to avoid over representation of ecotoxicological data from one particular species. The following major rules were used to aggregate data:
- If several chronic NOEC/L(E)C10 values based on the same toxicological endpoint were available for a given species, the values were averaged by calculating the geometric mean, re-sulting in the “species mean” NOEC/L(E)C10.
- If several (geometric mean) chronic NOEC/L(E)C10 values based on different toxicological endpoints were available for a given species, the lowest (geometric value) value was se-lected.
After the data aggregation step, only one ecotoxicity value (i.e., the geometric mean for the most sensitive endpoint) was assigned to a particular species.
Calculation of PNEC Using Statistical Extrapolation Methods
Estimation of the HC5 from the Species Sensitivity Distribution
When a large data set for different taxonomic groups is available, the PNEC can be calculated using a statistical extrapolation method. In this approach, the ecotoxicity data are ranked from low (most sensitive species) to high (least sensitive species). A species sensitivity distribution (SSD) was then constructed by applying an appropriate curve fitting distribution (usually a log-normal distribution) to the normalized high quality aggregated chronic toxicity data (Aldenberg & Jaworska, 2000). From the SSD, a 5th percentile value (at the median confidence interval) is calculated (i.e., median HC5) using the software program ETx, as described by Van Vlaardingen et al. (2004).
Selection of Appropriate Assessment Factor and Derivation of the PNEC
To account for uncertainty, an assessment factor (AF) may be applied to the median HC5. In general, such AFs vary between 1 and 5 and are determined on a case-by-case basis. The freshwater PNEC would therefore be calculated as follows:
freshwater PNEC = median HC5/AF
Based on the available chronic NOEC/L(E)C10 data, the following points were considered when determining the AF:
- The overall quality of the database and the endpoints covered (e.g., are all the compiled data representative of “true” chronic exposure?)
- The diversity of the taxonomic groups (Table 1) covered by the database (e.g., do the databases contain, at a minimum, organisms belonging to the eight taxonomic groups as defined by the 2001 London workshop?)
- The number of species (e.g., does the SSD cover at least 10 different L(E)C10/NOECs and preferably more than 15?)
- Use of bioavailability models and approach for bioavailability correction [e.g., do the bioavailability models (see Fact Sheet 4) allow the toxicity data for all species to be normalized?]
- Statistical extrapolation (e.g., how well does the SSD fit the toxicity data?)
- Comparisons between field and mesocosm studies and the PNEC (e.g., is the PNEC value protective for the effects observed in mesocosm/field studies?)
In the Nickel EU RA, no mesocosm/field data were available that allowed the determination of threshold concentrations of nickel in freshwaters under field conditions. All other identified criteria were fulfilled. Therefore, based on the weight of evidence, the Danish Rapporteur proposed to use an AF of 2.
||Fish (usually tested species like salmons, bluegill, channel catfish, etc.)
||A second family in the phylum Chordata (fish, amphibian, etc.)
||A crustacean (e.g., cladoceran, copepod, ostracod, isopod, amphipod, crayfish, etc.)
||An insect (e.g., mayfly, dragonfly, damselfly, stonefly, caddisfly, mosquito, midge, etc.)
||A family in a phylum other than Arthropoda or Chordata (e.g., Rotifera, Annelida, Mollusca, etc.)
||A family in any order of insect or any phylum not already represented
i The application of the quality screening criteria would also apply in case additional or new ecotoxicity data would be considered.