European perspective on metals´ bioavailability research and implementation of the Biotic Ligand Model (BLM) into regulatory frameworks Karel De Schamphelaere Bioavailability of metals seminar – 18 October 2007
BLM IN THE REAL WORLD Karel De Schamphelaere Bioavailability of metals seminar – 18 October 2007
Scientific EQS approach for metals EQS = HC5 based on Species Sensitivity Distribution (SSD) Metals (Cu, Zn, Ni, Cd) very data rich NOEC/EC10 available for species Potential pitfall: NOEC/EC10 obtained in test media with widely varying chemistry (= very different bioavailability) Generic/uncorrected SSD does not represent ‘intrinsic sensitivity’ alone but rather a mix of ‘intrinsic sensitivity’ + bioavailability effects Need models to perform bioavailability normalization of NOEC/EC10 to site/region specific water chemistry before SSD and HC5 estimation e.g., Biotic Ligand models (BLM)
MeOH + MeCO 3 Me- DOC pH [Me] on ‘biotic ligand’ Toxic effect WaterOrganism H+H+ pH Me 2+ Ca 2+ Na + Mg 2+ ‘biotic ligand’ e.g. gill, cell surface Speciation (WHAM) Intrinsic sensitivity Competition (log K’s) Log K CaBL Log K MgBL Log K NaBL Log K HBL
Overview of available models Standard test organisms CuZnNiCd AlgaeXXX- DaphniaXXXX FishXXXX Ceriodaphnia--X- Cu, Zn, Ni: BLM models or similar taking into account the effects of DOC, pH, hardness (Ca+Mg), Na, alkalinity Cd: Bioavailability correction based on hardness-toxicity relation for 3 species and 7 datapoints (applied to all species) HC5 (µg Cd/L) = 0.09 x (Hardness/50)
BLM’s are validated in field waters Factor 10 to 30 variability of toxicity > 90% of prediction errors < factor 2
What is normalization with BLM? Principle = NOEC alg with algae-BLM, NOEC invertebrate with Daphnia-BLM, NOEC fish/vertebrate with fish-BLM =Refinement compared to hardness-Cd toxicity correction NOEC species A (µg/L) Test water X (pH x, DOC x, Ca x ) Site water Y (pH Y, DOC Y, Ca y ) BLM NOEC species A [Me-BL] NOEC species A (µg/L) For site water Y BLM Intrinsic sensitivity
SSD and HC5 Plot normalized NOEC’s according to increasing probability
SSD and HC5 Plot normalized NOEC’s according to increasing probability Fit statistical distribution (SSD)
SSD and HC5 Plot normalized NOEC’s according to increasing probability Fit statistical distribution (SSD) Calculate HC5(50%) HC5 = 25 µg Zn/L
SSD and HC5 Plot normalized NOEC’s according to increasing probability Fit statistical distribution (SSD) Calculate HC5(50%) HC5 = 25 µg Zn/L HC5 = 168 µg Zn/L HC5 increases substantially with increasing pH, DOC and hardness Bioavailability matters!
REAL WORLD ISSUES
Real world issues Limited number of BLM’s (for standard species) Extrapolation to other species? (“non-BLM species”) Lab to field extrapolation? Species vs. communities? Conservatism? Models have boundaries What to do outside boundaries? Extrapolate BLM’s? How to implement in regulation? Consequences + practicalities
ISSUE 1 Limited number of BLM’s Extrapolation to other (non-BLM) species? A few examples
Same effect of pH on chronic toxiity of Cu 2+ for 4 species of algae (slope ~ 1.4)… …and 3 different endpoints (growth, biomass, P-uptake) Extrapolatable! Natural waters? BLM Cu algae De Schamphelaere & Janssen (2006) ES&T 40,
Typically: factor 10 to 30 variability in toxicity > 90% of prediction errors < factor 2 BLM Cu algae in natural waters
From Cu VRAR report (2007) Supports extrapolation of BLM’s across species Reduction of variability in NOEC data from literature Fish-BLM Daphnia BLM Alga-BLM
A single BLM can be used to effects of pH, hardness, and DOC on acute and chronic Ni toxicity to rainbow trout and fathead minnow Extrapolation possible! Ni-BLM fish Deleebeeck et al. (2007) Ecotoxicology and Environmental Safety 67: 1–13
Very similar pH slope for Zn among two algae species Can be extrapolated! Zn BLM algae De Schamphelaere et al. (2005) Environ Toxicol Chem 24: Wilde et al. (2006) Arch. Environ. Contam. Toxicol. 51: 174–185
Much evidence that Cu-BLM’s for all trophic levels can be accurately extrapolated (see also additional evidence in Cu-VRAR documents) Clear evidence that Ni-BLM for fish may be extrapolated to non-BLM fish Results of a comprehensive “spot-check” study indicate that BLM’s for other trophic levels may also be extrapolated (this issue is still under discussion at TC-NES) Clear evidence that Zn-BLM for algae may be extrapolated to other algae Although there is no toxicity-based evidence for invertebrate and fish Zn- BLM’s, extrapolation may possibly be justified on the basis of: Very similar mode of action (disruption of Ca-balance) Ca is most important protective cation BLM-constants (log K’s) of fish and Daphnia are very similar Clear need for toxicity-based research to test applicability of extrapolation Extrapolation: conclusions & outlook
ISSUE 2 LAB TO FIELD EXTRAPOLATION
Three high quality mesocosm studies Estimate HC5 based on NOEC values for the species within the mesocosm experiment = observed HC5 Estimate HC5 based on SSD with single-species literature toxicity data normalized to mesocosm chemistry (pH, DOC, Ca, …) = predicted HC5 Compare observed vs. predicted HC5 Example 1: Cu mesocosm data
From Cu VRAR (2007) – arrow reflects uncertainty due to non-equilibrium HC5(observed) from 3.4 to 19.6 µg/L Good agreement between observed and predicted HC5 SSD+BLM methodology for Cu seems appropriate for accurate protection in the field Example 1: Cu mesocosm data
Conducted for the UK Environment Agency Research consortium of Centre for Ecology and Hydrology (UK), UGent (B), Univ. Antwerp (B), Univ. Wageningen (NL) Monitoring of full chemistry, invertebrate and diatom community composition, metal bioaccumulation in invertebrates, Toxicity Identification Evaluation for reference and metal contaminated streams (n=35) Aims: To investigate if water chemistry and bioavailability should be taken into account when looking at ecological, community-level effects in the field To investigate if current and proposed EQS methodologies are adequate for protecting field communities Example 2: UK EQS project
Chemical analyses (dissolved metals, DOC, pH, major ions, alkalinity, etc.) UK EQS project - concept Physical site characterization (width, depth, stream velocity, etc.) RIVPACS MODEL Expected No. of TAXA present in stream Ecological analyses (invertebrates, diatoms) Observed No. of TAXA present in stream Predicted HC5 and % affected species BLM+SSD Observed/Expected No. of TAXA Agreement? Conservatism?
Chemistry clearly influenced how metals affect community composition Both speciation and competition effects seemed important The importance of metal mixtures in the field could not be dismissed Regression analysis suggested that ecological effects in non-acidic sites (pH>6) could best be explained in terms of contamination by Zn and/or Al and/or Cu and/or a mixture of these elements, although Cd could not be excluded either due to its correlation with Zn Under these circumstances: predictive capacity of Zn- BLM + SSD approach for effects observed in the field? UK EQS project – Main Results
Ecological effects are significantly correlated to exceedence of HC5(Zn) 7 sites correctly classified as non-impacted, 12 sites correctly classified as impacted, 6 false-negatives, 4 false-positives Mean (Zn/HC5) vs. field effects Preliminary calculations – do not quote ≥ 0.79 RIVPACS Class A quality
Interpretation In general: ecological effects in the field can be related to exceedence of thresholds (HC5) based on laboratory-based ecotoxicity data, normalized for bioavailability False-positives can be due to: Over-conservative HC5 Tolerance acquisition of local communities False-negatives can be due to: Under-conservative HC5 Temporary exceedence of HC5 (in two out of six cases) Toxicity contribution from other metals, including Al (mixture effects) In order to understand better the ecological effects of metal contamination in the field: mixture toxicity needs to be understood
Preliminary approach for metal mixtures toxicity in the field Assume that organisms consist of a set of binding sites relevant for accumulation and toxicity with which all metals and competing cations react (cf. BLM) in a similar way as humic acid reacts with all metals and cations Then: the total amount of all metal calculated with WHAM VI to be bound to HA (mol/g) could potentially be related to accumulation and effects The Toxicity Binding Model (TBM) Two examples: Metal accumulation in bryophytes Toxicity to P. subcapitata in the field samples
Metal accumulation in bryophytes Metals in bryophytes agrees fairly well with WHAM VI calculated metal binding to HA proof of principle that mixture-BLM is possible
Metal toxicity to algae in field samples Ftox = [metal bound to HA]/[metal’s specific toxicity] TBM approach is also promising for predicting metal mixture toxicity
Main conclusions UK EQS project Chemistry (both speciation and competition) seemed to be important for ecological effects of metals in the field As shown for Zn, ecological effects in the field can be related to exceedence of thresholds (HC5) based on laboratory-based ecotoxicity data, if normalized for bioavailability Metal mixtures in the field are a reality The TBM shows that BLM-like approaches might be valuable for taking mixture effects into account Final report expected soon (end 2007) Further information: UK Environment Agency (Paul Whitehouse)
ISSUE 3 MODELS HAVE BOUNDARIES
Boundaries within which bioavailability models for three trophic levels have been developed and/or validated CuZnNiCd pH range6 – – 8.2 Hardness range (mg/L) – 320*> 40 Ca range (mg/L) * Before Ni research with soft waters (lower hardness boundary was reduced to 6 mg CaCO 3 /L based on Ni-SOFT research (see further)
Cu toxicity to cladocerans in acidic waters Ni toxicity to cladocerans in soft waters Cd toxicity to Daphnia longispina in soft waters Three examples
Collected field waters and their inhabiting field cladocerans (water fleas) populations Toxicity test results in standard medium with these species were used to calibrate Cu-BLM Daphnia to sensitivity of field-species Predicted toxicity in natural waters with varying composition was compared with observed toxicity in natural waters Cu toxicity to field cladocerans
For normal sites (pH > 5.5): 27/28 LC50’s accurately predicted =further evidence in support of extrapolation Cu toxicity to field cladocera Bossuyt et al. (2004) Environ Sci Technol 38:
Cu toxicity to field cladocera For acidic sites (pH < 5.5): general overestimation of toxicity Further research required Bossuyt et al. (2004) Environ Sci Technol 38: For normal sites (pH > 5.5): 27/28 LC50’s accurately predicted =further evidence in support of extrapolation
Ni SOFT project Selected region for sampling: Both soft and hard close to each other Low anthropogenic input (N,P) Same climate Calcareous deposits for ‘hard’ region Collect cladocerans and algae from soft (H~6) and hard water (H~42) and test for chronic Ni toxicity in soft, moderately hard and hard water Deleebeeck et al. (2007) Aquat. Toxicol 84:
Ni SOFT project – hypotheses Cladocerans originating from soft water would be inherently more sensitive to Ni than those originating from hard water Cladocerans from soft water would be more protected against Ni toxicity by hardness than those from hard water
Ni SOFT project – design Chronic toxicity testing (reproduction, 10d to 21d) Species from soft water tested in Soft (S, hardness 6 mg/L) and Moderately Hard (MH, hardness 16 mg/L) water Species from hard water tested in Moderately Hard (MH) and Hard water (H, hardness 42 mg/L) Allows comparison of species sensitivity (comparison of EC50 in MH water) hardness effect (comparison of K CaBL and K MgBL estimated for soft and hard water species)
Chronic Ni toxicity to cladocerans Chronic EC50 (µg Ni/L) Speciesfromsoftmoderatehard Peracantha truncatasoft Ceriodaphnia quadrangulasoft Simocephalus serrulatussoft Ceriodaphnia quadrangulahard Ceriodaphnia pulchellahard Simocephalus vetulushard Daphnia longispinahard No significant difference in intrinsic sensitivity No significant difference in protective hardness effect Ni-BLM can be extrapolated down to hardness 6 mg/L
Cd SOFT project Hardness correction equation proposed in Cd RAR was only derived for hardness > 40 mg CaCO 3 /L Can equation be extrapolated to hardness as low as 5 mg CaCO 3 /L? Chronic toxicity testing (reproduction, 21d) with D. longispina In two Swedish soft waters with manipulated hardness
RAR hardness slope ( – dashed line) cannot be extrapolated to hardness < 40 mg CaCO3/L Hardness effect at hardness <50 is much lower (slope=0.1562=n.s.) Cd SOFT project
Conclusion extrapolation outside model boundaries Based on the given examples, any type of outcome may be expected from extrapolation outside model boundaries (accurate, overconservative, underconservative) Thus, extrapolation outside model boundaries will usually not be recommended without additional investigation for the specific local or regional abiotic conditions
IMPLEMENTATION Demonstration project in NL
NL issue Cu and Zn were considered nation-wide problematic substances Yellow, orange and red dots are sites where [Me] > EQS Baseline EQS not corrected for bioavailability (1.5 µg Cu/L, 9.4 µg Zn/L) Additional metal removal step from WWTP was planned nationally Large investments required while local water agencies wanted to invest mainly in ‘more important’ problems (eutrofication, habitat restoration) Thus: how large are true ecological risks if bioavailability is considered? Cu Zn
Monitoring campaign June 2006 – January 2007 Total and dissolved metal (Cu, Zn, Ni) TOC, DOC, pH, Ca, Mg, Na, K, Cl, SO4, alkalinity River basins# Sites# Samples / siteTotal # Samples Regge & Dinkel8648 Dommel HHSK6742 Velt &Vecht Hunze & Aa’s Vallei en Eem8648 Total
Chemistry summary (percentiles) 10%50%90% pH Hardness (mg CaCO 3 /L) Ca (mg/L) DOC (mg/L)
Metal measurements min10%50%90%max Dissolved Cu (µg/L) < 0,7124,28,3 Dissolved Ni (µg/L) < 123,89,833 Dissolved Zn (µg/L) < Despite careful discussions with and protocol transfer to people from local water agencies some difficulties noted Some laboratories acidified samples before filtration In many samples dissolved metal > total metal (varies among agencies) Cu (7-34%), Ni (2-50%), Zn (5-21%)
HC5 vs. baseline EQS min5%10%50%maxEQS* HC5-Cu (µg/L) HC5-Ni (µg/L) HC5-Zn (µg/L) HC5’s varied 10-fold (Cu, Ni) to 20-fold (Zn) HC5 is in most samples much higher than the baseline EQS * This is the baseline EQS, not corrected for bioavailability
HC5 vs. DOC Possibilities to develop simple equations (avoiding the use of complex BLM calculations + SSD fittings) EQS of metals without DOC measurement are worthless pH second most important
Compliance with HC5 vs. baseline EQS without bioav. correction with bioav. correction n>EQS%>EQSn>HC5%>HC5 Cu11658 %21.0 % Ni5126 %00.0 % Zn6332 %94.5 % Non-compliance with baseline EQS for 26-58% of samples Non-compliance with bioavailability-corrected HC5 for 0-4.5% of samples
Conclusion - implementation Analysis of dissolved metal concentrations is not as easy to implement as many people tend to believe A very different view about the “nation-wide metal problem” was obtained in NL when bioavailability is considered; the recommendation was to extend the analysis to all WFD monitoring stations of NL Bioavailability corrections might provide a more accurate picture of true ecological risks, thus avoiding “useless” investment of money that could be used for more important issues (e.g., eutrofication) Complex BLM + SSD calculation may be simplified without very much loss of accuracy…
General conclusions Validated bioavailability models are now available for Cu, Zn, Ni, (and Cd) Extrapolation of models across species: OK for Cu Available information for Ni under discussion at TCNES Some supportive information available for Zn, but more research recommended Extrapolation from lab to field: OK for Cu (mesocosms) Supportive information for Zn (UK project), mesocosm studies under investigation No data available for Ni Mixtures are a reality (mixture BLM seems possible – more research required) Extrapolation outside model boundaries: Variable outcomes; hence extrapolation not recommended without extra research Implementation in legal frameworks: Provides more accurate picture of true ecological risk – avoids wrong investments Training will be required (analytical issues, BLM+SSD calculations)