Ecotoxicity reference values for classification of data rich substances case: metals

1
Ecotoxicity reference values for
classificationof data rich substancescase
metals

• Dr. Frank Van Assche
• Environment Manager
• International Zinc Association - Europe
• 168, Avenue de Tervueren, B-1150 Brussels,
  Belgium

2
contents

• Data rich versus data poor issues
• Derivation of the Ecotoxicity Reference value
  (ERV)
• Data selection
• Data aggregation
• Derivation of the ERV
• Conclusions

3
Issues 1. Data rich versus data poor

• For many base metals extensive data sets are
  available on acute and chronic ecotoxicity
• Standard and non-standard tests
• Test protocols not always well defined
• Test conditions may not be relevant
• How to handle these extensive data sets?
• Grouping of data
• Derivation of the ERV

4
Issue ecotoxicity data at different pH

• For classification of metals and sparingly
  soluble metal compounds, combination of the
  ecotox data with T/D data requires ecotoxicity
  information between pH 6 and 8.5
• The vast majority of ecotox tests is carried out
  at pH 7.5-8.0
• Scarce data at other pHs
• ? ecotoxicity data at required pH ?

5
Data compilation selectionreliability

• Type of test
• For RA standard and non-standard tests
• for classification standard tests (e.g. OECD,
  ASTM) are preferred, if available
• Quality
• Description of test material methods (life
  stage of organisms, experimental design,)
• Description of physico-chemical test conditions
  (tolerance limits, bioavailability parameters,
  culture conditions)
• Chemical analysis (measured concentrations are
  preferred especially when effects conc. are close
  to background conc.)
• Concentration-effect relationship (statistics,
  test concentration interval ( factor of 2)

6
Data compilation selection reliability (2)

• Derivation of toxicity values
• L(E)Cx are preferred over NOEC
• Cut-off level (x) should represent a low effect
  percentile depends on the test variability
• No extrapolation below concentration range
  (introduces uncertainty)
• NOEC/L(E)Cx values should be estimated using
  appropriate statistics and experimental design
• Use of unbounded NOEC/LOEC could be considered in
  specific cases, e.g. when no other data are
  available

7
Data compilation selection relevancy

• Biological relevance of endpoints (survival,
  growth, are preferred)
• Relevancy of test substance (impurities!)
• Relevancy of species
• endemic and non-endemic could be used
• focus to standard species in classification
• Relevancy of test medium
• natural artificial test media, conditions must
  be related to transformation/dissolution media
  for classification
• relevancy of culture medium (metal background,
  Cb)
• Acclimatisation of test organisms prior to
  testing should be avoided.
• Cb in culture medium gt deficiency level

8
Changing zinc sensitivity of Raphidocelis
subcapitata as a function of culture conditions
Culture conditions prior to test
9
Data aggregation

• 1. Grouping of data
• grouping per species/endpoint
• grouping according to physico-chemical properties
  such as pH (or normalized using bioavailability
  models)
• 2. Geometric mean ( 2)
• 3. Lowest value based on different endpoints
• 4. Most sensitive life stage

10
Derivation of ERV-aggregation-lowest value
Daphnids tox data pH 6a
Fish tox data pH 6a
11
Derivation of ERV-lowest value-bioavailability
normalization
12
Is it correct to use the lowest ecotox value for
ranking of data-rich substances?
Because the data follow a distribution, the more
data, the higher the chance to have a lower
lowest value
1
2
3
Dataset 1
Dataset 2
4
1
2
5
6
3
7
9
8
4
1
2
5
6
3
11
2
10
7
Dataset 3
EC50
Need for a system for setting ERV that overrules
this effect
13
Derivation of the ERV use the distribution of
species sensitivities

• SSD percentiles are less dependent on number of
  tests
• ERV to be based on an X percentile of the SSD
• X percentile corresponding to lowest value at
  minimal datasets

14
Conclusions

• The derivation of a reference value for data-rich
  metals requires specific attention related to
• Data selection
• Data handling
• Data can be normalised to required pH to match
  T/D data
• The approach for ERV setting should be comparable
  with data-poor substances